New Beta Client v7! - ATI folders rejoice!

[lowfat]

So I am playing around w/ this for a bit since I am having a whole lot of VMware issues at the moment. Somes things I've noticed. I currently have two ATI clients going at the moment, it causes my computer to be completely unuseable. I get HUGE stuttering every ~ 20 - 30 seconds each of which lasts ~ 5 seconds. Each client is maxing out one of my 2600k @ 4.7Ghz threads, so around 15% CPU usage. Ass or PPD I have no idea since we can't use HFM.net to check PPD as all the work folders are kept in the same place now.

[Keywork]

What cards are you running Lowfat? And the only way to currently calculate PPD is to calculate your TPF from your logs. Let's hope we get some monitoring software updates in the next little while! (I'm looking at you HFM lol)

[lowfat]

I just did some math and if my calculations I am getting a whopping 4600PPD!! This is w/ unlocked HD6950s @ 975/1250. So it makes mey computer unusable and I get crappy PPD to boot.


Conclusion: AMD/ATI still sucks the big one at folding.

[LCB001]

Couple of interesting items;

The new and improved ATI FahCore-16 is not out yet, this is only a new client not new Cores.

VijayPande wrote:

ATI core 16 will be coming out shortly, hopefully this week.

This was posted by Bruce

I just want to highlight a few common questions that a lot of people will have:

• This is a new client, not new FahCores. With the exception of FahCore_16, you'll still get assignments of the same projects and will earn exactly the same PPD.
• FAHViewer only works with FahCores which produce a current.xyz file, so for most people it will show a demo protein.
• Third party monitoring programs such as HFM and FahMon and FCI have not yet been updated to work with V7.
• If you configure a Windows service in Win7/Vista, GPUs cannot fold. (Limitation by Microsoft).
• New WUs will be downloaded not long before the current WU finishes; That WU will start processing while the previous results are being uploaded.
• A "slot" is a new concept probably best defined by an example. If you have been running several V6 clients on a single OS, you will now be running one FAHClient with several slots.
• You will probably never use a command-line parameters. With FAHControl, you can change your configuration as needed.
• The more complex your installation, the more you'll need to read the documentation.

A lot of people seem to be having issues with the new FahViewer especially in Windose so I would avoid using it.

Stanford has opened the bug-tracking system for read-only access if you want to see if your problem is being addressed here; v7 Client, Open Tickets Ordered by Milestone and Priority

[sswilson]

New WUs will be downloaded not long before the current WU finishes; That WU will start processing while the previous results are being uploaded.

If nothing else comes out of the new client, this will make it all worthwhile. (No more losing PPD because the return server is out/laggy).

[Keywork]

Originally Posted by sswilson

If nothing else comes out of the new client, this will make it all worthwhile. (No more losing PPD because the return server is out/laggy).

This is one thing I've been really looking forward to, and so should every bigadv folder out there. The time lost on every WU to actually send it back creates a huge inefficiency in the whole process. The issue is only compounded by shitty upload speeds...

[SugarJ]

Has anyone set this up with bigadv?

[chrisk]

Originally Posted by SugarJ

Has anyone set this up with bigadv?

I got a bigadv on the old client. Its at 48%, but I'll ditch it to try v7 bigadv.

I'm going to give this a shot. From what I have read, bonus is not calculated on v7 viewer, so I'll take note of project/ppd with and without bonus for some comparison. Also, I think the PPD bonus calculators out there will still apply with frame times so a manual calculation should be no biggie.

[chrisk]

No luck so far. It can't get work from an assignment server, and getting a packet 'read error' message.

23:07:42:Server connection id=1 on 0.0.0.0:36330 from 127.0.0.1:54661
23:08:41:Connecting to assign3.stanford.edu:8080
23:08:41:News: Welcome to Folding@Home
23:08:41:Assigned to work server 171.64.65.54
23:08:41:Requesting new work unit for slot 00: READY smp:12 from 171.64.65.54
23:08:41:Connecting to 171.64.65.54:8080
23:08:42:ERROR: Exception: Failed to read response packet

[JMCD]

This is as far as my client gets.

23:13:04:WARNING: Failed to get assignment from 'assign-GPU.stanford.edu:8080': Empty work server assignment
23:13:04:ERROR: Exception: Could not get an assignment

No firewall running, windows is allowing it. Tried a reinstall, default folders and running as admin. Any ideas?

Edit: Nevermind, no idea why it's working now but it is. I can hear my 5870s squealing away.

Cores seem to be pulling about 12% cpu usage each. Overall system usage on my i7 920 @ 4.0 is 30%. Is there an easy way to add the SMP client into this bundle? With the current SMP even set at low priority it's bogging the cards down to 60-70% load when they're set higher.