How to optimise hash rate

[mgillr]

Hi, I have the following cards, no compile errors all running OK. However I can only achieve a hash rate of circa 6000h/s, seems low to me for these cards. Am i doing anything wrong here, is there a way to optimise the cards or have I failed to set up the config correctly?

• VERSIONS: XMRig/2.4.5 libuv/1.8.0 CUDA/7.50 gcc/5.4.0
• CPU: Intel(R) Xeon(R) CPU @ 2.20GHz x64 AES-NI
• GPU #0: Tesla P100-PCIE-16GB @ 1328/715 MHz 46x168 0x0 arch:60 SMX:56
• GPU please add the amd version #1: Tesla P100-PCIE-16GB @ 1328/715 MHz 46x168 0x0 arch:60 SMX:56
• GPU error with cuda_extra.cu #2: Tesla P100-PCIE-16GB @ 1328/715 MHz 46x168 0x0 arch:60 SMX:56
• GPU error when trying to mine in the background #3: Tesla P100-PCIE-16GB @ 1328/715 MHz 46x168 0x0 arch:60 SMX:56
• ALGO: cryptonight

[ghost]

Have you tried experimenting with different values e.g. 112x56 or 128x56?

[kimats]

try cuda 9.1?

[gordan-bobic]

P100 is an uncut version of GP102 (1080Ti, which is only slightly cut down). If you are getting 1500H/s out of a single card, that is almost suspiciously too good. I would expect nearer 1000 H/s out of one of those.