bfgminer/NEWS

Version 2.7.4 - August 23, 2012

- Perform select_pool even when not lagging to allow it to switch back if needed
to the primary.
- Simplify macros in output kernels avoiding apparent loops and local variables.
- Carry the needed bool over the work command queue.
- Move the decision to queue further work upstream before threads are spawned
based on fine grained per-pool stats and increment the queued count immediately.
- Track queued and staged per pool once again for future use.
- OpenCL 1.0 does not have native atomic_add and extremely slow support with
atom_add so detect opencl1.0 and use a non-atomic workaround.
- Pools: add RollTime info to API 'stats' and 'Stats' button in miner.php


Version 2.7.3 - August 22, 2012

- Minimise the number of getwork threads we generate.


Version 2.7.2 - August 22, 2012

- Pick worksize 256 with Cypress if none is specified.
- Give warning with sdk2.7 and phatk as well.
- Whitelist sdk2.7 for diablo kernel as well.
- Only keep the last 6 blocks in the uthash database to keep memory usage
constant. Storing more is unhelpful anyway.
- BFL Flash - always distribute source
- Increase kernel versions signifying changed APIs.
- BFL flash - include source in builds and more FPGA-README
- Check we haven't staged work while waiting for a curl entry before proceeding.
- Use atomic ops to never miss a nonce on opencl kernels, including nonce==0,
also allowing us to make the output buffer smaller.
- Remove compile errors/warnings and document compile/usage in FPGA-README
- bitforce-firmware-flash.c by Luke-jr
- Ignore the submit_fail flag when deciding whether to recruit more curls or not
since we have upper bounds on how many curls can be recruited, this test is
redundant and can lead to problems.
- API-README update cgminer version number
- API-README fix groups P: example mistake
- API-README add COIN and other edits
- gpu->hit should be reset on new work as well.
- Do not add time to dynamic opencl calculations over a getwork.
- miner.php allow 'coin' is custom pages


Version 2.7.1 - August 21, 2012

- Update windows build instructions courtesy of sharky.
- Increase max curls to number of mining threads + queue * 2, accounting for up
and downstream comms.
- Queue enough requests to get started.
- There is no point trying to clone_work in get_work() any more since we clone
on every get_work_thread where possible.
- There is no point subtracting 1 from maxq in get_work_thread.
- Only set lagging flag once there are no staged work items.
- select_pool does not switch back to the primary once lagging is disabled.
- miner.php allow page title to be defined in myminer.php
- Free work before retrying in get_work_thread.
- Increment total work counter under mutex lock.
- Increment the queued count after the curl is popped in case there's a delay
waiting on curls and we think we've queued work when in fact we're waiting
- API new command 'coin' with mining information
- Do the dynamic timing in opencl code over a single pass through scanhash to
make sure we're only getting opencl times contributing to the measured inte
- Increase curl reaping time to 5 minutes since comms between  curl requests can
be 2 mins apart with lots of rolltime.
- No need for extra variable in hash_push.
- Remove short options -r and -R to allow them to be reused and remove readme
entries for deprecated options.
- Avoid attempting to recursively lock the console mutex by disabling warnings
in gpu_fanpercent when fanspeed monitoring fails on windows. Debugged by l
- Deprecate the opt_fail_pause parameter, leaving a null placeholder for
existing configurations.
- Don't pause after failed getwork, set lagging flag and reassess.
- Add message to share if it's a resubmit.
- We should not be pausing in trying to resubmit shares.
- Get rid of the extending fail pause on failed connects since we discard work
after a period.
- get_work always returns true so turn it into a void function.
- get_work never returns false so get rid of fail pause loop.
- Get rid of pause and retry from get_upstream_work so we only do it from one
place.
- Deprecate the opt_retries feature as no one wants cgminer to automatically
abort. Leave a null placeholder for configurations that still have it.
- Reinstate fix ADL gpu-map not working when there are more ADL devices than
openCL patch by Nite69. Add virtual adl mapping for when none is specified o
- miner.php show summary Diff1 Shares total
- miner.php fix Work Utility totals
- miner.php format new Work Utility and Diff1 Shares
- API V1.17 show Work Utility and Diff1 Shares



Version 2.7.0 - August 18, 2012

- Introduce a new statistic, Work Utility, which is the number of difficulty 1
shares solved per minute. This is useful for measuring a relative rate of work
that is independent of reject rate and target difficulty.
- Implement a new pool strategy, BALANCE, which monitors work performed per pool
as a rolling average every 10 minutes to try and distribute work evenly over all
the pools. Do this by monitoring diff1 solutions to allow different difficulty
target pools to be treated equally, along with solo mining. Update the
documentation to describe this strategy and more accurately describe the
load-balance one.
- Getwork fail was not being detected. Remove a vast amount of unused variables
and functions used in the old queue request mechanism and redefine the getfail
testing.
- Don't try to start devices that don't support scrypt when scrypt mining.
- 0 is a valid return value for read so only break out if read returns -1.
- Consider us lagging only once our queue is almost full and no staged work.
- Simplify the enough work algorithm dramatically.
- Only queue from backup pools once we have nothing staged.
- Don't keep queueing work indefinitely if we're in opt failover mode.
- Make sure we don't opt out of queueing more work if all the queued work is
from one pool.
- Set lagging flag if we're on the last of our staged items.
- Reinstate clone on grabbing work.
- Grab clones from hashlist wherever possible first.
- Cull all the early queue requests since we request every time work is popped
now.
- Keep track of staged rollable work item counts to speed up clone_available.
- Make expiry on should_roll to 2/3 time instead of share duration since some
hardware will have very fast share times.
- Do the cheaper comparison first.
- Check that we'll get 1 shares' worth of work time by rolling before saying we
should roll the work.
- Simplify all those total_secs usages by initialising it to 1 second.
- Overlap queued decrementing with staged incrementing.
- Artificially set the pool lagging flag on pool switch in failover only mode as
well.
- Artificially set the pool lagging flag on work restart to avoid messages about
slow pools after every longpoll.
- Factor in opt_queue value into enough work queued or staged.
- Roll work whenever we can on getwork.
- Queue requests for getwork regardless and test whether we should send for a
getwork from the getwork thread itself.
- Get rid of age_work().
- 0 is a valid return value for read so only break out if read returns -1.
- Offset libusb reads/writes by length written as well in ztex.
- Cope with timeouts and partial reads in ztex code.
- fpga serial I/O extra debug (disabled by default)


Version 2.6.5 - August 15, 2012

- Don't try to get bitforce temperature if we're polling for a result to
minimise the chance of interleaved responses.
- Set memory clock based on memdiff if present from with engine changes,
allowing it to parallel manual changes from the menu as well.
- Increase the timeout on bitforce as per Paul Sheppard's suggestion to account
for throttling + work time + excess.
- Fix ADL gpu-map not working when there are more ADL devices than openCL.
Initial patch supplied by Nite69. Modified to suit.
- Windows' timer resolution is limited to 15ms accuracy. This was breaking
dynamic intensity since it tries to measure below this. Since we are repeatedly
sampling similar timeframes, we can average the gpu_us result over 5 different
values to get very fine precision.
- Fix harmless unused warnings in scrypt.h.
- api.c typo
- API allow display/change failover-only setting
- Check we are not lagging as well as there is enough work in getwork.
- Minimise locking and unlocking when getting counts by reusing shared mutex
lock functions.
- Avoid getting more work if by the time the getwork thread is spawned we find
ourselves with enough work.
- The bitforce buffer is cleared and hw error count incremented on return from a
failed send_work already so no need to do it within the send_work function.
- miner.php allow a custom page section to select all fields with '*' - e.g. to
create a STATS section on a custom page
- Escape " and \ when writing json config file
- miner.php optional single rig totals (on by default)


Version 2.6.4 - August 7, 2012

- Convert the serial autodetect functions to use int instead of char to
enumerate devices.
- Make the serial open timeout for BFL generically 1 second on windows.
- Deuglify windows autodetect code for BFL.
- There is no point zeroing temperature in BFL if we fail to get a response, and
we should register it as a HW error, suggesting throttling.
- Update SCRYPT README with information about HW errors.
- Use the scrypt CPU code to confirm results from OCL code, and mark failures as
HW errors, making it easier to tune scrypt parameters.
- We may as well leave one curl still available per pool instead of reaping the
last one.
- Need to recheck the pool->curls count on regaining the pool lock after the
pthread conditional wait returns.
- Display reaped debug message outside mutex lock to avoid recursive locking.
- Add specific information when ADL detects error -10 saying the device is not
enabled.
- api.c update API start message and include port number
- miner.php ignore arg when readonly
- miner.php allow pool inputs: delete, addpool, poolpriority


Version 2.6.3 - August 5, 2012

- Count likely throttling episodes on bitforce devices as hardware errors.
- Style cleanups.
- Use FTD2XX.DLL on Windows to autodetect BitFORCE SHA256 devices.
- Make pool_disabled the first in the enums == 0, fixing the pool enabled count
which compares if value is not enabled before enabling it.
- Correct writing of scrypt parameters to config file based on command line
parameters only.
- Use different variables for command line specified lookup gap and thread
concurrency to differentiate user defined versus auto chosen values.
- Queue a request on pool switch in case we have no work from the new pool yet.
- Display failover only mode in pool menu and allow it to be toggled live.
- Reinstate check for system queueing lag when the current pool's queue is maxed
out, there is no staged work, and the work is needed now.
- There is no need for pool active testing to be mandatory any more with queue
request changes.
- Fix harmless warnings.
- Check the current staged and global queued as well before queueing requests.
Discard stales before ageing work in the watchdog thread. Queue requests after
discarding and ageing work in watchdog thread. Display accurate global queued in
curses output. Reuse variable in age_work().
- The queueing mechanism has become a complex state machine that is no longer
predictable. Rewrite it from scratch watching only current queues in flight and
staged work available on a pool by pool basis.
- API remove unused warning in non-GPU compile
- api.c in linux allow to open a closed socket in TIME_WAIT
- Queue an extra request whenever staged work drops below mining thread count in
hash_pop.
- Update debian package configs to v2.6.2


Version 2.6.2 - August 3, 2012

- Scrypt mining does not support block testing yet so don't try to print it.
- Clear the bitforce buffer whenever we get an unexpected result as it has
likely throttled and we are getting cached responses out of order, and use the
temperature monitoring as a kind of watchdog to flush unexpected results.
- It is not critical getting the temperature response in bitforce so don't
mandatorily wait on the mutex lock.
- Check there is a cutoff temp actually set in bitforce before using it as a cut
off value otherwise it may think it's set to zero degrees.
- We dropped the temporary stopping of curl recruiting on submit_fail by
mistake, reinstate it.
- Make threads report in either side of the scanhash function in case we miss
reporting in when restarting work.
- Don't make mandatory work and its clones last forever.
- Make test work for pool_active mandatory work items to smooth out staged work
counts when in failover-only mode.
- Add debugging output when work is found stale as to why.
- Print the 3 parameters that are passed to applog for a debug line in
bitforce.c
- Clear bitforce buffer on init as previously.
- Add some headroom to the number of curls available per pool to allow for
longpoll and sendwork curls.
- Revert "Revert "Change BFL driver thread initialising to a constant 100ms
delay between devices instead of a random arrangement.""
- Revert "Remove bitforce_thread_init"
- Show the correct base units on GPU summary.
- Differentiate between the send return value being a bool and the get return
value when managing them in bitforce scanhash.
- 23a8c60 Revert "bitforce: Skip out of sending work if work restart requested"


Version 2.6.1 - July 30, 2012

- Display scrypt as being built in as well.
- Fix build warning about KL_SCRYPT when built without scrypt support.
- Remove the low hash count determinant of hardware being sick. A low hash rate
can be for poor network connectivity or scrypt mining, neither of which are due
to a sick device.
- api.c poolpriority changes


