Overclock.net banner
1 - 10 of 10 Posts

·
Registered
Joined
·
513 Posts
Discussion Starter · #1 ·
Holy cow I've beat my Asus board to death and it keeps on ticking.

Trying to find my max sustainable core speed for the CC I've run into an interesting situation. My C2 955 at 241x16 (3856 MHz) at 1.52v is completely stable if I trust linpack, folding, and prime. CPU gets up to about 58C in linpack, a bit less in the others.

Issue in linpack... was noticing odd drops in the gflops so I hooked up to the killawatt. Idle is typical, 126 watts, but in linpack that rises to about 344 watts. Enter the random drops to 150! Start CPU-Z, and sure enough, it's dropping to x4 multiplier when it gets deep into the cycle.

Something is thermally throttling and I don't think it's the CPU.

So my question is, anyone remember hooking up an oc'd thuban to a killawatt and reading the results? If a 4 GHz thuban will stay under a 200W draw I may have to seriously think about upgrading.

Side note... I guess I'm about at the limit of 4-pin CPU power, eh?
 

·
Registered
Joined
·
1,200 Posts
Are you sure you turned all the bios options off,
that put the cpu slower when its idling?

Maybe its conflicting with high overclock?
you could try it
smile.gif
 

·
OG AMD
Joined
·
9,023 Posts
That Asus board has Anti-surge protection in hardware monitor.Disable it,does it help?
 

·
Registered
Joined
·
513 Posts
Discussion Starter · #4 ·
Quote:
Originally Posted by prescotter;13048443
Are you sure you turned all the bios options off,
that put the cpu slower when its idling?

Maybe its conflicting with high overclock?
you could try it
smile.gif
Will try. Have been running with CnQ to keep degradation at a minimum.
Quote:
Originally Posted by Redwoodz;13048560
That Asus board has Anti-surge protection in hardware monitor.Disable it,does it help?
Will look for it, but I don't remember seeing that option.
 

·
Registered
Joined
·
513 Posts
Discussion Starter · #5 ·
CnQ/C1E disabled. Does not go to x4 multiplier on idle at all.

There is no anti-surge option anywhere in BIOS. I checked every screen. If it matters I'm sitting on BIOS rev 2002. Hardware monitor only lists cpu and mb temp, cpu/chassis/power fan speed, vcore/3.3v/5v/12v voltage, and q-fan.

Ran the linpack again. Same issue. Falls to 163 watts/x4 multi/964MHz. This is odd... the idle wattage is 171 watts without CnQ before and after test. It's going lower than idle during the test long enough for the CPU to drop 4 degrees C to 54 before rising again.

I suppose I have two real options, use a lower clock with this prescott-wannabe, or use another CPU. I'm guessing that thubans use less power seeing some people stable at or near 4GHz with 1.45v or less. Even though this board has a heatsinked 8+2 VRM setup it seems to have a real power limit.

If a thuban's efficiency would help this issue, I may upgrade; if not, I'll wait for BD and just swap the lot.
 

·
...when you walked in
Joined
·
6,439 Posts
Quote:
Originally Posted by Iroh;13057196
CnQ/C1E disabled. Does not go to x4 multiplier on idle at all.

There is no anti-surge option anywhere in BIOS. I checked every screen. If it matters I'm sitting on BIOS rev 2002. Hardware monitor only lists cpu and mb temp, cpu/chassis/power fan speed, vcore/3.3v/5v/12v voltage, and q-fan.

Ran the linpack again. Same issue. Falls to 163 watts/x4 multi/964MHz. This is odd... the idle wattage is 171 watts without CnQ before and after test. It's going lower than idle during the test long enough for the CPU to drop 4 degrees C to 54 before rising again.

I suppose I have two real options, use a lower clock with this prescott-wannabe, or use another CPU. I'm guessing that thubans use less power seeing some people stable at or near 4GHz with 1.45v or less. Even though this board has a heatsinked 8+2 VRM setup it seems to have a real power limit.

If a thuban's efficiency would help this issue, I may upgrade; if not, I'll wait for BD and just swap the lot.
You posted on my thread so you know I have the same issue you have(Or had anyway) I don't know what I did that made it go away but I do know that I reset my bios to default because I was getting pissed and just wanted to start from scratch.. and now it just doesn't do it anymore. Also, I have an Anti-Surge Protection option in my bios and I'm also running an Asus board. Bios Revision 1702, different board though. Even if you had it, I don't think it would make a difference. Mine is currently enabled.. will check on reboot but I'm pretty sure its enabled.
 

·
Registered
Joined
·
513 Posts
Discussion Starter · #7 ·
Ah you have a newer version of my board. Mine has the 785G chipset on it.

I only have been having this issue since I tried to push past 3.8 GHz. Anything under that doesn't cause this behavior. These boards really may have a lower electrical current limit than the high-end ones.
 

·
Registered
Joined
·
513 Posts
Discussion Starter · #9 ·
It doesn't matter as much with my CPU as other AMD's it seems. When I said 58C in linpack, that was referring to the socket/CPU temp. Core temp at that time was 60C.

They're always within 2-5C of one another and they seem to get closer the hotter it gets. Core temp is always higher than CPU temp.

Honestly I'm not sure which one AMD wants me to look at - when the CPU is first loaded, the core temp goes up faster and the CPU temp catches up in a few seconds; when unloaded, the CPU temp drops like a brick and core temps stay warm for a bit (the CPU temp will drop below 40C before the core temp gets down to 50C). So really, it's a toss-up to what I should be looking at. If the CPU temp is in the socket, then how in the world does it cool off faster being so much farther away from the cooler? Plus it doesn't help that the answer from AMD itself depends on who you talk to there.

Back to the OT...

Linpack always returns stable results even when this weirdness happens. Only the time is extended so GFlops calculated goes down.

This never happened when I only had 2 GB of ram because a full test would only last 21 seconds! Now that the usable memory went from 1300MB or so to about 4800MB, and the pass takes 173 seconds, it gets a whole lot hotter. If I set at say 3000 MB, it sometimes won't happen at all. Definitely a temperature issue, but I'm really not sure if its the board or the chip. Oh well. I need an excuse for a thuban anyway.
 
1 - 10 of 10 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top