DiabloMiner GPU Miner

[jcw9]

Hi Diablo, just catching up from IRC on my osx experience.

Code:

511 pyramid:~/Projects/DiabloMiner$ ./DiabloMiner-OSX.sh -u jcw9 -p notarealpasswordfolks -w 64
[12/14/10 2:56:50 PM] Started
[12/14/10 2:56:59 PM] Added ATI Radeon HD 5750 (#1) (10 CU, local work size of 64)
49951 khash/sec

This works great… but it seems to generally freeze the machine after a few minutes. Cursor is mobile and I was able to ssh in last time. The java process was no longer running when I sshed in. I'd like to help figure out ways to improve the stability. Any ideas?

[1715168272]

1715168272

[1715168272]

1715168272

[DiabloD3]

That usually indicates a fault in the OS somewhere. Since no one gives a shit about OpenCL on OSX (not even Apple), I'm not sure what to tell you. Any other OSX users want to chime in?

[tryptamine]

$ ./DiabloMiner-Linux.sh -u lolzuser -p mehpw -w 64
[12/15/10 6:39:36 PM] Started
[12/15/10 6:39:42 PM] Added Juniper (#1) (9 CU, local work size of 64)
111224 khash/sec


Listed as Juniper, it's a Radeon HD 4750.

[DiabloD3]

$ ./DiabloMiner-Linux.sh -u lolzuser -p mehpw -w 64
[12/15/10 6:39:36 PM] Started
[12/15/10 6:39:42 PM] Added Juniper (#1) (9 CU, local work size of 64)
111224 khash/sec


Listed as Juniper, it's a Radeon HD 4750.

Thats a 5750, actually. 5750s do about 116 mhash/sec and are Junipers.

[tryptamine]

True, my bad, a typo.

By the way, clocking core from 700MHz to 850MHz looks stable and you can get quite a bump in performance, 140 mhash/sec right now

[DiabloD3]

Update: Added dedicated getwork thread, now only gets new work once per 5 seconds instead of 3 times per gpu per 5 seconds.

This should improve performance on miner pools due to remote connection latency, and it should also improve performance on setups that have many miners connecting to a single LAN bitcoind.

Another update: Added -g/--getwork to change the number of seconds between getwork refresh, should help high pool latency issues, default remains 5.

Another another update: the sh scripts now set an environment variable that fixes SDK 2.2 and 2.3 CPU time usage. I still recommend 2.1, but at least it doesn't go apeshit anymore.

[russ]

I found I kept getting errors when running the miner...

I fixed the ELF_CLASS error by deleting the non *64.so libs and renaming them to *.so (just remove the 64!)

Then, I got an error about not being able to connect to a screen (it's a headless server);

I installed xvfb...ran this;

export DISPLAY=:1
Xvfb :1 -screen 0 1024x768x16 &

and it worked fine 

[DiabloD3]

I found I kept getting errors when running the miner...

I fixed the ELF_CLASS error by deleting the non *64.so libs and renaming them to *.so (just remove the 64!)

Then, I got an error about not being able to connect to a screen (it's a headless server);

I installed xvfb...ran this;

export DISPLAY=:1
Xvfb :1 -screen 0 1024x768x16 &

and it worked fine 

You didn't have to rename the libs. Setting DISPLAY alone fixes this (as long as you have an X server running).

[DiabloD3]

Update: Made the getwork clone from network thread method more paranoid, should help with low difficulty targets like pools and the testnet.

[Cdecker]

Does anybody have a comparison on how Linux against Windows machines are performing when running the GPU miners?

[DiabloD3]

Does anybody have a comparison on how Linux against Windows machines are performing when running the GPU miners?

No, but Linux runs faster.

[Cdecker]

Does anybody have a comparison on how Linux against Windows machines are performing when running the GPU miners?

No, but Linux runs faster.

Great news
I was already dreading I'd have to setup a Windows machine ^^

[russ]

I found I kept getting errors when running the miner...

I fixed the ELF_CLASS error by deleting the non *64.so libs and renaming them to *.so (just remove the 64!)

Then, I got an error about not being able to connect to a screen (it's a headless server);

I installed xvfb...ran this;

export DISPLAY=:1
Xvfb :1 -screen 0 1024x768x16 &

and it worked fine 

You didn't have to rename the libs. Setting DISPLAY alone fixes this (as long as you have an X server running).

I don't have an X server running, or didn't before xvfb.

Also, it did repeatedly crash before getting as far as crashing about X related things, moaning about the ELF CLASS. Renaming the libs fixed this.

[DiabloD3]

The ELFCLASS32 bug is, unfortunately, inside the AWT code inside Java. lwjgl rams into it because it initializes AWT even if I do not open a window. They are aware of the bug; Oracle has no plans on fixing it on their end (ANY app that tries to init AWT on a system without X running will do this).

[russ]

The ELFCLASS32 bug is, unfortunately, inside the AWT code inside Java. lwjgl rams into it because it initializes AWT even if I do not open a window. They are aware of the bug; Oracle has no plans on fixing it on their end (ANY app that tries to init AWT on a system without X running will do this).

Well, anyway I'd just thought posting up might help others in a situation like mine. I really wanted to get the GPU miner working, but don't want X...

Thanks for all your work on this project!

[DiabloD3]

Throwing up a dummy xvfb will indeed fix this. ATI users don't have this problem because X has to be running to use OpenCL

[DiabloD3]

Update: Forced all accesses to a getwork instance to be synchronous, should fix final network thread problems

[adulau]

I have tried DiabloMiner on MacOS X but I have some errors when DiabloMiner is creating the command queue:

[12/25/10 5:18:52 PM] Started
[12/25/10 5:18:52 PM] Added GeForce 9400M (#1) (2 CU, local work size of 256)
[12/25/10 5:18:54 PM] Added GeForce 9600M GT (#2) (4 CU, local work size of 256)
[12/25/10 5:18:54 PM] ERROR: [CL_DEVICE_NOT_AVAILABLE] : OpenCL Error : clCreateCommandQueue failed: Device 0x1020331f0 is not available.
[12/25/10 5:18:54 PM] ERROR: Failed to allocate queue
[12/25/10 5:18:54 PM] ERROR: [CL_INVALID_COMMAND_QUEUE] : OpenCL Error : clEnqueueWriteBuffer failed: invalid command queue 0x0.
Exception in thread "DiabloMiner Executor (GeForce 9600M GT (#2)/0)" java.lang.NullPointerException
        at org.lwjgl.opencl.APIUtil.releaseObjects(APIUtil.java:506)
        at org.lwjgl.opencl.APIUtil.releaseObjects(APIUtil.java:502)
        at org.lwjgl.opencl.CL10.clReleaseCommandQueue(CL10.java:565)
        at com.diablominer.DiabloMiner.DiabloMiner$DeviceState$ExecutionState.run(DiabloMiner.java:568)
        at java.lang.Thread.run(Thread.java:680)

I opened a ticket at github with more information  : https://github.com/Diablo-D3/DiabloMiner/issues#issue/3

Thank you for your work,

[DiabloD3]

For those wondering, I replied to the bug. OSX hybrid graphics are screwing up OpenCL apps.

[tryptamine]

BTW, the java is using 100% CPU in my computer, not 0% as advertised. I want my money back!

I'm using the last version. Do you want me to test anything?