Version 2.6.0 - July 29, 2012

- Display kilohash when suitable, but store the global mhash value still truly
in megahashes to not break the API output.
- Don't try and print curses output for devices that won't fit on the screen.
- Add scrypt documentation in the form of a separate readme.
- Fix build error without scrypt enabled.
- Limit total number of curls recruited per pool to the number of mining threads
to prevent blasting the network when we only have one pool to talk to.
- bitforce: Skip out of sending work if work restart requested
- Keep a counter of enabled pools and use that instead of iterating over the
pool list. Use that value to ensure we don't set the last remaining active pool
to the rejecting state.
- fpgautils: add support for 57.6 kBd serial
- miner.php add a socket RCV timeout for if cgminer is hung and the API thread
is still running
- Limit thread concurrency for scrypt to 5xshaders if shaders is specified.
- Simplify repeated use of gpus[gpu]. in ocl.c
- Find the nearest power of 2 maximum alloc size for the scrypt buffer that can
successfully be allocated and is large enough to accomodate the thread
concurrency chosen, thus mapping it to an intensity.
- Don't make opt_scrypt mandatory blocking with opencl code.
- Update kernel versions reflecting changes in the API.
- Make the thread concurrency and lookup gap options hidden on the command line
and autotune parameters with a newly parsed --shaders option.
- Fix target testing with scrypt kernel as it would have been missing shares
below target.
- Bugfix: Use a mutex to control non-curses output
- Simplify code to a single vprintf path for curses-less printing
- Move opt_quiet check to my_log_curses, so it works for curses-less builds
- Use log_generic for vapplog to cut down on code duplication
- Add space to log output now that there is more screen real estate available.
- BFL force all code to timeout to avoid hanging
- Bugfix: Copy argv[0] given to dirname()
- Always create the largest possible padbuffer for scrypt kernels even if not
needed for thread_concurrency, giving us some headroom for intensity levels.
- Use the detected maximum allocable memory on a GPU to determine the optimal
scrypt settings when lookup_gap and thread_concurrency parameters are not given.
- Check the maximum allocable memory size per opencl device.
- Add debugging output if buffer allocation fails for scrypt and round up
bufsize to a multiple of 256.
- Nonce testing for btc got screwed up, leading to no accepted shares. Fix it.
- Display size of scrypt buffer used in debug.
- Allow intensities up to 20 if scrypt is compiled in.
- Add name to scrypt kernel copyright.
- Allow lookup gap and thread concurrency to be passed per device and store
details in kernel binary filename.
- Ignore negative intensities for scrypt.
- Change the scale of intensity for scrypt kernel and fix a build warning.
- Correct target value passed to scrypt kernel.
- Use 256 output slots for kernels to allow 1 for each worksize.
- Test the target in the actual scrypt kernel itself saving further
calculations.
- Reinstate GPU only opencl device detection.
- Decrease lookup gap to 1. Does not seem to help in any way being 2.
- Fix build.
- Make pad0 and pad1 local variable in scrypt kernel.
- Constify input variable in scrypt kernel.
- Send correct values to scrypt kernel to get it finally working.
- Create command queue before compiling program in opencl.
- Detach pthread from within the api thread in case it is terminated due to not
being instantiated before pthread_cancel is called from main, leading to a
segfault.
- Debug output per thread hashrate is out by a factor of 1000.
- Initialise mdplatform.
- Find the gpu platform with the most devices and use that if no platform option
is passed.
- Allow more platforms to be probed if first does not return GPUs.
- Fix external scrypt algo missing.
- Limit scrypt to 1 vector.
- Handle KL_SCRYPT in config write.
- Get rid of stuff.
- Don't enqueuewrite buffer at all for pad8 and pass work details around for
scrypt in dev_blk.
- Set the correct data for cldata and prepare for pad8 fixes.
- Bugfix: Fix build without curses but with OpenCL
- Find the gpu platform with the most devices and use that if no platform option
is passed.
- Allow more platforms to be probed if first does not return GPUs.
- Get rid of spaces in arrays in scrypt kernel.
- Start with smaller amount of hashes in cpu mining to enable scrypt to return
today sometime.
- Show Khash hashrates when scrypt is in use.
- Free the scratchbuf memory allocated in scrypt and don't check if CPUs are
sick since they can't be. Prepare for khash hash rates in display.
- Add cpumining capability for scrypt.
- Set scrypt settings and buffer size in ocl.c code to be future modifiable.
- Cope with when we cannot set intensity low enough to meet dynamic interval by
inducing a forced sleep.
- Make dynamic and scrypt opencl calls blocking.
- Calculate midstate in separate function and remove likely/unlikely macros
since they're dependent on pools, not code design.
- bitforce: Use "full work" vs "nonce range" for kernel name
- Display in debug mode when we're making the midstate locally.
- Fix nonce submission code for scrypt.
- Make sure goffset is set for scrypt and drop padbuffer8 to something
manageable for now.
- Set up buffer8 for scrypt.
- Build fix for opt scrypt.
- Don't check postcalc nonce with sha256 in scrypt.
- Don't test nonce with sha and various fixes for scrypt.
- Make scrypt buffers and midstate compatible with cgminer.
- Use cgminer specific output array entries in scrypt kernel.
- Provide initial support for the scrypt kernel to compile with and mine scrypt
with the --scrypt option.
- Enable completely compiling scrypt out.
- Begin import of scrypt opencl kernel from reaper.
- bitforce_get_result returns -1 on error now.
- Check return value of read in BFgets
- Bugfix: Make our Windows nanosleep/sleep replacements standards-compliant
(which fixes nmsleep) and include compat.h for bitforce (for sleep)
- rpc: Use a single switch statement for both stringifications of cgpu->status
- Fix whitespace mangling.
- miner.php fix rig # when miners fail
- Only try to shut down work cleanly if we've successfully connected and started
mining.
- Use switch statement for cgpu->status and fix spelling.
- Abbrv. correction
- Bugfix: Don't declare devices SICK if they're just busy initialising
- Bugfix: Calculate nsec in nmsleep correctly
- Bugfix: Adapt OpenCL scanhash errors to driver API change (errors are now -1,
not 0)
- Remove superfluous ave_wait
- Put kname change for broken nonce-range back in
- Add average wait time to api stats
- Change BFL driver thread initialising to a constant 100ms delay between
devices instead of a random arrangement.
- Spelling typo.
- Time opencl work from start of queueing a kernel till it's flushed when
calculating dynamic intensity.
- Modify te scanhash API to use an int64_t and return -1 on error, allowing zero
to be a valid return value.
- Check for work restart after the hashmeter is invoked for we lose the hashes
otherwise contributed in the count.
- Remove disabled: label from mining thread function, using a separate
mt_disable function.
- Style changes.
- Missed one nonce-range disabling.
- Add average return time to api stats
- miner.php allow rig names in number buttons
- Remove bitforce_thread_init The delay thing does nothing useful... when long
poll comes around, all threads restart at the same time anyway.
- Change timeouts to time-vals for accuracy.
- fix API support for big endian machines
- Cope with signals interrupting the nanosleep of nmsleep.
- Use standard cfsetispeed/cfsetospeed to set baud rate on *nix
- miner.php split() flagged deprecated in PHP 5.3.0
- More BFL tweaks. Add delay between closing and reopening port. Remove buffer
clear in re-init Add kernel type (mini-rig or single)
- Make long timeout 10seconds on bitforce for when usleep or nanosleep just
can't be accurate...


Version 2.5.0 - July 6, 2012

- Fix --benchmark not working since the dynamic addition of pools and pool
stats.
- Make disabling BFL nonce range support a warning since it has to be explicitly
enabled on the command line now.
- miner.php allow renaming table headers
- Make bitforce nonce range support a command line option --bfl-range since
enabling it decrease hashrate by 1%.
- Add sanity checking to make sure we don't make sleep_ms less than 0 in
bitforce.
- The fastest minirig devices need a significantly smaller starting sleep time.
- Use a much shorter initial sleep time to account for faster devices and nonce
range working, and increase it if nonce range fails to work.
- Use nmsleep instead of usleep in bitforce.
- Provide a ms based sleep function that uses nanosleep to avoid the inaccuracy
of usleep on SMP systems.
- delay_time_ms is always set so need not be initialised in bitforce.
- Increase bitforce timeout to 10 seconds.
- Add more hysteresis and poll ~5 times to allow for timer delays in bitforce
devices.
- miner.php allow alternating line colours (off by default)
- Display the actual duration of wait when it is greater than the cutoff.
- Set nonce to maximum once we determine nonce range support is broken.
- Initial wait time is always known so no need to zero it beforehand in
bitforce.
- No point counting wait time until the work is actually sent to bitforce
devices.
- Use string comparison functions instead of explicit comparisons.
- Account for wait_ms time when nonce_range is in use on BFL.
- Split nonces up into 1/5 chunks when nonce range is supported.
- limit clear buffer iterations.
- Ad fd check to clear buffer.
- miner.php remove incorrect 'DATE' error message
- miner.php allow summary header in custom pages
- Disable nonce range support in BFL when broken support is detected.
- Restart_wait is only called with a ms value so incorporate that into the
function.
- Only try to adjust dev width when curses is built in.
- miner.php define custom sum fields as a simple array
- Fix off-by-one error in nonce increment in bfl.
- Use BE when setting nonce in bitforce nonce range work.
- Enable nonce range in the normal init sequence for bfl.
- Queue extra work at 2/3 differently depending on whether we're using nonce
range or not.
- Initially enable support for nonce range support on bfl, splitting nonces up
into 3/4 size and only disable it if it fails on work submit.
- Attempt to detect nonce range support in BFL by sending work requring its
support.
- Limit retrying on busy for up to BITFORCE_TIMEOUT_MS
- Attempt to initialise while bitforce device returns BUSY.
- Extend length of string that can be passed to BFL devices.
- Fix signedness warning.
- Adjust device width column to be consistent.
- Use cgpu-> not gpus[] in watchdog thread.
- Add api stats (sleep time)
- Timing tweaks Added long and short timeouts, short for detecting throttling,
long to give up totally. Reset sleep time when device re-initialised Still check
results after timeout Back up a larger time if result on first poll.
- Add API Notify counter 'Comms Error'
- Style police on api.c
- Do all logging outside of the bitforce mutex locking to avoid deadlocks.
- Remove applog call from bfwrite to prevent grabbing nested mutexes.
- Bitforce style changes.
- Minor style changes.
- Remove needless roundl define.
- Made JSON error message verbose.
- Fine-tune timing adjustment. Also remove old work_restart timing.
- Check for gpu return times of >= 0, not just 0, to fix intensity dropping to
-10.
- Restart is zeroed in the mining thread so no need to do it inside the bitforce
code.
- More improvements to comms. BFL return nothing when throttling, so should not
be considered an error. Instead repeat with a longer delay.
- Polling every 10ms there's not much point checking the pthread_cond_timedwait
as it just adds overhead. Simply check the value of work_restart in the bfl main
polling loop.
- Use a pthread conditional that is broadcast whenever work restarts are
required. Create a generic wait function waiting a specified time on that
conditional that returns if the condition is met or a specified time passed to
it has elapsed. Use this to do smarter polling in bitforce to abort work, queue
more work, and check for results to minimise time spent working needlessly.
- Add busy time to wait time.
- api.c put version up to 1.14
- Add tiny delay after writing to BFL Change BFL errors to something more human
readable Send work busy re-tries after 10ms delay


Version 2.4.4 - July 1, 2012

- Fix builds on non gnu platforms.
- api.c ensure old mode is always available when not using --api-groups + quit()
on param errors
- Implement rudimentary X-Mining-Hashrate support.
- Detect large swings in temperature when below the target temperature range and
change fan by amounts dependant on the value of tdiff.
- Adjust the fanspeed by the magnitude of the temperature difference when in the
optimal range.
- Revert "Restarting cgminer from within after ADL has been corrupted only leads
to a crash. Display a warning only and disable fanspeed monitoring."
- api.c fix json already closed
- implement and document API option --api-groups
- Put upper bounds to under 2 hours that work can be rolled into the future for
bitcoind will deem it invalid beyond that.
- define API option --api-groups
- api.c allow unwell devices to be enabled so they can be cured
- miner.php - fix/enable autorefresh for custom pages
- miner.php allow custom summary pages - new 'Mobile' summary
- Work around pools that advertise very low expire= time inappropriately as this
leads to many false positives for stale shares detected.
- Only show ztex board count if any exist.
- There is no need for work to be a union in struct workio_cmd
- fpgautils.c include a debug message for all unknown open errors
- Don't keep rolling work right up to the expire= cut off. Use 2/3 of the time
between the scantime and the expiry as cutoff for reusing work.
- Log a specific error when serial opens fail due to lack of user permissions
- Increase GPU timing resolution to microsecond and add sanity check to ensure
times are positive.
- Opencl code may start executing before the clfinish order is given to it so
get the start timing used for dynamic intensity from before the kernel is
queued.
- fpgautils.c - set BAUD rate according to termio spec
- fpgautils.c - linux ordering back to the correct way
- miner.php remove unneeded '.'s
- miner.php add auto refresh options
- miner.php add 'restart' next to 'quit'
- miner.php make fontname/size configurable with myminer.php
- Make the pools array a dynamically allocated array to allow unlimited pools to
be added.
- Make the devices array a dynamically allocated array of pointers to allow
unlimited devices.
- Dynamic intensity for GPUs should be calculated on a per device basis. Clean
up the code to only calculate it if required as well.
- Use a queueing bool set under control_lock to prevent multiple calls to
queue_request racing.
- Use the work clone flag to determine if we should subtract it from the total
queued variable and provide a subtract queued function to prevent looping over
locked code.
- Don't decrement staged extras count from longpoll work.
- Count longpoll's contribution to the queue.
- Increase queued count before pushing message.
- Test we have enough work queued for pools with and without rolltime
capability.
- As work is sorted by age, we can discard the oldest work at regular intervals
to keep only 1 of the newest work items per mining thread.
- Roll work again after duplicating it to prevent duplicates on return to the
clone function.
- Abstract out work cloning and clone $mining_threads copies whenever a rollable
work item is found and return a clone instead.
- api.c display Pool Av in json
- Take into account average getwork delay as a marker of pool communications
when considering work stale.
- Work out a rolling average getwork delay stored in pool_stats.
- Getwork delay in stats should include retries for each getwork call.
- Walk through the thread list instead of searching for them when disabling
threads for dynamic mode.
- Extend nrolltime to support the expiry= parameter. Do this by turning the
rolltime bool into an integer set to the expiry time. If the pool supports
rolltime but not expiry= then set the expiry time to the standard scantime.
- When disabling fanspeed monitoring on adl failure, remove any twin GPU
association. This could have been leading to hangs on machines with dual GPU
cards when ADL failed.
- modminer: Don't delay 2nd+ FPGAs during work restart
- Disable OpenCL code when not available.
- Fix openwrt crashing on regeneratehash() by making check_solve a noop.
- FPGA - allow device detect override without an open failure
- Fix sign warning.


