What's the max fsb you can get with your P5B, Thrax? It looks like my Gigabyte DS3 is around 520 fsb stable with the stock northbridge. But I can't bitch as the board only cost $110 and runs my E6300 like a charm.
I'll let you know after I mount my Thermalright HR-05 on there. Stock board at 1.55v nb, 2.15v RAM, the board has warm reset (Switch) POST problems at 510, but once it's in windows it can run orthos for 48 hours. I've never really tried to push it above 500x6, because I'm still working out the kinks in my cooling.
Yeah, I need to get off my ass and mount my HR-05 on my northbridge too. I think I have a bit more left in fsb headroom if I can cool the northbridge a little better than the stock setup. I have one of those 70mm fans mounted to it that come with the regular A64 stock cooler on it, so it shjould cool much better than the stock nb heatsink.
1.41 v vcore and Orthos temps are around 48-52 C (depending on room temps and at 3500 MHz) with the Tuniq Tower 120 and it's fan on high speed. I have a pretty decently clocking E6300 for an early production model. I have actually had it over 3600 MHz but started losing absolute stability then, which I think is more related to nb temps at that high a fsb speed. It is bootable into windows at around 528 fsb but it couldn't do any really hard tests such as Orthos at that fsb speed.
1) My case may be crap.
2) I have a convex CPU/HSF.
I'm running 1.25v (I can go lower) at 3GHz, and I'm IDLING at 50C. Granted, it is very warm today, but that's preposterous. Even when it's chilly, I'm idling at 46. I have a thermalright ultra-120! The extreme isn't that much damn better.
I noticed loads of difference in cooling ability with my U-120 after I lapped the sucker. Mine had a total crap finish on the base as delivered, with a big screwup on one side of the base (a ridge on the side) and it was also very concave. After lapping it down good and flat it then caught up with my TT-120 in cooling ability.
EDIT: I would absolutely recommend lapping any of Thermalright's heatsinks nowdays as their QC on base finish is in the crapper, IMO. Their designs are still top cabin though.
The locking mechanism (lid) on the socket can actually warp the IHS to some degree. I lapped my QX6700, and it didn't help at all until I removed the lid. My waterblock with a custom backplate "hard mounted" (no springs) is tightened down so much that it causes no problems to run it lid less anyway. Just something you might want to consider if you go through all the trouble of lapping it
It has been running 24/7 FAH since early this year with no issues at all. Vcore@1.5V and 400*9 (that's probably closer to 250W). Worst case temps for each core seems to be around 60C, while it's usally running at around 50-55C running FAH SMP.
Done with lapping. Amazing results, to say the least. I lapped the chip and the sink. I shaved more than <b>10C</b> off of both Core0 and Core1. It sucks, though, that Core1 idles 6C hotter than core0.
Is there anything I might be able to do about that, or is that just my luck of the draw?
I think that is normal Thrax. I don't know why, but all the C2D's I have used have done that. On the the quad, it's always the second core that runs hotter when idle (and also at "light" 100% load). Same thing on the allendale and conroe (possibly the second on the dual cores too, I don't remember).
Done with lapping. Amazing results, to say the least. I lapped the chip and the sink. I shaved more than <b>10C</b> off of both Core0 and Core1. It sucks, though, that Core1 idles 6C hotter than core0.
Is there anything I might be able to do about that, or is that just my luck of the draw?
Glad to hear it made a big difference, Thrax. Thermalright has some excellent engineering in their heatsinks but they definitely need to work on their base finishing process. Plus, the old "warped IHS" on Intel's processors has been a problem at times from way back to the Tualatin P3 days also. Some are relatively flat but others seem to be badly warped. I guess it's just due to mass production issues and as long as you aren't pushing your processor it doesn't affect it's stability. As for the core temp differences between the 2 cores, I see much the same thing with both my desktop C2D processors too. Strangely enough though, the processor in my Dell laptop (T7400) runs at nearly the same temps on both cores. I haven't taken it apart to see, but I assume that the mobile C2D procs are lidless like the P-M procs and if so, I guess the core temp difference is due to the IHS on the desktop Conroes.
Maybe they should look at the waterblock industry, and the new bowed bases. I have tested the old Apogee, and the new GT and GTX. With the flat base, the GT was maybe 2C better than the old, with the bowed base easily 5C (3C due to the bowed base alone). I don't see much of a difference going from the GT to GTX (both bowed) though. A 3-5C improvement is a lot for waterblocks. A problem might be allowing for some flexing in a HSF design, as this is what makes the bowed base mate so well (it flattens against the centre of the IHS as pressure is applied, and takes on the shape of the IHS to some degree).
