Who screwed me over..Intel or Tuniq??

commodore

Limp Gawd
Joined
Dec 29, 2002
Messages
297
Well i splurged on a whole new mid-range system with the sole intention of overclocking the hell out of it (now hell is what i'm getting literally!!)..

The config speaks for itself..

C2D E4500 @ 2.2Ghz
Abit IP35-E
MSI 8800GT 512 OC edition (@ DEFAULT CLOCKS 600/1500/1800)
Corsair XMS2 2GB DDR2 - 680
Corsair HX620 PSU
Tuniq Tower 120 w/ stock fan and Artic Silver 5

Now my overclock is 3.3Ghz @ 1.464 Vcore but @ 84'C on Core Temp with 2 instances of Prime 95 running!! :eek:
I thought the high temps were cuz of a badly seated HSF so i reseated it...4 friggin' times.. and yet the temps dropped by only 3'C which could be attributed to the 'setting' of the Artic Silver.

Now tell me what do i do..?? I cant moan over who screwed me with a concave/convex surface..

I was gonna lap the Tuniq Tower..but what if it is the Intel IHS to blame?? I dont even have a clue on lapping a IHS. How do i figure this one..overclocking overlords some insight please..
 

Ragenrok

Supreme [H]ardness
Joined
May 15, 2007
Messages
4,196
ok, no one screwed you over, you got exactly what you paid for. Does it run at stock speeds fine? If it does then you shouldn't be claiming anyone screwed you over.

now wiht that said, whats your stock temps? What program are you using to measure your temps? Have you tried with the stock heatsink to see what your temps are?
 

commodore

Limp Gawd
Joined
Dec 29, 2002
Messages
297
Nah .. didnt even give the stock HSF a glance. My rant was more out of frustration than anything else. I knew the Tuniq Tower can handle a moderately overclocked C2Q and so had carefully picked my config accordingly..

I'm using the Core Temp utility for temps.
@STOCK 2.2Ghz:
Idle: 44'C Load: 65'C

@O'Ced 3.3GHz:
Idle: 52' Load: 84'C

I used 2 instances of Prime 95 to check @ over an hour each time and fully stable. Point to note is that at normal full load usage (3dmark CPU tests etc.) my O'ced full load temps are 78'C.

guess i just pulled the short straw.. :rolleyes::rolleyes:
 

Ragenrok

Supreme [H]ardness
Joined
May 15, 2007
Messages
4,196
44 idle 64 load at stock is high, now see thats something to bitch about lol. Try putting your stock heatsink on and seeing what temps are, mabey check for any obvious issues on the surface of the chip and your tuniq.
 

Doodies

Limp Gawd
Joined
Dec 31, 2007
Messages
141
My e2180 @ 3.0 ghz and 1.41vcore is hitting 72c with Orthos and Idles around 40-42c using the new coretemp.
 

commodore

Limp Gawd
Joined
Dec 29, 2002
Messages
297
44 idle 64 load at stock is high, now see thats something to bitch about lol.
Tell me about it .. :( :(

I dont want to try the stock HSF solely because that means removing the Tuniq backplate and sticking the annoying spacer type tabs of the stock HSF in...
I'd rather go ahead and lap the HSF.. but would that help?? or is it better to do the IHS too in my case?

Will be using this method i suppose;
http://www.overclockers.com.au/wiki/How_To:_Lap_a_Heatsink
 

bicycle_wreck

[H]ard|Gawd
Joined
Nov 16, 2007
Messages
1,202
My e2180 @ 3.0 ghz and 1.41vcore is hitting 72c with Orthos and Idles around 40-42c using the new coretemp.
That's a bit high. My e2180 idles at 24* and under Prime95 its around 40*. Using uGuru and CoreTemp, specs in sig...

Have you tried a lower vcore?

OP, what kind of airflow are you getting? Also, if you google lapping, you'll get hundreds of hits with quite a few decent tutorials on it. There are even lapping kits that you can buy if you're unsure of what to purchase.

