NEWS 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270
  1. Version 1.3.0 - July 19, 2011
  2. - Massive infrastructure update to support pool failover.
  3. - Accept multiple parameters for url, user and pass and set up structures of
  4. pool data accordingly.
  5. - Probe each pool for what it supports.
  6. - Implement per pool feature support according to rolltime support as
  7. advertised by server.
  8. - Do switching automatically based on a 300 second timeout of locally generated
  9. work or 60 seconds of no response from a server that doesn't support rolltime.
  10. - Implement longpoll server switching.
  11. - Keep per-pool data and display accordingly.
  12. - Make sure cgminer knows how long the pool has actually been out for before
  13. deeming it a prolonged outage.
  14. - Fix bug with ever increasing staged work in 1.2.8 that eventually caused
  15. infinite rejects.
  16. - Make warning about empty http requests not show by default since many
  17. servers do this regularly.
  18. Version 1.2.8 - July 18, 2011
  19. - More OSX build fixes.
  20. - Add an sse4 algorithm to CPU mining.
  21. - Fix CPU mining with other algorithms not working.
  22. - Rename the poclbm file to ensure a new binary is built since.
  23. - We now are guaranteed to have one fresh work item after a block change and we
  24. should only discard staged requests.
  25. - Don't waste the work we retrieve from a longpoll.
  26. - Provide a control lock around global bools to avoid racing on them.
  27. - Iterating over 1026 nonces when confirming data from the GPU is old code
  28. and unnecessary and can lead to repeats/stales.
  29. - The poclbm kernel needs to be updated to work with the change to 4k sized
  30. output buffers.
  31. - longpoll seems to work either way with post or get but some servers prefer
  32. get so change to httpget.
  33. Version 1.2.7 - July 16, 2011
  34. - Show last 8 characters of share submitted in log.
  35. - Display URL connected to and user logged in as in status.
  36. - Display current block and when it was started in the status line.
  37. - Only pthread_join the mining threads if they exist as determined by
  38. pthread_cancel and don't fail on pthread_cancel.
  39. - Create a unique work queue for all getworks instead of binding it to thread 0
  40. to avoid any conflict over thread 0's queue.
  41. - Clean up the code to make it clear it's watchdog thread being messaged to
  42. restart the threads.
  43. - Check the current block description hasn't been blanked pending the real
  44. new current block data.
  45. - Re-enable signal handlers once the signal has been received to make it
  46. possible to kill cgminer if it fails to shut down.
  47. - Disable restarting of CPU mining threads pending further investigation.
  48. - Update longpoll messages.
  49. - Add new block data to status line.
  50. - Fix opencl tests for osx.
  51. - Only do local generation of work if the work item is not stale itself.
  52. - Check for stale work within the mining threads and grab new work if
  53. positive.
  54. - Test for idle network conditions and prevent threads from being restarted
  55. by the watchdog thread under those circumstances.
  56. - Make sure that local work generation does not continue indefinitely by
  57. stopping it after 10 minutes.
  58. - Tweak the kernel to have a shorter path using a 4k buffer and a mask on the
  59. nonce value instead of a compare and loop for a shorter code path.
  60. - Allow queue of zero and make that default again now that we can track how
  61. work is being queued versus staged. This can decrease reject rates.
  62. - Queue precisely the number of mining threads as longpoll_staged after a
  63. new block to not generate local work.
  64. Version 1.2.6 - July 15, 2011
  65. - Put a current system status line beneath the total work status line
  66. - Fix a counting error that would prevent cgminer from correctly detecting
  67. situations where getwork was failing - this would cause stalls sometimes
  68. unrecoverably.
  69. - Limit the maximum number of requests that can be put into the queue which
  70. otherwise could get arbitrarily long during a network outage.
  71. - Only count getworks that are real queue requests.
  72. Version 1.2.5 - July 15, 2011
  73. - Conflicting -n options corrected
  74. - Setting an intensity with -I disables dynamic intensity setting
  75. - Removed option to manually disable dynamic intensity
  76. - Improve display output
  77. - Implement signal handler and attempt to clean up properly on exit
  78. - Only restart threads that are not stuck waiting on mandatory getworks
  79. - Compatibility changes courtesy of Ycros to build on mingw32 and osx
  80. - Explicitly grab first work item to prevent false positive hardware errors
  81. due to working on uninitialised work structs
  82. - Add option for non curses --text-only output
  83. - Ensure we connect at least once successfully before continuing to retry to
  84. connect in case url/login parameters were wrong
  85. - Print an executive summary when cgminer is terminated
  86. - Make sure to refresh the status window
  87. Versions -> 1.2.4
  88. - Con Kolivas - July 2011. New maintainership of code under cgminer name.
  89. - Massive rewrite to incorporate GPU mining.
  90. - Incorporate original oclminer c code.
  91. - Rewrite gpu mining code to efficient work loops.
  92. - Implement per-card detection and settings.
  93. - Implement vector code.
  94. - Implement bfi int patching.
  95. - Import poclbm and phatk ocl kernels and use according to hardware type.
  96. - Implement customised optimised versions of opencl kernels.
  97. - Implement binary kernel generation and loading.
  98. - Implement preemptive asynchronous threaded work gathering and pushing.
  99. - Implement variable length extra work queues.
  100. - Optimise workloads to be efficient miners instead of getting lots of extra
  101. work.
  102. - Implement total hash throughput counters, per-card accepted, rejected and
  103. hw error count.
  104. - Staging and watchdog threads to prevent fallover.
