• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

lga775 temperature questions

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Stratus_ss

Overclockix Snake Charming Senior, Alt OS Content
Joined
Jan 24, 2006
Location
South Dakota
ok guys I was just wondering
I googled looking for what my 3 chips are spec'd to run at but didnt come up with much so I would like to know what each chip is spec'd to run at load

530J (currently with coolermaster silent running at 68 load)
820D (62 under one core 65-66 under both cores)
630 (low 50s on idle using zalmen cu-7000b)

now the only one I am really concerned with is the 530J. is that too hot for it? I am not using Artic Silver as I have recently run out (currently using goop from vantec that comes with the iceburq serries). But I have tried reseating the fan several times and the temps seem to be around the same. The chip idles at upper 40s which isnt horrid.

any help or comments would be great

thanks
 
Yeah, it's a little high. I tell folks to stay under mid-60's load temp for the Prescotts. Maybe you can improve case ventilation?
 
well i have 2 80s at the front for intake, 2 80s at the back blowing out and my psu has the fan at the bottom (assuming its taking air out?)

and one 80 on the case blowing directly on the mobo for the 530j

I should mention that the heat from the 820d is probably because its oc'd to 3.56 ghz on stock vcore and cooling (waiting for water cooling stuff if anyone has a mcp 350 they want to get rid of... ;))

I have only 2 80s on the 820d so there is room for improvement there but its not bad around 62 I am ok with that.

I tried running the 530j with the case off and that didnt seem to help at all either
 
can you link me to some specs for what intel thinks they should be run at (in terms of temp)
for my own curiosity...
also the 820 is the only chip that is oc'd
 
The big factor is voltage. The more vcore you run, the warmer it will run and can cause problems if you overheat. On default voltages, I think Intel says 70 degrees is ok. The Prescott will start thermal throttling at some point above 70 degrees.

While I agree, two 80mm fans in and two out sounds good... it depends on how much CFM these fans move. I've seen 80mm fans that don't flow much at all. You are right, PSU units have exhaust fans. I like to cut the fan mounts out. These perforated holes can sometimes cut down on about 40% of the air flow.
 
on the 530j the two on the front are 37 cfm and the back are 2 antecs at about 34 cfm
the one on the case is the strongest I believe it is 43 cfm

so you think the exhaust fan in the case is restricting airflow?
would reversing the flow help? i.e. the ones in the back blowing in? and the front blowing out?

the two blowing in are only obstructed by 1 74 gig raptor so I dont think that is much of a problem.
 
vcore has been left at stock on all of them

i suppose some specs would help I forgot that

630
Abit AAE8X
1 gig (single stick) ddr 2 533
Zalmen cu7000b
1 300 gig hd
1 XFX 6600gt
CDRW
420 watt enermax psu
Antec mid tower (5, 5.25 bays etc)

820D
P5WD2-P
2 gigs (2 X 1 gig) ddr2 533
1 X 74 gig raptor
1 X LG DVD-RW
1 eVGA 256 meg 6600gt
Same type case
450 watt psu

530 J
Intel D945SN
1 gig (single stick) ddr 2 533
coolermaster silent cooler
1 300 gig hd
1 XFX 6600gt
CDRW
420 watt enermax psu
Antec mid tower
 
Last edited:
sounds to me like you have a poorly seated heatsink. i was getting those kinds of temps with my 920, and it was the seating of the heatsink that caused it. try reseating, or lapping your chips.
 
Stratus_ss said:
i will try reseating it again
And while the HSF ***'y is off, you might try running a straight edge around the CPU IHS & the HS mating surfaces to assure they are true.

I used to have a 6" machinist's scale, which is actually just a small accurate 6" steal ruler, that is very straight, but unfortunately, I no longer know where it is.

But maybe you can conjure up something from around the house that's a known, accurate, straight edge.

All the reseating & TIM in the world won't due diddly, if the mating surfaces aren't true.

JMO,

Strat
 
Last edited:
Stratus_ss said:
this being the case what would I do to remedy this?

try lapping the heatsink. there is a stick on that somewhere, cant remember. also, get a razorblade and look through that. i looked on my big typhoon heatsink and its not straight either. gonna lap it before i give it to my brother.

also, what thermal compound are you using? if its not AS5, get some.
 
Stratus_ss said:
Stratcat said:
if the mating surfaces aren't true.
this being the case what would I do to remedy this?
You would use the process of LAPPING to make the component with the uneven surface flat.

The above link was the only one I could quickly find on the OCC frontpages, and is a somewhat deeper theoretical article than a more practical based one I saw there, a while ago.

