No whining.

[relic]

Other teams will do whatever they want, it's not our problem. If they turn in bad data they'll get deleted eventually, but I'm betting the data is fine and nothing will happen.
I'm sure they are going to run modded clients for multiple NVidia GPUs even if you don't like it. Just fold, like we've always done.

Get your [H] GPUs folding.
Have an Intel Quad and want to effectively double it's output? Add a cheap 8800GT and a GPU client.

In 6 months there will be something different to fret about, but for now it's GPU folding. Get 'er done.

 

[digital_exhaust]

Get 'er done.

/Glenfiddich to you relic.

Your absolutely right..... it's easy to get a little bent out of shape right now, but there's no need to....just do what you all do so well, and all well be good.

Fold On....and beers and tamales to you all

 

[Mitsimonsta]

/me digs Relic in the ribs, hands him a vegemite-laden Fosters*

All good guys.

 

[nomad8u]

/me passes Fosters around....

I stole some while reconning the enemy..

Whoops.. who let Mitsi in here..

All good indeed. Nice post relic!

 

[BillR]

Hey, just like the old days

Everything old is new again

 

[FLECOM]

im almost a little annoyed at how powerful the GPU client is, why should i maintain 100 machines at 400ppd with the console client when i can get one 8800gtx and do 4000ppd?

almost dosent seem worth the effort anymore

 

[The[H]ardenator]

im almost a little annoyed at how powerful the GPU client is, why should i maintain 100 machines at 400ppd with the console client when i can get one 8800gtx and do 4000ppd?

almost dosent seem worth the effort anymore

Thats is because the GPU is more suited to this kind of processing compared to the CPU

 

[SpoogeMonkey]

im almost a little annoyed at how powerful the GPU client is, why should i maintain 100 machines at 400ppd with the console client when i can get one 8800gtx and do 4000ppd?

almost dosent seem worth the effort anymore

Because the smp client will be out of beta sometime this decade

Maybe

 

[Jon855]

im almost a little annoyed at how powerful the GPU client is, why should i maintain 100 machines at 400ppd with the console client when i can get one 8800gtx and do 4000ppd?

almost dosent seem worth the effort anymore

Install a 8800 GT on every rig you have and win, win!

 

[relic]

im almost a little annoyed at how powerful the GPU client is, why should i maintain 100 machines at 400ppd with the console client when i can get one 8800gtx and do 4000ppd?

almost dosent seem worth the effort anymore

I completely understand.
It's a lot of effort, and only you can determine if i's worth it to you.
With me, your efforts will always be appreciated, no matter how you do it.

 

[Kendrak]

im almost a little annoyed at how powerful the GPU client is, why should i maintain 100 machines at 400ppd with the console client when i can get one 8800gtx and do 4000ppd?

almost dosent seem worth the effort anymore

For the science part, you still doing a heck of a lot of work!

Put some GPU's in where you can to keep the farm "current" but keep it going if you can.

Fold on!

 

[tedkj]

im almost a little annoyed at how powerful the GPU client is, why should i maintain 100 machines at 400ppd with the console client when i can get one 8800gtx and do 4000ppd?

almost dosent seem worth the effort anymore

I hear ya, however if it really is so that the points are distributed fairly according to the amount of work, well then .. it's just evolution.

/me wonders how long until Stanford releases a 3000x shader FoldingCard for PCI-Ex ;p

 

[FLECOM]

Install a 8800 GT on every rig you have and win, win!

LOL i wish

if you can find an 8800gt that fits in a low-profile dell and can run off the craptastic dell power supply, lmk lol

w/e im working on refurbrishing/redeploying my efforts

but as it seems the SMP client will be forever beta i will have to continue to use the console client...

 

[Crazy Larry]

LOL i wish

if you can find an 8800gt that fits in a low-profile dell and can run off the craptastic dell power supply, lmk lol

It's not a 8800gt, but it's the next best thing...a low profile 8600gt

 

[APOLLO]

It's not a 8800gt, but it's the next best thing...a low profile 8600gt

What does the 8600GT get PPD-wise?

 

[jws2346]

You hit the nail right on the head Mr relic.

It all boils down to what your personal beliefs are, if you feel comfortable with runnin' "modded clients for multiple NVidia GPUs" so be it, if you don't so be that also. There's no reason to "piss and moan" (whining) about it, just fold. IMO the bottom line is once the smoke clears, even if you have a "zillion" pernts it's all about finding a freakin' cure for some very !@#$%^ diseases.