EDIT: looks like you found one.
 

commodore

Limp Gawd
Joined
Dec 29, 2002
Messages
297
THave you tried a lower vcore? OP, what kind of airflow are you getting?
Yup didnt work .. also 1.464 Vcore seemed just a tad above average for a 3.3Ghz OC. The Tuniq fan is on 'high' along with 2 case fans blowing hot air out.

The weird thing is the base of the Tuniq Tower gets only very slightly warm @ o'ced full load. Like the heat is not being transferred.

I already got me the 200 to 2000 grit sandpaper... guess theres no turning back, unless you guys give me reason to think otherwise.
 

Ragenrok

Supreme [H]ardness
Joined
May 15, 2007
Messages
4,196
nope, if you feel your comfortable lapping it then go for it :D and let us know your results lol.
 

Clockz

Weaksauce
Joined
Dec 31, 2007
Messages
64
That's a bit high. My e2180 idles at 24* and under Prime95 its around 40*. Using uGuru and CoreTemp, specs in sig...
Is that the new CoreTemp you're using? It should give a +15C higher reading than uGuru if so (maybe?). :confused:

I'm just trying to figure out why there is such a wide variance in these processors with regard to temperatures. I'm not sure if it's just a software issue or if somehow some of these are running way colder than others. My E2200 idles at 45C and peaks at about 75-80C with two Orthos instances running at 2.7 Ghz at 1.27v (same as stock basically). Measured using the new CoreTemp.
 

commodore

Limp Gawd
Joined
Dec 29, 2002
Messages
297
nope, if you feel your comfortable lapping it then go for it :D and let us know your results lol.
No .. i dont feel comfortable lapping my spanking new HSF ..
But i dont have a choice...you see i'm a overclocker @ heart..;)
 

Ragenrok

Supreme [H]ardness
Joined
May 15, 2007
Messages
4,196
No .. i dont feel comfortable lapping my spanking new HSF ..
But i dont have a choice...you see i'm a overclocker @ heart..;)
rofl thats a good enough reason for me then :D

I'm replacing my zalman cnps9700 with watercooling because I want to, I have a clear case so it will up my computers WoW factor and im hoping for slightly higher clocks lol.
 

BillParrish

Supreme [H]ardness
Joined
Aug 25, 2006
Messages
7,520
@STOCK 2.2Ghz:
Idle: 44'C
You are on the right course. I do not know exactly why but that is far too hot at stock idle with your monster heatsink. I just built a 4500 box with an Arctic cooler pro 7 and was at high 20's idle, 60C full load but that was in a cool house. Lapp them both, the CPU is not as delicate as one would think, don't throw it around or anything but do not worry about getting it wet and lapp the IHS as well. When done clean it well with isopropal alcohol (if it says rubbing alcohol make sure the ingredients are just alcohol and water, no oils) and let it dry in a sunny window for an hour. I know you have remounted a bunch of times but pay special attention and make double damn sure the heatsink is sitting flat on the CPU and nothing is interfering with it seating properly.

I guess it is possible you got a dud core but even still those stock idle temps are way too high.

That heatsink should be solid when it is right, a gentle wiggle should not move it.
 

rainman

[H]ard|Gawd
Joined
Mar 9, 2000
Messages
1,171
Try Everest Ultimate and SpeedFan.



CoreTemp 0.95.4 and the new version supposedly report different temps for some processors...
 

ImageMD

n00b
Joined
Dec 11, 2007
Messages
31
I <3 my Tuniq. I didn't lap either HSF or CPU, and don't think I screwed it in with even pressure, and I'm getting 66-60-60-66 max temp (cpuid, coretemp) after 24 hours of P95 small FTT w/ 4 threads (0 errors).
 

Doodies

Limp Gawd
Joined
Dec 31, 2007
Messages
141
That's a bit high. My e2180 idles at 24* and under Prime95 its around 40*. Using uGuru and CoreTemp, specs in sig...

Have you tried a lower vcore?

OP, what kind of airflow are you getting? Also, if you google lapping, you'll get hundreds of hits with quite a few decent tutorials on it. There are even lapping kits that you can buy if you're unsure of what to purchase.

