Question on thermal power, stepping and architechture

[grey_1]

Title says it all. I want to learn more about exactly what I'm tinkering with by overclocking my 4400. I know the how, but my question is this.

My 4400 has (according to spec sheet) an E4, E6 stepping, at a stock speed of 2200 mghz. The x2 4800 is an E6 @ 2400. Both use 1.30/1.35 v.

What exactly does that E4, E6 stepping mean? And if I'm running at 2.4, is my 4400 essentially the same as a 4800? And the voltage..does 1.30/1.35 mean it may operate anywhere between the 2 out of box, or is that required voltage, or is that safe operating voltage? Almost forgot, would the thermal power go to the same 110 as well?

Thanks in advance for answers, I know these are n00bish questions, but google is turning up anything but what I want, and I like to know exactly how I busted my hardware. :D

thanks guys.

[J-Mag]

Yeah a 4400 running at 2.4ghz is essentially a 4800.

I think it was with the Venus line that AMD decided to release the same chip with different voltages. You can check to see what yours is running at with CPU-Z. http://www.cpuid.com/cpuz.php

I have run 1.55-1.6 for extended periods on my 90nm based A64.

[grey_1]

Yeah a 4400 running at 2.4ghz is essentially a 4800.

I think it was with the Venus line that AMD decided to release the same chip with different voltages. You can check to see what yours is running at with CPU-Z. http://www.cpuid.com/cpuz.php

I have run 1.55-1.6 for extended periods on my 90nm based A64.
Thx for the reply J-Mag. So far it seems super stable. i had it at 2500 for a few days, but I started getting weirdness, and wary of the voltage increases. But that was a hefty bump on yours, and I see some in others sigs. Maybe I'm being a bit paranoid.:p

Thanks again.

[J-Mag]

Thx for the reply J-Mag. So far it seems super stable. i had it at 2500 for a few days, but I started getting weirdness, and wary of the voltage increases. But that was a hefty bump on yours, and I see some in others sigs. Maybe I'm being a bit paranoid.:p

Thanks again.

You shouldn't be to wary of voltage increases under 1.5v, but sometimes higher voltage actually causes instability. Although it is more than likely to create more stability as long it is being cooled effectively.

I do know what you mean though as a specific overclock can work fine and then one day it becomes unstable. Sometimes this is because your cooling is becoming less effective (dust buildup, summer, etc..). Also some programs are more sensitive to specific subcomponent overclocks. It is also possible that a processor's transistors are experiencing electromigration and the chip could be dying. Although I doubt the later for your case because you haven't really pushed the voltage.

Recently I haven't even been overclocking because I hate crashes far more than the few fps I get over my stock 2.2ghz.

[Xion X2]

On my prior build w/ the X2 4400+, I had a Prime95 stable overclock at 2.65gHz by increasing the voltage to 1.5v. I cooled the chip w/ an Arctic Freezer 64.

From everything I've read/heard, you're pretty safe at 1.5v and under as long as your temps are good.

[grey_1]

1.5 it is then, if needed. I'm not even sure my instability was from lack of voltage or too tight timings or both. But you guys just gave me a dose of confidence for trying. I'm hoping for a stable 2.6 soon. Just 260 mghz (as of right now) gives a noticeably snappier machine. I like!:D

[J-Mag]

On my prior build w/ the X2 4400+, I had a Prime95 stable overclock at 2.65gHz by increasing the voltage to 1.5v. I cooled the chip w/ an Arctic Freezer 64.


Even if your overclock is prime stable, it could still be unstable in other apps, especially if you test with only one FFT type.

I prefer 3d apps because they load the whole system, CPU, GPU, Mem, various buses, etc...

[Xion X2]

Even if your overclock is prime stable, it could still be unstable in other apps, especially if you test with only one FFT type.

I prefer 3d apps because they load the whole system, CPU, GPU, Mem, various buses, etc...

That may be so, but in all the time I've primed systems I've never had an unstable one that went stable in Prime for 6hrs +. I would rather test it with Prime first before opening up all sorts of applications to test it with, because it can corrupt your files if you don't have a stable overclock. (or are you talking about a program like Orthos? Maybe I'm misunderstanding you here.)

I'll also run the Windows Memtest to make sure the RAM's stable as well.

[J-Mag]

That may be so, but in all the time I've primed systems I've never had an unstable one that went stable in Prime for 6hrs +.


Really? I have had it happen to me on several occasions.

What were you using Small, Large, or mixed FFTs?

Maybe there is one other factor we aren't looking at: Patience. I don't think I have ever let prime go 6hrs unless I passed out or something... LOL

Usually a couple three is max for me.


I would rather test it with Prime first before opening up all sorts of applications to test it with, because it can corrupt your files if you don't have a stable overclock.


Corrupt files FTL!

Well when you have primed for a while then you play a game and it keeps crashing you sorta loose the desire to waste time with prime (unless you set it while you are away or something)


(or are you talking about a program like Orthos? Maybe I'm misunderstanding you here.)


Nah. I use popular current gen games as my main testing methodology. Although I have dorked around with orthos.

[SlieTheSecond]

I agree with prime95. I've never had a problem with anything that prime95 never gave me an error for.

On the other hand I have benchmarked fine and then get prime95 errors after only a few minutes of it running.

[grey_1]

Whats best to use for prime? I just guessed blend was the harshest test.

[Xion X2]

Really? I have had it happen to me on several occasions.

What were you using Small, Large, or mixed FFTs?

Maybe there is one other factor we aren't looking at: Patience. I don't think I have ever let prime go 6hrs unless I passed out or something... LOL

Small FFT's, which burns the proc the most. And I usually let it run overnight and get a good 6-8 hours in.

Usually a couple three is max for me.

Yeah, I don't think that's enough. A good 6-8 should do the trick. Again, I've never had a machine that primed for 6-8 hours go unstable on me, including my current build. It hasn't crashed abnormally (other than normal Windows bugs, which of course happen occasionally) or gotten memory errors or disk read errors or anything else.

In fact, when I first build a system is when I look for my maximum overclock. I'll raise the speed/voltage and prime it, then rinse and repeat until I've reached the max I'm willing to go w/ it. Then after that is over, I'll do a fresh reformat of Windows. And if you have an unstable overclock at that point, Windows most likely won't install right. You'll get stupid IRQ errors and missing file errors all through the installation (this has happened w/ me before if I didn't prime stable first.)

Corrupt files FTL!

Yeah, you don't want those, which is why I prime it for a while first. If it doesn't fail under that intense testing (nothing else you run will heat your proc up or stress it as much, including gaming), then you're not going to get disk read errors, memory errors, IRQ errors, and all that fun stuff.

[Xion X2]

Whats best to use for prime? I just guessed blend was the harshest test.

Don't use blend. Use small FFT's and keep a close eye on your temps, because it will really heat it up.

All blend does is use memory; it doesn't stress your processor that much. Also make sure that you run two instances of prime from different locations on your drive. You'll need two windows open, and you'll click on the "affinity" option on the toolbar of each window under "advanced" or "options" (I forget which it is, and I'm in the process of a motherboard switch on my main build right now. On laptop at the moment.) Set affinity to 0 for one of your cores and 1 for the other. This assures that prime is running on both of your cores. Then torture test--> small FFT's.

And Windows Memtest is also a great program to test your memory stability along with your overclock.

http://hcidesign.com/memtest/

[grey_1]

^Alright, thanks for the info. Be a good project for the weekend.