I was running CUDA 6.10 from BOINC 7.0.75 with 301.24 NVidia drivers. That worked fine. I saw the NVidia Beta 304.79 mentioned at anandtech.com decided to give them a try. Definitely faster on some graphics software. I also upgraded to BOINC 7.0.28. After the two upgrades all 1388 tasks in the queue returned "Error while computing" in less then 24 hours. Ouch. Any thoughts? I did a reset. I'll probably downgrade the NVIdia drivers to 301.42 for now.

-Kurt

I think somone reported something similar a while back- the problem only occurs with the stock application. If you're running the Lunatics Optimised application there's no issue.

Edit- found the thread in question.

---

Grant
Darwin NT

I was running CUDA 6.10 from BOINC 7.0.75 with 301.24 NVidia drivers. That worked fine. I saw the NVidia Beta 304.79 mentioned at anandtech.com decided to give them a try. Definitely faster on some graphics software. I also upgraded to BOINC 7.0.28. After the two upgrades all 1388 tasks in the queue returned "Error while computing" in less then 24 hours. Ouch. Any thoughts? I did a reset. I'll probably downgrade the NVIdia drivers to 301.42 for now.
-Kurt

Depends on what's the error number.

---

TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.

I was running CUDA 6.10 from BOINC 7.0.75 with 301.24 NVidia drivers. That worked fine. I saw the NVidia Beta 304.79 mentioned at anandtech.com decided to give them a try. Definitely faster on some graphics software. I also upgraded to BOINC 7.0.28. After the two upgrades all 1388 tasks in the queue returned "Error while computing" in less then 24 hours. Ouch. Any thoughts? I did a reset. I'll probably downgrade the NVIdia drivers to 301.42 for now.
-Kurt

Depends on what's the error number.

Checking quickly before maintenance and eventual result purging:

Exit status 1 (0x1) Unknown error number
Incorrect function. (0x1) - exit code 1 (0x1)
CUFFT error in file 'd:/Projects/SETI/seti_boinc/client/cuda/cudaAcc_fft.cu' in line 62.

That does indeed sound like an incompatibility between "Application version SETI@home Enhanced v6.10 (cuda_fermi)" and the Beta driver.

I personally got a lot of -226 errors on the CPU and GPU.
Device not found on the latest driver on some cards.

---

TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.

Just on a side note, what have your found to be the optimal number of parallel SETI GPU tasks on the 680? (I am guessing its somewhere between 4 and 6 before adding more tasks doesn't gain any more throughput - i.e. the existing running tasks start to run longer with no actual increased overall throughput.)

Cheers

Mark

---

I was running CUDA 6.10 from BOINC 7.0.75 with 301.24 NVidia drivers. That worked fine. I saw the NVidia Beta 304.79 mentioned at anandtech.com decided to give them a try. Definitely faster on some graphics software. I also upgraded to BOINC 7.0.28. After the two upgrades all 1388 tasks in the queue returned "Error while computing" in less then 24 hours. Ouch. Any thoughts? I did a reset. I'll probably downgrade the NVIdia drivers to 301.42 for now.

-Kurt

Kurt, is this a typo? I thought that x.31 was the latest version, and that was still in beta? I am always for upgrading if there's a perf increase, but not sure if I'm brave enough to run beta Nvidia drivers. I've been burned in the past by them. Also not thrilled about the message screen being seperate in the last 6.x and all 7 versions. I'm pretty happy at this time with 6.10.60, last of the 'old style' versions. Are there any significant improvements on going newer?

---