While the link describes several varieties of lapping methods, IMO, the most common & easiest way to lap a CPU or HS, is to get some sandpaper in several grits, starting from, maybe, 200 (if the surface(s) are really off), to 600 (for a more mild problem). Then work your way thru several higher grits to around 1200 - 1500. Some peeps go to 2000 (if you can find it), which, IMHO, is overkill. Sandpaper is safest for a cpu, 'cuz it's non-conductive, but emery paper does a better (faster/easier) job due to it's being metallic...but Caveat Emptor if you wanna' try emery paper on ur cpu...I wouldn't. But emery is fine for the HS, if it's cleaned up well, afterwards, using an alcohol wash, or something similar. Also emery paper uses a different numbering scheme to specify its grit.

The sandpaper is quickly dipped in plain old water, then placed on a very thick piece of absolutely flat glass that's just had a quick rinse of water, too. The water will keep the sandpaper stuck to the glass, as long as it's reasonably wet.

Then you take the component that's not flat, and as squarely as possible, drag it accross the wet sandpaper. There! You're "Lapping"!!!

There's really a multitude of techniques (back & forth, circular, figure 8's), and each has their proponents.

I had a piece of glass cut for me that's 3/4" thick, & approximately 14" X 18". I got it from a window glass shop, where it was cut off a piece of scrap glass, for 7 bucks, U$D. The shop's owner got a kick out of my explanation for the glass' use, and I advised him on cleaning the air inlets to his store's problematic computer, & advised him to move it out from a very tightly enclosed space under the front counter.

With all due respects to the article's author, I personally wouldn't go too nuts w/the technical aspects of the article (like using lapping compound, or two surface lapping), but just get the glass & some sandpaper, & do one component at a time, like I mentioned. Unless you really want to get into it (I don't, but there are some real perfectionists out there...more power to 'em, & hey, if it's your thing, go for it).

This is NOT an inclusive post on lapping...It is merely a breif overview.

Before you do anything, I would try doing a better search of the forums (there was an excellent "single surface" sandpaper lapping article, somewhere, linked to the frontpage), & maybe Google around the net, too.

Remember: Lapping will void your warranty, unless the component already has a copper finish (your IHS is aluminum coated w/copper underneath, and some HS's are similarly coated). Actually, it technically should void all warranties, but if done correctly on a copper surface (i.e. a Zalman 7000cu), it then becomes an ethical issue on whether you should be able to do an RMA, if required, somewhere down the road, if a problem arises unrelated to your lapping (what-the-hay, if I clean up a manufacturer's defective surface finish, I figure I'm doing them a favor).

Also, if you lap your proc, you will loose all your info imprinted on the IHS! You'll have to record & remember what the lapped cpu is!

Anyway, if this is your problem, & you decide to go this route, good luck.

Strat
 
Last edited:
wow thanks for the info!
I dont think I will lap the cpu even if it is the problem for risk of damaging it
some food for thought
 
^^^

No Problem.

Good Luck whatever you decide.

That 530J is too hot, IMHO. It's hotter than my OC'd 3.2 S478 C0 Prescott was, when I had it running dual concurrent Distributed Computing clients, which is a similar cpu load to dual Prime95 (I also used a Zalman7000, too, but it was the 'cu' version - I later modded it to use a Delta fan).

Also, don't overlook the fact that temp monitoring can vary quite a bit between mobos & s/w proggys.

Here's a link to Panopsys' ThrottleWatch thermal throttling detection program.

It is possible your throttling at those full load temps, but don't realize it. Your proc should begin TM1 thermal throttling at ~70*C, give or take a few degrees, which is pretty close to where you currently are, and as I mentioned, temp monitoring can vary quite a bit.

Strat
 
well here is the update on the 530j
its still running a bit warm but its now down to 63
here is the wierd thing
I swapped cases and mobos

its the exact same case (i have 1 blue and 1 gray of the same case)
the grey one has the Asus, the blue has the Intel board
the only thing I can think of is that the heat sink in the Intel board is causing the problems because the blue case has 2 more fans in it then this one

the only other variable that is not the same (other then the amount of ram... the gray having 2 gigs more) is the PSU
in the gray case the psu exhaust fans are at either end
in the blue case the exhaust is at the bottom (facing the cpu) and the side
I am goint to swap psu's and will update on the situation
 
switching the psu resulted in virtually no change in load temps which leads me to believe that it has to be the massive heatsink on the intel board that is doing it
 
well I finally got the load temps of the 530 J down to about 59.
Here is what I did and I dont remember what exactly worked the best so I will list for others

1) I swapped case fans out with other ones to try and increase airflow
2) I opened the from of my case, took out the coverings on the 3.25 bays
3) I reapplied the paste but I used a cd paper sleeve like a straight edge and spread a very fine layer
4) Ran with the side of the case off (which didnt do any good)
5) reseated, reseated, reseated hsf several dozen times.
 
My 630 does (roughly) 42 degrees at idle and 58 degrees at peak. However, the thermal diode is mounted to the metal retention clip (the metal bitholding the CPU down), not to the heatsink or the IHS. A quick Ctrl-Alt-Del shoed me that the encoding program I was using was doing roughly 85-95% load (fluctuating) and the rest the system was using. Is that thermal throttling?
 
Back