jhProtominer - CPU and collisions/min charts

[sohueasy]

thank you and I will try later yvg1900

[yvg1900]

coreavx2 IS a version for Haswell.

There are cases with 4770 CPUs when avx/avx2 show no advantage to sse4. In this case your memory subsystem/setup is a bottleneck, not the CPU.

yvg1900

[sohueasy]

@yvg1900

I got a haswell i7 4770 which suppot AVX2. And I tried jhprotominer-yvg1900-M7h-win64-coreavx2 and jhprotominer-yvg1900-M7h-win64-corei7avx with the same configurations(8T 512M) but got no difference.

Would you please tell the difference between coreavx2 and corei7avx? And is there any possible to publish a version which support haswell avx2?

thx.

[r05]

To add to the list:

- Intel Core 2 Quad Q8300 - 70c/m on average after ~24 hours of use.
- Intel Pentium G860 - 50c/m on average after ~24 hours of use
- Intel Core 2 U9400 - 25c/m on average after ~24 hours of use

[yvg1900]

e is standard deviation of CPM value. e of 1.6 is relatively high and is not yet stabilized to make decisions on performance. In your case this is because of finetuning is still in progress and performance varies alot.

AV is algorithm variation. There are 32 significantly different algorithms.

LRT is last round time measured in milliseconds. Another performance metric used to select fastest AV.

After finetuning is complete, you can specify selected AV as a value of -F to skip finetuning for next runs.

yvg1900

[fulviata]

Hi, i dont understand what mins after collision/minut :

• e= 1.6465
• AV 12
• LRT=7775

i'm a little newby :-)
this is my algo mining:


thanks!

[yvg1900]

New version available - now matching new pool server configuration after it moved.

Authentic builds are:
https://www.dropbox.com/sh/jvp4wwek8jpohj7/RlW6hzYqTz
https://mega.co.nz/#F!h0tkXSxZ!f62uoUXogkxQmP2xO8Ib-g
http://sdrv.ms/18wfjYX
http://pan.baidu.com/s/1xOOWa

yvg1900

[RustySpork]

Using  Bulldozer Win64

AMD FX-6200 @ 3.9 GHz/8Gb   Averaged ~95   6 threads -m64

AMD FX-6200 @ 3.9 GHz/8Gb   126   6 threads  -m128

AMD FX-6200 @ 3.9 GHz/8Gb   132   6 threads  -m512

AMD FX-6200 @ 3.9 GHz/8Gb   151   6 threads  -m1024

*Please note: Stock clock is 3.8 GHz but results are very similar give or take 2 - 3 CPM average

[drmyfore]

Well, I had thought the bigger memory the better performance...
So what is the max memory per thread of your edition?

[yvg1900]

"f" vs "s" described in finetuning.txt, for your mach use corei7avx version, 512M is the best I think. Use M7gf in your case.

yvg1900

[drmyfore]

@yvg1900
yeah, I think I can achieve better results, so could you give me some suggestion?
I found there are 2 edition and which edition have 2 version (gs and gf)...
Need I test 18 times to get the fastest miner?

[yvg1900]

drmyfore,

You can achieve better results with outher versns I believe.

yvg1900

[drmyfore]

In fact, I don't know what edition I used... from this link http://pan.baidu.com/s/11kiAw
Xeon E5-2680v2 @ 3.1GHz     525     40 threads -m1024

[ptsomsk]

M7gs-win64-bulldozerv1
FX-8350 @ 5.2Ghz/16Gb   240   7 threads –m512
M7gs-win64-corei7avx
i-7 3770k  @ 4.5Ghz/16Gb   300   8 threads –m512

[lordfirefox]

Someone with a AMD FX-9590 needs to post their scores. I'm curious what the CPM would be.