8

I have 2 HD6950's which I wanted to use for ETH mining - but I have the well known DAG allocation problem.

I can't find a solution for this problem - that's why I'm asking here. And please don't tell me to just buy new cards...

The card has 2 Gb of ram which should usually be enough. I tried to set GPU allocation to 95 and 100% but without success.

I use Windows 7 64 bit. If it is 1,4 GB and my card has 2 GB it should work easy?

1
  • I use Windows 7 64 bit. If it is 1,4 GB and my card has 2 GB it should work easy? Commented Mar 12, 2016 at 23:18

3 Answers 3

4

It seems the HD6950 is limited to a 512MB allocation even with much more RAM onboard.

C:\Program Files\Ethereum 0.9.41\Release>ethminer --list-devices

Listing OpenCL devices.  
FORMAT: [deviceID] deviceName  
[0] Cayman  
        CL_DEVICE_TYPE: GPU  
        CL_DEVICE_GLOBAL_MEM_SIZE: 2147483648  
        CL_DEVICE_MAX_MEM_ALLOC_SIZE: 536870912  
        CL_DEVICE_MAX_WORK_GROUP_SIZE: 256

I sorted that out by replacing with files found here which load the DAG in chunks rather than as one monolithic file.

I'm getting a woeful hashrate of 1.2MH/s but maybe that's my 2008 vintage P35 motherboard :)

I'd love to know what numbers you end up getting.

3

Unfortunately, it might be impossible because of 2GB of RAM. Although current DAG is 1.4 GB in size and should fit the OpenCL implementation might not allow such big allocation for this GPU.

You might try the following:

  1. Try running from console without starting graphical desktop environment (if on Linux).
  2. Try benchmark ethminer -G -M that will use 1 GB DAG.
1
  • ethminer -G -M has the same result. "clcreatebuffer(-61)" GPU can't allocate the DAG in a single chunk. Bailing. Commented Mar 13, 2016 at 17:23
1

try setx GPU_MAX_ALLOC_PERCENT=100 if memory serves correctly on windows (on linux it's export GPU_MAX_ALLOC_PERCENT=100) to see if it solves your issue,

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.