Version 2.4.3 - June 14, 2012

- can_roll and should_roll should have no bearing on the cycle period within the
miner_thread so remove it.
- Check for strategy being changed to load balance when enabling LPs.
- Check that all threads on the device that called get_work are waiting on
getwork before considering the pool lagging.
- Iterate over each thread belonging to each device in the hashmeter instead of
searching for them now that they're a list.
- When using rotate pool strategy, ensure we only select from alive enabled
pools.
- Start longpoll from every pool when load balance strategy is in use.
- Add mandatory and block fields to the work struct. Flag any shares that are
detected as blocks as mandatory to submit, along with longpoll work from a
previously rejecting pool.
- Consider the fan optimal if fanspeed is dropping but within the optimal speed
window.
- Fix typo in some API messages (succeess/success)
- api.c MMQ stat bugs
- Bugfix: Fix warnings when built without libudev support
- Bugfix: slay a variety of warnings
- Bugfix: modminer: Fix unsigned/signed comparison and similar warnings
- API add ModMinerQuad support
- Bugfix: Honour forceauto parameter in serial_detect functions
- modminer: Temperature sensor improvements
- modminer: Make log messages more consistent in format
- Only adjust GPU speed up if the fanspeed is within the normal fanrange and
hasn't been turned to maximum speed under overheat conditions.
- ModMiner use valid .name
- New driver: BTCFPGA ModMiner
- Abstract generally useful FPGA code into fpgautils.c
- API add stats for pool getworks
- miner.php option to hide specific fields from the display
- miner.php add version numbers to the summary page
- Update debian configs to v2.4.2
- Add API and FPGA READMEs into Makefile to be included in source distribution.
- Icarus - fix unit64_t printf warnings


Version 2.4.2 - June 2, 2012

- API.class compiled with Java SE 6.0_03 - works with Win7x64
- miner.php highlight devs too slow finding shares (possibly failing)
- API update version to V1.11 and document changes
- API save default config file if none specified
- api.c save success incorrectly returns error
- api.c replace BUFSIZ (linux/windows have different values)
- Move RPC API content out of README to API-README
- Open a longpoll connection if a pool is in the REJECTING state as it's the
only way to re-enable it automatically.
- Use only one longpoll as much as possible by using a pthread conditional
broadcast that each longpoll thread waits on and checks if it's the current pool
before
- If shares are known stale, don't use them to decide to disable a pool for
sequential rejects.
- Restarting cgminer from within after ADL has been corrupted only leads to a
crash. Display a warning only and disable fanspeed monitoring.
- Icarus: fix abort calculation/allow user specified abort
- Icarus: make --icarus-timing hidden and document it in FPGA-README
- Icarus: high accuracy timing and other bitstream speed support
- add-MIPSEB-to-icarus-for-BIG_ENDIAN
- work_decode only needs swab32 on midstate under BIG ENDIAN
- add compile command to api-example.c
- save config bugfix: writing an extra ',' when no gpus
- Add dpkg-source commits


Version 2.4.1 - May 6, 2012

- In the unlikely event of finding a block, display the block solved count with
the pool it came from for auditing.
- Display the device summary on exit even if a device has been disabled.
- Use correct pool enabled enums in api.c.
- Import Debian packaging configs
- Ensure we test for a pool recovering from idle so long as it's not set to
disabled.
- Fix pool number display.
- Give cgminer -T message only if curses is in use.
- Reinit_adl is no longer used.
- API 'stats' allow devices to add their own stats also for testing/debug
- API add getwork stats to cgminer - accesable from API 'stats'
- Don't initialise variables to zero when in global scope since they're already
initialised.
- Get rid of unitialised variable warning when it's false.
- Move a pool to POOL_REJECTING to be disabled only after 3 minutes of
continuous rejected shares.
- Some tweaks to reporting and logging.
- Change FPGA detection order since BFL hangs on an ICA
- API support new pool status
- Add a temporarily disabled state for enabled pools called POOL_REJECTING and
use the work from each longpoll to help determine when a rejecting pool has
started working again. Switch pools based on the multipool strategy once a pool
is re-enabled.
- Removing extra debug
- Fix the benchmark feature by bypassing the new networking code.
- Reset sequential reject counter after a pool is disabled for when it is
re-enabled.
- Icarus - correct MH/s and U: with work restart set at 8 seconds
- ztex updateFreq was always reporting on fpga 0
- Trying harder to get 1.15y working
- Specifying threads on multi fpga boards extra cgpu
- Missing the add cgpu per extra fpga on 1.15y boards
- API add last share time to each pool
- Don't try to reap curls if benchmarking is enabled.


Version 2.4.0 - May 3, 2012

- Only show longpoll warning once when it has failed.
- Convert hashes to an unsigned long long as well.
- Detect pools that have issues represented by endless rejected shares and
disable them, with a parameter to optionally disable this feature.
- Bugfix: Use a 64-bit type for hashes_done (miner_thread) since it can overflow
32-bit on some FPGAs
- Implement an older header fix for a label existing before the pthread_cleanup
macro.
- Limit the number of curls we recruit on communication failures and with
delaynet enabled to 5 by maintaining a per-pool curl count, and using a pthread
conditional that wakes up when one is returned to the ring buffer.
- Generalise add_pool() functions since they're repeated in add_pool_details.
- Bugfix: Return failure, rather than quit, if BFwrite fails
- Disable failing devices such that the user can attempt to re-enable them
- Bugfix: thread_shutdown shouldn't try to free the device, since it's needed
afterward
- API bool's and 1TBS fixes
- Icarus - minimise code delays and name timer variables
- api.c V1.9 add 'restart' + redesign 'quit' so thread exits cleanly
- api.c bug - remove extra ']'s in notify command
- Increase pool watch interval to 30 seconds.
- Reap curls that are unused for over a minute. This allows connections to be
closed, thereby allowing the number of curl handles to always be the minimum
necessary to not delay networking.
- Use the ringbuffer of curls from the same pool for submit as well as getwork
threads. Since the curl handles were already connected to the same pool and are
immediately available, share submission will not be delayed by getworks.
- Implement a scaleable networking framework designed to cope with any sized
network requirements, yet minimise the number of connections being reopened. Do
this by create a ring buffer linked list of curl handles to be used by getwork,
recruiting extra handles when none is immediately available.
- There is no need for the submit and getwork curls to be tied to the pool
struct.
- Do not recruit extra connection threads if there have been connection errors
to the pool in question.
- We should not retry submitting shares indefinitely or we may end up with a
huge backlog during network outages, so discard stale shares if we failed to
submit them and they've become stale in the interim.


Version 2.3.6 - April 29, 2012

- Shorten stale share messages slightly.
- Protect the freeing of current_hash under mutex_lock to prevent racing on it
when set_curblock is hit concurrently.
- Change default behaviour to submitting stale, removing the --submit-stale
option and adding a --no-submit-stale option.
- Make sure to start the getwork and submit threads when a pool is added on the
fly. This fixes a crash when a pool is added to running cgminer and then
switched to.
- Faster hardware can easily outstrip the speed we can get work and submit
shares when using only one connection per pool.
- Test the queued list to see if any get/submits are already queued and if they
are, start recruiting extra connections by generating new threads.
- This allows us to reuse network connections at low loads but recuit new open
connections as they're needed, so that cgminer can scale to hardware of any
size.


Version 2.3.5 - April 28, 2012

- Restarting cgminer leads to a socket that can't be bound for 60 seconds, so
increase the interval that API binding waits to 30 seconds to minimise the
number of times it will retry, spamming the logs.
- Give a longpoll message for any longpoll that detects a block change, primary
or backup, and also display which pool it was.
- Decrease utility display to one decimal place.
- Small cosmetic output alignment.
- Add pool number to stale share message.
- Add space to log output now that there is more screen real estate available.
- Indentation clean up.
- Merge branch 'master' of github.com:ckolivas/cgminer
- Remove thread id display from rejected shares as well.
- Merge pull request #185 from Diapolo/diakgcn
- add goffset support for diakgcn with -v 1 and update kernel version
- Set have_longpoll to true when there is at least one pool with longpoll.
- Don't display the thread ID since it adds no useful information over the
device number.
- Don't display the first 8 bytes of a share since they will always be zero at
>= 1 difficulty.
- work->longpoll is reset across test_work_current so we need to recheck what
pool it belongs to.
- Use longpolls from backup pools with failover-only enabled just to check for
block changes, but don't use them as work.
- Start longpoll only after we have tried to extract the longpoll URL.
- Check for submitold flag on resubmit of shares, and give different message for
stale shares on retry.
- Check for submitold before submitstale.
- Don't force fresh curl connections on anything but longpoll threads.
- Create one longpoll thread per pool, using backup pools for those pools that
don't have longpoll.
- Use the work created from the longpoll return only if we don't have
failover-enabled, and only flag the work as a longpoll if it is the current
pool.
- This will work around the problem of trying to restart the single longpoll
thread on pool changes that was leading to race conditions.
- It will also have less work restarts from the multiple longpolls received from
different pools.
- Remove the ability to disable longpoll. It is not a useful feature and will
conflict with planned changes to longpoll code.
- Remove the invalid entries from the example configuration file.
- Add support for latest ATI SDK on windows.
- Export missing function from libztex.
- miner.php change socktimeoutsec = 10 (it only waits once)
- Bugfix: Make initial_args a const char** to satisfy exec argument type warning
(on Windows only)
- miner.php add a timeout so you don't sit and wait ... forever
- Create discrete persistent submit and get work threads per pool, thus allowing
all submitworks belonging to the same pool to reuse the same curl handle, and
all getworks to reuse their own handle.
- Use separate handles for submission to not make getwork potentially delay
share submission which is time critical.
- This will allow much more reusing of persistent connections instead of opening
new ones which can flood routers.
- This mandated a rework of the extra longpoll support (for when pools are
switched) and this is managed by restarting longpoll cleanly and waiting for a
thread join.
- miner.php only show the current date header once
- miner.php also add current time like single rig page
- miner.php display rig 'when' table at top of the multi-rig summary page
- README - add some Ztex details
- api.c include zTex in the FPGA support list
- api.c ensure 'devs' shows PGA's when only PGA code is compiled
- cgminer.c sharelog code consistency and compile warning fix
- README correct API version number
- README spelling error
- api.c combine all pairs of sprintfs()
- api.c uncomment and use BLANK (and COMMA)
- Code style cleanup
- Annotating frequency changes with the changed from value
- README clarification of 'notify' command
- README update for API RPC 'devdetails'
- api.c 'devdetails' list static details of devices
- Using less heap space as my TP-Link seems to not handle this much


Version 2.3.4 - April 25, 2012

- Extensively document the cause of GPU device issues and the use of --gpu-map.
- Support for share logging
- Detect poorly performing combination of SDK and phatk kernel and add verbose
warning at startup.
- Icarus update to new add_cgpu()
- Icarus driver working with Linux and Windows
- api.c fix unused variable compile warning
- Display all OpenCL devices when -n is called as well to allow debugging of
differential mapping of OpenCL to ADL.
- Add a --gpu-map option which will allow arbitrarily mapping ADL devices to
OpenCL devices for instances where association by enumeration alone fails.
- Increase upper limit on number of extra items to queue as some FPGA code can't
yet reliably keep many devices busy.
- Display configuration file information when -c option is passed and only when
file exists on loading default config file.
- Display configuration file loaded, if any, and debug output if configuration
file parsing failed.
- Add missing ztex header to Makefile for distribution.
- Document long-form COM port device names on Windows, required to specify
serial ports above 9
- Include ztex bitstreams firmware in distribution and install if configured in.
- Style police on driver-ztex.c
- work_restart should only be changed by cgminer.c now
- Shut down the api cleanly when the api thread is cancelled. This should allow
the api socket to be closed successfully to next be reopened with app_restart.
- Make a union for cgpu device handles, and rename "device" to "device_ztex"
since it's Ztex-specific
- Initialise name variable.
- Remove unnecessary check for variable that always has memory allocated.
- Bugfix: Missing "break" no-op in default case
- Make the status window and log window as large as can fit on startup,
rechecking to see if it can be enlarged after the fact. This allows any number
of devices to be displayed provided the window is made long enough without
corrupting the output.
- Style police on libztex.c.
- API add removepool like the screen interface
- api.c escape required characters in return strings + pools returns the
username
- Set lp_path to NULL after free for consistency.
- Removing dmalloc import left behind by mistake
- Fixing leak in resp_hdr_cb
- miner.php warning highlight GPU stats if they are zero (e.g. ADL not enabled)
- miner.php highlight any device that isn't 'Enabled'
- miner.php highlight any Status that isn't 'Alive'
- miner.php optionally support multiple rigs
- Initial Ztex support 1.15x board.


