Wow they have really harder requirements on that SANTI Batch it seems. I run one successfully with 50mV overvoltage on the 560ti 384core. Until now, all cards run with +25mV and computed successfully with this setting, Santis too. But only this card needs more. But +50mV needs serious cooling.. im nearly at full fanspeed and over 80degress with open case and no extra heating in the flat.

---

DSKAG Austria Research Team: http://www.research.dskag.at

Over 80C is too hot for me. I like mine 70C max.

---

BOINC <<--- credit whores, pedants, alien hunters

Wow they have really harder requirements on that SANTI Batch it seems.

That's what I'm thinking too. Had another one that caused the machine to reboot continuously until I caught it. This time a SANTI_MARwtcap. The only way to stop the cycle is to abort the WU. Lowered the clocks yet again and the next one is running fine. In fact that machine is currently showing 20 valid and the 1 error WU that caused constant bluescreens. NVIDIA GeForce GTX 650 Ti (1024MB) driver: 331.82.

Locked again on a workunit, i stopped again with this card on gpugrid and changed it back to einstein :/ will try again in one or two weeks.

---

DSKAG Austria Research Team: http://www.research.dskag.at

After 17 days a Santi resulted in one 660 to down clock. And only 2 errors in 19 days. But today my rig with to 660's was booted when I found it. After logging in it booted immediately when BOINC started a few times, so I went to Windows in safe mode where I have some more time to abort the task. As I didn't know which one, I aborted both. Its now happily crunching again.

---

Greetings from TJ

Too many errors for me......stopping here.
Task
click for details
Show names Work unit
click for details Sent Time reported
or deadline
explain Status Run time
(sec) CPU time
(sec) Credit Application
7614815 5043955 2 Jan 2014 | 20:22:26 UTC 3 Jan 2014 | 3:05:04 UTC Error while computing 23,648.57 20,821.78 --- Long runs (8-12 hours on fastest card) v8.14 (cuda42)
7614651 5043020 2 Jan 2014 | 21:42:39 UTC 3 Jan 2014 | 4:35:29 UTC Aborted by user 20,147.56 20,074.19 --- Long runs (8-12 hours on fastest card) v8.14 (cuda42)
7613690 5043565 2 Jan 2014 | 16:43:59 UTC 2 Jan 2014 | 23:04:48 UTC Completed and validated 22,184.66 18,799.57 20,550.00 Short runs (2-3 hours on fastest card) v8.15 (cuda42)
7613447 5043367 2 Jan 2014 | 16:43:59 UTC 2 Jan 2014 | 20:22:26 UTC Error while computing 12,559.07 12,065.65 --- Short runs (2-3 hours on fastest card) v8.15 (cuda42)
7612704 5042731 2 Jan 2014 | 4:29:25 UTC 2 Jan 2014 | 13:59:13 UTC Completed and validated 33,819.15 17,980.88 20,550.00 Short runs (2-3 hours on fastest card) v8.15 (cuda42)
7612537 5040031 2 Jan 2014 | 4:20:27 UTC 2 Jan 2014 | 4:29:25 UTC Error while computing 274.96 127.56 --- Short runs (2-3 hours on fastest card) v8.15 (cuda42)

Just now I found my rig with two 660's frozen. No Idea when it happened, even ctrl-alt-del didn't work. After booting immediately message that the graphics drives has crashed and recovered, three times in a row and then it booted itself again. After three attempts I got BOINC to stop. I am now installing the latest beta driver but that should not be necessary and it ran for more then a month with the 331.82 driver.

I have not a lot of joy with the 660 since I bought two of them in summer.
I don't like these Santi's.

---

Greetings from TJ

Again I found my rig with two 660's frozen.
After 5 boots I managed to get rid of all the Santi's and switch over to LR only.

It is or an AMD CPU has problems with Santi or GPU's lower tan 7XX have problems with Santi's or a combination of the two has problems with Santi's.
I guess none of the above as there a re many AMD rigs and many more rigs with 660 that does there terrible Santi's. But my RAC is rapidly falling this way.

Edit: another Santi crash, so will become 7 boots eventually. I start to hate these Santi's!

Edit 2: Yes I got two Nathan's on the 660's, so I can go sleep this night and not watch my system continuously.

---

Greetings from TJ

TJ, you should lower the GPU frequency of those 660s, or increase the GPU voltage by 12mV.

