XMR-STAK-CPU-DEV .......this is odd... long but interesting post.

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
I decided to play around with xmr-stak-cpu-dev, and something very odd is going on.

Output of the older version I normally run just to show you what the normal performance is.

@LinuxBeast:~/xmr-stak-cpu-master$ bin/xmr-stak-cpu
-------------------------------------------------------------------
XMR-Stak-CPU mining software, CPU Version.
Based on CPU mining code by wolf9466 (heavily optimized by myself).
Brought to you by fireice_uk under GPLv3.

Configurable dev donation level is set to 0.0 %

You can use following keys to display reports:
'h' - hashrate
'r' - results
'c' - connection
-------------------------------------------------------------------
[2017-06-12 22:25:02] : Starting single thread, affinity: 2.
[2017-06-12 22:25:02] : Starting single thread, affinity: 3.
[2017-06-12 22:25:02] : Starting single thread, affinity: 4.
[2017-06-12 22:25:02] : Starting single thread, affinity: 5.
[2017-06-12 22:25:02] : Starting single thread, affinity: 6.
[2017-06-12 22:25:02] : Starting single thread, affinity: 7.
[2017-06-12 22:25:02] : Starting single thread, affinity: 8.
[2017-06-12 22:25:02] : Starting single thread, affinity: 9.
[2017-06-12 22:25:02] : Starting single thread, affinity: 10.
[2017-06-12 22:25:02] : Starting single thread, affinity: 11.
[2017-06-12 22:25:02] : Starting single thread, affinity: 12.
[2017-06-12 22:25:02] : Starting single thread, affinity: 13.
[2017-06-12 22:25:02] : Starting single thread, affinity: 14.
[2017-06-12 22:25:02] : Starting single thread, affinity: 15.
[2017-06-12 22:25:02] : Starting single thread, affinity: 16.
[2017-06-12 22:25:02] : Starting single thread, affinity: 17.
[2017-06-12 22:25:02] : Starting single thread, affinity: 18.
[2017-06-12 22:25:02] : Starting single thread, affinity: 19.
[2017-06-12 22:25:02] : Starting single thread, affinity: 20.
[2017-06-12 22:25:02] : Starting single thread, affinity: 21.
[2017-06-12 22:25:02] : Starting single thread, affinity: 22.
[2017-06-12 22:25:02] : Starting single thread, affinity: 23.
[2017-06-12 22:25:02] : Starting single thread, affinity: 24.
[2017-06-12 22:25:02] : Starting single thread, affinity: 25.
[2017-06-12 22:25:02] : Starting single thread, affinity: 26.
[2017-06-12 22:25:02] : Starting single thread, affinity: 27.
