560 GTX vs. Fermi/Core_17.fah/FahCore_17.exe

[suprleg]

I was wracking my brain trying to solve this problem when it dawned on me it might not be an Nvidia driver problem at all. It popped into my head that this 3rd gen Intel chip has Intel HD graphics built on the chip.
I uninstalled the Intel graphics drivers for the chip and "0x17" wu's are running fine.
Thanks to all who helped....don't know why it didn't dawn on me earlier.

[7im]

That's strange. Why could the onboard driver interfere with the discreet driver? (not looking for a response)

But I am curious what the Intel chip model is, and what the onboard video chip model is called. Could come up again.

[suprleg]

Intel® Core™ i5-3450 Processor, Sandy Bridge, (6M Cache, up to 3.50 GHz)
Intel® HD Graphics 2500
Windows7 pro
Nvidia GTX560
Churning out "0x17" wu's.....coincidence, maybe, but I think not.

[PantherX]

On my laptop, I have installed the Intel Driver and Nvidia ones without any issues and could fold FahCore_15 and FahCore_17 WUs without any issue. Laptop specifications:
Intel Core i7-3840QM
Intel HD Graphics 4000 - Driver version 10.18.10.3496 (15.33.18.64.3496)
Nvidia GTX 675M - Driver version 9.18.13.3523 (335.23)
Windows 8 64-bit

[bruce]

That's strange. Why could the onboard driver interfere with the discreet driver? (not looking for a response)

You may not be looking for one, but you're going to get one anyway.

Excellent detective work, suprleg. (Technically, the graphics software, or in your case, firmware, is essentially independent of the code supporting OpenCL, though it may not appear that way.)

When drivers are installed for ATI/NV/Intel, they're generally packaged with component software that intercepts OpenCL API calls and ultimately converts the call to specific instructions that drive their GPU hardware. As long as the code is either 100% independent or any shared code works for any OpenCL call, you won't run into problems, but we've seen similar reports (rarely enough that I didn't think of it) where apparently some part of the Microsoft OpenCL code is improperly intercepting something that should be going through an ATI or NV code path. [Just my unproven interpretation.]

Since the Intel OpenCL code is not supported, removing whatever part of the Intel code isn't needed and/or reinstalling the NV code to restore whatever was (probably) clobbered by Intel will probably fix everything.

[7im]

That's assuming Intel was installed 2nd.

[P5-133XL]

Was changing the opecl-index tried? Theoretically, each opencl device/driver should have a different index value. I could easily see the v7 client getting confused and using the Intel CPU OpenCL driver instead of the Nvidia one. Changing the opencl-index should have fixed that problem.