  105. - Stale and reject share guarding.
  106. - Autodetection of new blocks without longpoll.
  107. - Dynamic setting of intensity to maintain desktop interactivity.
  108. - Curses interface with generous statistics and information.
  109. - Local generation of work (xroll ntime) when detecting poor network
  110. connectivity.
  111. Version 1.0.2
  112. - Linux x86_64 optimisations - Con Kolivas
  113. - Optimise for x86_64 by default by using sse2_64 algo
  114. - Detects CPUs and sets number of threads accordingly
  115. - Uses CPU affinity for each thread where appropriate
  116. - Sets scheduling policy to lowest possible
  117. - Minor performance tweaks
  118. Version 1.0.1 - May 14, 2011
  119. - OSX support
  120. Version 1.0 - May 9, 2011
  121. - jansson 2.0 compatibility
  122. - correct off-by-one in date (month) display output
  123. - fix platform detection
  124. - improve yasm configure bits
  125. - support full URL, in X-Long-Polling header
  126. Version 0.8.1 - March 22, 2011
  127. - Make --user, --pass actually work
  128. - Add User-Agent HTTP header to requests, so that server operators may
  129. more easily identify the miner client.
  130. - Fix minor bug in example JSON config file
  131. Version 0.8 - March 21, 2011
  132. - Support long polling: http://deepbit.net/longpolling.php
  133. - Adjust max workload based on scantime (default 5 seconds,
  134. or 60 seconds for longpoll)
  135. - Standardize program output, and support syslog on Unix platforms
  136. - Suport --user/--pass options (and "user" and "pass" in config file),
  137. as an alternative to the current --userpass
  138. Version 0.7.2 - March 14, 2011
  139. - Add port of ufasoft's sse2 assembly implementation (Linux only)
  140. This is a substantial speed improvement on Intel CPUs.
  141. - Move all JSON-RPC I/O to separate thread. This reduces the
  142. number of HTTP connections from one-per-thread to one, reducing resource
  143. usage on upstream bitcoind / pool server.
  144. Version 0.7.1 - March 2, 2011
  145. - Add support for JSON-format configuration file. See example
  146. file example-cfg.json. Any long argument on the command line
  147. may be stored in the config file.
  148. - Timestamp each solution found
  149. - Improve sha256_4way performance. NOTE: This optimization makes
  150. the 'hash' debug-print output for sha256_way incorrect.
  151. - Use __builtin_expect() intrinsic as compiler micro-optimization
  152. - Build on Intel compiler
  153. - HTTP library now follows HTTP redirects
  154. Version 0.7 - February 12, 2011
  155. - Re-use CURL object, thereby reuseing DNS cache and HTTP connections
  156. - Use bswap_32, if compiler intrinsic is not available
  157. - Disable full target validation (as opposed to simply H==0) for now
  158. Version 0.6.1 - February 4, 2011
  159. - Fully validate "hash < target", rather than simply stopping our scan
  160. if the high 32 bits are 00000000.
  161. - Add --retry-pause, to set length of pause time between failure retries
  162. - Display proof-of-work hash and target, if -D (debug mode) enabled
  163. - Fix max-nonce auto-adjustment to actually work. This means if your
  164. scan takes longer than 5 seconds (--scantime), the miner will slowly
  165. reduce the number of hashes you work on, before fetching a new work unit.
  166. Version 0.6 - January 29, 2011
  167. - Fetch new work unit, if scanhash takes longer than 5 seconds (--scantime)
  168. - BeeCee1's sha256 4way optimizations
  169. - lfm's byte swap optimization (improves via, cryptopp)
  170. - Fix non-working short options -q, -r
  171. Version 0.5 - December 28, 2010
  172. - Exit program, when all threads have exited
  173. - Improve JSON-RPC failure diagnostics and resilience
  174. - Add --quiet option, to disable hashmeter output.
  175. Version 0.3.3 - December 27, 2010
  176. - Critical fix for sha256_cryptopp 'cryptopp_asm' algo
  177. Version 0.3.2 - December 23, 2010
  178. - Critical fix for sha256_via
  179. Version 0.3.1 - December 19, 2010
  180. - Critical fix for sha256_via
  181. - Retry JSON-RPC failures (see --retry, under "minerd --help" output)
  182. Version 0.3 - December 18, 2010
  183. - Add crypto++ 32bit assembly implementation
  184. - show version upon 'minerd --help'
  185. - work around gcc 4.5.x bug that killed 4way performance
  186. Version 0.2.2 - December 6, 2010
  187. - VIA padlock implementation works now
  188. - Minor build and runtime fixes
  189. Version 0.2.1 - November 29, 2010
  190. - avoid buffer overflow when submitting solutions
  191. - add Crypto++ sha256 implementation (C only, ASM elided for now)
  192. - minor internal optimizations and cleanups
  193. Version 0.2 - November 27, 2010
  194. - Add script for building a Windows installer
  195. - improve hash performance (hashmeter) statistics
  196. - add tcatm 4way sha256 implementation
  197. - Add experimental VIA Padlock sha256 implementation
  198. Version 0.1.2 - November 26, 2010
  199. - many small cleanups and micro-optimizations
  200. - build win32 exe using mingw
  201. - RPC URL, username/password become command line arguments
  202. - remove unused OpenSSL dependency
  203. Version 0.1.1 - November 24, 2010
  204. - Do not build sha256_generic module separately from cpuminer.
  205. Version 0.1 - November 24, 2010
  206. - Initial release.