End of GPU1 folding? [Yes]

[road-runner]

No, if its not doing anything to help find cures cut the chord is the way I see it...

[RAH]

I really can not see any reason to complain. Which is something I love to do.
I assumed from one of your first post Mr. P. that the GPU1 was dead.
Though some said "NO", including mods, I just knew it was to be.
The same with the QMD client. As was explained, IIRC, as it was put, no one there to carry on.
If it was a major break though, someone would have been brought in to carry on.
But wasn't it a start to the SMP client? It was SMP aware correct?

Things do go on. I still use P4s for SMP folding. As long as I can keep the DL in line, I figure it OK.
I know, that one day, things will change. At my age, it happens to fast some times.

I still say, all in all, I think you people are doing a great job. Having to handle all the different
aspects of running the program. Do wish that the Stanford IT team, would get all the servers
working better, but we can't have everything. Or can we!!

No matter, I wish you and the team, the best of luck, with all new clients, programs, etc.

One thing. You have your blog. Use it to the best for the contributors, so at least they
can think, that what they do matters. We know it does. But some don't.

Happy MEMORIAL DAY! From a VET.

[toTOW]

ok, thanks for the feedback.

Would anyone else like to see GPU1 continued under the current circumstances?

You can turn it off ... the only board I used to fold (x1650) is now retired as I replaced the CPU with a dual core which runs SMP

[Nonymoussurfer]

I'm philosophically against the idea of running it for the sake of running it....

I feel the same way. Pull the plug. I wasn't aware that all GPU1 associated data had been deemed bunk. From the info I'd previously read, I had assumed that some useful science would be salvagable from all that GPU1 crunching. Such is the nature of research I suppose. You win some, you lose some. In a pure research capacity, one might expect to fail 99% of the time... but that remaining 1% makes it all worthwhile. Of course, I believe just the widespread beta release of the GPU1 client was a success in and of itself-even if only to prove that it was the wrong way to GPU fold.

On another note... There's too much moaning & groaning going on here. It is a privilege to be able to contribute to such a grand and noble endeavor. Perhaps some day (perhaps some day soon), we'll be able to talk about how we were a part of the research project that discovered why people get alzheimer's/cancer/etc... and finally showed medicine a way to go about effectively treating it.

Points are pay? You do realize you still can't buy a cup of coffee with all those points. As a silly protest of this silly idea, I've decided to fold under the names of some of my teammates for awhile.

No sense folding for naught though. I will go ahead & turn off my remaining GPU1 crunchers until I can come up with some newer hardware. I'm overdue for some equipment shuffling anyways. I'll soon have a 2nd GPU2 cruncher up & running.

[John Naylor]

There was supposed to be a paper coming out based on the GPU1 research... but even if that is no longer the case, I think that all the work done on it was worthwhile as the springboard to the new GPU2 client... I just hope my 8600GT is fast enough to run the nvidia version

[Ivoshiee]

There was supposed to be a paper coming out based on the GPU1 research... but even if that is no longer the case, I think that all the work done on it was worthwhile as the springboard to the new GPU2 client... I just hope my 8600GT is fast enough to run the nvidia version

I am sure that it has not changed and there will be a paper based on GPU1 work.

[maarten_vaandrager]

I wouldn't feel as good about folding If i would doubt its usefulness.
Please make those GPU clocks count and I'll keep crunching!

[Oldhat]

It's a shame to see the GPU1 client shutdown, but I was only running mine until the current work units were completed.
If there is no scientific value to be gained though, I would say shut it down.

I may be in a minority here, but I was under the impression that SMP, GPU and PS3 all returned different work units
of different value to the project. I decided that I wanted to fold on each of these platforms due to that reason.

Personally I felt from the hints/comments that Professor Pande was dropping in his blog in January, there was a chance
the new client wouldn't support the newer cards.

http://folding.typepad.com/news/2008/01/index.html

Here are some code development updates on some important client/cores

GPU core: we've got the GPU core running in house and we found and fixed some bugs in our QA stage. We're now
continuing QA to see if we find any more bugs. Right now, the GPU core is running on all new ATI cards, so we're
excited to roll it out.

[TriWheel]

I feel Flathead74 speaks for a lot of donators. I bought a total of six 3870s in December (two for my brother), after the November announcement of an imminent
GPU2 client. Over the next four months of waiting for GPU2 I ended up selling all but two of them, at a great loss to me and great gain to the Chinese economy. This month I ended up picking up two more Diamond 3870s after seeing how well my original 3870s where finally doing. One.day.later I find out that the 8800 series will fold soon, very soon, thus removing the last thing holding me back from buying superior NVidia videocards. Add this to the stomach churning notion that all the electricity and cooking away during the last summer and all 600,000 GPU1 points earned have basically been discovered to be a total write off should tell you how I feel about the project right now without getting into expletives and colorful metaphors.