Another thing, I don't personally know whether the dual gpu folding produces bad data or not (I don't think it does). If it does it'll "all come out in the wash" anyway (be discounted), Stanford ain't exactly a bunch of dummies.

 

[relic]

You hit the nail right on the head Mr relic.

It all boils down to what your personal beliefs are, if you feel comfortable with runnin' "modded clients for multiple NVidia GPUs" so be it, if you don't so be that also. There's no reason to "piss and moan" (whining) about it, just fold. IMO the bottom line is once the smoke clears, even if you have a "zillion" pernts it's all about finding a freakin' cure for some very !@#$%^ diseases.

Another thing, I don't personally know whether the dual gpu folding produces bad data or not (I don't think it does). If it does it'll "all come out in the wash" anyway (be discounted), Stanford ain't exactly a bunch of dummies.

hehe.
Since it hasn't been mentioned yet in this thread... (how did we miss this golden opportunity?)

All you [H]ardasses need to: shaddup and fold.
However we will accept a slight modification due to OCAU's ramp.

Shaddup and fold...MORE.

"Shaddup and fold"
copyright 2001, [H]ardOCP
all rights reserved

 

[lelliott731]

Since the science is the most important part for me, I'm not touching that setup till Stanford does it themselves, because to waste cycles on data that could be invalidated would suck.

FLECOM, I feel your pain, I've been working on setting up all these boxes to try & compete, and once the nVidia multi-GPU client comes out officially, with an Asus X48 board with a Q6600 CPU & four 8800GTS 512MB you can get 20,000PPD from that machine alone. If you loaded it up with four 9800GTX+'s, you'd get 24,000PPD. That's a heck of a lot of Q6600 boxes all by themselves, and a whole heck of a lot of 400PPD machines.

 

[lelliott731]

Now if only cases were big enough for 4 GTX280's, you could have a single box producing 35,200PPD with a single Q6600. And that would be the same producition as ALL of my boxes put together.

I wonder if the points are going to even out soon. Because like Stanford says, the GPU can only do a very select amount of science, whereas CPU's can do any type of science.

 

[Jon855]

Double post much?

 

[lelliott731]

Yeah, I had an issue, so since I haven't yet been able to figure out how to delete a post, I just added another thought But you caught me Jon855!

 

[Jon855]

Yeah, I had an issue, so since I haven't yet been able to figure out how to delete a post, I just added another thought But you caught me Jon855!

Wait... That was no double post
It must be that god-darned Spike drinks.

 

[lelliott731]

No, it was a double post, but I went back and edited the second post and added another thought I had. So don't worry, the Spike drink isn't killing you yet!

 

[Jon855]

No, it was a double post, but I went back and edited the second post and added another thought I had. So don't worry, the Spike drink isn't killing you yet!

Oh smart. Now I know I was not seeing double, thank god at least for that. Now I'm waiting for this cancerous drink to mess me up.

Sorry for the derailment.

 

[APOLLO]

Now if only cases were big enough for 4 GTX280's, you could have a single box producing 35,200PPD with a single Q6600.

Are 280s larger than other nVidia cards? If four other dual-slot cards would fit, I think four 200-series should also fit. I have at least three cases that could accommodate that many cards, but no motherboard that has more than two 16x PCI-E slots. Not that I would even think of purchasing a GTX280 at such ridiculous prices when for the same 'investment' one can acquire 3-4 8800GT cards for double the production (or more).

And that would be the same producition as ALL of my boxes put together.

Points production perhaps, but it's questionable if it would be equivalent in science production, which is more important. It's one of the reasons I believe Stanford will normalize the points structure like they did to some extent for the SMP client.

I wonder if the points are going to even out soon. Because like Stanford says, the GPU can only do a very select amount of science, whereas CPU's can do any type of science.

I wouldn't be surprised, actually, I'm expecting it. Everybody jumping full bore on the nVidia bandwagon might be in for a bit of a surprise later. Stanford doesn't yet know how much validity there is in the results to date, and probably their rationale behind the evaluation of the GPU2 WUs on the high side. Everything is very new and very beta.

If you compare the value of the defunct GPU1 WUs (330pts), the GPU2 WUs are worth approximately 30% more. Once the GPU2 infrastructure is fully in place (read attract sufficient players) and Stanford has a better idea of the science it can produce, we may see a reevaluation of the point structure on the order of 20% or greater. Would it then be worth running that many GPU2 clients and the associated power consumption concerns?

Already there's reports in the FCF that some people have seen a significant drop in PPD with just the latest client release (b8). Thus, barring the release of a new GPU architecture, I imagine we have already witnessed the acme of production for GPU2.