Cuda error: Kernel [pme_fill_charges_accumulate] failed in file 'fillcharges.cu' in line 73 : unknown error.

OS: Vista32
Driver: 182.06
GPU: GTX 295

325502
324165
322648
322054
321100

could this error be from the new driver?

---

could this error be from the new driver?

Considering that most of those work units failed rather quickly, a corrupt or broken driver installation seems a good possibility. For an example, see the thread Continual computing errors.

Alternately, if it's a new GTX 295 then the card itself could be faulty as all those errors occurred on device 0. If the driver and card can flawlessly complete all work units on device 1, then the card itself is likely malfunctioning.

Edit:
I just checked the machine information for those tasks, and noticed that the GTX 295 only shows up as a single device when it should show up as two. Something's awry with the configuration.

It looks odd, but you have to make sure that the drivers install for each of the Cores in the GPU ... so, in my case I really had to see the install happen 4 times (2 GTX 295 cards) ...

You can make sure of course by looking in the device tab to make sure that all the devices are "live" and for some drivers and Vista you have to extend the desktop to the other GPU core to make it usable.

I just checked the machine information for those tasks, and noticed that the GTX 295 only shows up as a single device when it should show up as two. Something's awry with the configuration.

System config is good now. I found out that I have to extended it to the HDMI port not the 2nd DVI.

I think BOINC v6.6.0 is also part the problem. I switch to v6.5.0 now.

When I replaced the HDMI port with dummy one (hdmi->dvi=>>dvi->vga with 75ohms on pins 4 and 5), it was working for awhile then lost contact.

Anyone know how to create HDMI dummies?

---