[2017-06-12 22:25:02] : Starting single thread, affinity: 28.
[2017-06-12 22:25:02] : Starting single thread, affinity: 29.
[2017-06-12 22:25:02] : Starting single thread, affinity: 30.
[2017-06-12 22:25:02] : Starting single thread, affinity: 31.
[2017-06-12 22:25:02] : Connecting to pool pool.minexmr.com:3333 ...
[2017-06-12 22:25:02] : Connected. Logging in...
[2017-06-12 22:25:02] : Difficulty changed. Now: 200007.
[2017-06-12 22:25:02] : New block detected.
HASHRATE REPORT
| ID | 2.5s | 60s | 15m | ID | 2.5s | 60s | 15m |
| 0 | 40.2 | (na) | (na) | 1 | 39.0 | (na) | (na) |
| 2 | 39.4 | (na) | (na) | 3 | 40.0 | (na) | (na) |
| 4 | 35.9 | (na) | (na) | 5 | 35.7 | (na) | (na) |
| 6 | 36.0 | (na) | (na) | 7 | 35.6 | (na) | (na) |
| 8 | 30.3 | (na) | (na) | 9 | 37.9 | (na) | (na) |
| 10 | 38.3 | (na) | (na) | 11 | 23.4 | (na) | (na) |
| 12 | 34.3 | (na) | (na) | 13 | 34.7 | (na) | (na) |
| 14 | 39.2 | (na) | (na) | 15 | 38.3 | (na) | (na) |
| 16 | 40.2 | (na) | (na) | 17 | 39.3 | (na) | (na) |
| 18 | 39.3 | (na) | (na) | 19 | 40.0 | (na) | (na) |
| 20 | 34.9 | (na) | (na) | 21 | 35.7 | (na) | (na) |
| 22 | 36.5 | (na) | (na) | 23 | 35.6 | (na) | (na) |
| 24 | 36.6 | (na) | (na) | 25 | 37.8 | (na) | (na) |
| 26 | 38.7 | (na) | (na) | 27 | 25.1 | (na) | (na) |
| 28 | 34.7 | (na) | (na) | 29 | 34.1 | (na) | (na) |
-----------------------------------------------------
Totals: 1086.8 (na) (na) H/s
Highest: 1086.2 H/s
HASHRATE REPORT
| ID | 2.5s | 60s | 15m | ID | 2.5s | 60s | 15m |
| 0 | 40.2 | (na) | (na) | 1 | 39.3 | (na) | (na) |
| 2 | 39.6 | (na) | (na) | 3 | 40.0 | (na) | (na) |
| 4 | 35.7 | (na) | (na) | 5 | 35.7 | (na) | (na) |
| 6 | 35.7 | (na) | (na) | 7 | 35.5 | (na) | (na) |
| 8 | 37.8 | (na) | (na) | 9 | 37.8 | (na) | (na) |
| 10 | 38.2 | (na) | (na) | 11 | 25.1 | (na) | (na) |
| 12 | 34.2 | (na) | (na) | 13 | 34.7 | (na) | (na) |
| 14 | 36.8 | (na) | (na) | 15 | 39.7 | (na) | (na) |
| 16 | 40.2 | (na) | (na) | 17 | 39.2 | (na) | (na) |
| 18 | 39.6 | (na) | (na) | 19 | 40.0 | (na) | (na) |
| 20 | 33.9 | (na) | (na) | 21 | 35.7 | (na) | (na) |
| 22 | 36.5 | (na) | (na) | 23 | 35.6 | (na) | (na) |
| 24 | 30.0 | (na) | (na) | 25 | 37.8 | (na) | (na) |
| 26 | 38.7 | (na) | (na) | 27 | 36.1 | (na) | (na) |
| 28 | 34.7 | (na) | (na) | 29 | 34.0 | (na) | (na) |
-----------------------------------------------------
Totals: 1097.8 (na) (na) H/s
Highest: 1092.6 H/s


