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. We are using CAL now (ATI's hardware abstraction layer) and that seems to make life a lot easier, and also should make running a GPU client a lot easier from the point of view of donors, as the driver issues and complexities should now be resolved. We are still looking into an NVIDIA client. The NVIDIA GPUs are very different to program, so a port isn't a simple thing to do. We are looking into this, though.
SMP core: right now, SMP on Linux and OSX is behaving fairly well, whereas Windows is giving some issues. This is perhaps not a surprise, since the SMP code must use MPI, which has its origins on UNIX and is a newcomer to Windows. We have been working with Windows MPI developers to improve the situation, but they tell us this isn't a simple fix. Since we are in the business of studying proteins, not writing MPI libraries for Windows, we will wait until the MPI experts improve the Windows MPI before we make any claims of improvement there.
Finally, beta clients will be expiring soon, and we are in the process of QA for new clients. We will also extend the expiration deadlines in the future clients to give some more time, and since the clients are appearing to be maturing.
Go on, we're waiting for a new client more than for a second coming!
Posted by: Galahad | January 25, 2008 at 01:14 PM
Has anyone checked/fixed the issues denying the 32bit FAH SMP client and cores for Linux?
Posted by: Ivoshiee | January 25, 2008 at 02:13 PM
Glad to hear it! Many of us bought new Radeons around Christmas that we're dying to add to the fight. :) A client still in beta is fine by me.
One question - can you comment on CPU usage generated by the new GPU code? Does the main processor need to perform part of the calculations, or can the newer GPUs accomplish most/all of the crunching on their own?
Posted by: Tim | January 25, 2008 at 03:41 PM
Does the use of CAL help to speed up the calculations or does it hinder performance instead?
Posted by: Ivoshiee | January 25, 2008 at 04:47 PM
The single biggest issue I'm looking forward to seeing resolved is the Windows SMP client stalling out if network connections are changed/lost. This pretty much excludes wireless-based home networks. I believe this is an MPI issue, is there any hope of this being resolved in the next Windows SMP client?
Posted by: Clint | January 25, 2008 at 05:12 PM
When can we expect an Nvidia client?
Posted by: Hooty | January 26, 2008 at 01:44 AM
Hey Hooty, I don't think this post is saying you can expect it at all, not in the true sense of the word at least.
Anyway, I can't wait for the new GPU client. This Windows SMP client is giving me real a headache!
Thanks for the update, they are appreciated. Keep up the good work =)
Posted by: ruisu | January 26, 2008 at 08:48 AM
If i read this 'between the lines' the Nvidia client will NOT be ready for a VERY long time..
I really hope i'm wrong here.
Yeah nvidia, come on, the G80's are waiting to help science !
Posted by: Peter | January 27, 2008 at 05:25 AM
Can't wait to see how the R600s perform on the new GPU client, they seem to be designed for massive scale. Would it be possible to run both GPU client and SMP client on one dual core machine? It would be great if SMP and GPU clients were merged, it would probably speed up the research tremendously.
Posted by: Ibrahim | January 28, 2008 at 01:31 AM
Hahahah, the 3870 X2's are out. Amazing performance, stellar text, and, uh... we need to ask the Letterman Question: Will It Fold?
Posted by: J Willaford | January 28, 2008 at 12:11 PM
I'm excited to hear that the new client is running on the new ATI cards pretty well, but today the new 3870 X2 cards were released and was wondering if that card was included in testing or not.
Posted by: Shawn | January 28, 2008 at 12:51 PM
Shawn, in my opinion 3870 x2 will fold, but I guess it will be necessary to disable xfire mode on them. and you will need dual core cpu at least for them, unless CAL helped them to overcome this issue :)
anyway, great news, and we are waiting :)
Posted by: muziqaz | January 28, 2008 at 02:04 PM
Regarding:
"The single biggest issue I'm looking forward to seeing resolved is the Windows SMP client stalling out if network connections are changed/lost. This pretty much excludes wireless-based home networks. I believe this is an MPI issue, is there any hope of this being resolved in the next Windows SMP client?"
I am running the Windows SMP client with a home wireless network, and while it's given me some problems, I wouldn't say it's impossible.
The problem you are experiencing is probably coming from when you exit range of your wireless router and then reenter range. Also I think that this is probably more likely to happen if you exit range for a substantial period of time (based on anecdotal evidence). So if you can at least manage to stay put a bit or at least stay in range (or out of range) of your router for an extended period of time, it should be workable. (If you move around a whole bunch, though, then maybe the problem is unfixable.)
Also, I'm guessing you've already discovered this, but all you really need to do is pull up Task Manager and make sure the processes are running when you have finished moving. You can easily do this by clicking the CPU column to sort by CPU usage (the FahCore_a1.exe processes will be towards the top). If not, close the console window by opening it and hitting Ctrl-C, wait a bit for cleanup, and restart.
Posted by: Stephen Dewey | January 30, 2008 at 06:00 AM
"We will also extend the expiration deadlines in the future clients to give some more time, and since the clients are appearing to be maturing."
Is that the expiration date of the client itself, or the expiration deadline of the WorkUnit?
Posted by: Adrian | January 30, 2008 at 09:09 AM
Is there ANYthing being fixed by downloading and using these new Client versions ???
Why isn't there a version history with fixes list like there are with most programs and their new releases ???
Posted by: noorman (team 734 Mod) | January 31, 2008 at 02:10 AM
So, when are we going to stop working with binaries produced in March 2007? The SMP client has *critical* bugs that have existed for more than a year, an ETERNITY in this field. This kind of stale software doesn't give me high hopes for the scientific side of this process. Are the people who use the data generated by these clients just as.. stale?
Posted by: Firestarter | February 02, 2008 at 03:24 AM
@ Firestarter
Sorry - *critical* bugs? What are these?
I'm afraid I don't call 10 months in beta an ETERNITY either...
Posted by: Adrian | February 04, 2008 at 11:09 AM
@ Firestarter
Sorry - *critical* bugs? What are these?
I'm afraid I don't call 10 months in beta an ETERNITY either...
Posted by: Adrian | February 04, 2008 at 11:09 AM
@ Firestarter
Sorry - *critical* bugs? What are these?
I'm afraid I don't call 10 months in beta an ETERNITY either...
Posted by: Adrian | February 04, 2008 at 11:10 AM
I have problems with 6.01beta3. 1 of 10 startup attemts it fails. its important to me couse im folding at night in d other room. Without monitor, cant see if its on or not.
Posted by: Paul | February 19, 2008 at 05:15 AM