F@H SMP - How does it fair as stabilty testing tool these days?

[Assimilator1]

Early this year I used F@H to stabilty test my quad rig (& ran it for a while after just to crunch too ), I found it would fail where P95 would pass for 24hrs ,not only that but it failed pretty quickly too! :Q ,anyway I dropped the FSB & raised the vcore & then it ran F@H fine.
My point is it was a quick, reliable & stringent stabilty tester, is that still the case with the latest clients? I thought I read somewhere around here that people had started to get EUEs for a while even though nothing had changed on the rig & then the EUEs just dissappeared again (IIRC). Was that with some now expired clients? Are EUEs currently only due h/w errors or can client errors still(?) cause it?

I also plan to compare it to OCCT if F@H SMP is reliable, anyone compared the 2 for stability testing?

Btw for those who are interested, my Q6600 was running at 3GHz (F@H stable) in my MSI Nvidia 650 mbrd, currently its running at 3.35 GHz in my new Asus P5Q Pro (P45) mbrd , though I've yet to properly test it. It will go higher too but temps get a bit on the high side.

 

[Drsignguy]

I have tested all my rigs using OCCT, Prime and Intel. Each of those put out more heat than the SMP clients that are loaded. Each Quad rig I own (3) uses 2 clients + a Gpu. The E7200 uses 1 smp and a Gpu and what I have found is that it was the Core ( smp ) in the program, not the H/W that causes the EUE's. Just lately, probably within a month, I have been testing my Q6600 ( not in sig ) and it has shown a few EUE's. I thought it was the Rig but after a while, it sent the results back, loaded another client and then it has been working flawlessly since. I know it was stable but I wanted to confirm and I was right, it was the client that has caused the problems.

But to reassure your comment about stability, Yes, Smp clients are very demanding and if your Rig(s) are NOT stable, it will pick it up pretty quickly. Just info from what I have picked up on since using Smp's.

 

[Foxery]

OCCT is a better test for a new rig due to two main caveats:
1) OCCT compares its calculations against known answers, so it catches errors immediately. F@H is computing unknown data, and it may take hours for enough bad data to build up to a crash.
2) F@H also has work units which are bad by nature, and will fail on any machine. i.e. It will give you false positives.

 

[Assimilator1]

Ah ok, so my vague recollections were right then, you can't entirely trust F@H alone for testing. Although I didn't know bad WUs could cause problems too.
I will still run it for a bit after my rig passes a 24hr OCCT test.

Thx for the info guys.

 

[Insidious]

If fah doesn't crash over a long time, then you are pretty safe to call it a stable setup, but like the posts above this one indicate, the opposite doesn't necessarily hold true.

-Sid

 

[Assimilator1]

Hey hi Sid , where abouts are you these days in F@H?

 

[Markfw]

Originally posted by: Assimilator1
Hey hi Sid , where abouts are you these days in F@H?

In third. right behind me !

 

[Insidious]

Originally posted by: Markfw900

Originally posted by: Assimilator1
Hey hi Sid , where abouts are you these days in F@H?

In third. right behind me !

I prefer to think of it as right in front of the guy in fourth. :laugh:

HiYa right badk at ya AS1

-Sid

 

[Assimilator1]

lol And nice going you being right near the top!

Hey Mark
Just the man , what vcores have you got your Q6600s at?
Currently mine's at ~1.34v loaded running DPAD/SETI @ 3.35 GHz.
At 3.4 GHz it's going to need at least 1.35v (maybe more), though it's starts to get a bit warm then, running OCCT the 2 hottest cores peak at 67C (with ambient of 25C). Though that is around 5-6C hotter than DC programs get it.
Also am I right in saying the def vcore range for Q6600s is 1.25-1.3v?

 

[Markfw]

well, I just looked at the one I am posting from, 1.36@3.33 (just took it down, as I just got a BSOD) My vid is 1.275, and current temps are 37-45c (its cool in this room right now, 65f?)

 

[Assimilator1]

Yeah that's a bit chilly isn't it? (IIRC 68F is 21C which is 'normal' room temp).

What vcore have you set your 2 Q6600s @ ~3.5GHz to?