More SMP Problems

[Wheresatom]

Ok, so I have been working on Overclocking my computer recently(to no avail). This has involved a bit of shutting down and looking around in the BIOS, and trying to get myself familiar with it. Upon re-starting my computer i saw that I had downloaded a new WU and re-started in the SMP. Is this right? Shouldn't it pick up where it left off?

To be honest I am not real happy with the turn off/on cycle of my computer. I do not know what has happened but it is acting funny.

 

[APOLLO]

The SMP client is highly sensitive to OC, to such an extent that no other application may notice it. Shutting down the client can crash the WU, prompting a new DL from Stanford that will invariably be the same WU issued to you again. My advice is to run the system stock for the duration of a single WU just to be certain there are no other issues affecting your system besides clock speed.

Nice badge, BTW.

 

[budec]

What I've found out (the hard way, by trial and error) is that if the SMP client has problems (it seems like any little hiccup that can occur) it'll delete the current work unit and start "fresh" (by downloading a new one).

This happened to me a couple times so far.

 

[Wheresatom]

Thanks for the quick replies. I actually haven't overclocked anything yet. I am not sure if my particular board just has a confusing BIOS or what, but I can't seem to figure out some of the settings, so that is slow going, and I have little time.

Is there a correct sequence to shutting down my computer? Should I go into task manager and stop the service or should I just go to start and shut down?

 

[SmokeRngs]

Running the SMP client as a service is not recommended and this is one of the reasons why. As a service, it doesn't always shut down properly. If you're running a single SMP client in Windows, the only "good" way to shut it down with little risk is by hitting CTRL-C when the terminal window is active. Then you need to wait for a bit to make sure all the cores have stopped running. At that point it's safe to restart the machine. The Windows client seems to take longer to shut down the cores than the Linux client. At least, that's how it used to be. I haven't checked it out lately.

 

[Wheresatom]

Excuse me for a stupid question, but how do I run the SMP in a terminal window? How would I open that now that I have it set up as a service?

 

[Tigerbiten]

If your running as a service, your best bet is to backup the whole folding folder before you stop the service/shut down.
That way if it does fail to restart properly, You can stop the service, replace the work folder & queue.dat file from the back up, restart the service.
Look in this thread by Killer on how to auto-backup your work.

Luck ..............

 

[Wheresatom]

Could me running the SMP as a service and not shutting it down correctly have anything to do with the problems I have been having mentioned in this thread?

I don't see how they could be related, but the problems started at the exact same shut down, so who knows.

 

[Wheresatom]

Ok, so I reset the CMOS and the restart issue went away. I have turned in over half of my WU's with this computer on stock, and 3 or 4 since I went Vista.

I have entered a new FSB and re-started. FahMon shows my log as saying
"extra sse boost ok"

Then nothing. No 0% completed or anything. FAH is still showing up in task manager 4 times, taking all 100% of my processor to the cleaners, but I don't think it is actually computing anything that is going to do me any good. The FahMon window has a yellow square at the top left corner. Here is my log as of now.

Launch directory: C:\FAH
Service: C:\FAH\fah
Arguments: -svcstart -forceasm 

Launched as a service.
Entered C:\FAH to do work.

Warning:
 By using the -forceasm flag, you are overriding
 safeguards in the program. If you did not intend to
 do this, please restart the program without -forceasm.
 If work units are not completing fully (and particularly
 if your machine is overclocked), then please discontinue
 use of the flag.