EDIT: looks like you found one.
If you use the new coretemp 0.96 it will add about +15c to your load and idle temps, so yours would be 39 idle and 55 load which is still very good. Also are you using dual prime 95, when I do dual prime my load is around 69c.
 

bicycle_wreck

[H]ard|Gawd
Joined
Nov 16, 2007
Messages
1,202
My temps: 24* idle / 40* load (Prime95 - 100% on both cores for 8 hours)

If you use the new coretemp 0.96 it will add about +15c to your load and idle temps, so yours would be 39 idle and 55 load which is still very good. Also are you using dual prime 95, when I do dual prime my load is around 69c.
EDIT: I misread your post, sorry.

I am using CoreTemp 0.95.4, and uGuru. I have been in the BIOS to confirm idle temps, and they are spot-on. PC Wizard '08 also shows the same temps, and there is no 15* degree off-set lunacy happening here. ;)

Yes, I am running two worker threads in Prime95, and maxing out both cores to 100%, and the machine stable for 8+ hours (indefinitely).
 

Skrying

2[H]4U
Joined
Jan 10, 2005
Messages
2,851
You're not getting those temperatures, I'm sorry. Use the new CoreTemp people have said. You're basically not doing so because you most likely know your current temps are wrong, which they are.
 

Clockz

Weaksauce
Joined
Dec 31, 2007
Messages
64
I am using CoreTemp 0.95.4, and uGuru. I have been in the BIOS to confirm idle temps, and they are spot-on. PC Wizard '08 also shows the same temps, and there is no 15* degree off-set lunacy happening here. ;)
All of those are wrong because they don't handle the newer CPU's with a 100C Tjunction Max temperature. Your E2180 has a 100C Tjuncton Max. They incorrectly think it's 85C like the old chips.

In other words, even your BIOS is wrong because it's not using the right Tjunction Max value for your CPU. The new CoreTemp 0.96 is the only thing I have seen that is correct because it will use a 100C Tjunction Max.

This has been explained in other threads but in boils down to the fact that the temperature is reported from the CPU as a delta from Tjunction Max and if the software doesn't use the the correct value for each chip then the reported absolute temperature will be wrong.
 

bicycle_wreck

[H]ard|Gawd
Joined
Nov 16, 2007
Messages
1,202
You're not getting those temperatures, I'm sorry. Use the new CoreTemp people have said. You're basically not doing so because you most likely know your current temps are wrong, which they are.
What's with the attitude? If I knew there was a way to fix my temp readings, I would have done it. I am not putting on some charade for the sake of processor temps. I REALLY don't care enough to get into some stupid argument on a web forum about the temperatures of my microprocessor.

All of those are wrong because they don't handle the newer CPU's with a 100C Tjunction Max temperature. Your E2180 has a 100C Tjuncton Max. They incorrectly think it's 85C like the old chips.

In other words, even your BIOS is wrong because it's not using the right Tjunction Max value for your CPU. The new CoreTemp 0.96 is the only thing I have seen that is correct because it will use a 100C Tjunction Max.

This has been explained in other threads but in boils down to the fact that the temperature is reported from the CPU as a delta from Tjunction Max and if the software doesn't use the the correct value for each chip then the reported absolute temperature will be wrong.
That makes total sense. Is this likely to be remedied in a BIOS update, or can it be set manually within the BIOS? I looked, but didn't see the option. Thanks for giving me the how's and why's rather than jumping on some bandwagon.
 

LOUISSSSS

Limp Gawd
Joined
Dec 29, 2007
Messages
193
i think the 85C at 1.4vcore is normal. u know 1.4vcore is rather high on any air cooling right.

my q6600 even hits 69C-ish @ 1.33vcore and 399FSB after running 6 hours of Prime 95.

drop the vcore down to 1.35ish and watch ur temps go down.
 

Skrying

2[H]4U
Joined
Jan 10, 2005
Messages
2,851
What's with the attitude? If I knew there was a way to fix my temp readings, I would have done it. I am not putting on some charade for the sake of processor temps. I REALLY don't care enough to get into some stupid argument on a web forum about the temperatures of my microprocessor.
Because people have told you to update the program you're using yet you haven't. It's like you're simply ignoring it but keep parading around your temperatures. It's sad.
 

