Painful days ahead for NVIDIA GPU

[EvilAlchemist]

Ohh .. this is just too funny .....

If some of you guys are just venting that is fine ... but some of you are just crazy ..

You have no idea what I have been going through on the ATI side ..

Dan Ensign & Mike Houston have been working their butts off trying to get the ATI client stable ... so I don' want to hear about you sudden EUE and quitting the project..

I have been getting EUE's for 4 weeks .... they just now figured out the cause and released a fix in beta....

I have been running WU on my cards for 1400 PPD ... yeah ...

If you think your 4000+ ppd cards are running "slow" get over it ....

I think Stanford is working as hard as they can to fix problems .. it takes time.
If you really wanna see what goes on .. go sign up for their beta testing.
I think then you will understand some of your misconceptions.

If you don't believe me, ask Sunin and Xilikon!
They are in there with me everyday posting about errors and trying tons of new clients before you ever see them. If you want to make it better, then help us test.

I know a lot of you know me very well, and i rarely get mad, but this had started to annoy me. Dan & Mike have pm'ed me , sometimes at very early in the moning, with solutions and ways to get my clients working, so don't ever say they are not working on problems or trying to screw over the community.

 

[nomad8u]

Ohh .. this is just too funny .....

If some of you guys are just venting that is fine ... but some of you are just crazy ..

You have no idea what I have been going through on the ATI side ..

Dan Ensign & Mike Houston have been working their butts off trying to get the ATI client stable ... so I don' want to hear about you sudden EUE and quitting the project..

I have been getting EUE's for 4 weeks .... they just now figured out the cause and released a fix in beta....

I have been running WU on my cards for 1400 PPD ... yeah ...

If you think your 4000+ ppd cards are running "slow" get over it ....

I think Stanford is working as hard as they can to fix problems .. it takes time.
If you really wanna see what goes on .. go sign up for their beta testing.
I think then you will understand some of your misconceptions.

If you don't believe me, ask Sunin and Xilikon!
They are in there with me everyday posting about errors and trying tons of new clients before you ever see them. If you want to make it better, then help us test.

I know a lot of you know me very well, and i rarely get mad, but this had started to annoy me. Dan & Mike have pm'ed me , sometimes at very early in the moning, with solutions and ways to get my clients working, so don't ever say they are not working on problems or trying to screw over the community.

Evil, first of all, if I've made you mad, I apologize. While not all of your post was directed at my comments, some appear to have been so I feel the need and reserve the right to respond.

First of all no one denies the fact that that the ATI guys have gotten jerked on the points since the NV client came out. I've commented on the fact a couple of times. But this (or rather my post) wasn't about the ATI client/WU, it was about the current release of the NV WU. You're absolutely right, I have no idea what you've been going through on the ATI side. But apparently you've been going through that as a Beta tester. All well and good as you have advanced notice and choice of whether or not you wish to submit to/participate in that.

Point IS, I didn't sign up as a beta tester, even though I chose to run the GPU client on the NV side as a Beta when it first came out. I fully expected and dealt with the problems (and yes reported some in the process) But the NV client was released and removed from Beta status. So are ALL the F@H clients to be treated as Beta even though they're listed as release? Not a positive process IMHO.

And as far as testing the WU before releasing them in a production environment, I find it hard to believe these WU on the NV side were "fully" tested by someone running other than stock cards before being released on the general "production environment" public without problems.

So this isn't about points or the decrease in points if you're referring to MY post. This is about the fact that if I knowingly run a Beta client/WU, I'll determine whether or not I'm willing to accept a 10, 20 30% or more failure rate (numbers for illustrative purposes only) and the corresponding waste of my time and power. The point is I'm doing it knowingly AND willingly and NOT because Dan Ensign or Mike Houston decided it'd be ok to wait until the work is being distributed to let someone know after the fact that these new WUs may cause some instability problems and may stress the cards more than had been previously thought.

The issue was avoided so they didn't have to hear the complaining about the points. I get it. But if down clocking cards could have potentially avoided the "Unstable Unit" errors I think a heads up would have been in order. The time, effort and power wasted in the process were just that. Not too much to ask IMO. If they truly didn't know, then again IMO they didn't test effectively. Either way, all my original comments stand.

However, I intended to piss no one off with my comments, so in the interest of team unity and harmony, I'll quit posting regarding this. I do truly apologize if I upset you as that wasn't my intention. I have too much respect for you to do that. Hence the reason I wanted to make these final comments public. Even though I've been with the team a short time (I'm sure you can appreciate THAT) I have too much love and respect for the members to let this sink to a pissing match.

