A question as to the adviseability of trying to lap my AMD cpu.
Having read the guide here for lapping cpu's and heatsinks I thought I'd give it a go with my AMD Phenom II X4 B50.... My temps have seemed high from the building of this system and when I went to reapply my TIM with AS5 I could tell that the cpu and heatsink both were slightly concave. The guide here addresses Intel cpu's and I'd hate to start with this amd if they were not as receptive to being lapped.
I'd follow the same procedure with possibly adding the disassembling of an old dead mb to get the cpu socket to use to hold the cpu for lapping, mount 600 grit dry on plate of glass, and follow thru to 1500....or wherever it looked best.
It's OK to lap your AMD CPU. Say goodbye to your warranty, though.
If you're going to lap your CPU, might as well lap your cooler too. This makes sure that their surfaces are evenly matched, as some coolers expect the CPU surface to be imperfect, and make a cooler base that is "compatible" with that shape.
I lapped an AMD cpu and an Ultra 120 cpu heatsink. No problem. Just don't expect any massive drops in temperature. Typical drop seems to be about 2 degrees celsius. Mine dropped just a little under 2C.
You mentioned using Artic Silver 5. I hope you realise that is an older thermal compound that requires 200 hours of operation and a lot of powering on and off in order for the compound to cure and and stabilize.
Last time I checked there were over 80 thermal compounds available in a variety of categories. Everyone has their favorite. I used IC 7 Carat Diamond for the past two years. It works for me. No curing required. Just apply and run Prime 95 at 100% load for an hour.
You mentioned your temps "seemed" high. Have you actually used an application or utility to measure temps during a stress test such as Prime 95?
Thanks for the info on the AC5, hadn't realized it would take that long to become stable temp wise. Might look into that IC 7 stuff, or any of the myrid others out there. I would prefer one that "sets" right off.
CPU is an unlocked Phenom II X4 B50 BE. clock is at 3.410 at 1.392 volts. Idle temps with a Rosewell RCX-Z4 are 33c. 8+ hours of Prime95 on a blend test woulkd see a max of 63c, constant of about 60c. I haven't monitored while gaming as I haven't had much time to game the last week or so, but will hit 51c during a 3DMark06 run with background programs running at the same time.
With the unlocked cores I don't receive actual core temps, just the cpu temp reading from the board. While running as a 2 core the actual core temps would run about 9c less than the cpu temp reading.
I might just make a couple of passes on some 600 gaper and then evaluate how concave it actually is, might not be worth all the trouble if not "significant". ... and yes, I will also be doing the heat sink.
Due to the unique shape and sizes of the particles in Arctic Silver 5's conductive matrix, it will take a up to 200 hours and several thermal cycles to achieve maximum particle to particle thermal conduction and for the heatsink to CPU interface to reach maximum conductivity. (This period will be longer in a system without a fan on the heatsink or with a low speed fan on the heatsink.) On systems measuring actual internal core temperatures via the CPU's internal diode, the measured temperature will often drop 2C to 5C over this "break-in" period. This break-in will occur during the normal use of the computer as long as the computer is turned off from time to time and the interface is allowed to cool to room temperature. Once the break-in is complete, the computer can be left on if desired.
"So by my estimation of this statement it would take almost a year of normal use to properly cure the AC5 compound, or almost nine days of continuous power cycles to meet their recommendation."
I also like yhe IC Diamond 7 karat stuff.....make SURE to read application instructions which includes :
a) warming it up (I used cup of hot water)
b) letting it air out
Expect to lap the CPU for anywhere from 1-3 hours, depending on how many grades you use/polishing. I remember someone here on the forums got impatient from lapping and just took the whole CPU heatsink off.
I agree with Kokin, the lapping process if you want to do it properly will take a good chunk of your afternoon. 1-3 hours seems accurate and you will see an improvement of 1-3C. It will be up to you to decide if it is worth it. I typically lap my heatsinks (If they are not polished), but only lap the processors that will be going for max overclock.
You probably will find more dramatic temp improvements by looking at your cooling setup for your case. Ideally, you should have more cfm's on the intake than the outtake (positive pressure) or more cfms on the exhaust than the intake (=negative pressure).
Having equal number of fans and cfm's on the intake and exhaust seems like it would be efficient, but this creates dead spots in your case where heat can build up and is not removed by the fans. If you want airflow to move in your case, its all about pressure.
Also, you may look at the heatsink in particular. I had my Thermalright Ultra 120 HS pointed at my rear exhaust in my Antec 900, but I wasn't pleased with the temps. I moved it 90 degrees and pointed it toward the 200mm blowhole fan and temps dropped about 2C. I also removed the side panel intake fan which surprisingly improved things as well about 3C if I remember correctly. I believe it was messing with airflow front to rear. Removing a fan can cause a dramatic effect.
I would test some things out with regards to the case airflow before lapping your CPU, as you will not see a vast improvement. At 1.4V I would expect the idle temps you mentioned, and even the prime 95 torture temps doesn't seem unexpected. However as you mentioned those are CPU temps and not core temps, which will be about 10C higher on avg.
My X4 940 @ 3.5 Ghz idles at 33-35C and goes into the low 50's under gaming load, but I imagine if i ran prime 95 for 8 hours it would get pretty toasty. AOD is capable of getting the cores to 60C in a matter of minutes. I believe I am operating in the 1.35V range.