Version 2.3.3 - April 15, 2012

- Don't even display that cpumining is disabled on ./configure to discourage
people from enabling it.
- Do a complete cgminer restart if the ATI Display Library fails, as it does on
windows after running for some time, when fanspeed reporting fails.
- Cache the initial arguments passed to cgminer and implement an attempted
restart option from the settings menu.
- Disable per-device status lines when there are more than 8 devices since
screen output will be corrupted, enumerating them to the log output instead at
startup.
- Reuse Vals[] array more than W[] till they're re-initialised on the second
sha256 cycle in poclbm kernel.
- Minor variable alignment in poclbm kernel.
- Make sure to disable devices with any status not being DEV_ENABLED to ensure
that thermal cutoff code works as it was setting the status to DEV_RECOVER.
- Re-initialising ADL simply made the driver fail since it is corruption over
time within the windows driver that's responsible. Revert "Attempt to
re-initialise ADL should a device that previously reported fanspeed stops
reporting it."
- Microoptimise poclbm kernel by ordering Val variables according to usage
frequency.


Version 2.3.2 - March 31, 2012

- Damping small changes in hashrate so dramatically has the tendency to always
make the hashrate underread so go back to gentle damping instead.
- Revert the crossover of variables from Vals to W in poclbm kernel now that
Vals are the first declared variables so they're used more frequently.
- Vals variables appearing first in the array in poclbm is faster.
- Change the preferred vector width to 1 for Tahiti only, not all poclbm
kernels.
- Use a time constant 0.63 for when large changes in hashrate are detected to
damp change in case the large change is an aliasing artefact instead of a real
chang
- Only increment stale counter if the detected stales are discarded.
- Attempt to re-initialise ADL should a device that previously reported fanspeed
stops reporting it.
- Move the ADL setup and clearing to separate functions and provide a reinit_adl
function to be used when adl fails while running.
- Use slightly more damping on the decay time function in the never-ending quest
to smooth off the hashmeter.
- Set the starting fanspeed to a safe and fairly neutral 50% when autofan is
enabled.
- Provide locking around updates of cgpu hashrates as well to prevent multiple
threads accessing data fields on the same device.
- Display the beginning of the new block in verbose mode in the logs.
- Reinstate old diablo kernel variable ordering from 120222, adding only goffset
and vector size hint. The massive variable ordering change only helped one SDK
on
- Change the version number on the correct kernels.
- api.c devicecode/osinfo incorrectly swapped for json
- Add extensive instructions on how to make a native windows build.
- Update version numbers of poclbm and diablo kernels as their APIs have also
changed.
- Use global offset parameter to diablo and poclbm kernel ONLY for 1 vector
kernels.
- Use poclbm preferentially on Tahiti now regardless of SDK.
- Remove unused constant passed to poclbm.
- Clean up use of macros in poclbm and use bitselect everywhere possible.
- Add vector type hint to diablo kernel.
- Add worksize and vector attribute hints to the poclbm kernel.
- Spaces for non-aligned variables in poclbm.
- More tidying of poclbm.
- Swap Vals and W variables where they can overlap in poclbm.
- More tidying of poclbm.
- Tidy up first half of poclbm.
- Clean up use of any() by diablo and poclbm kernels.
- Minor variable symmetry changes in poclbm.
- Put additions on separate lines for consistency in poclbm.
- Consolidate last use of W11 into Vals4 in poclbm.
- Change email due to SPAM
- api.c miner.php add a '*' to the front of all notify counters - simplifies
future support of new counters
- miner.php add display 'notify' command
- Small change to help arch's without processor affinity
- Fix bitforce compile error
- api.c notify should report disabled devices also - of course
- API returns the simple device history with the 'notify' command
- code changes for supporting a simple device history
- api.c Report an OS string in config to help with device issues
- api.c fix Log Interval - integer in JSON
- api.c config 'Device Code' to show list of compiled devices + README
- api.c increase buffer size close to current code allowable limit
- removed 8-component vector support from kernel, as this is not supported in
CGMINER anyway
- forgot to update kernel modification date, fixed ;)
- reordered an addition in the kernel, which results in less instructions used
in the GPU ISA code for GCN
- miner.php: option for readonly or check privileged access
- Ignore reduntant-with-build options --disable-gpu, --no-adl, and --no-restart
- miner.php: ereg_replace is DEPRECATED so use preg_replace instead
- Make curses TUI support optional at compile-time.
- Bugfix: AC_ARG_WITH provides withval instead of enableval
- miner.php split devs output for different devices
- api.c: correct error messages
- icarus.c modify (regular) timeout warning to only be debug
- icarus.c set the windows TODO timeout
- Allow specifying a specific driver for --scan-serial
- optimized nonce-check and output code for -v 2 and -v 4
- Bugfix: Check for libudev header (not just library) in configure, and document
optional dependency
- Add API support for Icarus and Bitforce
- Next API version is 1.4 (1.3 is current)
- README/api.c add "When" the request was processed to STATUS
- Bugfix: ZLX to read BitFORCE temp, not ZKX -.-
- Use libudev to autodetect BitFORCE GPUs, if available
- Use the return value of fan_autotune to set fan_optimal instead of passing it
as a pointer.
- Pass the lasttemp from the device we're using to adjust fanspeed in twin
devices.
- fix the name to 3 chars, fix the multi-icarus support
- Bugfix: "-S auto" is the default if no -S is specified, and there is no such
delay in using it
- README add information missing from --scan-serial
- Update README RPC API Version comment
- Bugfix: Allow enabling CPU even without OpenCL support
- Change failed-to-mine number of requested shares messge to avoid segfault on
recursive calling of quit().
- Get rid of extra char which is just truncated in poclbm kernel.
- only small code formating changes
- removed vec_step() as this could lead to errors on older SDKs
- unified code for generating nonce in kernel and moved addition of base to the
end -> faster

Version 2.3.1 - February 24, 2012

- Revert input and output code on diakgcn and phatk kernels to old style which
worked better for older hardware and SDKs.
- Add a vector*worksize parameter passed to those kernels to avoid one op.
- Increase the speed of hashrate adaptation.
- Only send out extra longpoll requests if we want longpolls.
- API implement addpool command
- API return the untouched Total MH also (API now version 1.3)
- Add enable/disablepool to miner.php example and reduce font size 1pt


Version 2.3.0 - February 23, 2012

- Consider extra longpoll work items as staged_extra so as to make sure we queue
more work if queueing regular work items as longpolls.
- Use diablo kernel on all future SDKs for Tahiti and set preferred vector width
to 1 on poclbm kernel only.
- Explicitly type the constants in diakgcn kernel as uint, to be in line with
poclbm kernel.
- Reset all hash counters at the same time as resetting start times to get
accurate hashrates on exiting which is mandatory for benchmarking.
- Report thread out before it starts to avoid being flagged as sick when waiting
for the first work item.
- Don't disable and re-enable devices as they may recover and in the meantime
have their status set to OFF.
- API new commands enablepool and disablepool (version already incremented)
- Tolerate new-format temperature readings for bitforce
- Modify cgminer.c pool control to allow API to call it
- Bugfix: Fix BitFORCE driver memory leak in debug logging
- Extra byte was being unused in poclbm leading to failure on some platforms.
- Explicitly type the constants in poclbm kernel as uint.
- Don't save 'include' when saving the configuration
- Allow configuration file to include another recursively
- Use the SDK and hardware information to choose good performing default
kernels.
- Move phatk kernel to offset vector based nonce bases as well.
- Add a --benchmark feature which works on a fake item indefinitely to compare
device performance without any server or networking influence.
- Allow writing of multiple worksizes to the configuration file.
- Allow writing of multiple vector sizes to the configuration file.
- Allow writing of multiple kernels to the configuration file.
- Allow multiple different kernels to be chosen per device.
- Allow the worksize to be set per-device.
- Allow different vectors to be set per device.
- If we're well below the target temperature, increase gpu engine speed back to
maximum in case we have gotten lost between profiles during an idle period.
- We should be setting the value of fan_optimal, not its address.
- As all kernels will be new versions it's an opportunity to change the .bin
format and make it simpler. Specifying bitalign is redundant and long can be l.
- Use any() in kernel output code.
- Put the nonce for each vector offset in advance, avoiding one extra addition
in the kernel.
- Reset times after all mining threads are started to make estimating hashrates
easier at startup.
- Bugfix: allow no-exec (NX) stack
- Fix minor warning.
- fix the bitforce.c code style follow 1TBS
- fix icarus.c compile warning
- small changes to speedup no vec for AMD 898.1 OCL runtime
- Update licensing to GPL V3.
- Reset the longpoll flag after it's been used once to prevent it restarting
work again.
- Begin import of DiabloMiner kernel.
- Modify API debug messages to say API instead of DBG
- When API shuts down cgminer don't kill itself
- Don't make rolled work from the longpoll be seen as other longpoll work items.
- API add 'privileged' command so can verify access level
- Set the lp_sent variable under lock since there will almost always be a race
on setting this variable, potentially leading to multiple LPs being sent out.
- API restrict access to all non display commands by default
- Update API version to 1.2 for new 'Log Interval'
- API add --log Interval to 'config' reply
- --api-allow special case 0/0 means all


Version 2.2.7 - February 20, 2012

- Send out extra longpolls when we have switched pools and the longpoll thread
is still bound to the old one. This is particularly useful with p2pool where
longpolls do not correlate with main bitcoin block change and would have led to
high reject rates on failover.
- Store whether a work item is the result of a longpoll or not in struct work
and use it to help determine block changes directly from the work longpoll bool.
- Keep track of when a longpoll has been sent for a pool and if the current pool
is requesting work but has not sent a longpoll request, convert one of the work
items to a longpoll.
- Store the longpoll url in the pool struct and update it from the pool_active
test in case it changes. This is to allow further changes to longpoll management
on switching pools.
- Re-check for a longpoll supporting pool every 30 seconds if none is found
initially.
- Report threads as busy waiting on getwork on startup to avoid them being
flagged sick on startup during slow networking.
- Allow devices that are disabled due to overheating to be flagged as recovering
instead of disabling them and re-enable them if they're below ideal temperatures
- Tahiti prefers worksize 64 with poclbm.
- No need to expressly retain the opencl program now that the zero binary issue
is fixed. This actually fixes cgminer to work with the latest SDK included with
the ATI catalyst driver 12.2.
- Show error code on any opencl failure status.
- Add detection for version 898.1 SDK as well but only give SDK 2.6 warning once
on startup instead of with each device initialisation.
- Always use a fresh connection for longpoll as prolonged persistent connections
can fail for many reasons.
- Keep track of intended engine clock speed and only adjust up if it's higher
than the last intended speed. This avoids setting the clock speed to one
relative to a lower profile one by mistake.
- Use gpu-memdiff on startup if an engine clockspeed is set and a memdiff value
is set.
- Revert "Adjust engine speed up according to performance level engine setting,
not the current engine speed." - ineffectual.
- Freeze the queues on all threads that are sent the pause message to prevent
them trying to start up again with saved pings in their queues.
- Updates to diakgcn kernel/
- Consolidate all screen updates to the watchdog thread and touch both windows
before refresh.
- Curses will be disabled in clean_up so don't do it early in kill_work, and
disable_adl so that GPU settings may be restored to normal in case shutting down
curses leads to instability on windows.
- Stop the mining threads before trying to kill them.
- Plain refresh() does not give reliably screen updates so get rid of all uses
of it.
- First release with working diakgcn kernel.

Version 2.2.6 - February 16, 2012

- Provide warning on each startup about sdk 2.6
- Fix unused warnings on win32.
- bitforce: Simplify BFopen WIN32 ifdef/else
- Fix initialization warning with jansson 1.3
- bitforce: Cleanup extraneous TODO that isn't needed
- Move tcsetattr (and new tcflush) into *nix BFopen to simplify things a bit
- Add message explaining 2nd thread disabling for dynamic mode and how to tune
it.
- Move logwindow down once number of devices is known.
- Automatically choose phatk kernel for bitalign non-gcn ATI cards, and then
only select poclbm if SDK2.6 is detected.
- Allow the refresh interval to be adjusted in dynamic intensity with a
--gpu-dyninterval parameter.
- Make curses display visible right from the beginning and fix the window sizes
so the initial messages don't get lost once the status window is drawn.
- The amount of work scanned can fluctuate when intensity changes and since we
do this one cycle behind, we increment the work more than enough to prevent
repeati
- bitforce: Set a 30 second timeout for serial port on Windows, since the
default is undefined
- Use PreVal4addT1 instead of PreVal4 in poclbm kernel.
- Import PreVal4 and PreVal0 into poclbm kernel.
- Import more prepared constants into poclbm kernel.
- Keep variables in one array but use Vals[] name for consistency with other
kernel designs.
- Replace constants that are mandatorily added in poclbm kernel with one value.
- Remove addition of final constant before testing for result in poclbm kernel.
- Hand optimise variable addition order.
- Hand optimise first variable declaration order in poclbm kernel.
- Radical reordering machine based first pass to change variables as late as
possible, bringing their usage close together.
- fix strcpy NULL pointer if env HOME unset.
- bitforce: Disable automatic scanning when at least one device is specified
manually
- Unroll all poclbm additions to enable further optimisations.


