Gigabyte EP43-UD3L stuck in POST

[eccerr0r]

Re: Gigabyte EP43-UD3L stuck in POST

Read my response above again.

Didn't see anything relevant to the issue at hand and it's not overheating CPUs, but rather why there's a temperature anomaly...

In addition to C1E state that Chaos mentioned below, I think most (or all?) the 45 nm CPUs also come with "soldered" heatspreaders on the CPU core, whereas many 65 nm chips didn't.

Now this [C1E] may explain things... except that I found that without disabling C1E on my "new" C2Q it would crash on the DQ45CB when it idles but seems just fine on the EP43-UD3L... Despite disabling the C1E its idle temperature sensors still reports temperatures similar to my "old" C2Q so I'm still happy with it... (Using the DQ45CB because I didn't want to power a GPU which isn't necessary for its use...)

The idle temperatures are _irrelevant_.

But it is relevant in finding an explanation for something that does not seem physically possible at idle -- unless the idle power consumption is simply higher for the 65nm parts despite being half the cores. But in any case, as I expect the machine to frequently be idle waiting for jobs so it's good enough - though just getting a 45nm duo would have been sufficient. But I'm waiting on my RAM so that I can pound this machine as it is part of the gcc/clang compute farm...

[momaka]

Re: Gigabyte EP43-UD3L stuck in POST

Didn't see anything relevant to the issue at hand and it's not overheating CPUs, but rather why there's a temperature anomaly...

Well, the IHS not being soldered on the older CPUs means you will always get higher temperatures - both at idle and under load. The CPU core simply will not be able to transfer its heat as efficiently as a CPU with soldered IHS.

You are right, however, that the difference in idle temps is a bit too large... so the IHS coupling technique alone may not be the only reason why... and looks like the C1e may be the answer here indeed.

But it is relevant in finding an explanation for something that does not seem physically possible at idle -- unless the idle power consumption is simply higher for the 65nm parts despite being half the cores.

Which it can be...
For this, a good ol' Kill-a-Watt used on the same system with the same PSU and motherboard (i.e. only swap CPUs) truly shows who uses less power and when.

But I'm waiting on my RAM so that I can pound this machine as it is part of the gcc/clang compute farm...

... which is why you should make sure the CPU stays cool properly under heavy load and not just being "good enough" while idling.

[eccerr0r]

Re: Gigabyte EP43-UD3L stuck in POST

So far my old c2q has been running for many, many years 24/7 with its miniature stock HSF despite being pounded on quite a bit... it's my main compute server (shell box, NAS/RAID, web server, mail server, etc., etc.). Surprised it stays so cool in general. It even survived being in two MBs, it formerly was in this EP43-UD3L before the board initially failed.

This "new" (again new to me) c2q of course has an unknown history... and so far the same tests it runs hotter than my old c2q - despite it has the "better" heatsink, but being in the smaller µATX case might do something.

Now the only cpu I do somewhat worry about is my i7 that I frequently get it to trip BIOS temperature warning so that it starts beeping, with its stock HS/F... but I think the c2q's seem fine (other than the c2d, but it's still much better than the p4.) Thinking about getting more RAM for it so it gets even less time to swap for a breather... but offloading some compute to the (slower) C2Qs hopefully at least reduce total time at full bore.

[ChaosLegionnaire]

Re: Gigabyte EP43-UD3L stuck in POST

well, i've had my chance to get my hands on some dual core conroe cpus that came bundled together with the motherboard i bought and it looks like either the bios of some boards display the temperature of the conroes 10°C higher than normal or the non-soldered heatspreader makes the conroes poorer at heat dissipation. not sure which because evidence points in both directions.

firstly, penryn seems to have more power saving features than conroe. looking at cpu world, the e6700, one of the first core 2 duo cpus to be released, has only 4 low power features: enhanced speedstep technology, stop grant state, halt state and extended halt state. while the e8600 has double that, 8 low power features: enhanced speedstep technology, stop grant state, halt state, extended halt state, extended stop grant state, sleep state, deep sleep state and deeper sleep state.

furthermore, certain conroe steppings have better optimizations to lower the idle power from 22w, 12w and 8w. so if u have a hot idling conroe cpu, u may have an earlier stepping that idles at 22w.

next, i've read that earlier on when core 2 was first released, intel didnt publish the tjunction figures for their conroe cpus so ppl were left guessing exactly what the right cpu core temps were. so the core temp readings for conroes werent accurate for some software programs that read the sensors so u may have to take it with a pinch of salt.

for me, the core temp and tcase readings were the same for my e6750 cpu which wasnt right... the ambient was 27°C and at idle, the cpu tcase was 38°C with the cpu undervolted to 1.0625v. im using a third party heatpipe cooler, no way it idles that hot above ambient while undervolted. im more inclined to believe the cpu tcase is 28°C not 38°C. the cpu core temp also often runs around 10°C higher than the tcase temp so im even more inclined to believe the tcase is actually 28°C.

i've also read that the penryns actually have bugged and inaccurate cpu core sensors. it either displays the core temp too high or too low than it really is. so if your quad penryn runs unusually cool when overvolted and/or overclocked, u have a bugged core sensor that displays the core temp too low than it really is.

but on the other hand, when stress testing, the tcase temp can reach 60°C but when i touch the heatpipes on the cpu cooler, they are not even warm at all. so it might be the crappy thermal paste intel used under the heatspreader on the dual core conroes making them hotter.

i'll have to run more tests and post back with any updates or corrections, if any.