Boost the voltage and produce more heat or get crunching on Linux. Take a look at my results. I have errors but 99% of those are tasks I aborted because I played with stuff and ended up with too many tasks in my cache or other reasons. I have two 670 and one 660Ti on Linux and they almost never crash SANTI tasks. I'm running the stock clock speeds and if I keep the temps below 70C the clock boost thing kicks in regularly. They hardly ever crash on any task and if they do the OS doesn't hang, BOINC continues running, another GPUgrid task downloads and starts and life carries on.

---

BOINC <<--- credit whores, pedants, alien hunters

Again I found my rig with two 660's frozen.
After 5 boots I managed to get rid of all the Santi's and switch over to LR only.

It is or an AMD CPU has problems with Santi or GPU's lower tan 7XX have problems with Santi's or a combination of the two has problems with Santi's.
I guess none of the above as there a re many AMD rigs and many more rigs with 660 that does there terrible Santi's. But my RAC is rapidly falling this way.

I have had those problems too. The main reason seems to be that the 660s were bumping up against their power limit, causing them to be starved for current on the tough portions of the hardest work units. Increasing the power limit to 110% by using Nvidia Inspector has largely solved the problem for me on the two cards (a Zotac and a Gigabyte) that I now use for GPUGrid, without the need for any other changes:
http://www.gpugrid.net/results.php?hostid=159002&offset=0&show_names=1&state=0&appid=

But they are often overclocked too much at the factory for the work here, and on another of my Zotac 660s I also have had to reduce the clocks a little (GPU clock from 993 MHz to 950 MHz, and memory clock from 3004 to 2804 MHz) and also bump up the core voltage (from 1.162 to 1.175 volts). For some reason on the Zotacs the software control utilites (such as Nvidia Inspector or MSI Afterburner) do not work to change the voltage, and I had to modify the BIOS with Kepler BIOS Tweaker, and then flash it into the video card with nvflash. You can use GPU-Z to first make a copy of your present BIOS that you then modify (keep a copy of your old BIOS as a backup). If you don't want to deal with that, just reduce the clock frequency, first on the GPU clock and then on the memory clock if necessary, until it is stable.

Hi Guys,

Good advise. I lowered the clocks per Zoltans advise and will see what happens.
If not stable I will increase the voltage little.
Problem is that the rig was stable for 17 days continuously, that is wondering me.

I let them run at factory settings, at first months ago. They are both form EVGA. Their maximal fan speed is 74%.

---

Greetings from TJ

You can start a snowball effect with system freezes and BSOD's if you don't run checkdisk after getting those errors, too many orphaned files or wrong time stamps and such just causes more and more problems. If you don't do that, the errors your getting now could be related to it.

Your cracking me up TJ, you think your BSOD's and freezes are because you have an AMD system? I think the drought were having in California is because of the Intel CPU in my laptop (gotta put on my tin foil hat).

No BSOD but again a frozen system. This time by a Noelia on my 660's rig. I did manage to get the clocks down with Pricison X from EVGA, but after a while they boost automatically again. Trying to do it with MSI Afterburner, shows only one card, there is nowhere I click that I can see the settings of my second card.
Well from August these 660's are trouble some to me, so I will buy a second 780Ti as it, even in Windows7, does the same as two 660's is less time to replace the 660's.

---

Greetings from TJ

In afterburner, click Settings (bottom right corner of the left pane) and then you can change the GPU under the General Tab.

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

History quiz: Does anyone remember when MS ballyhooed long and loudly that they had solved the "black screen of death"? Remember the solution?

They made the color a user configurable option then told everyone if they get a black screen of death it's their own damn fault.

---

BOINC <<--- credit whores, pedants, alien hunters

In afterburner, click Settings (bottom right corner of the left pane) and then you can change the GPU under the General Tab.

Thanks skgiven, found it and used it!

---

Greetings from TJ

Just had to move my 560ti onto the short runs

Even got the Dyson out expecting a GPU full of fluff but no it's the Work Unit's

Just 2 successful long runs since 24th Dec


There's a problem with the current batch. I'll just have to be patient.

Maybe this time next week


Dave

Your clocks are too high. Try 1644Mhz for the processor & 2004Mhz for the memory.

'GPUgrid stresses the parts other projects can't reach'

Probably would be useful if people having a problem with a self overclocked card would return it to stock clocks before reporting problems with WU.