Version 2.2.5 - February 13, 2012

- Make output buffer write only as per Diapolo's suggestion.
- Constify nonce in poclbm.
- Use local and group id on poclbm kernel as well.
- Microoptimise phatk kernel on return code.
- Adjust engine speed up according to performance level engine setting, not the
current engine speed.
- Try to load a binary if we've defaulted to the poclbm kernel on SDK2.6
- Use the poclbm kernel on SDK2.6 with bitalign devices only if there is no
binary available.
- Further generic microoptimisations to poclbm kernel.
- The longstanding generation of a zero sized binary appears to be due to the
OpenCL library putting the binary in a RANDOM SLOT amongst 4 possible binary
locations. Iterate over each of them after building from source till the real
binary is found and use that.
- Fix harmless warnings with -Wsign-compare to allow cgminer to build with -W.
- Fix missing field initialisers warnings.
- Put win32 equivalents of nanosleep and sleep into compat.h fixing sleep() for
adl.c.
- Restore compatibility with Jansson 1.3 and 2.0 (api.c required 2.1)
- Modularized logging, support for priority based logging
- Move CPU chipset specific optimization into device-cpu


Version 2.2.4 - February 11, 2012

- Fix double definition of A0 B0 to zeroA zeroB.
- Retain cl program after successfully loading a binary image. May decrease
failures to build kernels at startup.
- Variable unused after this so remove setting it.
- BFI INT patching is not necessarily true on binary loading of files and not
true on ATI SDK2.6+. Report bitalign instead.
- Various string fixes for reject reason.
- Generalize --temp-cutoff and implement support for reading temperature from
BitFORCE FPGAs
- Change message from recovered to alive since it is used on startup as well as
when a pool has recovered.
- Start mining as soon as any pool is found active and rely on the watchpool
thread to bring up other pools.
- Delayed responses from testing pools that are down can hold up the watchdog
thread from getting to its device testing code, leading to false detection of
the GPU not checking in, and can substantially delay auto gpu/auto fan
management leading to overheating. Move pool watching to its own thread.
- Bugfix: BitFORCE index needs to be static to count correctly
- Space out retrieval of extra work according to the number of mining threads.
- Make shutdown more robust. Enable the input thread only after the other
threads exist. Don't kill off the workio thread and use it to exit main() only
if there is an unexpected problem. Use kill_work() for all anticipated shutdowns
where possible. Remove unused thread entry.
- Change poclbm version number.
- One array is faster than 2 separate arrays so change to that in poclbm kernel.
- Microoptimisations to poclbm kernel which increase throughput slightly.
- Import diablominer kernel. Currently disabled as not working.
- Import diapolo kernel. Currently disabled as not working.
- Conflicting entries of cl_kernel may have been causing problems, and
automatically chosen kernel type was not being passed on. Rename the enum to
cl_kernels and store the chosen kernel in each clState.
- Set cl_amd_media_ops with the BITALIGN flag and allow non-bitselect devices to
build.
- ALlow much longer filenames for kernels to load properly.
- Allow different kernels to be used by different devices and fix the logic fail
of overcorrecting on last commit with !strstr.
- Fix kernel selection process and build error.
- queue_phatk_kernel now uses CL_SET_VARG() for base-nonce(s), too
- added OpenCL >= 1.1 detection code, in preparation of OpenCL 1.1 global offset
parameter support
- Use K array explicitly to make it clear what is being added.
- Work items have a tendency to expire at exactly the same time and we don't
queue extra items when there are plenty in the queue, regardless of age. Allow
extra work items to be queued if adequate time has passed since we last
requested work even if over the limit.
- Discard work when failover-only is enabled and the work has come from a
different pool.
- Missing include to build on newer mingw32.
- Move from the thread safe localtime_r to regular localtime which is the only
one supported on newer pthread libraries on mingw32 to make it compile with the
newer ming. Thread safety is of no importance where localtime is used in this
code.
- Define in_addr_t in windows if required
- sys/wait.h not required in windows
- Allow API to restrict access by IP address
- Add pool switching to example miner.php
- Display X-Reject-Reason, when provided
- Remove the test for whether the device is on the highest profil level before
raising the GPU speed as it is ineffectual and may prevent raising the GPU
speed.
- Remove unnecessary check for opt_debug one every invocation of applog at
LOG_DEBUG level and place the check in applog().


Version 2.2.3 - February 6, 2012

- Revert "Rewrite the convoluted get_work() function to be much simpler and roll
work as much as possible with each new work item." This seems to cause a race on
work in free_work(). Presumably other threads are still accessing the structure.


Version 2.2.2 - February 6, 2012

- Provide support for the submitold extension on a per-pool basis based on the
value being detected in a longpoll.
- Don't send a ping to a dynamic device if it's not enabled as that will just
enable it for one pass and then disable it again.
- Rewrite the convoluted get_work() function to be much simpler and roll work as
much as possible with each new work item.
- Roll as much work as possible from the work returned from a longpoll.
- Rolling work on each loop through the mining thread serves no purpose.
- Allow to stage more than necessary work items if we're just rolling work.
- Replace divide_work with reuse_work function used twice.
- Give rolled work a new ID to make sure there is no confusion in the hashtable
lookups.
- Remove now-defunct hash_div variables.
- Remove unused get_dondata function.
- Silence ADL warnings.
- Silence unused parameter warnings.
- Stagger the restart of every next thread per device to keep devices busy ahead
of accessory threads per device.
- Deprecate the --donation feature. Needlessly complex, questionable usefulness,
depends on author's server and a central pool of some kind, and was not heavily
adopted.
- It's devices that report back now, not threads, update message.
- Continue auto-management of fan and engine speeds even if a device is disabled
for safety reasons.
- No need to check we're highest performance level when throttling GPU engine
speed.
- Abstract out tests for whether work has come from a block that has been seen
before and whether a string is from a previously seen block.
- Probe but don't set the timeout to 15 seconds as some networks take a long
time to timeout.
- Remove most compiler warnings from api.c
- Add last share's pool info in cgpu_info
- Allow the OpenCL platform ID to be chosen with --gpu-platform.
- Iterate over all platforms displaying their information and number of devices
when --ndevs is called.
- Deprecate main.c
- Some networks can take a long time to resolve so go back to 60 second timeouts
instead of 15.
- Only enable curses on failure if curses is desired.
- Fix warnings in bitforce.c
- Bugfix: Need to open BitForce tty for read-write
- Fix various build issues.
- Modularize code: main.c -> device-cpu + device-gpu
- Fix phatk kernel not working on non-bitalign capable devices (Nvidia, older
ATI).
- Update poclbm kernel for better performance on GCN and new SDKs with bitalign
support when not BFI INT patching. Update phatk kernel to work properly for non
BFI INT patched kernels, providing support for phatk to run on GCN and non-ATI
cards.
- Return last accepted share pool/time for devices
- Display accepted share pool/time for CPUs
- Bug intensity always shows GPU 0
- Update example web miner.php to use new API commands


Version 2.2.1 - January 30, 2012

NOTE - The GPU Device reordering in 2.2.0 by default was considered a bad idea
so the original GPU ordering is used by default again unless reordering is
explicitly requested.

- Fix bitforce failing to build into cgminer.
- Add missing options to write config function.
- Add a --gpu-reorder option to only reorder devices according to PCI Bus ID
when requested.
- Fix for midstate support being broken on pools that supported no-midstate
work by ensuring numbers are 32 bits in sha2.c
- Set virtual GPUs to work when ADL is disabled or all mining will occur on GPU
0.
- Add information about paused threads in the menu status.
- Disable all but the first thread on GPUs in dynamic mode for better
interactivity.
- Set the latest network access time on share submission for --net-delay even if
we're not delaying that submission for further network access.
- Clear adl on exiting after probing values since it may attempt to overclock.
- As share submission is usually staggered, and delays can be costly, submit
shares without delay even when --net-delay is enabled.
- Display GPU number and device name when ADL is successfully enabled on it.
- Display GPU ordering remapping in verbose mode.
- Don't fail in the case the number of ADL and OpenCL devices do not match, and
do not attempt to reorder devices unless they match. Instead give a warning
about
- Display error codes should ADL not return ADL_OK in the more critical function
calls.
- Fix unused warning.
- Fix compile warnings in api.c
- Add extensive ADL based device info in debug mode.
- Make --ndevs display verbose opencl information as well to make debugging
version information easier.
- Display information about the opencl platform with verbose enabled.
- Explicitly check for nvidia in opencl platform strings as well.


Version 2.2.0 - January 29, 2012

NOTE: GPU Device order will change with this release with ATI GPUs as cgminer
now can enumerate them according to their Bus ID which means the values should
now correlate with their physical position on the motherboard.

- Default to poclbm kernel on Tahiti (7970) since phatk does not work, even
though performance is sub-standard so that at least it will mine successfully by
defau
- Retain cl program after every possible place we might build the program.
- Update ADL SDK URL.
- Fix potential overflow.
- Map GPU devices to virtual devices in their true physical order based on
BusNumber.
- Change the warning that comes with failure to init cl on a device to be more
generic and accurate.
- Advertise longpoll support in X-Mining-Extensions
- Detect dual GPU cards by iterating through all GPUs, finding ones without
fanspeed and matching twins with fanspeed one bus ID apart.
- Do not attempt to build the program that becomes the kernel twice. This could
have been leading to failures on initialising cl.
- Some opencl compilers have issues with no spaces after -D in the compiler
options.
- Allow intensity up to 14.
- Use calloced stack memory for CompilerOptions to ensure sprintf writes to the
beginning of the char.
- Whitelist 79x0 cards to prefer no vectors as they perform better without.
- Adjust fan speed gently while in the optimal range when temperature is
drifting to minimise overshoot in either direction.
- Detect dual GPU cards via the indirect information of - 1st card has a fan
controller. 2nd card does not have a fan controller, cards share the same device
name
- Instead of using the BFI_INT patching hack on any device reporting
cl_amd_media_ops, create a whitelist of devices that need it. This should enable
GCN architec
- Fixed API compiling issue on OS X
- Add more explanation of JSON format and the 'save' command
- Return an error if using ADL API commands when it's not available
- Read off lpThermalControllerInfo from each ADL device.
- Add ADL_Overdrive5_ThermalDevices_Enum interface.
- Add API commands: config, switchpool, gpu settings, save
- Implement socks4 proxy support.
- Fix send() for JSON strings
- Introduce a --net-delay option which guarantees at least 250ms between any
networking requests to not overload slow routers.
- Generalise locking init code.
- Allow invalid values to be in the configuration file, just skipping over them
provided the rest of the file is valid JSON. This will allow older configurat
- Allow CPU mining explicitly enable only if other mining support is built in.
- BitForce FPGA support
- Configure out building and support of all CPU mining code unless
--enable-cpumining is enabled.
- Allow parsed values to be zero which will allow 0 values in the config file to
work.
- Advertise that we can make our own midstate, so the pool can skip generating
it for us
- Refactor the CPU scanhash_* functions to use a common API. Fixes bugs.
- Don't consider a pool lagging if a request has only just been filed. This
should decrease the false positives for "pool not providing work fast enough".
- Invalidating work after longpoll made hash_pop return no work giving a false
positive for dead pool. Rework hash_pop to retry while finds no staged work u
- Remove TCP_NODELAY from curl options as many small packets may be contributing
to network overload, when --net-delay is enabled.
- Refactor miner_thread to be common code for any kind of device
- Simplify submit_nonce loop and avoid potentially missing FOUND - 1 entry.
Reported by Luke-Jr.
- Micro-optimisation in sha256_sse2 code courtesy of Guido Ascioti
guido.ascioti@gmail.com
- Refactor to abstract device-specific code


Version 2.1.2 - January 6, 2012

- If api-description is specified, save it when writing the config file
- Adjust utility width to be constant maximum as well.
- Add percent signs to reject ratio outputs
- Should the donation pool fail, don't make the fallover pool behave as though
the primary pool is lagging.
- Use an alternative pool should the donation getwork fail.


Version 2.1.1 - January 1, 2012

- Include API examples in distribution tarball.
- Don't attempt to pthread_join when cancelling threads as they're already
detached and doing so can lead to a segfault.
- Give more generic message if slow pool at startup is the donation pool.
- Continue to attempt restarting GPU threads if they're flagged dead at 1 min.
intervals.
- Don't attempt to restart sick flagged GPUs while they're still registering
activity.
- Make curl use fresh connections whenever there is any communication issue
in case there are dead persistent connections preventing further comms from
working.
- Display pool in summary if only 1 pool.
- Adjust column width of A/R/HW to be the maximum of any device and align them.


Version 2.1.0 - December 27, 2011

- Major infrastructure upgrade with RPC interface for controlling via sockets
encoded with/without JSON courtesy of Andrew Smith. Added documentation for
use of the API and sample code to use with it.
- Updated linux-usb-cgminer document.
- Rewrite of longpoll mechanism to choose the current pool wherever possible to
use for the longpoll, or any pool that supports longpoll if the current one
does not.
- Display information about longpoll when the chosen server has changed.
- Fix the bug where longpoll generated work may have been sent back to the
wrong pool, causing rejects.
- Fix a few race conditions on closing cgminer which caused some of the crashes
on exit.
- Only adjust gpu engine speed in autotune mode if the gpu is currently at the
performance level of that being adjusted.
- Various fixes for parsing/writing of configuration files.
- Do not add blank lines for threads of unused CPUs.
- Show which pool is unresponsive on startup.
- Only show GPU management menu item if GPUs are in use.
- Align most device columns in the curses display.