Peace

 

[EvilAlchemist]

Evil, first of all, if I've made you mad, I apologize.

No need for that .... i never take anything said here personal.
We are a team , and allowed to have different opinions.
Keep up the good work and post as often as you want .....

you just better keep running Folding@Home

 

[Xilikon]

nomad8u and EvilAlchemist :

Like EA said, I'm in the closed beta testing so I saw the effect of those WU before. I made a lot of reports about those WU, including the EUE at the end of a WU which is irritating (this happen consistently with 2x 8800GS, not with the rest so it's probably caused by a bad driver). I told them I cannot stand this instability anymore so I pulled them off beta but in a twisted timing, they decide to release it to the public at the same time. IMHO, it's a few days too early but they seems to cannot wait to get the new results of those units.

The only thing is pray that they don't send too much of those units (right now, I get a 50/50 mix of old and new units) and to have Scott release the new changes asap.

 

[aldamon]

Evil, first of all, if I've made you mad, I apologize.

Don't let him fool you. Apologies will not be not enough. You'll need bail money, too.

 

[Sunin]

Good news is not a single EUE on any of my machines from the new WUs!!! So all I can say is test, retest and put alot of fans in your cases both front and back. I average 6-10fans per case, well worth it for the worry free effects!

 

[Wheresatom]

I too have been trying to go over my logs looking for EUEs with no problems.

I fear that this might become the straw that breaks the camels back for some. This new WU thing really hasn't been a super big deal. If it were not for some of the previous history of the project, i don't think we would be hearing much dissent.

 

[drgnfang]

Based on the weird point totals I'm seeing for a last couple of updates I think my boxen started to EUE early this morning. It'll have to wait until I get home before I can look at it though.

 

[jws2346]

Hmm..., Many people are getting a little disgusted and distrustful with the direction of F@H (GPU1 fiasco, constant expirations with a short previous client expiration and then time extentions and then a later model of the same sorry client scheech, etc). Now if a person vents his/her anger at the people controlling F@H and with Stanford seeming to ignore the old "the customers always right" deal (that sayin's been around forever and volunteers are something like customers) If you don't believe that some of the Stanford management don't give a "rats patoot" about it's volunteers (I said some, not all) , just read some of the sarcastic posts by 7up and Kotow ( err..., I meant 7im and Totow) If a person living in a "free society" wants to switch to another DC type program that's just "tough doggy poo poo" for the "higher uppers". The interesting thing is Stanford should try and get their "poop in one bag" and be more particular in their testing methods concerning new WU's and/or clients (once again the GPU1 fiasco, the SMP 2.22 bullcrap, etc) and try to release a thoroughly, in the wild, tested WU or client before the presently used client expires or you're issued new crumby WU's (you know, like give more warning and more time before the current client expires or tell you to downclock your OC'ed GPU card for the new WU's, d'oh )

I ain't sayin' I'm a' gonna' leave F@H. ATM I have two quads doin' the Windows GPU2 client and their CPU's are doin' the WCG client, but until Stanford straightens out a little, like getting an already Linux capable GPU2 DC client released (thanks Shellnut2 et al) or gets better at "quality control" and stops the release of "alpha" WU's/clients I'm goin' be very leery of anything coming out of Stanford in the way of information (their forum, etc) or outa' the mouths or posts of any F@H fanbois/fangals.

Further more, anyone that doesn't like what I've tried to get across in my own personal opinions or is of the mentality "Stanford can't do anything wrong" they can just "puff on a big ole' donkey peter"

OBTW, I don't wanna' hear the "they're doin' the best they can" bullpoop because that tells me their best just ain't good enough.

Man, it's simply amazing how "politically correct" I'm getting (censoring my swearing)

DC for who you want and with what you've got

 

[Tigerbiten]

From Dan over on the folding forums about the new work-units being buggy.
Quote ...........

"That's what I'm thinking. I'm pulling these new guys for some additional analysis. We've seen the simulation temperature climb which suggests some instability in the integrator* that is a problem for more atoms.

*The integrator is the thingy that makes it go. Kind of like a car engine."

So maybe its back to the test protien for a bit more free bonus points.

Luck ...........

 

[Pooch666]

Speaking of drivers, the 177.35 is stable on Vista Ultimate 64, anything later is not. On Vista Ultimate 32, the latest drivers are perfectly stable. This is on my machines, can't speak for others. Practically everyone overclocks at least the shaders for Folding. Where was Stanford? The forums are full of information on overclocking. Didn't they know????????