This is with my dual E5-2667 V3 ES pc that has 35MB L3 cache per cpu so I can run all 32 threads mining but I usually only run 30 to keep my desktop responsive.

The important part of the config.txt from the older xmr-stak-cpu:
"cpu_threads_conf" : [

{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 2 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 3 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 4 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 5 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 6 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 7 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 8 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 9 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 10 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 11 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 12 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 13 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 14 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 15 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 16 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 17 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 18 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 19 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 20 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 21 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 22 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 23 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 24 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 25 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 26 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 27 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 28 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 29 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 30 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 31 },
],



Now I installed the latest developer version of xmr-stak-cpu.

It installed fine and when I ran it for the first time it did the auto config deal where it decides what the config.txt settings for the cpu_thread_config.

It recommended this:
"cpu_threads_conf" :
[
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 0 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 1 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 2 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 3 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 4 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 5 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 6 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 7 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 8 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 9 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 10 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 11 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 12 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 13 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 14 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 15 },
],

I have a dual Xeon motherboard so I assumed it choked on that so I changed it and added the usual 30 threads.

Output of xmr-stak-cpu-dev:
You can use following keys to display reports:
'h' - hashrate
'r' - results
'c' - connection
-------------------------------------------------------------------
[2017-06-12 22:16:11] : Starting double thread, affinity: 2.
[2017-06-12 22:16:11] : Starting double thread, affinity: 3.
[2017-06-12 22:16:11] : Starting double thread, affinity: 4.
[2017-06-12 22:16:11] : Starting double thread, affinity: 5.
[2017-06-12 22:16:11] : Starting double thread, affinity: 6.
[2017-06-12 22:16:11] : Starting double thread, affinity: 7.
[2017-06-12 22:16:11] : Starting double thread, affinity: 8.
[2017-06-12 22:16:11] : Starting double thread, affinity: 9.
[2017-06-12 22:16:11] : Starting double thread, affinity: 10.
[2017-06-12 22:16:11] : Starting double thread, affinity: 11.
[2017-06-12 22:16:11] : Starting double thread, affinity: 12.
[2017-06-12 22:16:11] : Starting double thread, affinity: 13.
[2017-06-12 22:16:11] : Starting double thread, affinity: 14.
[2017-06-12 22:16:11] : Starting double thread, affinity: 15.
[2017-06-12 22:16:11] : Starting double thread, affinity: 16.
[2017-06-12 22:16:11] : Starting double thread, affinity: 17.
[2017-06-12 22:16:11] : Starting double thread, affinity: 18.
[2017-06-12 22:16:11] : Starting double thread, affinity: 19.
[2017-06-12 22:16:11] : Starting double thread, affinity: 20.
[2017-06-12 22:16:11] : Starting double thread, affinity: 21.
[2017-06-12 22:16:11] : Starting double thread, affinity: 22.
[2017-06-12 22:16:11] : Starting double thread, affinity: 23.
[2017-06-12 22:16:11] : Starting double thread, affinity: 24.
[2017-06-12 22:16:11] : Starting double thread, affinity: 25.
[2017-06-12 22:16:11] : Starting double thread, affinity: 26.
[2017-06-12 22:16:11] : Starting double thread, affinity: 27.
[2017-06-12 22:16:11] : Starting double thread, affinity: 28.
[2017-06-12 22:16:11] : Starting double thread, affinity: 29.
[2017-06-12 22:16:11] : Starting double thread, affinity: 30.
[2017-06-12 22:16:11] : Starting double thread, affinity: 31.
[2017-06-12 22:16:11] : Connecting to pool pool.minexmr.com:3333 ...
[2017-06-12 22:16:11] : Connected. Logging in...
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : hwloc: memory pinned
[2017-06-12 22:16:12] : Difficulty changed. Now: 200007.
[2017-06-12 22:16:12] : New block detected.
RESULT REPORT
You haven't found any results yet.
HASHRATE REPORT
| ID | 2.5s | 60s | 15m | ID | 2.5s | 60s | 15m |
| 0 | 26.7 | (na) | (na) | 1 | 26.7 | (na) | (na) |
| 2 | 27.0 | (na) | (na) | 3 | 26.8 | (na) | (na) |
| 4 | 24.7 | (na) | (na) | 5 | 25.2 | (na) | (na) |
| 6 | 23.0 | (na) | (na) | 7 | 22.6 | (na) | (na) |
| 8 | 23.4 | (na) | (na) | 9 | 23.5 | (na) | (na) |
| 10 | 23.8 | (na) | (na) | 11 | 23.6 | (na) | (na) |
| 12 | 22.1 | (na) | (na) | 13 | 22.4 | (na) | (na) |
| 14 | 33.2 | (na) | (na) | 15 | 32.8 | (na) | (na) |
| 16 | 26.4 | (na) | (na) | 17 | 25.8 | (na) | (na) |
| 18 | 26.6 | (na) | (na) | 19 | 26.6 | (na) | (na) |
| 20 | 23.8 | (na) | (na) | 21 | 25.1 | (na) | (na) |
| 22 | 22.9 | (na) | (na) | 23 | 22.3 | (na) | (na) |
| 24 | 23.4 | (na) | (na) | 25 | 23.1 | (na) | (na) |
| 26 | 23.1 | (na) | (na) | 27 | 23.0 | (na) | (na) |
| 28 | 21.9 | (na) | (na) | 29 | 22.1 | (na) | (na) |
-----------------------------------------------------
Totals: 743.6 (na) (na) H/s
Highest: 743.9 H/s
[2017-06-12 22:16:37] : Result accepted by the pool.