Version 2.0.8 - November 11, 2011

- Make longpoll do a mandatory flushing of all work even if the block hasn't
changed, thus supporting longpoll initiated work change of any sort and merged
mining.
- Byteswap computed hash in hashtest so it can be correctly checked. This fixes
the very rare possibility that a block solve on solo mining was missed.
- Add x86_64 w64 mingw32 target
- Allow a fixed speed difference between memory and GPU clock speed with
--gpu-memdiff that will change memory speed when GPU speed is changed in
autotune mode.
- Don't load the default config if a config file is specified on the command
line.
- Don't build VIA on apple since -a auto bombs instead of gracefully ignoring
VIA failing.
- Build fix for dlopen/dlclose errors in glibc.


Version 2.0.7 - October 17, 2011

- Support work without midstate or hash1, which are deprecated in bitcoind 0.5+
- Go to kernel build should we fail to clCreateProgramWithBinary instead of
failing on that device. This should fix the windows problems with devices not
initialising.
- Support new configuration file format courtesy of Chris Savery which can write
the config file from the menu and will load it on startup.
- Write unix configuration to .cgminer/cgminer.conf by default and prompt to
overwrite if given a filename from the menu that exists.


Version 2.0.6 - October 9, 2011

- Must initialise the donorpool mutex or it fails on windows.
- Don't make donation work interfere with block change detection allowing
donation to work regardless of the block chain we're mining on.
- Expire shares as stale with a separate timeout from the scantime, defaulting
to 120 seconds.
- Retry pools after a delay of 15 seconds if none can be contacted on startup
unless a key is pressed.
- Don't try to build adl features without having adl.
- Properly check shares against target difficulty - This will no longer show
shares when solo mining at all unless they're considered to be a block solve.
- Add altivec 4 way (cpu mining) support courtesy of Gilles Risch.
- Try to use SSL if the server supports it.
- Display the total solved blocks on exit (LOL if you're lucky).
- Use ADL activity report to tell us if a sick GPU is still busy suggesting it
is hard hung and do not attempt to restart it.


Version 2.0.5 - September 27, 2011

- Intensity can now be set to dynamic or static values per-device.
- New donation feature --donation sends a proportion of shares to author's
account of choice, but is disabled by default!
- The hash being displayed and block detection has been fixed.
- Devices not being mined on will not attempt to be ADL managed.
- Intensity is now displayed per GPU device.
- Make longpoll attempt to restart as often as opt_retries specifies.
- We weren't rolling work as often as we could.
- Correct some memory management issues.
- Build fixes.
- Don't mess with GPUs if we don't have them.


Version 2.0.4 - September 23, 2011

- Confused Longpoll messages should be finally fixed with cgminer knowing for
sure who found the new block and possibly avoiding a rare crash.
- Display now shows the actual hash and will say BLOCK! if a block is deemed
solved.
- Extra spaces, which would double space lines on small terminals, have been
removed.
- Fan speed change is now damped if it is already heading in the correct
direction to minimise overshoot.
- Building without opencl libraries is fixed.
- GPUs are autoselected if there is only one when in the GPU management menu.
- GPU menu is refreshed instead of returning to status after a GPU change.


Version 2.0.3 - September 17, 2011

- Various modes of failure to set fanspeeds and adl values have been addressed
and auto-fan should work now on most hardware, and possibly other values
which previously would not have worked.
- Fixed a crash that can occur on switching pools due to longpoll thread races.
- Use ATISTREAMSDKROOT if available at build time.
- Fanspeed management is returned to the driver default on exit instead of
whatever it was when cgminer was started.
- Logging of events deemed WARNING or ERR now will display even during
periods where menu input is being awaited on.


Version 2.0.2 - September 11, 2011

- Exit cleanly if we abort before various threads are set up or if they no
longer exist.
- Fix a rare crash in HASH_DEL due to using different mutexes to protect the
data.
- Flag devices that have never started and don't allow enabling of devices
without restarting them.
- Only force the adapter speed to high if we've flagged this device as being
managed.
- Flag any devices with autofan or autogpu as being managed.
- Use a re-entrant value to store what fanspeed we're trying to set in case the
card doesn't support small changes.     Force it to a multiple of 10% if it
fails on trying to speed up the fan.
- Do not bother resetting values to old ones if changes to GPU parameters report
failure, instead returning a failure code only if the return value from get()
differs.
- Remove redundant check.
- Only display supported values from fanspeed on change settings.
- Missing bracket from output.
- Display fan percentage on devices that only support reporting percent and not
RPM.
- Properly substitute DLOPEN flags to build with ADL support when -ldl is needed
and not when opencl is not found.


Version 2.0.1 - September 9, 2011

- Fix building on 32bit glibc with dlopen with -lpthread and -ldl
- ByteReverse is not used and the bswap opcode breaks big endian builds. Remove
it.
- Ignore whether the display is active or not since only display enabled devices
work this way, and we skip over repeat entries anwyay.
- Only reset values on exiting if we've ever modified them.
- Flag adl as active if any card is successfully activated.
- Add a thermal cutoff option as well and set it to 95 degrees by default.
- Change the fan speed by only 5% if it's over the target temperature but less
than the hysteresis value to minimise overshoot down in temperature.
- Add a --no-adl option to disable ADL monitoring and GPU settings.
- Only show longpoll received delayed message at verbose level.
- Allow temperatures greater than 100 degrees.
- We should be passing a float for the remainder of the vddc values.
- Implement accepting a range of engine speeds as well to allow a lower limit to
be specified on the command line.
- Allow per-device fan ranges to be set and use them in auto-fan mode.
- Display which GPU has overheated in warning message.
- Allow temperature targets to be set on a per-card basis on the command line.
- Display fan range in autofan status.
- Setting the hysteresis is unlikely to be useful on the fly and doesn't belong
in the per-gpu submenu.
- With many cards, the GPU summaries can be quite long so use a terse output
line when showing them all.
- Use a terser device status line to show fan RPM as well when available.
- Define max gpudevices in one macro.
- Allow adapterid 0 cards to enumerate as a device as they will be non-AMD
cards, and enable ADL on any AMD card.
- Do away with the increasingly confusing and irrelevant total queued and
efficiency measures per device.
- Only display values in the log if they're supported and standardise device log
line printing.


Version 2.0.0 - September 6, 2011

Major feature upgrade - GPU monitoring, (over)clocking and fan control for ATI
GPUs.

New command line switches:
--auto-fan-     Automatically adjust all GPU fan speeds to maintain a target
temperature
--auto-gpu-     Automatically adjust all GPU engine clock speeds to maintain
a target temperature
--gpu-engine <arg>  Set the GPU engine (over)clock in Mhz - one value for all or
separate by commas for per card.
--gpu-fan <arg>     Set the GPU fan percentage - one value for all or separate
by commas for per card.
--gpu-memclock <arg> Set the GPU memory (over)clock in Mhz - one value for all
or separate by commas for per card.
--gpu-powertune <arg> Set the GPU powertune percentage - one value for all or
separate by commas for per card.
--gpu-vddc <arg>    Set the GPU voltage in Volts - one value for all or separate
by commas for per card.
--temp-hysteresis <arg> Set how much the temperature can fluctuate outside
limits when automanaging speeds (default: 3)
--temp-overheat <arg> Set the overheat temperature when automatically managing
fan and GPU speeds (default: 85)
--temp-target <arg> Set the target temperature when automatically managing fan
and GPU speeds (default: 75)

- Implement ATI ADL support for GPU parameter monitoring now and setting later
(temp, fan, clocks etc.).
- Check for the presence of the ADL header files in ADL_SDK.
- Import adl_functions.h from amd overdrive ctrl.
- Implement a setup function that tries to detect GPUs that support the ADL and
link in the parameters into the gpus struct.
- Put a summary of monitoring information from the GPU menu.
- Implement changing memory speed and voltage on the fly.
- Implement fan speed setting.
- Minor corrections to set fan speed by percentage.
- Make sure to read off the value in RPM only.
- Implement auto fanspeed adjustment to maintain a target temperature and
fanspeed below 85%, with an overheat check that will speed the fan up to 100%.
- Add an --auto-fan command line option to allow all GPUs to have autofan
enabled from startup.
- Add a gpu autotune option which adjusts GPU speed to maintain a target
temperature within the bounds of the default GPU speed and any overclocking set.
- Avoid a dereference if the longpoll thread doesn't exist.
- Clean up by setting performance profiles and fan settings to startup levels on
exit.
- Add a small amount of hysteresis before lowering clock speed.
- Allow target, overheat and hysteresis temperatures to be set from command
line.
- Combine all stats collating into one function to avoid repeating function
calls on each variable.
- Add gpu statistics to debugging output via the watchdog thread.
- Implement menus to change temperature limits.
- Implement setting the GPU engine clock speed of all devices or each device as
a comma separated value.
- Implement setting the GPU memory clock speed of all devices or each device as
a comma separated value.
- Implement setting the GPU voltage of all devices or each device as a comma
separated value.
- Implement setting the GPU fan speed of all devices or each device as a comma
separated value.
- Add support for monitoring powertune setting.
- Implement changing of powertune value from the GPU change settings menu.
- Get the value of powertune in get_stats.
- Implement setting the GPU powertune value of all devices or each device as a
comma separated value.
- Remove the safety checks in speed setting since confirmation is done first in
the menu, then show the new current values after a short pause.
- Force the speed to high on startup and restore it to whatever the setting was
on exit.
- Add temperature to standard output where possible and use more compact output.
- Move and print at the same time in curses to avoid random trampling display
errors.
- Update the status window only from the watchdog thread, do not rewrite the top
status messages and only refresh once all the status window is complete,
clearing the window each time to avoid corruption.
- Set a safe starting fan speed if we're automanaging the speeds.
- Provide locking around all adl calls to prevent races.
- Lower profile settings cannot be higher than higher profile ones so link any
drops in settings.
- Add new needed text files to distribution.
- Queue requests ignoring the number of staged clones since they get discarded
very easily leading to false positives for pool not providing work fast enough.
- Include libgen.h in opt.c to fix win32 compilation warnings.
- Fix compilation warning on win32.
- Add the directory name from the arguments cgminer was called from as well to
allow it running from a relative pathname.
- Add a --disable-adl option to configure and only enable it if opencl support
exists.
- Retry before returning a failure to get upstream work as a failure to avoid
false positives for pool dead.
- Retry also if the decoding of work fails.
- Use the presence of X-Roll-Ntime in the header as a bool for exists unless N
is found in the response.


Version 1.6.2 - September 2, 2011

- Add --failover-only option to not leak work to backup pools when the primary
pool is lagging.
- Change recommendation to intensity 9 for dedicated miners.
- Fix the bouncing short term value by allowing it to change dynamically when
the latest value is very different from the rolling value, but damp the change
when it gets close.
- Use the curses_lock to protect the curses_active variable and test it under
lock.
- Go back to requesting work 2/3 of the way through the current scantime with
CPU mining as reports of mining threads running out of work have occurred with
only 5 seconds to retrieve work.
- Add start and stop time scheduling for regular time of day running or once off
start/stop options.
- Print summary on quit modes.
- Put some sanity checks on the times that can be input.
- Give a verbose message when no active pools are found and pause before
exiting.
- Add verbose message when a GPU fails to initialise, and disable the correct
GPU.
- Cryptopp asm32 was not correctly updated to the incremental nonce code so the
hash counter was bogus.
- Get rid of poorly executed curl check.
- If curl does not have sockopts, do not try to compile the
json_rpc_call_sockopt_cb function, making it possible to build against older
curl libraries.
- Most people expect /usr/local when an unspecified prefix is used so change to
that.
- Rename localgen occasions to getwork fail occasions since localgen is
unrelated now.


Version 1.6.1 - August 29, 2011

- Copy cgminer path, not cat it.
- Switching between redrawing windows does not fix the crash with old
libncurses, so redraw both windows, but only when the window size hasn't
changed.
- Reinstate minimum 1 extra in queue to make it extremely unlikely to ever have
0 staged work items and any idle time.
- Return -1 if no input is detected from the menu to prevent it being
interpreted as a 0.
- Make pthread, libcurl and libcurses library checks mandatory or fail.
- Add a --disable-opencl configure option to make it possible to override
detection of opencl and build without GPU mining support.
- Confusion over the variable name for number of devices was passing a bogus
value which likely was causing the zero sized binary issue.
- cgminer no longer supports default url user and pass so remove them.
- Don't show value of intensity since it's dynamic by default.
- Add options to explicitly enable CPU mining or disable GPU mining.
- Convert the opt queue into a minimum number of work items to have queued
instead of an extra number to decrease risk of getting idle devices without
increasing risk of higher rejects.
- Statify tv_sort.
- Check for SSE2 before trying to build 32 bit SSE2 assembly version. Prevents
build failure when yasm is installed but -msse2 is not specified.
- Add some defines to configure.ac to enable exporting of values and packaging,
and clean up output.
- Give convenient summary at end of ./configure.
- Display version information and add --version command line option, and make
sure we flush stdout.
- Enable curses after the mining threads are set up so that failure messages
won't be lost in the curses interface.
- Disable curses after inputting a pool if we requested no curses interface.
- Add an option to break out after successfully mining a number of accepted
shares.
- Exit with a failed return code if we did not reach opt_shares.
- The cpu mining work data can get modified before we copy it if we submit it
async, and the sync submission is not truly sync anyway, so just submit it sync.


