Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Cuda 9 #66

Open
wants to merge 878 commits into
base: master
Choose a base branch
from
Open

Cuda 9 #66

wants to merge 878 commits into from

Conversation

kalleknast
Copy link

keccak fails. I compiled with both CUDA 8.0 and 9.1 but got the same error.


./ccminer -a keccak -d 0 -o stratum+tcp://pool-us.smartcash.cc:3333 -u xxx.w0 -p x

*** ccminer 2.2.3 for nVidia GPUs by tpruvot@github ***
    Built with the nVidia CUDA Toolkit 9.1 64-bits

  Originally based on Christian Buchner and Christian H. project
  Include some kernels from alexis78, djm34, djEzo, tsiv and krnlx.

BTC donation address: 1AJdfCpLWPNoAMDfHF1wD5y8VgKSSTHxPo (tpruvot)

[2017-12-23 08:43:17] Starting on stratum+tcp://pool-us.smartcash.cc:3333
[2017-12-23 08:43:17] NVML GPU monitoring enabled.
[2017-12-23 08:43:17] 1 miner thread started, using 'keccak' algorithm.
[2017-12-23 08:43:17] Stratum difficulty set to 1024 (8.00000)
Cuda error in func 'scanhash_keccak256' at line 81 : unknown error.

The file is keccak256.cu and the line 81 is:
CUDA_SAFE_CALL(cudaMalloc(&d_hash[thr_id], throughput * 64));
right below //really useful?

However, the KlausT ccminer works. It uses a different keccak256.cu

tpruvot and others added 30 commits February 18, 2016 11:22
This reverts commit 4e76991.

proper v1.7.3 for yiimp
the extranonce is already placed after header in job.coinbase
there was a weird job id problem not seen on yiimp stratum
+ remove hardcoded stales check on suprnova...
multiple 1.7.3 preview versions were compiled during the decred stratum
tests. Please use this newer version.
Signed-off-by: Tanguy Pruvot <[email protected]>
without param or 0 = no vote, valid choices will start to 1..

stratum + getwork
+ ifdef the 4WAY commented code...
haval256 is now 2x faster, but sha512 perf depends a lot on cuda version...
tpruvot and others added 28 commits August 13, 2017 21:04
main improvement is on echo, based on alexis sib kernel work

tested on SM 3.0 and more recent
only made a few changes to keep algos in the alpha order
+ fix the fugue leak..

Also update sib algo with this improvement
Same hash as keccak, just different pool settings (sha256d and diff factor 256)
and increase default intensity for these recent cards
Improved Lyra2Re2 Performance by 1 %
old keccak256_gpu_hash_32 kernel commented to reduce binary size

compat. not yet tested on old cards
disable the heavy algo by default for build compat

This is the only algo using thrust which is generally broken on new cuda releases
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.