Intel Core 2 Duo E8400

[Aphax]

I've just swapped my ABIT IP35 for an MSI P35 Neo2-FIR and I still have the exact same issue, I am now leaning more towards broken CPU, and unfortunately since I never bothered to properly test it before I started OCing it I will never know if that's what made it broken.. oh well, lesson learned I guess

Thanks for the suggestions by the way curaga, will give cpuburn-in a try

[curaga]

I've used cpuburn-in for cpu heat testing, and when 8h of that passed, I went on to GCC and Glibc's torture tests, which stress the cpu actually even more

[M1AU]

Aphax, I'm not sure about this, but maybe you really broke you CPU while overclocking. Though I'm currently using an Abit AN8 SLI mainboard myself and I'm not really that pleased with it.
Actually I can't get my new 1 TB hard drive to format correctly with ext3. Every time I tried it, the system just hangs up. Well while searching for the problem, I read about some linux related bugs, but it also could be a hardware (especially mainboard) problem I guess.

Even though I have to thank Abit for there absolutely broken Windows(tm) drivers to bringing me to the linux platform at all! Back than in 2006 I didn't got Windows XP working stable on this Abit mainboard. Every few days the system crashed completely.
After a few weeks of research, I came to the conclusion that the awful drivers caused the problem. I installed Ubuntu Linux and everything worked just as it should, though I had to learn the whole thing from scratch for the following few month :P


Well, with a pricey mainboard, you can also have the same troubles, but this time I want to try that out and invest a lil bit more as usually.

[Aphax]

I bought this CPU a month ago or so together with an Abit IP35 mainboard (running Gentoo Linux with kernel 2.6.25.4). Sadly I could never get it past 3.6GHz or it would run unstable. It seemed to be a problem with the FSB, since setting it to 410+ and using a low multiplier so the CPU itself ran at 3GHz or below would still leave it unstable, even after increasing vtt and mch voltages a little and playing with the GLT Ref. setting.

Anyway, 3.6GHz was of course still a decent overclock, and it is probably what I would have left it at anyway considering I didn't want to put too high of a voltage on it (for 3.6Ghz I only had to increase vcore from the default 1.225 to 1.265) or have to resort to extreme cooling. I have an Arctic Freezer 7 Pro on it at the moment which keeps it nicely cool (cores are ~42C under load, PWM around ~50C under load).

Unfortunately dispite being able to run Prime95 (yeah I know, on Windows.. I should find myself a decent Linux stress-tester) for 8 hours and passing 3 memtest86+ 2.01 passes I would get a sporadic lock up (once every few days or once a week). Even more depressing is that I still got this after reverting to standard settings.

A few days ago I finally found a way to reliably reproduce the lock up (still back on the stock 3GHz) when it happened to me when I was trying to compile Wine 1.1.0 through Gentoo's portage. On the console it gave me a Machine Check Exception (see http://pastebin.com/m10e2a503) which from googling for a bit tells me this really is a pure HW issue as the error suggests.

The really weird thing is that it absolutely only happens when I try to compile Wine (tested both 1.0 and 1.1.0), and always during the "make depend" command (which is actually done before the compiling) and always the same type of Machine Check Exception (sadly running the numbers through mcelog didn't give me anything useful), but not always at exactly the same directory in the source tree. I also couldn't reproduce the hang if I just compiled it by hand (not using portage) using the same ./configure arguments that portage uses and then running "make depend". Yet through portage it always does it.. I think I've tried over 20 times now. I've also done a full system upgrade recompiling things like Inkscape, Evolution, vlc, mplayer, and dozens of other packages without any problems whatsoever.

Anyway, after trying a ton of things like running newer (2.6.25.8) and older (2.6.24.x) kernels, disabled all unneeded kernel modules and not running X, underclocking the cpu+memory, increasing memory timings, swapping the memory, disconnecting all unneeded hardware, swapping the videocard, increasing voltages a tad (to see if it just needed more juice), updating to the latest BIOS (which explicitly added 45nm support), and a few more things I probably forgot, I finally managed to make it not crash after I swapped in a Q6600.

So all this seems to suggest either I borked my CPU, maybe by putting a too high voltage on it when I was trying to get it to 4GHz (never went over 1.3v vcore though) or it just doesn't work well with my mobo, or some odd CPU bug is being triggered. I'm not sure what to do now, the issue seems so specific I find it hard to justify cashing out for a new CPU... I guess next time I should run the CPU at defaults for at least a few weeks so that I know for sure it wasn't caused by my overclocking and I can still RMA it.

Oh well, apologies for the wall of text, this has been on my mind the last few days and when I read this article I just had to write it all down somewhere, and this seemed as good a place as any

[M1AU]

About the cooler, yesterday a friend recommended me the Scythe NINJA series cooler and I guess I'm just going for a passive cooled setup.
With a few case fans (which I already have mounted), it should be possible without problems.

[sabriah]

Phoronix: Intel Core 2 Duo E8400

The processor from the Wolfdale series we are looking at today under Linux is the Core 2 Duo E8400.

http://www.phoronix.com/vr.php?view=12551

I bought this nice CPU three weeks ago!

Good to hear it was a 10/10 choice!

Thanks for creating that test suite. It really is informative thanks to its wide selection of apps!

It was somewhat surprising that 2500 inserts in SQLite didn't gain that much. I guess it is hard drive bound. What would be more more CPU intensive? Sorting 100,000 records? Or perhaps find all duplicate records in a 100,000 record file? I've done that with 2,000,000 records using MySQL, but then it didn't fit the memory.

Great tests!

.

[rbmorse]

I've got a Gigabyte EP-35DS3R with an E8400 and 4GB of Corsair XMS2 (PC2-6400) running Ubuntu 8.04. Without at doubt it's the most solid, stable and reliable PC I've ever had.

Stock Intel CPU cooler works great...the E8400 runs about 40c with the fan turning a very quiet 1,200 RPM.

The only note I'd add is if you're using the stock cooler, mount it on the CPU/motherboard before you mount the motherboard in the case. That way you can visually confirm the pushpins are fully deployed and locked...they don't offer a lot of tactile feedback.

[M1AU]

Uhm, why do you want to go with such an expensive mainboard? Do you need SLI/CrossFire? If you don't, just go with one of the cheaper boards. I am very happy with my Gigabyte P35-DS3R. When I bought it, it was at about 80?, the P45 version should be at about the same price. The board works really well under Linux, everything I tested is supported.
The successor of the board (GigaByte GA-EP45-DS3R) is available for ~110?, the "non-raid" edition for <100?. I don't see a real need to go for a motherboard that is almost 50? more expensive.

Well, I made a few really bad experiences in the past with cheap ones.
But yes, a mainboard for approximately EUR 100 will normally also do it.

[ivanovic]

Uhm, why do you want to go with such an expensive mainboard? Do you need SLI/CrossFire? If you don't, just go with one of the cheaper boards. I am very happy with my Gigabyte P35-DS3R. When I bought it, it was at about 80€, the P45 version should be at about the same price. The board works really well under Linux, everything I tested is supported.
The successor of the board (GigaByte GA-EP45-DS3R) is available for ~110€, the "non-raid" edition for <100€. I don't see a real need to go for a motherboard that is almost 50€ more expensive.

[M1AU]

Thanks d2kx, I will have a look at these!

Edit:
That cooler looks damn nice! I'll take it :-)