Version 1.6.0 - August 26, 2011

- Make restarting of GPUs optional for systems that hang on any attempt to
restart them.     Fix DEAD status by comparing it to last live time rather than
last attempted restart time since that happens every minute.
- Move staged threads to hashes so we can sort them by time.
- Create a hash list of all the blocks created and search them to detect when a
new block has definitely appeared, using that information to detect stale work
and discard it.
- Update configure.ac for newer autoconf tools.
- Use the new hashes directly for counts instead of the fragile counters
currently in use.
- Update to latest sse2 code from cpuminer-ng.
- Allow LP to reset block detect and block detect lp flags to know who really
came first.
- Get start times just before mining begins to not have very slow rise in
average.
- Add message about needing one server.
- We can queue all the necessary work without hitting frequent stales now with
the time and string stale protection active all the time.     This prevents a
pool being falsely labelled as not providing work fast enough.
- Include uthash.h in distro.
- Implement SSE2 32 bit assembly algorithm as well.
- Fail gracefully if unable to open the opencl files.
- Make cgminer look in the install directory for the .cl files making make
install work correctly.
- Allow a custom kernel path to be entered on the command line.
- Bump threshhold for lag up to maximum queued but no staged work.
- Remove fragile source patching for bitalign, vectors et. al and simply pass it
with the compiler options.
- Actually check the value returned for the x-roll-ntime extension to make sure
it isn't saying N.
- Prevent segfault on exit for when accessory threads don't exist.
- Disable curl debugging with opt protocol since it spews to stderr.


Version 1.5.8 - August 23, 2011

- Minimise how much more work can be given in cpu mining threads each interval.
- Make the fail-pause progressively longer each time it fails until the network
recovers.
- Only display the lagging message if we've requested the work earlier.
- Clean up the pool switching to not be dependent on whether the work can roll
or not by setting a lagging flag and then the idle flag.
- Only use one thread to determine if a GPU is sick or well, and make sure to
reset the sick restart attempt time.
- The worksize was unintentionally changed back to 4k by mistake, this caused a
slowdown.


Version 1.5.7 - August 22, 2011

- Fix a crash with --algo auto
- Test at appropriate target difficulty now.
- Add per-device statics log output with --per-device-stats
- Fix breakage that occurs when 1 or 4 vectors are chosen on new phatk.
- Make rolltime report debug level only now since we check it every work
item.
- Add the ability to enable/disable per-device stats on the fly and match
logging on/off.
- Explicitly tell the compiler to retain the program to minimise the chance of
the zero sized binary errors.
- Add one more instruction to avoid one branch point in the common path in the
cl return code. Although this adds more ALUs overall and more branch points, the
common path code has the same number of ALUs and one less jmp, jmps being more
expensive.
- Explicitly link in ws2_32 on the windows build and update README file on how
to compile successfully on windows.
- Release cl resources should the gpu mining thread abort.
- Attempt to restart a GPU once every minute while it's sick.
- Don't kill off the reinit thread if it fails to init a GPU but returns safely.
- Only declare a GPU dead if there's been no sign of activity from the reinit
thread for 10 mins.
- Never automatically disable any pools but just specify them as idle if they're
unresponsive at startup.
- Use any longpoll available, and don't disable it if switching to a server that
doesn't have it. This allows you to mine solo, yet use the longpoll from a pool
even if the pool is the backup server.
- Display which longpoll failed and don't free the ram for lp_url since it
belongs to the pool hdr path.
- Make the tcp setsockopts unique to linux in the hope it allows freebsd et. al
to compile.


Version 1.5.6 - August 17, 2011

- New phatk and poclbm kernels. Updated phatk to be in sync with latest 2.2
courtesy of phateus. Custom modified to work best with cgminer.
- Updated output buffer code to use a smaller buffer with the kernels.
- Clean up the longpoll management to ensure the right paths go to the right
pool and display whether we're connected to LP or not in the status line.


Version 1.5.5 - August 16, 2011

- Rework entirely the GPU restart code. Strike a balance between code that
re-initialises the GPU entirely so that soft hangs in the code are properly
managed, but if a GPU is completely hung, the thread restart code fails
gracefully, so that it does not take out any other code or devices. This will
allow cgminer to keep restarting GPUs that can be restarted, but continue
mining even if one or more GPUs hangs which would normally require a reboot.
- Add --submit-stale option which submits all shares, regardless of whether they
would normally be considered stale.
- Keep options in alphabetical order.
- Probe for slightly longer for when network conditions are lagging.
- Only display the CPU algo when we're CPU mining.
- As we have keepalives now, blaming network flakiness on timeouts appears to
have been wrong.     Set a timeout for longpoll to 1 hour, and most other
network connectivity to 1 minute.
- Simplify output code and remove HW errors from CPU stats.
- Simplify code and tidy output.
- Only show cpu algo in summary if cpu mining.
- Log summary at the end as per any other output.
- Flush output.
- Add a linux-usb-cgminer guide courtesy of Kano.


Version 1.5.4 - August 14, 2011

- Add new option: --monitor <cmd> Option lets user specify a command <cmd> that
will get forked by cgminer on startup. cgminer's stderr output subsequently gets
piped directly to this command.
- Allocate work from one function to be able to initialise variables added
later.
- Add missing fflush(stdout) for --ndevs and conclusion summary.
- Preinitialise the devices only once on startup.
- Move the non cl_ variables into the cgpu info struct to allow creating a new
cl state on reinit, preserving known GPU variables.
- Create a new context from scratch in initCQ in case something was corrupted to
maximise our chance of succesfully creating a new worker thread. Hopefully this
makes thread restart on GPU failure more reliable, without hanging everything
in the case of a completely wedged GPU.
- Display last initialised time in gpu management info, to know if a GPU has
been re-initialised.
- When pinging a sick cpu, flush finish and then ping it in a separate thread in
the hope it recovers without needing a restart, but without blocking code
elsewhere.
- Only consider a pool lagging if we actually need the work and we have none
staged despite queue requests stacking up. This decreases significantly the
amount of work that leaks to the backup pools.
- The can_roll function fails inappropriately in stale_work.
- Only put the message that a pool is down if not pinging it every minute. This
prevents cgminer from saying pool down at 1 minute intervals unless in debug
mode.
- Free all work in one place allowing us to perform actions on it in the future.
- Remove the extra shift in the output code which was of dubious benefit. In
fact in cgminer's implementation, removing this caused a miniscule speedup.
- Test each work item to see if it can be rolled instead of per-pool and roll
whenever possible, adhering to the 60 second timeout. This makes the period
after a longpoll have smaller dips in throughput, as well as requiring less
getworks overall thus increasing efficiency.
- Stick to rolling only work from the current pool unless we're in load balance
mode or lagging to avoid aggressive rolling imitating load balancing.
- If a work item has had any mining done on it, don't consider it discarded
work.


Version 1.5.3 - July 30, 2011

- Significant work went into attempting to make the thread restart code robust
to identify sick threads, tag them SICK after 1 minute, then DEAD after 5
minutes of inactivity and try to restart them. Instead of re-initialising the
GPU completely, only a new cl context is created to avoid hanging the rest of
the GPUs should the dead GPU be hung irrevocably.
- Use correct application name in syslog.
- Get rid of extra line feeds.
- Use pkg-config to check for libcurl version
- Implement per-thread getwork count with proper accounting to not over-account
queued items when local work replaces it.
- Create a command queue from the program created from source which allows us
to flush the command queue in the hope it will not generate a zero sized binary
any more.
- Be more willing to get work from the backup pools if the work is simply being
queued faster than it is being retrieved.


Version 1.5.2 - July 28, 2011

- Restarting a hung GPU can hang the rest of the GPUs so just declare it dead
and provide the information in the status.
- The work length in the miner thread gets smaller but doesn't get bigger if
it's under 1 second.     This could end up leading to CPU under-utilisation and
lower and lower hash rates.     Fix it by increasing work length if it drops
under 1 second.
- Make the "quiet" mode still update the status and display errors, and add a
new --real-quiet option which disables all output and can be set once while
running.
- Update utility and efficiency figures when displaying them.
- Some Intel HD graphics support the opencl commands but return errors since
they don't support opencl. Don't fail with them, just provide a warning and
disable GPU mining.
- Add http:// if it's not explicitly set for URL entries.
- Log to the output file at any time with warnings and errors, instead of just
when verbose mode is on.
- Display the correct current hash as per blockexplorer, truncated to 16
characters, with just the time.


Version 1.5.1 - July 27, 2011

- Two redraws in a row cause a crash in old libncurses so just do one redraw
using the main window.
- Don't adjust hash_div only up for GPUs. Disable hash_div adjustment for GPUs.
- Only free the thread structures if the thread still exists.
- Update both windows separately, but not at the same time to prevent the double
refresh crash that old libncurses has.     Do the window resize check only when
about to redraw the log window to minimise ncurses cpu usage.
- Abstract out the decay time function and use it to make hash_div a rolling
average so it doesn't change too abruptly and divide work in chunks large enough
to guarantee they won't overlap.
- Sanity check to prove locking.
- Don't take more than one lock at a time.
- Make threads report out when they're queueing a request and report if they've
failed.
- Make cpu mining work submission asynchronous as well.
- Properly detect stale work based on time from staging and discard instead of
handing on, but be more lax about how long work can be divided for up to the
scantime.
- Do away with queueing work separately at the start and let each thread grab
its own work as soon as it's ready.
- Don't put an extra work item in the queue as each new device thread will do so
itself.
- Make sure to decrease queued count if we discard the work.
- Attribute split work as local work generation.
- If work has been cloned it is already at the head of the list and when being
reinserted into the queue it should be placed back at the head of the list.
- Dividing work is like the work is never removed at all so treat it as such.
However the queued bool needs to be reset to ensure we *can* request more work
even if we didn't initially.
- Make the display options clearer.
- Add debugging output to tq_push calls.
- Add debugging output to all tq_pop calls.


Version 1.5.0 - July 26, 2011

- Increase efficiency of slow mining threads such as CPU miners dramatically. Do
this by detecting which threads cannot complete searching a work item within the
scantime and then divide up a work item into multiple smaller work items.
Detect the age of the work items and if they've been cloned before to prevent
doing the same work over. If the work is too old to be divided, then see if it
can be time rolled and do that to generate work. This dramatically decreases the
number of queued work items from a pool leading to higher overall efficiency
(but the same hashrate and share submission rate).
- Don't request work too early for CPUs as CPUs will scan for the full
opt_scantime anyway.
- Simplify gpu management enable/disable/restart code.
- Implement much more accurate rolling statistics per thread and per gpu and
improve accuracy of rolling displayed values.
- Make the rolling log-second average more accurate.
- Add a menu to manage GPUs on the fly allowing you to enable/disable GPUs or
try restarting them.
- Keep track of which GPUs are alive versus enabled.
- Start threads for devices that are even disabled, but don't allow them to
start working.
- The last pool is when we are low in total_pools, not active_pools.
- Make the thread restart do a pthread_join after disabling the device, only
re-enabling it if we succeed in restarting the thread. Do this from a separate
thread so as to not block any other code.This will allow cgminer to continue
even if one GPU hangs.
- Try to do every curses manipulation under the curses lock.
- Only use the sockoptfunction if the version of curl is recent enough.


Version 1.4.1 - July 24, 2011

- Do away with GET for dealing with longpoll forever. POST is the one that works
everywhere, not the other way around.
- Detect when the primary pool is lagging and start queueing requests on backup
pools if possible before needing to roll work.
- Load balancing puts more into the current pool if there are disabled pools.
Fix.
- Disable a GPU device should the thread fail to init.
- Out of order command queue may fail on osx. Try without if it fails.
- Fix possible dereference on blank inputs during input_pool.
- Defines missing would segfault on --help when no sse mining is built in.
- Revert "Free up resources/stale compilers." - didn't help.
- Only try to print the status of active devices or it would crash.
- Some hardware might benefit from the less OPS so there's no harm in leaving
kernel changes that do that apart from readability of the code.

