python OpenCL bitcoin miner

[OVerLoRDI]

Does anyone have an estimate about how much bandwidth poclbm uses for mining?  I'm worried that a x1 slot might not provide enough bandwidth for mining.

[Garrett Burgwardt]

I remember reading somewhere that it doesn't affect it at all. But I can't find where it was so take that with a grain of salt.

[aistto]

aistto, see http://blog.zorinaq.com/?e=11

Initially, you can start the miner on first device, then connect your monitor to second device to make it available. If this seems too difficult, see above link for a way to keep second device always active. This is a windows only problem.

I switched on crossfire (otherwise i didn't see second gpu)

on 1-st GPU I ran last version  with 100-250 kh/s
on 2-nd GPU I ran 20101126           50-150 kh/s
iI have problem with connection as well.

winXP, Catalyst 11.2, ATI Stream 2.2,  with crossfire bridge.

[qet]

Hi, I'm getting the following error in windows. (XP sp3, fresh install today. )

http://sc3.cc/pics/bitcoinerror.JPG

Any ideas?
I have tried installing .net and C++ redistributable.

Much appreciated, thanks.



edit:   after randomly installing various DirectX, C++ and .Net 2, 3, 4 versions, its now working.
Who knows which one you actually need??

[qet]

sorry to post again, but now i'm encountering another problem.
i've got a 5970 card.
I run poclbm on GPU1 and i get no trouble.

but on GPU2 it crashes after a few seconds with "verification failed. check hardware"

so I tried underclocking the gpu to as low as it would go. the hash rate was down accordingly when i re-ran poclbm, but it still errored within 10 seconds.

Any ideas?

Thanks again

Edit:
i tried adjusting the worksize to a low number:  "-w 8" . this reduced the MH/s to 38 MH/s   (normal is 260) 
however the program still exited with "verification failed" after a minute or two.

don't tell me this card is stuffed?

[qed]

Is poclbm periodically checking for the correct generation of ashes or do i have to make a stability test after changing voltage/frequency?

[JWU42]

Are you running the card at stock for now while debugging?

[qed]

Are you running the card at stock for now while debugging?

Me? My cards are actually on underclock and undervoltage. With the standard GPu at 1.1V (stock voltage) my 2xHD6950 are stable at 920 MHz (800MHz is the stock clock).
The power disippated is heavily related to the voltage and way less on the gpu clock, so there is no point for me to keep the 2 cards at 1.1V. The cards are way quieter.

[qet]

Is poclbm periodically checking for the correct generation of ashes or do i have to make a stability test after changing voltage/frequency?

qed, if you read my post then you'd know that it does check.
if your card isn't stable then it will crash with "Verification failed. check hardware"

[qet]

And just to add another issue:
i ran poclbm overnight and in the morning it had stopped working with an error something like "failed to communicate with RPC server" ...
does this mean the bitcoin server crashed some time overnight?

[qed]

Is poclbm periodically checking for the correct generation of ashes or do i have to make a stability test after changing voltage/frequency?

qed, if you read my post then you'd know that it does check.
if your card isn't stable then it will crash with "Verification failed. check hardware"

It's not that simple.

[Fuma]

Hi
Been  using this miner.
and i did some changes in my GPU configuration.
i got a hd5870 and 8800GTX running together.
the miner recognize the 5870 but lists cuda as the interface(?) instead of opencl
and it never starts.

tried changing the number of device in bat file. basically tried everything i could on my own...
i know the double drivers must confuse the miner...but is there a way to go around this?

Edit :
the platform command fix this issue.
but now i'm getting connecting to rpc server problems.  
hope it's not related.

[nster]

Hi
Been  using this miner.
and i did some changes in my GPU configuration.
i got a hd5870 and 8800GTX running together.
the miner recognize the 5870 but lists cuda as the interface(?) instead of opencl
and it never starts.

tried changing the number of device in bat file. basically tried everything i could on my own...
i know the double drivers must confuse the miner...but is there a way to go around this?

Edit :
the platform command fix this issue.
but now i'm getting connecting to rpc server problems.  
hope it's not related.

I couldn't get my GT240 and 6870 top play nice together, so only my 6870 mines. I get ~280Mhash/s with 1040 core and 300 mem. keep in mind my 6870 OCs really badly, 975 core in games = instant BSOD. Try using MSI afterburner

[Fuma]

I couldn't get my GT240 and 6870 top play nice together, so only my 6870 mines. I get ~280Mhash/s with 1040 core and 300 mem. keep in mind my 6870 OCs really badly, 975 core in games = instant BSOD. Try using MSI afterburner

I get ~315Mhash with 5870 on stock frequencies. w/o the 8800gtx together they make on stock about ~345Mhash
so i dont let the 8800gtx to do nothing -not worth the electricity.
but they work well mining and gaming.

[qed]

I couldn't get my GT240 and 6870 top play nice together, so only my 6870 mines. I get ~280Mhash/s with 1040 core and 300 mem. keep in mind my 6870 OCs really badly, 975 core in games = instant BSOD. Try using MSI afterburner

Then you are generating invalid hases at that frequency. The fact that your card is not is not giving you a BOSD while mining doesnt mean it's stable.

[nster]

I couldn't get my GT240 and 6870 top play nice together, so only my 6870 mines. I get ~280Mhash/s with 1040 core and 300 mem. keep in mind my 6870 OCs really badly, 975 core in games = instant BSOD. Try using MSI afterburner

Then you are generating invalid hases at that frequency. The fact that your card is not is not giving you a BOSD while mining doesnt mean it's stable.

how could I check for invalid hashes?

[Grinder]

how could I check for invalid hashes?

poclbm will say "Verification failed. check hardware". You probably don't need to worry, at least I've never seen this from overclocking. Either it generates valid hashes, or the computer freezes. I've got a 5850 running fine at 930 MHz (default is 725). It actually didn't crash before it reached 970, but I had to flash it to put it in a Linux server so I can't really fine tune it.

[deMangler]

how could I check for invalid hashes?

poclbm will say "Verification failed. check hardware". You probably don't need to worry, at least I've never seen this from overclocking. Either it generates valid hashes, or the computer freezes. I've got a 5850 running fine at 930 MHz (default is 725). It actually didn't crash before it reached 970, but I had to flash it to put it in a Linux server so I can't really fine tune it.

I mine with overclocked NVidias. Using poclpm.
I slowly tweak them as high as I can with riva, and I do get "Verification failed. check hardware" as I approach (or rather slightly cross) their limit.  I find it is a good test of how high I can overclock my cards.
dM

[qed]

how could I check for invalid hashes?

poclbm will say "Verification failed. check hardware". You probably don't need to worry, at least I've never seen this from overclocking. Either it generates valid hashes, or the computer freezes. I've got a 5850 running fine at 930 MHz (default is 725). It actually didn't crash before it reached 970, but I had to flash it to put it in a Linux server so I can't really fine tune it.

That's not true.

[Grinder]

That's not true.

Guess my computer's just is really good at fooling Slush's pool, then.