O/C X2 4400+ : Getting nowhere

[Markbnj]

I think it has a lot to do with components other than the CPU. The motherboard, PSU, and ram, for example.

It may be that my chip would run stable at lower vcore if I got the ht and dim voltage just right. Not sure. I will play with it some more this evening. I am pretty happy, though, with 2.53ghz @ 1.475. If it is stable there I probably won't try to push it higher. I may tighten up the ram timings a bit though.

 

[onn2000]

It looks like it's your RAM. Either that or the memory controller of the CPU just doesn't work with such RAM speeds very well.

 

[Markbnj]

Yeah, it was running all day while I was at work, and when I came home I found it had rebooted and had a system error in the event log. Had to lower the ram timings back down for now 🙁/

Not sure where to go with it at this point. Gonna do some testing at 2.53 ghz with the divided ram speed and see if it is prime stable. If so I can fool around with the ram and try to get it stable.

 

[Pabster]

Have you tried giving the RAM some more juice?

Some of the high-performance modules are rated as high as 2.9 factory and I believe OCZ is rated at 2.8?

Edit: Oops I see you are using Corsair, not sure why I thought OCZ 😛

Regardless, try going to 2.8 or 2.9 and see if that makes it stable.

 

[Markbnj]

Ok, I messed around with it all night again, and I think I may have finally found a stable combination:

HTT : 250 mhz x 10 (2500 mhz) @ 1.475 vCore
HT Link: 4X
DIMM mult.: 166 (208 mhz) @ 3-3-3-8 (SPD)
All other voltages stock

Idle: 42c
Load (p95 x 2, small FFT): 65c

Crystalmark: 50117

I like this, even though it is a hair slower than my 230 x 11 setup, because it allows me to run the HT link and DIMMs at full rated speed. Fully loaded temp is still higher than I would like, but I ain't replacing the Zalman for 42c. The AS5 hasn't fully cured yet either.

I'll be damned if I can stabilize it one iota faster than this though. I tried everything.

Edit: I did try boosting the vDIMM as far as it would go, was getting 2.8v, no luck with Prime 95. Same thing with HT link voltage. I tried different multiples, including 9 x 270, 9 x 280, and 10 x 260. I wish this thing had higher multipliers, because I have a feeling 12 x 220 would be sweet. But in general this setup doesn't seem to want to run faster than 2.5-ish ghz.

 

[Velk]

Originally posted by: Markbnj
I like this, even though it is a hair slower than my 230 x 11 setup, because it allows me to run the HT link and DIMMs at full rated speed. Fully loaded temp is still higher than I would like, but I ain't replacing the Zalman for 42c. The AS5 hasn't fully cured yet either.

I'll be damned if I can stabilize it one iota faster than this though. I tried everything.

Edit: I did try boosting the vDIMM as far as it would go, was getting 2.8v, no luck with Prime 95. Same thing with HT link voltage. I tried different multiples, including 9 x 270, 9 x 280, and 10 x 260. I wish this thing had higher multipliers, because I have a feeling 12 x 220 would be sweet. But in general this setup doesn't seem to want to run faster than 2.5-ish ghz.

That actually mirrors my findings with the same memory. No amount of voltage made any difference to memory stability.

I would be pretty concerned about that load temperature though - I certainly would not run it over 60 myself. You might consider dropping it to 220x11 with 1:1 memory and stock vcore. It's unlikely you will be able to notice the difference in anything other than benchmarks, and should result in significantly lower temperatures.

Up to you of course.

 

[onn2000]

The X2s maximum temps according to AMD is 65 degrees celcius. Don't push it, please! Either that or get a better cooler. Is your case open?

 

[Markbnj]

Nah, case is closed. It only hit 65c when under full load from two copies of Prime 95. I will wait until the AS5 is cured and see where it ends up. I also put another fan in the top of the case tonight, but it isn't having a huge effect. I'm using a Zalman 7000cu, which is what CompUSA had when I got the bug. Maybe I should put a big typhoon in there.

Oh, and I forgot to set the ram command timing back to 1T from when I was testing:

Crystalmark: 52361

You know, the fact is, even running at the stock 2.2 ghz this box is way fast, and although I do notice the 300 mhz difference it isn't _massive_. But I just can't stop myself from trying to make it run faster 🙂. I was more worried about voltage anyway, but I'm under 1.5 so feel pretty decent about it. If I can play BF2 and keep the cores under 55c I'll feel relatively secure, at least until the AS5 is cured.

 

[seanp789]

Mark, you will get better memory performance at stock ddr400 or below, with 1T and 2-2-2 timing. at DDR460 and timing that loose it would not be worth it, you would have to be pushing ddr530+ to make those timings worth while.


Also, I have a Big Typhoon, I would highly recommend it or the SI-120. You will see your load temps drop at least 10C.

Originally posted by: Markbnj
Here's where I ended up tonight after a long evening of tweaking...

cpu: htt=230mhz x 11 (2530mhz), 1.475v
ht: 4x, +.01v
dimm: 1:1, 230mhz, 3-3-3-7, 2T, +.01v (2.70v)

idle = 40-42c
load (p95 x 2, small FFT) = 64c

Not great temps, but the AS5 is still curing.

CrystalMark: 51889

Seems stable, but I only ran the last iteration of Prime95 x 2 for 30 mins. I'll exercise it more tomorrow night.

 

[Markbnj]

Hi, you're saying I should reduce the timings? I'm running at SPD now (3-3-3-8) at 1T. What would you try?

 

[kyp275]

what he's suggesting is for you to use a memory divider to get the ram running at a lower speed, but with tighter timings.

only way to find out which runs faster would be to bench each settings, though personally I don't think there will be much of a difference in real-world app.

 

[Markbnj]

Ah, ok I misread that. The memory is running pretty damn close to stock now, 208 vs 200. Timings are SPD, but I didn't think of slowing it down just to try tighter timings. Might be worth a shot to see what happens.