So now I am forced to offload my useless X1900XT CF set and my X1950XTX CF set for less then the cost of one 3870, while knowing all they did was burn a couple of uranium candles and produced lots of heat. It's not like I kept them so long because I like to stay old school. 8800GTX launch buyers are now my unqualified heros, I wish I had joined that club instead of this expensive ATi folding endevour.

[VijayPande]

PS Thanks to Oldhat for reminding me of this post
http://folding.typepad.com/news/2008/01 ... opmen.html
where I did mention we're using CAL for GPU2/ATI and mention the existence of the GPU2/NVIDIA project. I know it's not much, but there wasn't much else to say at that time, and hopefully it shows that we have been talking about this for a while (the post is 5 months ago).

[alancabler]

... all they did was burn a couple of uranium candles and produced lots of heat...

That statement is clearly inaccurate, but can be explained by human nature, I suppose.

Thank you, Professor Pande.
This is exactly the type of information sharing from Pande Group that donors hunger for and appreciate when given.

If anyone can conceive of a better resolution of the GPU1 endeavor, then by all means post your idea.

[Flathead74]

Points are pay? You do realize you still can't buy a cup of coffee with all those points. As a silly protest of this silly idea, I've decided to fold under the names of some of my teammates for awhile.

Yes, points are pay, among their other uses.
I'm not the only one that feels this way, either.
Perhaps you do not fully understand the different ways in which the points system motives folks to participate.

As far as you condemning the idea as "silly", you are entitled to your opinion just as everyone else is.
Please, enjoy your protest.

I have been partaking of the privilege to help with this project since 2001; How about you?

[7up1n3]

No, if its not doing anything to help find cures cut the chord is the way I see it...

Yeah, all annoyances aside - this is the bottom line. We're here for the science. Whilst I'd certainly prefer a solution that allows for GPU1 to be scientifically relevant ... if it isn't it isn't. Goodbye and thanks for providing an assist in the development of GPU Folding.

[silveradocyn]

I think the timing of the GPU1 shutdown was very unfortunate. It would have been better if the news about issues with the science of the GPU projects had been known to the community sooner. I bought 3 different cards for the GPU1 project this year. I made the assumption that although they may not generate the points of the delayed GPU2 client, they would still be usable, and I wanted to help the project.
Now I have a XT1900 I bought used, but never got into production, a XT1800, and an AGP 1650, all collecting dust. As I bought these for a philanthropy, I would gladly donate them to a philanthropy if I knew they would be used somewhere.
If I would have known there was an issue with the GPU1 client, I would not have spent "a bunch" of money on these cards this year, and instead would have been able to invest in faster processor, or an additional HD3870.
By timing the news of the GPU1 client with the release of the GPU2 client, it looks like the Pande Group may have been with holding the information from the donor community, until the new client was out, unfortunately some of us were mis-spending funds in that period.

Most of us understand the science and exploration of the project is in multiple disiplines. Although the the end product is in the biological areas, there is also significant computer science work being done. Including how to expand the distributed computing, and how to use distributed GPU processing for a project like this. Exploration frequently finds dead-ends. Most of us understand that. It just feels like the community could have had better information that would have been a benefit to the project and the doners, that there may have been a brick wall approaching on that path.

- - Silveradocyn

[bruce]

Personally I felt from the hints/comments that Professor Pande was dropping in his blog in January, there was a chance
the new client wouldn't support the newer cards.

http://folding.typepad.com/news/2008/01/index.html

Here are some code development updates on some important client/cores

GPU core: we've got the GPU core running in house and we found and fixed some bugs in our QA stage. We're now
continuing QA to see if we find any more bugs. Right now, the GPU core is running on all new ATI cards, so we're
excited to roll it out.

At the time that he said that, I understood that the new client might not support the older cards, but I believed that the GPU1 client was still doing scientifically useful work. From that perspective, I believed that FAH would continue to support the GPU1 client for the x1900 cards and that the GPU2 client would provide additional features for the HD2xxx (and beyond) cards. FAH can still accept scientifically valid results from a PII running Windows 98 and the same would probably be true for the CPU1 client -- if the science was actually useful.

I do not understand the nature of the problems with the DirectX approach or what it does to the results, but I do applaud Stanford for taking the extra effort to rewrite the interface to use CAL/CUDA.

Science is often like that. You propose a theory (that FAH can use GPUs by way of DX) and you spend a lot of time to prove your theory. You observe mixed results. You propose a new theory (that FAH needs CAL/CUDA to exploit GPUs) so you spend that much more time to prove that theory -- and it turns out that it also disproves the earlier theory. Now that's real progress.