743.9 H/s WTH???



So I changed it to what the autoconfig told me to copy and paste to config.txt:
"cpu_threads_conf" :
[
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 0 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 1 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 2 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 3 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 4 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 5 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 6 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 7 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 8 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 9 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 10 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 11 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 12 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 13 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 14 },
{ "low_power_mode" : true, "no_prefetch" : true, "affine_to_cpu" : 15 },
],

Then ran it:
-------------------------------------------------------------------
[2017-06-12 22:49:48] : Starting double thread, affinity: 0.
[2017-06-12 22:49:48] : Starting double thread, affinity: 1.
[2017-06-12 22:49:48] : Starting double thread, affinity: 2.
[2017-06-12 22:49:48] : Starting double thread, affinity: 3.
[2017-06-12 22:49:48] : Starting double thread, affinity: 4.
[2017-06-12 22:49:48] : Starting double thread, affinity: 5.
[2017-06-12 22:49:48] : Starting double thread, affinity: 6.
[2017-06-12 22:49:48] : Starting double thread, affinity: 7.
[2017-06-12 22:49:48] : Starting double thread, affinity: 8.
[2017-06-12 22:49:48] : Starting double thread, affinity: 9.
[2017-06-12 22:49:48] : Starting double thread, affinity: 10.
[2017-06-12 22:49:48] : Starting double thread, affinity: 11.
[2017-06-12 22:49:48] : Starting double thread, affinity: 12.
[2017-06-12 22:49:48] : Starting double thread, affinity: 13.
[2017-06-12 22:49:48] : Starting double thread, affinity: 14.
[2017-06-12 22:49:48] : Starting double thread, affinity: 15.
[2017-06-12 22:49:48] : Connecting to pool pool.minexmr.com:3333 ...
[2017-06-12 22:49:48] : Connected. Logging in...
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : hwloc: memory pinned
[2017-06-12 22:49:48] : Difficulty changed. Now: 200007.
[2017-06-12 22:49:48] : New block detected.
HASHRATE REPORT
| ID | 2.5s | 60s | 15m | ID | 2.5s | 60s | 15m |
| 0 | 71.2 | (na) | (na) | 1 | 71.8 | (na) | (na) |
| 2 | 77.7 | (na) | (na) | 3 | 77.4 | (na) | (na) |
| 4 | 78.2 | (na) | (na) | 5 | 79.3 | (na) | (na) |
| 6 | 69.0 | (na) | (na) | 7 | 69.7 | (na) | (na) |
| 8 | 71.0 | (na) | (na) | 9 | 62.6 | (na) | (na) |
| 10 | 75.8 | (na) | (na) | 11 | 74.4 | (na) | (na) |
| 12 | 74.8 | (na) | (na) | 13 | 75.8 | (na) | (na) |
| 14 | 67.1 | (na) | (na) | 15 | 66.4 | (na) | (na) |
-----------------------------------------------------
Totals: 1162.3 (na) (na) H/s
Highest: 1162.6 H/s

1162.6 H/s running 16 threads!!!
Look at the per thread increase!!
From mid 30's H/s per core to almost double that?

WTH!!???






I do not understand this.......

I'm thinking of trying to run two instances of xmr-stak-cpu-dev one on cores 0-15 on one and 16-32 on another. LOL
 
Last edited:

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Well I tried running two instances.

And the performance sucked, the performance is great if I run only one at a time about 1200 H/s but it takes a nosedive with two.

I'm not surprised but I still do not understand whats going on.
 
Last edited:

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Light bulb goes off, well at least its flickering. ;)

I'm starting to get a vague, very vague, idea whats going on.

I'm a reluctant programmer, only dive into stuff that interest me, but it looks like the cache memory is pinned to the numa nodes.

And I think each of the cores are running two threads instead of one.

hwlocMemory.hpp file

#pragma once

#include "console.h"

#ifndef CONF_NO_HWLOC

#include <hwloc.h>

/** pin memory to NUMA node
*
* Set the default memory policy for the current thread to bind memory to the
* NUMA node.
*
* @param puId core id
*/
void bindMemoryToNUMANode( int puId )
{
int depth;
hwloc_topology_t topology;
hwloc_obj_t obj;

hwloc_topology_init(&topology);
hwloc_topology_load(topology);

depth = hwloc_get_type_depth(topology, HWLOC_OBJ_PU);

for( int i = 0;
i < hwloc_get_nbobjs_by_depth(topology, depth);
i++ )
{
hwloc_obj_t pu = hwloc_get_obj_by_depth(topology, depth, i);
if( pu->os_index == puId )
if( 0 > hwloc_set_membind_nodeset(
topology,
pu->nodeset,
HWLOC_MEMBIND_BIND,
HWLOC_MEMBIND_THREAD)
)
printer::inst()->print_msg(L0, "hwloc: can't bind memory");
else
{
printer::inst()->print_msg(L0, "hwloc: memory pinned");
break;
}
}
}
#else

void bindMemoryToNUMANode( int )
{
}

#endif
 
Last edited:

