Heads Up! Some new GPU work released from beta!

[Tobit]

Just a heads up folks, although many of you will see little if any difference, as there are some new WUs that were just released from beta to general population a couple hours ago. Some of you may see these as "unknown" on your FAHMon, FAHSpy, or HFM.net.. simply just issue a "download projects" when these make it to the general psummary page or reconfigure your monitoring client to pull from psummaryC if you haven't already.

The projects in question are 5781-5786 and there should be no performance impact on your client. These are all NVIDIA GPU WUs worth 783 points each. While these were in beta, my clients crunched these with no issues unlike what we saw with p10101. I will include some benchmarks below.

 Project ID: 5782
 Core: GROGPU2
 Credit: 783
 Frames: 100

 Name: ganymede - GTX 260
 Path: \\ganymede\fah\gpu0
 Number of Frames Observed: 200

 Min. Time / Frame : 00:01:15 - 9,020.2 PPD
 Avg. Time / Frame : 00:01:15 - 9,020.2 PPD

 Name: io - 8800GT
 Path: C:\fah\gpu0\
 Number of Frames Observed: 200

 Min. Time / Frame : 00:02:17 - 4,938.0 PPD
 Avg. Time / Frame : 00:02:18 - 4,902.3 PPD

 Name: io - 9600GSO
 Path: C:\fah\gpu1\
 Number of Frames Observed: 200

 Min. Time / Frame : 00:02:33 - 4,421.6 PPD
 Avg. Time / Frame : 00:02:34 - 4,392.9 PPD

 

[Vaulter98c]

Hmm, thanks for the heads up
Your a beta guy right? Any word on how long these 10101s will run? Between the crazy power draw and heat issues (not to mention low point values) I forsee a lot of problems come warmer months, these could actually damage hardware
Posted via [H] Mobile Device

 

[Tobit]

p10101's are in gen-pop so they are here til the project is completed unless Vijay determines otherwise. There hasn't been any discussion about removing them from gen-pop. He had thought forcing a core upgrade would solve these issues but some are still having issues it would appear.

I really wish I knew what the answer was for these. These have run flawless for me since inception and none of my cards are running significantly warmer nor seeing any major difference in power draw. I am not doubting you all because I've read of similar issues on other forums. However, those having problems do appear to be in the minority compared to the success stories. I will mention this again but these have much higher atom count than what we are used to so there is probably not much that can be done. What I think is happening is, because of the high atom count, these WU units are sitting in VRAM a lot longer than other WUs and this is causing a lot of the heat and stress reports.

 

[APOLLO]

What I think is happening is, because of the high atom count, these WU units are sitting in VRAM a lot longer than other WUs and this is causing a lot of the heat and stress reports.

Is the additional heat emanating from the memory chips and not the GPU?

 

[Tobit]

I'm unable to confirm or refute my theory unfortunately and that, honestly, is all it is. I have no idea what is going on with these for some people. Unfortunately, there hasn't been a mass of people complaining about it to indicate there is a major problem. Hell, these even run fine for me on Linux via WINE.

 

[APOLLO]

I'm unable to confirm or refute my theory unfortunately and that, honestly, is all it is. I have no idea what is going on with these for some people. Unfortunately, there hasn't been a mass of people complaining about it to indicate there is a major problem. Hell, these even run fine for me on Linux via WINE.

They definitely run fine. I don't think I received one EUE yet, to my knowledge. However, temperature monitors are reporting higher temps, there's no question about it. It's tolerable from now until spring, there's no doubt. What concerns some of us is the warmer months later in the year. A lot of folders are running single-slot OEM coolers, and these aren't usually great at cooling even for regular WUs. I'm going to replace some of my single-slot G92 cards, but hoped it would be a more gradual phase out...

 

[Tobit]

However, temperature monitors are reporting higher temps, there's no question about it.

I agree. I do see a temperature increase of 5C on one of my cards. However, this 5C isn't going to be enough of an impact, for me, this summer to cause problems. Feed my some data with real numbers folks and I will pass it along. Just making comments like "these run too hot" generally fall on deaf ears at Stanford... I think their primary language is science before geek.

 

[APOLLO]

I agree. I do see a temperature increase of 5C on one of my cards. However, this 5C isn't going to be enough of an impact, for me, this summer to cause problems. Feed my some data with real numbers folks and I will pass it along. Just making comments like "these run too hot" generally fall on deaf ears at Stanford... I think their primary language is science before geek.

I'll try to get temperatures on my single-slots tomorrow. I'm not worried about the cards with heftier coolers. They always run cooler even with higher OCs.

I realize Stanford won't do anything about these WUs because of extra heat, and maybe if I was in their shoes I'd do the same. They need results because results produce answers, and that's the priority in the project. I'm just glad that my older cards can still process the newer WUs. I do hope that they continue what they have been doing and that is to alternate different types of WUs to give our hardware a bit of a rest from the additional stress.

 

[Tobit]

I do hope that they continue what they have been doing and that is to alternate different types of WUs to give our hardware a bit of a rest from the additional stress.

and hopefully these new projects just released from beta tonight will help with that a bit. Again, these new WUs should be fine.. these aren't very much different than the 787 pointers that have been out for a long time.

 

[Vaulter98c]

oo, I'm getting more than the 353s with these

 

[capreppy]

My GPU2 clients run in waves. I'll have either all 1888's, all 353's, or all 548's. It would ease heat issues if I could actually get a mix (offset each other). The 548's do run hotter. Based on Sunin's post, they suck up a hell of a lot of power as well!!!

I was monitoring my rigs more frequently when I was home for two weeks during the holidays, but now that I am back on the road, I can't monitor them as frequently as my client blocks out logmein.com (why I have absolutely no clue) so I can only check on them in the morning or evening from the hotel.