Quote from: dave_bbp on April 13, 2020, 05:09:40 PM

Quote from: afeno on April 13, 2020, 01:55:05 PM

I have two servers exactly identical. Same CPU, same ram, same HDD, same ISP.
Once is giving me 1300 H/s and the other one is 4300 H/s

Server1
[2020-04-13 15:28:33.072] speed 10s/60s/15m 1287.2 1370.7 1374.7 H/s max 1440.7 H/s

Server2
[2020-04-13 15:29:02.897] speed 10s/60s/15m 4397.6 4414.0 4384.7 H/s max 4607.8 H/s

Both servers are showing this when startig :

Code:

* ABOUT bbprig/5.10.0 gcc/7.5.0
* BBP + XMR - Welcome to the future of orphan charity
* To donate to the XMRig devs, send XMR here 48edfHu7V9Z84YzzMa6fUueoELZ9ZRXq9VetWzYGzKt52XU5xvqgzYnDK9URnRoJMk 1j8nLwEVsaSWJ4fhdUyZijBGUicoD (or BTC here) 1P7ujsXeX7GxQwHNnJsRMgAdNkFZmNVqJT
* LIBS libuv/1.18.0 OpenSSL/1.1.0g hwloc/1.11.2
* HUGE PAGES supported
* 1GB PAGES disabled
* CPU Intel(R) Xeon(R) CPU X5660 @ 2.80GHz (2) x64 AES
L2:3.0 MB L3:24.0 MB 12C/24T NUMA:2
* MEMORY 1.5/15.7 GB (9%)
* DONATE 10%
* ASSEMBLY auto:intel
* POOL #1 foundation.biblepay.org:3001 algo auto
* COMMANDS hashrate, pause, resume
* OPENCL disabled
* CUDA disabled
[2020-04-13 14:41:56.278] net use pool foundation.biblepay.org Orphan Charity
[2020-04-13 14:41:56.278] net new job from foundation.biblepay.org diff 75000 algo rx/0 height 2075831
[2020-04-13 14:41:56.280] msr msr kernel module is not available
[2020-04-13 14:41:56.280] rx init datasets algo rx/0 (24 threads) seed b2e5f3d303dff783...
[2020-04-13 14:41:56.282] rx #0 allocated 2080 MB huge pages 0% (2 ms)
[2020-04-13 14:41:56.282] rx #1 allocated 2080 MB huge pages 0% (2 ms)
[2020-04-13 14:41:56.283] rx #0 allocated 256 MB huge pages 0% +JIT (1 ms)
[2020-04-13 14:41:56.283] rx -- allocated 4416 MB huge pages 0% 0/2208 (3 ms)
[2020-04-13 14:41:59.565] rx #0 dataset ready (3282 ms)
[2020-04-13 14:42:00.307] rx #1 dataset ready (742 ms)
[2020-04-13 14:42:00.307] cpu use profile * (24 threads) scratchpad 2048 KB
[2020-04-13 14:42:00.781] cpu READY threads 24/24 (24) huge pages 0% 0/24 memory 49152 KB (474 ms)

Any idea why such H/s difference in the two servers?

That doesn't seem right. Are you sure you are starting them with the same command/script from the same operating system? I guess you have ubuntu, plz check with "htop" for the CPU load.

btw: some tipps for xmrig:

use the number of physical threads your System has, which is 12 (2x6), not 24
for slightly better performance enable "hugepages" in your system or start the miner as root (xmrig can configure RAM itself)
for even better performance you could enable "1GB-hugepages" (just google xmrig 1gb-hugepages ubuntu), but this would require a lot more RAM than you currently have (at least 40 GB)

Thanks for the Tips. With hugepages enabled and 12 threads Im getting some more H/s. But I still have a huge difference between the two identical servers. Both running the same command. And htop shows 99-100% CPU utilization in both servers.... Very strange....

EDIT:
With 12 cores and hugepages enabled, the difference is even bigger:

Server1:
[2020-04-14 11:07:12.505] speed 10s/60s/15m 1655.8 1657.6 1654.0 H/s max 2204.9 H/s

Server2:
[2020-04-14 11:05:46.498] speed 10s/60s/15m 6188.6 6223.6 6235.3 H/s max 6273.5 H/s

Any idea what I can check? As said, both servers are showing 12 cores at 100% cpu utilization. (htop)

afeno

Posts

Merits