- 14 Oct, 2017 1 commit
-
-
fireice-uk authored
-
- 13 Oct, 2017 2 commits
-
-
psychocrypt authored
remove the gpu_thread_num safety
-
psychocrypt authored
Keep consistency between class and struct
-
- 12 Oct, 2017 1 commit
-
-
fireice-uk authored
-
- 11 Oct, 2017 6 commits
-
-
fireice-uk authored
-
fireice-uk authored
fix cuda 7.5 compile
-
fireice-uk authored
fix CMake find package output
-
psychocrypt authored
fix error: `identifier "__builtin_ia32_monitorx" is undefined`
-
fireice-uk authored
remove early returns
-
fireice-uk authored
improve AMD auto suggestion
-
- 10 Oct, 2017 7 commits
-
-
psychocrypt authored
Minor doc fixes
-
psychocrypt authored
fix #3 - remove QUIET from find package for OpenCL and CUDA - remove pascal architecture from default CUDA_ARCH for CUDA < 8.0
-
psychocrypt authored
- use always a multiple of the number of compute units - add more device information to the config file
-
psychocrypt authored
Early returns within a kernel can have unexpected behavior, this strongly depends on the opencl runtime compiler. To avoid errors all early returns are removed. There is no negative effect(e.g.performance) if the threads stay alive up to the end of the kernel. One source for the early return is: http://al-key-opencl.blogspot.de/2014/09/be-careful-not-to-mix-early-return-and.html This pull request also fix a race condition where all fill the shared memory. Also a bug introduced with #16 is fixed, because of the early return not was possible that the last block works with an wrong initilized shared memory (result should be a wrong hash if the result target size is valid).
-
fireice-uk authored
-
fireice-uk authored
avoid possible illegal memory access
-
fireice-uk authored
fix travis OSX build
-
- 08 Oct, 2017 1 commit
-
-
psychocrypt authored
add `brew update` to avoid that `brew tap homebrew/science` crash
-
- 06 Oct, 2017 5 commits
-
-
fireice-uk authored
fix wrong timer for stats
-
psychocrypt authored
fix #21, hash values randomly not available. The reason is that I changed the lables but forget to change all timer. A GPU needs sometimes ore than 2.5s to finish one hash round, therefore the results are only visible randomly.
-
psychocrypt authored
Is is possible that the number of results are greater than the result output array, in this case invalid memory can be access within the device and on the host side.
-
fireice-uk authored
fix invalid kernel call parameter
-
psychocrypt authored
Port the crash fix from xmr-stak-cpu
-
- 05 Oct, 2017 8 commits
-
-
psychocrypt authored
check that the number of global threads is a multiple of the workgroup size
-
psychocrypt authored
Fix large pages on Windows and make them more user-friendly
-
fireice-uk authored
Increase the scratchpad size to make sure we don't malform JSON
-
fireice-uk authored
-
fireice-uk authored
-
fireice-uk authored
fix amd auto suggestion
-
fireice-uk authored
fix broken nicehash support
-
psychocrypt authored
Major cleanup and speedup of soft aes
-
- 04 Oct, 2017 9 commits
-
-
fireice-uk authored
-
fireice-uk authored
-
fireice-uk authored
-
psychocrypt authored
The number of threads within a kernel must be a multiple of the worksize. If not it can crash on some systems.
-
psychocrypt authored
Platform index zero was always used to search for the device detection. Fix: use correct detected AMD platform index
-
https://github.com/fireice-uk/xmr-stakfireice-uk authored
Merge branch 'topic-large-pages-win' of https://github.com/fireice-uk/xmr-stak into topic-large-pages-win
-
fireice-uk authored
-
fireice-uk authored
Topic cleanup
-
fireice-uk authored
-