Version 1.4.0 - July 23, 2011

- Feature upgrade: Add keyboard input during runtime to allow modification of
and viewing of numerous settings such as adding/removing pools, changing
multipool management strategy, switching pools, changing intensiy, verbosity,
etc. with a simple keypress menu system.
- Free up resources/stale compilers.
- Kernels are safely flushed in a way that allows out of order execution to
work.
- Sometimes the cl compiler generates zero sized binaries and only a reboot
seems to fix it.
- Don't try to stop/cancel threads that don't exist.
- Only set option to show devices and exit if built with opencl support.
- Enable curses earlier and exit with message in main for messages to not be
lost in curses windows.
- Make it possible to enter server credentials with curses input if none are
specified on the command line.
- Abstract out a curses input function and separate input pool function to allow
for live adding of pools later.
- Remove the nil arguments check to allow starting without parameters.
- Disable/enable echo & cbreak modes.
- Add a thread that takes keyboard input and allow for quit, silent, debug,
verbose, normal, rpc protocol debugging and clear screen options.
- Add pool option to input and display current pool status, pending code to
allow live changes.
- Add a bool for explicit enabling/disabling of pools.
- Make input pool capable of bringing up pools while running.
- Do one last check of the work before submitting it.
- Implement the ability to live add, enable, disable, and switch to pools.
- Only internally test for block changes when the work matches the current pool
to prevent interleaved block change timing on multipools.
- Display current pool management strategy to enable changing it on the fly.
- The longpoll blanking of the current_block data may not be happening before
the work is converted and appears to be a detected block change.     Blank the
current block be
- Make --no-longpoll work again.
- Abstract out active pools count.
- Allow the pool strategy to be modified on the fly.
- Display pool information on the fly as well.
- Add a menu and separate out display options.
- Clean up the messy way the staging thread communicates with the longpoll
thread to determine who found the block first.
- Make the input windows update immediately instead of needing a refresh.
- Allow log interval to be set in the menu.
- Allow scan settings to be modified at runtime.
- Abstract out the longpoll start and explicitly restart it on pool change.
- Make it possible to enable/disable longpoll.
- Set priority correctly on multipools.     Display priority and alive/dead
information in display_pools.
- Implement pool removal.
- Limit rolltime work generation to 10 iterations only.
- Decrease testing log to info level.
- Extra refresh not required.
- With huge variation in GPU performance, allow intensity to go from -10 to +10.
- Tell getwork how much of a work item we're likely to complete for future
splitting up of work.
- Remove the mandatory work requirement at startup by testing for invalid work
being passed which allows for work to be queued immediately.     This also
removes the requirem
- Make sure intensity is carried over to thread count and is at least the
minimum necessary to work.
- Unlocking error on retry. Locking unnecessary anyway so remove it.
- Clear log window from consistent place. No need for locking since logging is
disabled during input.
- Cannot print the status of threads that don't exist so just queue enough work
for the number of mining threads to prevent crash with -Q N.
- Update phatk kernel to one with new parameters for slightly less overhead
again.     Make the queue kernel parameters call a function pointer to select
phatk or poclbm.
- Make it possible to select the choice of kernel on the command line.
- Simplify the output part of the kernel. There's no demonstrable advantage from
more complexity.
- Merge pull request #18 from ycros/cgminer
- No need to make leaveok changes win32 only.
- Build support in for all SSE if possible and only set the default according to
machine capabilities.
- Win32 threading and longpoll keepalive fixes.
- Win32: Fix for mangled output on the terminal on exit.


Version 1.3.1 - July 20, 2011

- Feature upgrade; Multiple strategies for failover. Choose from default which
now falls back to a priority order from 1st to last, round robin which only
changes pools when one is idle, rotate which changes pools at user-defined
intervals, and load-balance which spreads the work evenly amongst all pools.
- Implement pool rotation strategy.
- Implement load balancing algorithm by rotating requests to each pool.
- Timeout on failed discarding of staged requests.
- Implement proper flagging of idle pools, test them with the watchdog thread,
and failover correctly.
- Move pool active test to own function.
- Allow multiple strategies to be set for multipool management.
- Track pool number.
- Don't waste the work items queued on testing the pools at startup.
- Reinstate the mining thread watchdog restart.
- Add a getpoll bool into the thread information and don't restart threads stuck
waiting on work.
- Rename the idlenet bool for the pool for later use.
- Allow the user/pass userpass urls to be input in any order.
- When json rpc errors occur they occur in spits and starts, so trying to limit
them with the comms error bool doesn't stop a flood of them appearing.
- Reset the queued count to allow more work to be queued for the new pool on
pool switch.

Version 1.3.0 - July 19, 2011

- Massive infrastructure update to support pool failover.
- Accept multiple parameters for url, user and pass and set up structures of
pool data accordingly.
- Probe each pool for what it supports.
- Implement per pool feature support according to rolltime support as
advertised by server.
- Do switching automatically based on a 300 second timeout of locally generated
work or 60 seconds of no response from a server that doesn't support rolltime.
- Implement longpoll server switching.
- Keep per-pool data and display accordingly.
- Make sure cgminer knows how long the pool has actually been out for before
deeming it a prolonged outage.
- Fix bug with ever increasing staged work in 1.2.8 that eventually caused
infinite rejects.
- Make warning about empty http requests not show by default since many
servers do this regularly.


Version 1.2.8 - July 18, 2011

- More OSX build fixes.
- Add an sse4 algorithm to CPU mining.
- Fix CPU mining with other algorithms not working.
- Rename the poclbm file to ensure a new binary is built since.
- We now are guaranteed to have one fresh work item after a block change and we
should only discard staged requests.
- Don't waste the work we retrieve from a longpoll.
- Provide a control lock around global bools to avoid racing on them.
- Iterating over 1026 nonces when confirming data from the GPU is old code
and unnecessary and can lead to repeats/stales.
- The poclbm kernel needs to be updated to work with the change to 4k sized
output buffers.
- longpoll seems to work either way with post or get but some servers prefer
get so change to httpget.


Version 1.2.7 - July 16, 2011

- Show last 8 characters of share submitted in log.
- Display URL connected to and user logged in as in status.
- Display current block and when it was started in the status line.
- Only pthread_join the mining threads if they exist as determined by
pthread_cancel and don't fail on pthread_cancel.
- Create a unique work queue for all getworks instead of binding it to thread 0
to avoid any conflict over thread 0's queue.
- Clean up the code to make it clear it's watchdog thread being messaged to
restart the threads.
- Check the current block description hasn't been blanked pending the real
new current block data.
- Re-enable signal handlers once the signal has been received to make it
possible to kill cgminer if it fails to shut down.
- Disable restarting of CPU mining threads pending further investigation.
- Update longpoll messages.
- Add new block data to status line.
- Fix opencl tests for osx.
- Only do local generation of work if the work item is not stale itself.
- Check for stale work within the mining threads and grab new work if
positive.
- Test for idle network conditions and prevent threads from being restarted
by the watchdog thread under those circumstances.
- Make sure that local work generation does not continue indefinitely by
stopping it after 10 minutes.
- Tweak the kernel to have a shorter path using a 4k buffer and a mask on the
nonce value instead of a compare and loop for a shorter code path.
- Allow queue of zero and make that default again now that we can track how
work is being queued versus staged. This can decrease reject rates.
- Queue precisely the number of mining threads as longpoll_staged after a
new block to not generate local work.


Version 1.2.6 - July 15, 2011

- Put a current system status line beneath the total work status line
- Fix a counting error that would prevent cgminer from correctly detecting
situations where getwork was failing - this would cause stalls sometimes
unrecoverably.
- Limit the maximum number of requests that can be put into the queue which
otherwise could get arbitrarily long during a network outage.
- Only count getworks that are real queue requests.


Version 1.2.5 - July 15, 2011

- Conflicting -n options corrected
- Setting an intensity with -I disables dynamic intensity setting
- Removed option to manually disable dynamic intensity
- Improve display output
- Implement signal handler and attempt to clean up properly on exit
- Only restart threads that are not stuck waiting on mandatory getworks
- Compatibility changes courtesy of Ycros to build on mingw32 and osx
- Explicitly grab first work item to prevent false positive hardware errors
due to working on uninitialised work structs
- Add option for non curses --text-only output
- Ensure we connect at least once successfully before continuing to retry to
connect in case url/login parameters were wrong
- Print an executive summary when cgminer is terminated
- Make sure to refresh the status window

Versions -> 1.2.4

- Con Kolivas - July 2011. New maintainership of code under cgminer name.
- Massive rewrite to incorporate GPU mining.
- Incorporate original oclminer c code.
- Rewrite gpu mining code to efficient work loops.
- Implement per-card detection and settings.
- Implement vector code.
- Implement bfi int patching.
- Import poclbm and phatk ocl kernels and use according to hardware type.
- Implement customised optimised versions of opencl kernels.
- Implement binary kernel generation and loading.
- Implement preemptive asynchronous threaded work gathering and pushing.
- Implement variable length extra work queues.
- Optimise workloads to be efficient miners instead of getting lots of extra
  work.
- Implement total hash throughput counters, per-card accepted, rejected and
  hw error count.
- Staging and watchdog threads to prevent fallover.
- Stale and reject share guarding.
- Autodetection of new blocks without longpoll.
- Dynamic setting of intensity to maintain desktop interactivity.
- Curses interface with generous statistics and information.
- Local generation of work (xroll ntime) when detecting poor network
connectivity.

Version 1.0.2

- Linux x86_64 optimisations - Con Kolivas
- Optimise for x86_64 by default by using sse2_64 algo
- Detects CPUs and sets number of threads accordingly
- Uses CPU affinity for each thread where appropriate
- Sets scheduling policy to lowest possible
- Minor performance tweaks

Version 1.0.1 - May 14, 2011

- OSX support

Version 1.0 - May 9, 2011

- jansson 2.0 compatibility
- correct off-by-one in date (month) display output
- fix platform detection
- improve yasm configure bits
- support full URL, in X-Long-Polling header

Version 0.8.1 - March 22, 2011

- Make --user, --pass actually work

- Add User-Agent HTTP header to requests, so that server operators may
  more easily identify the miner client.

- Fix minor bug in example JSON config file

Version 0.8 - March 21, 2011

- Support long polling: http://deepbit.net/longpolling.php

- Adjust max workload based on scantime (default 5 seconds,
  or 60 seconds for longpoll)

- Standardize program output, and support syslog on Unix platforms

- Suport --user/--pass options (and "user" and "pass" in config file),
  as an alternative to the current --userpass

Version 0.7.2 - March 14, 2011

- Add port of ufasoft's sse2 assembly implementation (Linux only)
  This is a substantial speed improvement on Intel CPUs.

- Move all JSON-RPC I/O to separate thread.  This reduces the
  number of HTTP connections from one-per-thread to one, reducing resource
  usage on upstream bitcoind / pool server.

Version 0.7.1 - March 2, 2011

- Add support for JSON-format configuration file.  See example
  file example-cfg.json.  Any long argument on the command line
  may be stored in the config file.
- Timestamp each solution found
- Improve sha256_4way performance.  NOTE: This optimization makes
  the 'hash' debug-print output for sha256_way incorrect.
- Use __builtin_expect() intrinsic as compiler micro-optimization
- Build on Intel compiler
- HTTP library now follows HTTP redirects

Version 0.7 - February 12, 2011

- Re-use CURL object, thereby reuseing DNS cache and HTTP connections
- Use bswap_32, if compiler intrinsic is not available
- Disable full target validation (as opposed to simply H==0) for now

Version 0.6.1 - February 4, 2011

- Fully validate "hash < target", rather than simply stopping our scan
  if the high 32 bits are 00000000.
- Add --retry-pause, to set length of pause time between failure retries
- Display proof-of-work hash and target, if -D (debug mode) enabled
- Fix max-nonce auto-adjustment to actually work.  This means if your
  scan takes longer than 5 seconds (--scantime), the miner will slowly
  reduce the number of hashes you work on, before fetching a new work unit.

Version 0.6 - January 29, 2011

- Fetch new work unit, if scanhash takes longer than 5 seconds (--scantime)
- BeeCee1's sha256 4way optimizations
- lfm's byte swap optimization (improves via, cryptopp)
- Fix non-working short options -q, -r

Version 0.5 - December 28, 2010

- Exit program, when all threads have exited
- Improve JSON-RPC failure diagnostics and resilience
- Add --quiet option, to disable hashmeter output.

Version 0.3.3 - December 27, 2010

- Critical fix for sha256_cryptopp 'cryptopp_asm' algo

Version 0.3.2 - December 23, 2010

- Critical fix for sha256_via

Version 0.3.1 - December 19, 2010

- Critical fix for sha256_via
- Retry JSON-RPC failures (see --retry, under "minerd --help" output)

Version 0.3 - December 18, 2010

- Add crypto++ 32bit assembly implementation
- show version upon 'minerd --help'
- work around gcc 4.5.x bug that killed 4way performance

Version 0.2.2 - December 6, 2010

- VIA padlock implementation works now
- Minor build and runtime fixes

Version 0.2.1 - November 29, 2010

- avoid buffer overflow when submitting solutions
- add Crypto++ sha256 implementation (C only, ASM elided for now)
- minor internal optimizations and cleanups

Version 0.2 - November 27, 2010

- Add script for building a Windows installer
- improve hash performance (hashmeter) statistics
- add tcatm 4way sha256 implementation
- Add experimental VIA Padlock sha256 implementation

Version 0.1.2 - November 26, 2010

- many small cleanups and micro-optimizations
- build win32 exe using mingw
- RPC URL, username/password become command line arguments
- remove unused OpenSSL dependency

Version 0.1.1 - November 24, 2010

- Do not build sha256_generic module separately from cpuminer.

Version 0.1 - November 24, 2010

- Initial release.