[16:08:41] - Ask before connecting: No
[16:08:41] - User name: Wheresatom (Team 33)
[16:08:41] - User ID: 527BB9385EEE6437
[16:08:41] - Machine ID: 1
[16:08:41] 
[16:08:41] Loaded queue successfully.
[16:08:41] - Preparing to get new work unit...
[16:08:41] + Attempting to get work packet
[16:08:41] - Connecting to assignment server
[16:08:41] - Successful: assigned to (171.64.65.64).
[16:08:41] + News From Folding@Home: Welcome to Folding@Home
[16:08:42] Loaded queue successfully.
[16:08:42] - Error: Attempt #1  to get work failed, and no other work to do.
             Waiting before retry.
[16:08:47] + Attempting to get work packet
[16:08:47] - Connecting to assignment server
[16:08:48] - Successful: assigned to (171.64.65.64).
[16:08:48] + News From Folding@Home: Welcome to Folding@Home
[16:08:48] Loaded queue successfully.
[16:09:00] + Closed connections
[16:09:00] 
[16:09:00] + Processing work unit
[16:09:00] Core required: FahCore_a1.exe
[16:09:00] Core found.
[16:09:00] Working on Unit 08 [March 14 16:09:00]
[16:09:00] + Working ...
[16:09:00] 
[16:09:00] *------------------------------*
[16:09:00] Folding@Home Gromacs SMP Core
[16:09:00] Version 1.74 (March 10, 2007)
[16:09:00] 
[16:09:00] Preparing to commence simulation
[16:09:00] - Ensuring status. Please wait.
[16:09:17] - Assembly optimizations manually forced on.
[16:09:17] - Not checking prior termination.
[16:09:22] - Expanded 2445951 -> 12916317 (decompressed 528.0 percent)
[16:09:23] - Data doesn't match checksum.
[16:09:23] - Starting from initial work packet
[16:09:23] c
[16:09:23] Warning:  check for stray files
[16:09:23] - Starting from initial work packet
[16:09:23] s
[16:09:23] - Starting from initial work packet
[16:09:24] 77)
[16:09:24] 
[16:09:24] ect: 2653 (Run 31, Clone 69, Gen 77)
[16:09:24] 
[16:09:25] .
[16:09:25] Entering M.D.
[16:09:25] ations on if available.
[16:09:25] Entering M.D.
[16:09:33]  files
[16:09:33] Extra SSE boost OK.
[16:09:33] iting local files
[16:09:37] Extra SSE boost OK.

EDIT:

I am dumb and impatient. Ignore this. All is well. Lets just see if the SMP runs all night (thus making my current overclock stable).

 

[Sunin]

You probably should test it via prime95 version 25 for an hour or so that you don't cause errors in the WU. Monitoring temps via coretemp version 97

Also run memtest to ensure your memory is not at issue.

 

[Kendrak]

You probably should test it via prime95 version 25 for an hour or so that you don't cause errors in the WU. Monitoring temps via coretemp version 97

Also run memtest to ensure your memory is not at issue.

That will help.... but there is no true equal for folding stress.

Well maybe folding and playing Supreme Commander at the same time

 

[Wheresatom]

Ok, I feel a little sheepish guys. I went to bed last night before reading Sunin's post. I folded all night though, no hang ups.

I went from 2.13 GHz to 2.66 GHz. I left the voltage on auto. The board has set that to about 1.3 volts. When I left for work this morning I was plugging along at 54°C still on the stock HSF and bedroom at 22.2°C (72°F).

I guess if it has been running for about 9 hours on folding it should be pretty stable right?

One question is if the voltage (read on CPU-Z) is 1.3 volts, is that before or after Vdroop? If I wanted to go into the BIOS and manually try to set that down a notch or two to squeeze some temperature out of the OC would I want to set it to around 1.3 to start with and work down? Or am I going to have to start a little higher than that.

 

[Kendrak]

It is after Vdroop (90% sure)

I have my mobo set to 1.38, but CPU-Z show a hair over 1.3

 

[Sunin]

Ok, I feel a little sheepish guys. I went to bed last night before reading Sunin's post. I folded all night though, no hang ups.

I went from 2.13 GHz to 2.66 GHz. I left the voltage on auto. The board has set that to about 1.3 volts. When I left for work this morning I was plugging along at 54°C still on the stock HSF and bedroom at 22.2°C (72°F).

I guess if it has been running for about 9 hours on folding it should be pretty stable right?

One question is if the voltage (read on CPU-Z) is 1.3 volts, is that before or after Vdroop? If I wanted to go into the BIOS and manually try to set that down a notch or two to squeeze some temperature out of the OC would I want to set it to around 1.3 to start with and work down? Or am I going to have to start a little higher than that.

Don't feel sheepish if it worked all night your probably safe... I primed mine for 1hr and then double smp'd it and ran it over night like you did having coretemp monitor temps and did a find on the log highest it hit was 55c so I said good go for it... most it hits during the day is 57... so your fine... relax and fold [H]arder

 

[Wheresatom]

I have a Zalman CNPS 9500 that I will be putting on next week. I wanted to get a good perspective of what the improvement was so I am going to run this OC for a few days and keep a log of temperatures, PPD, clock speed etc. I started this log about a week ago so I have a little bit of data started just for comparison sake.

Maybe after I put the Zalman on I can push it a little higher.