What I don't understand is that the Ati cards are running this project right now...Wouldn't it make more sense to start the new GPU project with them since they are already proven to work and add the Cuda in the near future ?? All your problems between gpu and cpu users are solved..........Doing it this way you are causing more problems because you are going to be adding more users (nvidia card) before you move the ati cards to the new project.......

You are probably right. The only problem is that I don't have the code for the MilkyWay_gpu project to do the ATI port (a new application will be required for MW_gpu).

As long as there is no application (and no WU for it and a working website), ATI cards can't start on the new GPU project ;)

I'm afraid they know exactly that.....servers won't be stressed. Plenty time to fine tune Cpu side and Nvidia GPU side.

Time will tell..

---

What I don't understand is that the Ati cards are running this project right now...Wouldn't it make more sense to start the new GPU project with them since they are already proven to work and add the Cuda in the near future ?? All your problems between gpu and cpu users are solved..........Doing it this way you are causing more problems because you are going to be adding more users (nvidia card) before you move the ati cards to the new project.......

You are probably right. The only problem is that I don't have the code for the MilkyWay_gpu project to do the ATI port (a new application will be required for MW_gpu).

As long as there is no application (and no WU for it and a working website), ATI cards can't start on the new GPU project ;)

I'm afraid they know exactly that.....servers won't be stressed. Plenty time to fine tune Cpu side and Nvidia GPU side.

Time will tell..

Time is telling that we have a problem for both CPU and ATI-GPU here in that we can't get work even though, apparently, there is plenty available. I would have thought that some resources both from the project and that volunteered by Cluster Physik should be going into addressing this, eg, migrating ATI to MilkyWay_GPU. But instead all resource seems to be going into something that doesn't help current crunchers.

---

I don't think that is the case -- I believe the GPU project will need a different (new) application to run. Cluster made the ATI application to run on MilkyWay off of the .19 base. The GPU application will have a different start point.

Since Cluster has made the ATI app, that will be carried over and Travis is doing the CUDA app, so both will work.

---

Your RAC is drooping.

---

me@rescam.org

Your RAC is drooping.

Mine has drooped by 50% since 4 weeks ago or so. Then it was 150,000+ - now it's just 75,000

---

Yep same here, I'm down from 3300 to 2000. Oh well hoping they get the other site up soon, that will help out all? I hope. ;-P

---

Well folks. Before you can run the script to make the project, you have to build the boinc system. Which, to say the least, is not pleasant on FreeBSD systems. Anywho, to update on the status, Travis is working on the GPU app. He has been working on it for the past 2 weeks. He has to translate the entire code from C into CUDA, which obviously isn't trivial. He told me the other day that he's got some functionality out of it but it's not yet complete.

A BOINC project can align a program with CPU, and if the user enables CUDA access then the CUDA app is used instead (like with SETI). If ATi cards are going to be used, then a CPU enabled app would have to be offered on the MW_GPU site at some point, otherwise only CUDA enabled devices would get WUs.

Plus it also means that you can get some results with ATi cards while the CUDA app is being developed; Cluster Physik is doing an awesome job and I bet he'd love some code to work on :)

Now, back to my own CUDA work...

---

I don't think that is the case -- I believe the GPU project will need a different (new) application to run. Cluster made the ATI application to run on MilkyWay off of the .19 base. The GPU application will have a different start point.

Since Cluster has made the ATI app, that will be carried over and Travis is doing the CUDA app, so both will work.

I had thought that it was going to carry over, but I see it isn't now.

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

Any kind of official news on this would be very much appreciated. Please! Over two weeks without a single piece of news is way too long.

... Over two weeks without a single piece of news is way too long.

However, far from unusual in BOINC world. Sadly. I have seen this happen over and over again on various projects.

---

This is the only project where the Admins normally do keep the crunchers up to date. This GPU angle, using CUDA rather than Cluster Physics ATI client, it the first time there has been silence for so long.

I think they must have a range of priorities which are taking precedence, as well as the CUDA client transfer taking longer than thought.

Mind - all this is pure speculation on my part.

---

Go away, I was asleep

This GPU angle, using CUDA rather than Cluster Physics ATI client,

Do you have some inside knowledge that nobody else does?

Last I heard they were going to support both GPU apps.

The project to develop their own for CUDA, and still support third party CAL.

---

.

Yawn.....

Yawn.....

Yea...


haha... :p

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

Naked Alarm Clock

100% kid friendly...

My understanding (supposition mostly) is that the CUDA client (at least to start) will be done by the RPI folks, and that the ATI client will be developed by contributors here, but ONLY AFTER they get the Milkyway_GPU application release to work with. Until then, the ATI GPU effort is going nowhere. For that matter, at a guess, when the first CUDA application shows up from the RPI folks, I will be very surprised if it is not in need of serious optimization.

This GPU angle, using CUDA rather than Cluster Physics ATI client,

Do you have some inside knowledge that nobody else does?

Last I heard they were going to support both GPU apps.

The project to develop their own for CUDA, and still support third party CAL.

---

So it looks like I just keep on doing 'normal' MW on my Radeons with Cluster's client for a while... and GPUGrid on my 9800s.

---

Cheers,

PeterV

.

So it looks like I just keep on doing 'normal' MW on my Radeons with Cluster's client for a while... and GPUGrid on my 9800s.

I expect I'll be doing the same, except I only have just the one 9600.

---

SNAFU

---

4870 GPU
4870 GPU

Any new news anyone about how MW GPU is developing?
Did Travis inadvertently created a Black Hole in which he got sucked in and disapeared from the surface of the earth?

Martin