miner vs pool hashrate

[dustofdeath]

So i noticed that pools seem to report  half of what im runnign at - for example if im at 340kh/s they show that im at 130-160kh/s

So i run at low settings - getting 130kh/s... and it  reports that im running at 100- 130kh/s.

What could be breaking that?

[1713519447]

1713519447

[1713519447]

1713519447

[1713519447]

1713519447

[1713519447]

1713519447

[1713519447]

1713519447

[1713519447]

1713519447

[pontiacg5]

It really depends, some pools won't reliably show hashrates that low.

What are you using to mine with? Are you showing a lot of rejected shares or hardware errors?

[dustofdeath]

noticed it with diff pools and diff  coins iv tried not a specific one.

Dont see any HW errors in cgminer nether... ut perhaps it shwos wrong hashrate?
Since for low load its quite close to real value.

HD7850 (vtx3d hd7850 1gb gddr5 xtreme edition  1ghz gpu /1225 mem 256 bit 1024 shaders)  - close to standard 7950 gpu-s, running with:
cgminer --scrypt -o stratum+tcp://pool.d2.cc:3333 -u x.1 -p x -w 256 --thread-concurrency 8192 -I 17 -g 1  --shaders 1024

[dustofdeath]

yeah more or less... and thats weird.

Also.. anyoen else had problems with GPU_MAX_ALLOC_PERCENT=100 oo it simply doesnt do anything.. i cant go beyond the 8k.

Even tho the card has 1gb of ram the cgminer cries that the value is above ~512mb range. And i did check environmental variables and it is set there. Oo Any switch to use for cgminer or such?

[pontiacg5]

Few things...

First, what are you showing for a WU/m in CGminer? It should be really close to your hashrate. As long as that is true your pool should be paying based on what you are seeing it hash at.



Second, my 7870 really likes to run a high thread concurrency when running with a high intensity. You may get a better hashrate with these settings.

Low TC and intensity, but dual threads.

--shaders 1024 --thread-concurrency 8192 --gpu-threads 2 --work-size 256 --lookup-gap 2 -g 2 -i 13

Or high TC and intensity and single thread...

--shaders 1024 --thread-concurrency 12288 --gpu-threads 1 --work-size 256 --lookup-gap 2 -g 1 -i 19

If you are running windows launch from a bat file, like this...

""-setx GPU_USE_SYNC_OBJECTS 1
-setx GPU_MAX_ALLOC_PERCENT 100
cgminer  --scrypt -o stratum -u red0 -p 123 --shaders 1280 --intensity 18 -g 1 --worksize 256 --lookup-gap 2 --thread-concurrency 15232 --gpu-threads 1 --gpu-engine 1000 --gpu-memclock 1500 --verbose""

[iCEBREAKER]

Try setting your share difficulty to one for troubleshooting/benchmarking purposes.

The higher the difficulty, the larger the actual vs estimated hashrate discrepancy variance may be.

[dustofdeath]

cant get it higher then 8k for tc.
The mac alloc is done.

It seems like its the driver.  Could be that cgminer doesnt work well with the latest 13.9 :/

dual thread approach will net 40kh/s + lag vs 340kh/s with -g 1

[pontiacg5]

That makes sense, I only get ~420kh/s from the 7870 and it certainly likes single thread/high intensities.

I have some r9 280x cards with the newest beta driver working just fine, though these seem to run better dual thread/low TC and intensity.

My 7870 machine is still on 13.4 because I'm lazy, though. Might be worth a try   Seems like you should be able to get at least 350-375 from that card...

You used the setx command right? I think that will severely limit the max TC, if you forget it or don't have it set right.

[dustofdeath]

ah it wanted a reboot for the  setx command to actually apply.

 --thread-concurrency 10240 -I 16 -g 1 --lookup-gap 2

seems to be the most optimal - around 340 WU.

Higher intensity drops the wu .. even tho there is no HW or anything.