aboutsummaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAge
* Improve host-side checks, fix MSB bug, misc tidyingDavid Phillips2018-04-24
| | | | | | | | | | Host-side check now uses from-scratch libssl SHA to confirm GPU work. Bug when most significant bit of the exponent is set turns out to be because of libssl's storage of such exponents requiring 5 bytes. This cannot be dynamically adjusted without adjusting the pre-calculated value which is hardcoded into the CL kernel for performance reasons (INITIAL_DATA_LEN)
* Fix OOB accessDavid Phillips2018-04-24
| | | | Thanks to alyptik for spotting this one
* Remove unnecessary variable and commentsDavid Phillips2018-04-24
|
* Further improvementsDavid Phillips2018-04-24
|
* Misc performance improvements in CL kernelDavid Phillips2018-04-24
|
* Initial dump of SHA CL portDavid Phillips2018-04-24
|
* Add sand-leek-cl to gitignoreDavid Phillips2018-04-24
|
* Fix incorrect grouping in device-side ROLDavid Phillips2017-11-19
|
* Resize device result bufferDavid Phillips2017-10-03
|
* Dump WIP of opencl portDavid Phillips2017-10-03
|
* Add more targets to -clDavid Phillips2017-10-02
|
* Add opencl CFLAGS and targets to makefileDavid Phillips2017-10-02
|
* Remove partly broken sand-leek-cl target from allDavid Phillips2017-10-01
|
* Factor out magic sha1 chunk size constantDavid Phillips2017-10-01
|
* Add NIH SHA1 implementation mockup for portingDavid Phillips2017-10-01
| | | | | | | | | | | Mocked this up in host code to become familiar with the algorithm, and to iron out most bugs before writing/porting to OpenCL. For laughs, I switched sand-leek to use this implementation instead of OpenSSL's. Performance decrease is about 90% with no change in CFLAGS, and about 60% with -O3 instead of -O2. Of course this holds no weight, since the implementation is only meant to be ported to OpenCL and run on graphics cards where SIMD will skyrocket performance.
* Import my CL trampoline from fractal-gen projectDavid Phillips2017-09-27
|
* Tidy unnecessary magic constants awayDavid Phillips2017-09-20
|
* Fix accidental regression with never-zero loop countDavid Phillips2017-09-14
|
* Clean up test runner scriptDavid Phillips2017-09-12
|
* Free bignum_e at exit, don't leak that memoryDavid Phillips2017-09-12
| | | | Chur alyptik for reminding me
* Don't use GNU extensions on rmDavid Phillips2017-08-23
|
* Thread count can never be zero, don't worry about /0 from itDavid Phillips2017-08-09
|
* Remove division by zero if no loops completeDavid Phillips2017-08-09
|
* Change from sh to bash for pushdDavid Phillips2017-07-27
|
* Add more tests, pull simple test runner from other projectDavid Phillips2017-07-27
|
* Add more function docs to onion_base32.cDavid Phillips2017-07-03
|
* Correct equality check in test.shDavid Phillips2017-06-27
|
* Add test recipe to travis runDavid Phillips2017-06-27
|
* Add test recipe to MakefileDavid Phillips2017-06-27
|
* Expand \r to \n before parsing in test.shDavid Phillips2017-06-27
|
* Don't trample worker thread output with performance stringDavid Phillips2017-06-27
|
* Add basic sanity check scriptDavid Phillips2017-06-27
|
* Correct old outdated error messageDavid Phillips2017-06-27
|
* Add travis build label to readmeDavid Phillips2017-06-27
|
* Fix LDFLAGS orderDavid Phillips2017-06-27
|
* Fix indentation in .travis.ymlDavid Phillips2017-06-27
|
* Add .travis.yml for CIDavid Phillips2017-06-27
|
* Change char array to unsigned to avoid (pedantic) overflowDavid Phillips2017-06-26
|
* Remove outdated key sanity commentDavid Phillips2017-06-24
|
* Update benchmarkDavid Phillips2017-06-20
|
* Remove debug spin haltDavid Phillips2017-05-29
|
* Free alloced objects, move volatile to start of declDavid Phillips2017-05-29
|
* Remove unnecessary include from key_update.hDavid Phillips2017-05-29
|
* Remove bitmask and raw_len computation out of tight loopDavid Phillips2017-05-29
|
* Roll back two accidental changes in last commitDavid Phillips2017-05-28
|
* Openssl-1.0 compatDavid Phillips2017-05-28
|
* More benchmarksDavid Phillips2017-05-28
|
* Update benchmarksDavid Phillips2017-05-28
|
* Tidy the raw_len switch slightlyDavid Phillips2017-05-28
|
* Revert "Change raw_len from hard-coded to equation"David Phillips2017-05-28
| | | | | | This reverts commit ff244210acaf78beb6530c9f9d64a5f6965fdc75. This actually made the algorithm slightly slower
generated by cgit v1.1 at 2024-11-30 23:07:38 +0000