Try an experiment.... Set the Affinity for a few processes to Core1, see if that evens out temps. It may be that Core0 is actually "loaded" slightly, I mean the CPU is never really totally "Idle" in XP, right?
If that works, there really isn't an Intel driver like the AMD one that helps with the Multi-Processing...but didn't I just read about a patch or some "flaw" in the C2D processor that M$ released a patch for?..... hmmmm.
I continue to have fantastic success. As the thermal paste continues to cure, I'm now getting 32C/38C for Core0/1. Another couple degrees! I wish Core 1 wasn't so hot.
Lasse, you mentioned removal of the retention plate. In my case, I don't think that would help. The surface of the chip clears the plate, and the sink is fully screwed into its bracket long before removal of the plate would allow me to get a few extra millimetres of pressure.
Another great benefit is that my ΔT has been reduced by 5-6c. It used to have a 10C spread between idle and load, and now it's only 4-5C. This is great!
On many boards, the IHS on the CPU itself becomes bent from the uneven pressure applied by the retention bracket (copper is soft). It's not the cooler base hitting the bracket. In most cases, you can just pull the lid off. Look carefully, and you will see how (you have to do it at a certain angle, and it's just as easy to put it back. I'm guessing it's not a big problem for you, since you got great improvement by lapping the IHS, but you should check it out
Can't hurt. I'm going to dismantle my rig completely this week and unmount the mobo from the tray to mount my new HR-05. I'm hitting the limit of the OEM cooling capabilities, and I'm itching to push the system a little harder and work out the warm reset kinks.
I'll give it a shot. Core 2s are cheap, and I might get one of the new G0-stamped chips if things go to hell in a handcart..
Comments
1.41 v vcore and Orthos temps are around 48-52 C (depending on room temps and at 3500 MHz) with the Tuniq Tower 120 and it's fan on high speed. I have a pretty decently clocking E6300 for an early production model. I have actually had it over 3600 MHz but started losing absolute stability then, which I think is more related to nb temps at that high a fsb speed. It is bootable into windows at around 528 fsb but it couldn't do any really hard tests such as Orthos at that fsb speed.
1) My case may be crap.
2) I have a convex CPU/HSF.
I'm running 1.25v (I can go lower) at 3GHz, and I'm IDLING at 50C. Granted, it is very warm today, but that's preposterous. Even when it's chilly, I'm idling at 46. I have a thermalright ultra-120! The extreme isn't that much damn better.
EDIT: I would absolutely recommend lapping any of Thermalright's heatsinks nowdays as their QC on base finish is in the crapper, IMO. Their designs are still top cabin though.
It has been running 24/7 FAH since early this year with no issues at all. Vcore@1.5V and 400*9 (that's probably closer to 250W). Worst case temps for each core seems to be around 60C, while it's usally running at around 50-55C running FAH SMP.
Is there anything I might be able to do about that, or is that just my luck of the draw?
EDIT: Great results though
Glad to hear it made a big difference, Thrax. Thermalright has some excellent engineering in their heatsinks but they definitely need to work on their base finishing process. Plus, the old "warped IHS" on Intel's processors has been a problem at times from way back to the Tualatin P3 days also. Some are relatively flat but others seem to be badly warped. I guess it's just due to mass production issues and as long as you aren't pushing your processor it doesn't affect it's stability. As for the core temp differences between the 2 cores, I see much the same thing with both my desktop C2D processors too. Strangely enough though, the processor in my Dell laptop (T7400) runs at nearly the same temps on both cores. I haven't taken it apart to see, but I assume that the mobile C2D procs are lidless like the P-M procs and if so, I guess the core temp difference is due to the IHS on the desktop Conroes.
Try an experiment.... Set the Affinity for a few processes to Core1, see if that evens out temps. It may be that Core0 is actually "loaded" slightly, I mean the CPU is never really totally "Idle" in XP, right?
If that works, there really isn't an Intel driver like the AMD one that helps with the Multi-Processing...but didn't I just read about a patch or some "flaw" in the C2D processor that M$ released a patch for?..... hmmmm.
Another great benefit is that my ΔT has been reduced by 5-6c. It used to have a 10C spread between idle and load, and now it's only 4-5C. This is great!
(As if swirling a $200 chip on a some sandpaper wasn't bad enough
I'll give it a shot. Core 2s are cheap, and I might get one of the new G0-stamped chips if things go to hell in a handcart..