Saphire 7950s can't go over intensity 13 without HW errors. [[SOLVED]]

[Sancho]

 I have a rig with 3 Sapphire 7950s - brand new.
Specifically - SAPPHIRE 100352-3L Radeon HD 7950 3GB 384-bit GDDR5 PCI Express 3.0 x16 HDCP Ready CrossFireX Support Video Card
4GB RAM

Running cgminer 2.11.4
MSI Afterburner Beta

Current settings - best I've been able to run stable.

setx gpu_max_alloc_percent_100
setx gpu_use_sync_objects 1
cgminer.exe --scrypt -i 13 -g 2 -w 256 --thread-concurrency 8192

(followed by correct pool settings)

When I increase the intensity beyond 13 it gives me hardware errors and inaccurate hashing power.

When I increase the thread concurrency, cgminer will start but the GPUs will be disabled and it says GPU 0 Faulure, disabling! Thread 1 being disabled
Error -5: Enqueueing kernel onto command queue. (c1EnqueueNDRangeKernel)

It gives the same error message for GPU 1 as well.

Please help, with the current settings I'm not getting more than 300 khash/s per card.

Thank you!

[how-high]

howmuch ram?

[Sancho]

4GB.

I have two other rigs with 7950s and no issues like this - only the brand is different.

[how-high]

change -g 2 to -g 1

[Wipeout2097]

setx gpu_max_alloc_percent_100
setx gpu_use_sync_objects 1
cgminer.exe --scrypt i-13 -g 2 -w 256 --thread-concurrency 8192

setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1
cgminer.exe --scrypt -I 18 -g 1 -w 256 --thread-concurrency 21712 --gpu-memclock 1500 --gpu-engine 1000

[Sancho]

I tried all of that and it reports the same error I quoted in the original post.

The GPU fails and it talks about enqueueing kernels.

No idea what to do.

[Sancho]

I think I found the problem. One of the RAM sticks wasn't in all the way!

Doh!


Thanks all!