Jetlag

New Member
Jun 7, 2016
17
4
3
43
In your new configuration "low_power_mode" is set to true, which is a different algorithm that uses twice the cache. You might try 17 or 18 threads and see what happens. Then you could set "low_power_mode" to false and try 30-32 threads and see that gets you. "no_prefetch" was also changed from false to true. So you could change that and see what numbers you get.
 

MiniKnight

Well-Known Member
Mar 30, 2012
3,072
973
113
NYC
That's why I let others be pioneers on this stuff.

You're also using non-standard chips. Maybe that's a problem?
 

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
In your new configuration "low_power_mode" is set to true, which is a different algorithm that uses twice the cache. You might try 17 or 18 threads and see what happens. Then you could set "low_power_mode" to false and try 30-32 threads and see that gets you. "no_prefetch" was also changed from false to true. So you could change that and see what numbers you get.

I did not even catch that, the low power mode or the no prefetch running the recommended 16 threads that is recommended by the first run of the program, I have already changed changed it use 30 threads I just copied and pasted over from my older config.tx which is set to "low_power mode" is false and "no_prefetch" is also false but had far worse performance at ~750 H/s.

From my first post:
The important part of the config.txt from the older xmr-stak-cpu:
"cpu_threads_conf" : [

{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 2 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 3 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 4 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 5 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 6 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 7 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 8 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 9 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 10 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 11 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 12 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 13 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 14 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 15 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 16 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 17 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 18 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 19 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 20 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 21 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 22 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 23 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 24 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 25 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 26 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 27 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 28 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 29 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 30 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 31 },
],

And that got me less than 750 H/s.

So off to my config.txt.
 
Last edited:

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Low_power mode: true and no_prefetch: true with 16 threads:
HASHRATE REPORT
| ID | 2.5s | 60s | 15m | ID | 2.5s | 60s | 15m |
| 0 | 61.4 | 61.8 | 60.2 | 1 | 44.9 | 61.4 | 61.9 |
| 2 | 61.2 | 66.5 | 67.7 | 3 | 63.6 | 64.6 | 66.9 |
| 4 | 61.0 | 65.2 | 66.2 | 5 | 56.1 | 65.3 | 68.2 |
| 6 | 53.5 | 60.6 | 60.4 | 7 | 60.2 | 60.5 | 60.6 |
| 8 | 62.6 | 69.1 | 68.1 | 9 | 50.7 | 67.3 | 66.8 |
| 10 | 76.2 | 75.0 | 73.5 | 11 | 70.3 | 72.4 | 71.6 |
| 12 | 73.0 | 74.1 | 73.1 | 13 | 75.1 | 74.6 | 73.2 |
| 14 | 67.3 | 67.1 | 65.6 | 15 | 52.0 | 65.9 | 64.9 |
-----------------------------------------------------
Totals: 989.0 1071.4 1068.7 H/s
Highest: 1206.0 H/s
Note that its not as consistent as the older xmr-stack-cpu version.

Now adding more threads, 17:

HASHRATE REPORT
| ID | 2.5s | 60s | 15m | ID | 2.5s | 60s | 15m |
| 0 | 46.4 | 45.4 | (na) | 1 | 51.8 | 52.1 | (na) |
| 2 | 57.1 | 58.9 | (na) | 3 | 60.0 | 60.3 | (na) |
| 4 | 62.3 | 61.1 | (na) | 5 | 61.3 | 61.7 | (na) |
| 6 | 55.3 | 52.0 | (na) | 7 | 55.8 | 51.8 | (na) |
| 8 | 54.9 | 66.6 | (na) | 9 | 56.9 | 66.5 | (na) |
| 10 | 71.4 | 73.5 | (na) | 11 | 72.2 | 72.1 | (na) |
| 12 | 71.7 | 72.0 | (na) | 13 | 70.7 | 72.6 | (na) |
| 14 | 57.8 | 64.3 | (na) | 15 | 49.8 | 64.8 | (na) |
| 16 | 46.2 | 45.4 | (na) |
-----------------------------------------------------
Totals: 1001.7 1041.1 (na) H/s
Highest: 1106.4 H/s

A little less.