 

[Xilikon]

Actually, Vijay made more comments about that. He think that the check mechanism to make sure units is done in a valid fashion is a bit too strict, hence the big amount of EUE even if it isn't really a EUE. This compounded by the big increase in temp made them pull those units for the time being until a new core is available to fix those issues.

 

[HighYield]

Even if everyone overclocks which certainly doesn't happen, how would the Stanford group be able to reasonably predict how their WU will work on X number of different MOBOs, Y different GPUs, Z different power supplies, and N different cooling solutions all with variable levels of OC and extra voltage applied? Wouldn't the best we can hope for is the WUs work ok on stock boxen? Maybe an actual programmer could comment on this (I am not...unless A/C Basic on an Amiga 20 something years ago counts )

 

[metallicafan]

Ohh .. this is just too funny .....
Dan Ensign & Mike Houston have been working their butts off trying to get the ATI client stable ... so I don' want to hear about you sudden EUE and quitting the project..

I have been using the ATI client since day one. It has been one of the easiest installing, easy configuring, and most stable client I have ever used. Props to Dan and Mike for a job well done.

I think Stanford is working as hard as they can to fix problems .. it takes time.
If you really wanna see what goes on .. go sign up for their beta testing.
I think then you will understand some of your misconceptions.

Lets not blow this out of proportion guys. Evil is right. The guys at Stanford are surely working hard on this. Its in their best interest to do so. Even well tested software projects have bugs when they come out. If they didn't software patches wouldn't exist.

I think we all need to remember that the [H]orde is a force to be reckoned with. We are more or less donating a supercomputer to medical research. The more WUs we tear through the closer we get to cures. Points drops are community wide and uncontrollable on our part. There is only one way to battle points drops: Add more processing power to the [H]orde!!

Fold Hard!

 

[Kendrak]

There is only one way to battle points drops: Add more processing power to the [H]orde!!

Fold Hard!

Truer words have rarely been spoken.

I added a 9800GX2 tonight to make up for the drop in points on the other 9800GX2.

 

[PRIME1]

face it the point system on F@H has been broken since day 1 and they dont really care to find a good solution because there is no "perfect" solution

The points are not the main goal of folding anyway (the points are for fun).

Although I must say I just fired up my GTX260 and

It's da only way ta fold.

 

[FLECOM]

The points are not the main goal of folding anyway (the points are for fun).

Although I must say I just fired up my GTX260 and

It's da only way ta fold.

your post literally made me

how i read your post:

1) points don't matter
2) man all these gpu points are awesome!

half the threads in this forum prove that points DO matter... nobody would buy $5000 in video cards just for FAH if points didn't matter, be real man, don't lie to yourself

i haven't, which is why I'm focusing my efforts on WCG now

 

[PRIME1]

1) points don't matter

To me finding a cure for these diseases is why I run the program. How selfish of me.

2) man all these gpu points are awesome!

I did say the points were fun. However it's not why I run Folding@home.

half the threads in this forum prove that points DO matter... nobody would buy $5000 in video cards just for FAH if points didn't matter, be real man, don't lie to yourself

To each their own I guess.

 

[jws2346]

your post literally made me

how i read your post:

1) points don't matter
2) man all these gpu points are awesome!

half the threads in this forum prove that points DO matter... nobody would buy $5000 in video cards just for FAH if points didn't matter, be real man, don't lie to yourself

i haven't, which is why I'm focusing my efforts on WCG now

Very well put FLECOM

While I agree with the science part being the most important in the final goal of finding a cure (IMO the people that contantly bring it up sound more noble ), IMHO the point system makes the effort for finding a cure (it sometimes takes a long time and we're only helping) with any DC type program (F@H, WCG, etc) is the point awards. The awards for some people (I said some people, myself included ) are what makes the long and tedious proccess of finding that illusive cure much more competitive and fun.

I really find it hard to believe anyone would spend huge amounts of money, huge amounts of time, sometimes the DC'ers comfort (heat, custom space, beefed up electrical circuts, extention cords, etc) and sometimes his/her personal relationships are threatened (ie wife, husband, parents, children, GF, boss, etc), to work and post harmoniouly together (different sexes, different interests, different backgrounds, different nationallities, etc) toward a common goal on any DC program (F@H, WCG, etc) and do all of the above without trying to kill each other (dipshite wars) IMHO there's just no way in hell it could be done without the point awards.

Oh yeah FLECOM, I like that "nobody would buy $5000 in video cards just for FAH if points didn't matter" statement and AFAIK true.

DC'n with what I've got