I'm just curious what rule of thumb you guys have adapted towards folding with an OCd box. I'm currently doing the bulk of my folding on my main rig, it's an NF7-S and Barton 2500+ running at 210 FSB (2.3 gigs) on watercooling. I've been using the -forceasm flag, but if it's somehow corrupting the results, I certainly don't want to do that. How do you guys guage when and when not to use that flag. Also, what indications should I be looking for when verifying that WUs are completed correctly?
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
-
Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.
Folding and overclocking
- Thread starter cgrant26
- Start date
Tigerbiten
[H]ard|DCer of the Month - February 2007/January 2
- Joined
- Sep 24, 2002
- Messages
- 5,028
Run Memtest86 to check that your memory is stable.
The only good result from that is no errors in around three passes.
After that it a quick run of the Prime95 tourture test.
If you pass both of them then your not overclocked far enough......
The flag I use on all my boxen.
It just there so that if the system crashes or shuts down due to a power cut then F@H restarts with the SSE boost on.
As for the client stopping early.
I expect to get less than 1 every 100 work units due to Stanford releasing new work units from beta testing that dont work in general use.
Look at your FAHlog.txt file.
It should tell you the core status as each work unit ends.
Folding@home Core Shutdown: FINISHED_UNIT
CoreStatus = 64 (100)
Luck.............
The only good result from that is no errors in around three passes.
After that it a quick run of the Prime95 tourture test.
If you pass both of them then your not overclocked far enough......
The flag I use on all my boxen.
It just there so that if the system crashes or shuts down due to a power cut then F@H restarts with the SSE boost on.
As for the client stopping early.
I expect to get less than 1 every 100 work units due to Stanford releasing new work units from beta testing that dont work in general use.
Look at your FAHlog.txt file.
It should tell you the core status as each work unit ends.
Folding@home Core Shutdown: FINISHED_UNIT
CoreStatus = 64 (100)
Luck.............
Dark Ember
[H]ard|Gawd
- Joined
- Jul 30, 2001
- Messages
- 2,021
Basically, if I'm overclocking any signifigant amount, you want to make sure that you can Prime95 torture test for at least 24 hours before you start folding. Why? Because that test will tell you whether you are doing floating point math correctly at that speed. If your computer is not passing prime, then you could be sending bogus data back to stanford, which means that they'll will have to send the same unit out again for someone else to do.
BillR
Born Again Cynic
- Joined
- Feb 17, 2002
- Messages
- 18,535
Very often the difference between stable and unstable is a one-clock cycle change to the FSB. I doubt you will notice the change.
I too have a Shuttle, cute little bugger it is too. Mine has the heat pipe cooling system. I made one simple change to the case. I cut out the back fan grill and added a sucker fan to the internal blower fan. Dropped 8c off my temps and now I run a Barton 2500 as a 3200, totally stable. I used a simple case fan on the backside and it added no noise.
Rule of thumb..Over clock till she quits, then back it down a notch or two
I too have a Shuttle, cute little bugger it is too. Mine has the heat pipe cooling system. I made one simple change to the case. I cut out the back fan grill and added a sucker fan to the internal blower fan. Dropped 8c off my temps and now I run a Barton 2500 as a 3200, totally stable. I used a simple case fan on the backside and it added no noise.
Rule of thumb..Over clock till she quits, then back it down a notch or two
If you can run Prime95 for 24 hours on your overclocked settings then you are completely stable. It is the best CPU torture test out. If you have a weak spot in your setup, it will let you know. To bad it cant tell you the hardware that failed!
After passing Prime95 full folding ahead!
After passing Prime95 full folding ahead!
Carnival Forces
Supreme [H]ardness
- Joined
- Feb 9, 2003
- Messages
- 4,297
oc it as far as you can (testing along the way w/ Prime95 for about 2 hrs each)
once you reach the level where you get errors, go back to the setting before it, and test Prime95 for 24hrs
no errors = good to go
once you reach the level where you get errors, go back to the setting before it, and test Prime95 for 24hrs
no errors = good to go
If FAH detects too much deviation, it will stop working on the current unit and you'll get partial credit for it.
I was overclocking my Athlon64, or at least trying to push it past 2100, and at around 2300 FAH shut down. I am keeping it at 2100 for now, it has always been stable at that speed.
I was overclocking my Athlon64, or at least trying to push it past 2100, and at around 2300 FAH shut down. I am keeping it at 2100 for now, it has always been stable at that speed.
Dark Ember
[H]ard|Gawd
- Joined
- Jul 30, 2001
- Messages
- 2,021
Buckus said:If FAH detects too much deviation, it will stop working on the current unit and you'll get partial credit for it.
I was overclocking my Athlon64, or at least trying to push it past 2100, and at around 2300 FAH shut down. I am keeping it at 2100 for now, it has always been stable at that speed.
Thats why you want to prime first. There could be some deviation to give bad results, but not enough for the core to detect, in which case you would be giving bad data back to Stanford without the core crashing.