Now 18 threads:
HASHRATE REPORT
| ID | 2.5s | 60s | 15m | ID | 2.5s | 60s | 15m |
| 0 | 36.8 | 36.8 | (na) | 1 | 36.4 | 36.4 | (na) |
| 2 | 55.7 | 52.0 | (na) | 3 | 49.1 | 50.2 | (na) |
| 4 | 47.5 | 49.1 | (na) | 5 | 49.6 | 52.7 | (na) |
| 6 | 42.5 | 40.1 | (na) | 7 | 45.9 | 42.4 | (na) |
| 8 | 53.6 | 59.8 | (na) | 9 | 53.8 | 59.5 | (na) |
| 10 | 58.0 | 64.5 | (na) | 11 | 67.2 | 63.8 | (na) |
| 12 | 69.6 | 65.1 | (na) | 13 | 70.0 | 64.9 | (na) |
| 14 | 64.5 | 59.4 | (na) | 15 | 51.5 | 57.9 | (na) |
| 16 | 36.7 | 36.9 | (na) | 17 | 36.3 | 36.2 | (na) |
-----------------------------------------------------
Totals: 924.8 927.8 (na) H/s
Highest: 1032.1 H/s

Even a little less.

Defiantly need to play around with it more.
 

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Well I re-ran it with the old config copied over and it's better this time:
"cpu_threads_conf" :
[
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 2 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 3 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 4 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 5 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 6 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 7 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 8 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 9 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 10 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 11 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 12 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 13 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 14 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 15 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 16 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 17 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 18 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 19 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 20 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 21 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 22 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 23 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 24 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 25 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 26 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 27 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 28 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 29 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 30 },
{ "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 31 },
],

HASHRATE REPORT
| ID | 2.5s | 60s | 15m | ID | 2.5s | 60s | 15m |
| 0 | 38.3 | (na) | (na) | 1 | 38.6 | (na) | (na) |
| 2 | 39.0 | (na) | (na) | 3 | 39.5 | (na) | (na) |
| 4 | 35.4 | (na) | (na) | 5 | 35.3 | (na) | (na) |
| 6 | 36.7 | (na) | (na) | 7 | 36.5 | (na) | (na) |
| 8 | 39.5 | (na) | (na) | 9 | 38.4 | (na) | (na) |
| 10 | 38.6 | (na) | (na) | 11 | 39.3 | (na) | (na) |
| 12 | 35.2 | (na) | (na) | 13 | 35.2 | (na) | (na) |
| 14 | 39.5 | (na) | (na) | 15 | 36.5 | (na) | (na) |
| 16 | 39.5 | (na) | (na) | 17 | 38.6 | (na) | (na) |
| 18 | 39.0 | (na) | (na) | 19 | 39.4 | (na) | (na) |
| 20 | 35.0 | (na) | (na) | 21 | 35.2 | (na) | (na) |
| 22 | 36.8 | (na) | (na) | 23 | 36.4 | (na) | (na) |
| 24 | 39.4 | (na) | (na) | 25 | 38.5 | (na) | (na) |
| 26 | 38.1 | (na) | (na) | 27 | 39.2 | (na) | (na) |
| 28 | 35.2 | (na) | (na) | 29 | 35.1 | (na) | (na) |
-----------------------------------------------------
Totals: 1127.0 (na) (na) H/s
Highest: 1122.8 H/s


I guess the crappy performance the first time was an anomaly and I now understand whats going on.
 
Last edited:

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Lesson learned, don't mess with something when you should have went to bed after having only a couple of hours of sleep the night before.

You just might waste time and miss the obvious.
 

spfoo

Member
May 23, 2017
102
16
18
I noticed that in there was "Starting DOUBLE thread" in the dev version versus "Starting SINGLE thread" in the other in your outputs. Maybe the dev version doubled the thread count?