python OpenCL bitcoin miner

[zimpixa]

Failover failed cuz there were no server change (it didnt react on such message)

[bitcoinBull]

It prints newlines on every hash rate update when run in lxterminal (linuxcoin).  

Anyone experience the same?  Know why?

EDIT: was told this is because of the width of the terminal window.  There is some extra whitespace after the hash rate text which wraps around to form blank lines if the the terminal window isn't wide enough.

[RaTTuS]

Same for me in both windows and linux

[jgarzik]

...and how does the client inform the server that it supports this feature?

There is version in user-agent, server can determine miner and act accordingly.

Sorry, this is not scalable approach.  Please re-think.

If all miners implement this, then each pool server must implement version checking code for every supported miner.  Each pool server must update their source code for each updated or new miner.

In contrast, if a miner client provides "X-Feature-Supported" in HTTP headers, then there is a simple check in the server that never needs to be updated.

User-Agent is the wrong approach.

[Andrey]

Got a problem yesterday, when I checked the miner which had 4 servers for main + fallover it couldn't connect to any of them, although connection was up.
When I restarted the client - everything started to work.

Please check, I think some resource is getting exhausted upon reconnects to fallover servers.

[brunoshady]

I have a doubt.



I have one rig with 2 5850...

When I set the GPU1 (device 1) with frames=1, I get about 348/349 mhash constantly...
so when I turn on the GPU0, no matter with frames, I get lower mhash and the GPU1 drops its mhash...


is that normal?

both cards have the exactly same configuration and they are the same model...
in my mind I though I could set both cards with frames=1 and get the 348/349 mhash individually.. besides the desktop lag

[Mobius]

I have a doubt.



I have one rig with 2 5850...

When I set the GPU1 (device 1) with frames=1, I get about 348/349 mhash constantly...
so when I turn on the GPU0, no matter with frames, I get lower mhash and the GPU1 drops its mhash...


is that normal?

both cards have the exactly same configuration and they are the same model...
in my mind I though I could set both cards with frames=1 and get the 348/349 mhash individually.. besides the desktop lag

when you run ./poclbm.py by itself in the /poclbm directory, what devices is it showing?

[brunoshady]

3 devices... both the 2 gpu and the cpu




ps: its ubuntu 11...

[Mobius]

3 devices... both the 2 gpu and the cpu




ps: its ubuntu 11...

Do you have the crossfire cable installed, I keep each of my cards at frames=10, 3 cards on one rig, I try to keep the total of all the frames= to add up to 30.

If you do any other frames= combo do you see that much impact?

[brunoshady]

3 devices... both the 2 gpu and the cpu




ps: its ubuntu 11...

Do you have the crossfire cable installed, I keep each of my cards at frames=10, 3 cards on one rig, I try to keep the total of all the frames= to add up to 30.

If you do any other frames= combo do you see that much impact?

they are not crossfired...

GPU1 (device1) is with frames 1, and GPU0 with 10... if I lower the frame on the GPU0, GPU1 is affected.. I guess even with 10 GPU1 is affected considering frame 30 normal

[Mobius]

3 devices... both the 2 gpu and the cpu




ps: its ubuntu 11...

Do you have the crossfire cable installed, I keep each of my cards at frames=10, 3 cards on one rig, I try to keep the total of all the frames= to add up to 30.

If you do any other frames= combo do you see that much impact?

they are not crossfired...

GPU1 (device1) is with frames 1, and GPU0 with 10... if I lower the frame on the GPU0, GPU1 is affected.. I guess even with 10 GPU1 is affected considering frame 30 normal

What size PSU do you have?

[twmz]

If you set aggression (in phoenix) too high or -f (in poclbm) too low, my experience is that it interferes with multiple GPU mining.  The miner gets bogged down with the overhead of being too aggressive and can't keep the GPUs fed with work to do. 

I would try a higher -f until you find the point at which you can mine on all GPUs without the second GPU affecting the first GPU's hash rate.

[zimpixa]

Have the same isssue. Settings around -f35 for main card (with monitor) and -f2 for second one are best for me, but ofc loosing about 20MHash compared to 2x singlecard-rigs.

[SeXy]

Hello.

POCLBM crashes almost instantly on my setup (2x5970+ 1x5850 + W7 x64)

if i remove the 5850 everything works fine

ideas?

[Mobius]

Hello.

POCLBM crashes almost instantly on my setup (2x5970+ 1x5850 + W7 x64)

if i remove the 5850 everything works fine

ideas?

Is your power supply big enough

5970=300watts *2 + 5870=200watts *1 +175watts overhead (Board, Fans, Processors, Disk Drives) =975watts @ 85%eff = 1150watt supply

[SeXy]

1250w enermax

[NetTecture]

Some comments and questions

* What happened to options like AGGRESSION etc.? I can not enter them anymore.
* Can we as next step get multi GPU support? So that one can define multiple GPU or just something like -d all and it uses all GPU? Quite a lot of people I think run multi card configs with the same card everywhere, makes everything easier.
* Is there a way to get rid of the new measurement output and put it back to "line by line"? Easier for those starting the miner from a control program that then parses the output Not sure I overlooked that.

Besides that - good thing.

[brunoshady]

Some comments and questions

* What happened to options like AGGRESSION etc.? I can not enter them anymore.
* Can we as next step get multi GPU support? So that one can define multiple GPU or just something like -d all and it uses all GPU? Quite a lot of people I think run multi card configs with the same card everywhere, makes everything easier.
* Is there a way to get rid of the new measurement output and put it back to "line by line"? Easier for those starting the miner from a control program that then parses the output Not sure I overlooked that.

Besides that - good thing.

aggression = frames


and the line by line, try the --verbose mode

[SeW900]

Hey guys, the new Catalyst drivers come with support for OpenCL and AMD APP. So is the installation of the SDK still required, or is it not?

[Luke-Jr]

Latest poclbm mainline breaks with the new difficulty!

Bugfix: http://gitorious.org/~Luke-Jr/bitcoin/luke-jrs-poclbm/commit/7fdc508a1070f5540046f2d7649eed15503177ad
To merge:

Code:

git fetch https://git.gitorious.org/~Luke-Jr/bitcoin/luke-jrs-poclbm.git bugfix_difficulty_error && git merge FETCH_HEAD

If this saves you time and money, consider dropping me some bitcoin 1AgqndFUg2sDSxW8NbzPyHvUibW6FHqQeA