Clockz

Weaksauce
Joined
Dec 31, 2007
Messages
64
That makes total sense. Is this likely to be remedied in a BIOS update, or can it be set manually within the BIOS? I looked, but didn't see the option. Thanks for giving me the how's and why's rather than jumping on some bandwagon.
Yeah, I imagine this will be fixed in a BIOS update at some point along with the rest of the software. I don't know if there is a way to tell the abit BIOS about the difference but for other software if it lets me adjust the temperature then I just tell it to add 15C.
 

bicycle_wreck

[H]ard|Gawd
Joined
Nov 16, 2007
Messages
1,202
Because people have told you to update the program you're using yet you haven't. It's like you're simply ignoring it but keep parading around your temperatures. It's sad.
I didn't ignore anyone. I was being told that CoreTemp was off, so I offered temps from three additional sources, but no one offered an explanation as to why those would be skewed as well. Clockz's answer was what I was looking for, and would have shut me up pretty quick. Making the assumption that I am 'parading' my temps is simply absurd, I just wanted an explanation, and I got one.
 

Doodies

Limp Gawd
Joined
Dec 31, 2007
Messages
141
Okay, so here is a little more insight into what is going on. At the coretemp forum, the guy that created the program is saying that in the next release the tjmax for the e2xxx and e4xxx M0 stepping cpus is going back to 85c, and that it was correct. Although there is another guy that is saying that the new temperatures are correct though.. so I'm still a little confused..

here is the thread http://www.alcpu.com/forums/viewtopic.php?t=362

You can set coretemp to show the delta between the tjmax and actual temp, which shouldn't go lower than 15-20c, this temperature should be the same in both versions of coretemp.
 

Clockz

Weaksauce
Joined
Dec 31, 2007
Messages
64
Okay, so here is a little more insight into what is going on. At the coretemp forum, the guy that created the program is saying that in the next release the tjmax for the e2xxx and e4xxx M0 stepping cpus is going back to 85c, and that it was correct. Although there is another guy that is saying that the new temperatures are correct though.. so I'm still a little confused..

here is the thread http://www.alcpu.com/forums/viewtopic.php?t=362

You can set coretemp to show the delta between the tjmax and actual temp, which shouldn't go lower than 15-20c, this temperature should be the same in both versions of coretemp.
Thanks for posting that link, I had not seen that thread. There is surely a lot of confusion so who knows. The fact is the official Intel specs list chips like the M0 E2180 as having a 100C Tjunction Max. Unless Intel is wrong about their own chip (they very well could be) then the temperature should be calculated based on that. Yes it seems high to me too but this is the official Intel specification we're talking about here.

I also noticed them talking about the Intel Thermal Specification in that thread. That specification is based on the Tcase temperature which is not the same thing as the core temperature. The Tcase temp is typically 10-15C lower than the core (the difference from Tjunction to Tcase is also in the official specs for each CPU).

There are two main temperatures we are concerned with:
1. Tjunction Max is what temperature the CPU will shut itself down at.
2. The Thermal Specification which is the maximum Tcase temperature recommended by Intel.

By far the best reading on this topic is here:
http://www.tomshardware.com/forum/221745-29-core-temperature-guide

And yes, having CoreTemp show the raw delta value is 100% accurate no matter what because that is just the raw value straight from the CPU. Although that isn't going to tell you the actual temperature because it is just the distance from the current temperature to when the CPU will shut down and nothing more.

Edit:
Looks like that temperature guide has been deleted or something. Only temporarily hopefully. Another copy is here:
http://www.xcpus.com/forums/power-supplies-cases-cooling/7680-core-2-quad-duo-temperature-guide.html
 

Glow

Supreme [H]ardness
Joined
Jul 20, 2002
Messages
4,834
remount and paste your tuniq, it screws in right? do top right then bottom left, then top left then bottim right. You got to keep the pressure equal just like on a cars wheel. But those temps dont sound right at all, not to mention that voltage seems rather high.
 

lamboman

Gawd
Joined
Jul 17, 2007
Messages
598
As a guide to what is more normal, my C2D E4300 is now at 3ghz (update sig soon) and it idles at 35 and loads at 61, which is in a warm room.

You might have not fitted your cooler on tightly enough. Did you put on an excessive amount of thermal paste?
 

