Cuda 8 #59
Open
Cuda 8 #59
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
rename the var and only show warnings with --debug
5% improvement by the vshl asm swap functions, mixed shl+add inst., Add also xchg(x, y) func and XCHG(x, y) define in cuda_helper for later use... other jh changes are mainly for the beauty of the code... Signed-off-by: Tanguy Pruvot <tanguy.pruvot@gmail.com>
i had to clean it... lot of unused defines...
This allow to get 69 regs used (tested on linux) 69 or 72 make the compiler to use 64 regs which is not enough on the 750Ti for optimal performance...
ccminer.cpp begins to be too big...
shavite is faster, echo doesn't really change due to the reg. overload This changes allow custom lauchbounds without other code changes and improve the portability against different devices. also set a minimum throughput to 1024 for these algos (shared mem req. size)
made for linux and require libpci-dev (optional) if libpci is not installed, card's vendor names are not handled... Note: only a few vendor names were added, common GeForce vendors. Signed-off-by: Tanguy Pruvot <tanguy.pruvot@gmail.com>
Fedora libpci devel package is named pciutils-devel Debian/Ubuntu is libpci-dev Slackware lib requires libz... i added the dependency in configure.ac
_WIN32 is also defined on x64 builds... + fix compat/getopt dos2unix (there was mixed CRLF)
make a lot of cpu validation errors on windows, to be double checked in the next version... This reverts commit 1187a6e.
for the day nvidia will fix their nvmlDeviceGetPciInfo api..
so... pciutils branch is no more required on windows :p
fallbacks on v2 nvml exports and check api implementation on call
--pstate 0 is an alternative to clocks values to set the device in P0 mode, 1 will set appl. clocks to default P1 clocks, 2 to default P2.. --plimit 150W really works on the 9xx and allow to limit/reduce the gpu usage Note: nvml interface is only available on linux and windows x64 Signed-off-by: Tanguy Pruvot <tanguy.pruvot@gmail.com>
and show log on start if set/changed without --debug
--keep-clocks option prevent reset clocks on exit...
Supcoin seems.... dead and the algo was not supported on all devices
Used by Chaincoin and Flaxscript
seen on yiimp benchs, PNY duplicate could be for usb keys only
also query the led level and fix OV delta output note: the Gigabyte 1080 G1 seems not compatible with the current nvapi
you can now use --led=0xFF00FF to change the color/level (windows only) for non-rvb standard nvidia devices, you can use a level value (0 to 100) ps: i really need to find how to turn them off on linux, not found yet...
and rename RVB to RGB, french typo...
disabled by default, require --led=100 on nvidia compatible devices Gigabyte RGB led hack is not perfect for the moment, can fail and require a reboot.
sample usage if GPU #0 is RVB and GPU #1 a generic led: ccminer --led=0x00ff00,100,mining to put the led on while the gpus are scanning
rewrote almost properly ;)
1 MH/s reached on the 1070 ...
without that, no way to read sizes > 4GB on x86 binaries
Signed-off-by: Tanguy Pruvot <tanguy.pruvot@gmail.com>
sha512/ripemd swab this was preview 2
drop ripemd.cu and remove unused kernels
also reduce the few errors on coins using shared mem
zotac cards doesnt allow a real RGB value, so i converted manually the card seems to handle it, but seems to lock sometimes for some time... colors can be tested with "ccminer -d 2 --led=0x1f1f00 -Dn" but... unsure the card handle it fast enough to be fully usable
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
hi tpruvot
i am new to github i want to use your ccminer in linux 16.04 and gpu NVIDIA® Tesla® P100 GPUs
please give me your ccminer with my gpu
and i have the eror on my linux
make all-recursive
make[1]: Entering directory '/home/waqas/ccminer'
Making all in compat
make[2]: Entering directory '/home/waqas/ccminer/compat'
make[3]: Entering directory '/home/waqas/ccminer/compat'
make[3]: Nothing to be done for 'all-am'.
make[3]: Leaving directory '/home/waqas/ccminer/compat'
make[2]: Leaving directory '/home/waqas/ccminer/compat'
make[2]: Entering directory '/home/waqas/ccminer'
nvcc -gencode=arch=compute_50,code="sm_50,compute_50" -gencode=arch=compute_52,code="sm_52,compute_52" -I/usr/l
ocal/cuda/include -I. --ptxas-options="-v" --maxrregcount=128 -o equi/cuda_equi.o -c equi/cuda_equi.cu
/bin/bash: nvcc: command not found
Makefile:2669: recipe for target 'equi/cuda_equi.o' failed
make[2]: *** [equi/cuda_equi.o] Error 127
make[2]: Leaving directory '/home/waqas/ccminer'
Makefile:2201: recipe for target 'all-recursive' failed
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory '/home/waqas/ccminer'
Makefile:654: recipe for target 'all' failed
make: *** [all] Error 2