Slade

2[H]4U
Joined
Jun 9, 2004
Messages
2,539
I'd inspect the distribution pattern of the thermal paste on your heatsink after you've mounted it. When you remove it you should find that it is quite difficult to remove directly upward and has to be "slid" off if there was good interface between the heatsink and cpu.

If it is simple to just yank off by simply pulling up, then look for air gaps, thicker spots, areas with no pressure, areas where pressure was applied to say only the top of the chip.

That's too high of a temp at idle, even the stock heatsink, in all it's convex glory shouldn't give off temps that high.

For reference, I have a c2q @ 3.6ghz on air hitting as high as 65C under load for all cores. No way in hell is 70+ anywhere near normal for ANY cpu.
 

commodore

Limp Gawd
Joined
Dec 29, 2002
Messages
297
I'd inspect the distribution pattern of the thermal paste on your heatsink after you've mounted it. When you remove it you should find that it is quite difficult to remove directly upward and has to be "slid" off if there was good interface between the heatsink and cpu. If it is simple to just yank off by simply pulling up, then look for air gaps, thicker spots, areas with no pressure, areas where pressure was applied to say only the top of the chip.
In fact it is impossible to yank off my HSF upward..i have used a thin layer of Artic Silver 5 in between.
 

commodore

Limp Gawd
Joined
Dec 29, 2002
Messages
297
Well guys i have just lapped my Tuniq Tower and the IHS of my E4500. No visible scratches on both and i can clearly see my reflection on both surfaces. Seems like i managed to do a decent job ..

My temps dropped by only 2'C ..no much difference. Very disappointing to say the least.

My new temps @ 3.3Ghz and 1.464 Vcore

Core Temp 0.95.4 temps : (shows Tj = 85'C)
Idle: 36'C Load: 66'C

Core Temp 0.96 temps : (shows Tj=100'C)
Idle : 51'C Load: 81'C
 

TheRapture

Supreme [H]ardness
Joined
May 31, 2000
Messages
6,871
And yes, having CoreTemp show the raw delta value is 100% accurate no matter what because that is just the raw value straight from the CPU. Although that isn't going to tell you the actual temperature because it is just the distance from the current temperature to when the CPU will shut down and nothing more.

So many people refuse to try using coretemp this way...don't know why, it's accurate and does not change with BIOS revisions and/or software updates...
 

Clockz

Weaksauce
Joined
Dec 31, 2007
Messages
64
My new temps @ 3.3Ghz and 1.464 Vcore

Core Temp 0.95.4 temps : (shows Tj = 85'C)
Idle: 36'C Load: 66'C

Core Temp 0.96 temps : (shows Tj=100'C)
Idle : 51'C Load: 81'C
These temperatures are very close to what I get on my E2200. Using CoreTemp 0.96 I get about 45°C idle and 78°C with two Orthos running CPU-stress (small FFT). The raw delta to shutdown from the CPU reports 22°C. This is with a room temperature of about 18°C or less which is pretty cool.

I'm only at 1.27v and 2.7 Ghz (almost stock speed here). If I crank the voltage to even 1.3 my temps go up +5°C and if I go up to 3 Ghz they go up another 3 to 4°C. So I would say your cooler is working way better than my stock one (as it should).

Maybe we both have something funky going on though with the contact area between the heatsink. Or there is a possibility the heatspreader on our CPU's is messed up. Who knows.
 

commodore

Limp Gawd
Joined
Dec 29, 2002
Messages
297
Maybe we both have something funky going on though with the contact area between the heatsink. Or there is a possibility the heatspreader on our CPU's is messed up. Who knows.
No chance of the HSF or IHS being concave/convex.. these temps are for real and not gonna change for me unfortunately.

Reason being ...if i had botched the lapping the temps should have hit the ceiling..

I have to wonder though .. does everyone use Core Temp 0.95.4 to check temps cuz they seem awful low even @ 1.5 Vcore in some instances.
 

TheRapture

Supreme [H]ardness
Joined
May 31, 2000
Messages
6,871
Coretemp .96 is the latest........94.x is pretty old......why don't you try the new one?
 
Top