**Official e8400/e8500 Retail OC Thread

[PhilDoc]

Yeh I know I even made sure that the heatsink was making contact with the cpu before installing in my case and everything looked right.

The base of the TT 120 Xtreme is very cool to the touch and is probably in high 20's.

According to Everest utility it reads like this

CPU 30c
CPU #1 / Core #1 61c
CPU #2 / Core #1 61c

So I assume the first one is reading the actual cpu temp from the bios.

Unfortunately the core sensors on the cpu are not accurate. It'll probably be fixed by a later revision, but we got stuck. I'm going by cpu temp, which reads about 28C idle/55C load for me. My core temps are around 50C idle/65C load, both cores, regardless of why I try. Guess I should consider myself lucky considering there's not a big delta between the cores.

[Mav451]

guys if the screen freezes at boot up (but no BSOD), is that a proc problem or RAM? thanks

Yeah I've had that happen. I was testing to see if my 500GB was louder than my 2 intakes, so I disconnected the power cable. However, upon the following POST I got a blank screen with a funny white line shape.

I have a funny feeling it was still looking for the 500GB, so i disconnected the SATA cable as well. POSTed instantly, and here I am. It's odd, cuz I remember being able to pull 4pins for the PATAs, and BIOS not complaining at all. Guess it's a minor difference with SATAs.

[Zucker2k]

@Unclewebb, is there a way you could measure the offset between core temp and cpu temp so at least we can have an estimate of the difference between the two?

[jcniest5]

Okay, so I got another E8400 earlier. I believe it's the last day MicroCenter offer with the $189.99 price. On their website and the current sale catalog, it's listed as $199.99 now. I was talking with a sale person and he said it's suppose to be $199.99 and I was like oh no, but when I took it to the cashier, it still rang up as $189.99 so I was like sweet...main reason was to get it before the price goes too high like other vendors/stores.

Anyway, I saw a few packdates, 1/15/08 (all Q746A) and 1/29/08 with a few Q746A and only one Q748A223 (that I saw of and bought).

Anyway, I haven't opened it just yet. I want to make sure it's a good stepping/packdate before open it as once it's opened, it's unreturnable. I also bought an E1200 (Dual Core Celeron) just for the heck of it. But I may take it back to exchange for an E2160.

[unclewebb]

... is there a way you could measure the offset between core temp and cpu temp?

The reason Intel went to digital on chip thermal sensors was to get away from the inaccuracy of temperature readings from the on board diodes. I know the CPU temp reading on my P5B can vary by 10C depending on if I re-boot or resume from stand by mode. CPU temp readings can also change from one bios version to the next and from one board to the next.

The on chip DTS is consistent and individually calibrated. CoreTemp and SpeedFan are the only two programs that are reading this data correctly from the chip in real time. I've tested various other programs, including TAT, that take the raw data and averages it and does other sorts of bastardizations to it.

Running CoreTemp and subtracting 10C off of the displayed readings should be giving everyone some very accurate core temperatures so why bother screwing around with anything else?

My E8400 survived the heat baking. I think I'll make something close to this my 24/7 setting and see if it degrades any.



I was planning to EBay off my E6400 Conroe core but it's built like a tank compared to these new fragile 45nm chips so I'm keeping it. If Penryn can't reliably run 4.0 GHz then might as well dump it. All this worrying about degrading chips is taking the fun out of overclocking.

Edit: Almost forgot. When experimenting with my chip I noticed that when Orthos fails it is always Core0 that dies first. For me this is also the core that has a temperature sensor that gets stuck at 67C away from the throttling point. That is far enough away that it won't cause a problem for me but I was just wondering if anyone else with a stuck sensor has noticed Orthos crapping out on the same core? I think stuck sensors could be a warning sign of a second rate chip that is more likely to degrade and fail even if the core voltage never goes over 1.40 volts.

[ThatGuy16]

I have come so close to saying screw it and setting the bios to default after reading all of this...

But im still rockin 4ghz, i think at my voltage i shouldn't have a problem..

[Big Lar]

The reason Intel went to digital on chip thermal sensors was to get away from the inaccuracy of temperature readings from the on board diodes. I know the CPU temp reading on my P5B can vary by 10C depending on if I re-boot or resume from stand by mode. CPU temp readings can also change from one bios version to the next and from one board to the next.

The on chip DTS is consistent and individually calibrated. CoreTemp and SpeedFan are the only two programs that are reading this data correctly from the chip in real time. I've tested various other programs, including TAT, that take the raw data and averages it and does other sorts of bastardizations to it.

Running CoreTemp and subtracting 10C off of the displayed readings should be giving everyone some very accurate core temperatures so why bother screwing around with anything else?

My E8400 survived the heat baking. I think I'll make something close to this my 24/7 setting and see if it degrades any.



I was planning to EBay off my E6400 Conroe core but it's built like a tank compared to these new fragile 45nm chips so I'm keeping it. If Penryn can't reliably run 4.0 GHz then might as well dump it. All this worrying about degrading chips is taking the fun out of overclocking.

Edit: Almost forgot. When experimenting with my chip I noticed that when Orthos fails it is always Core0 that dies first. For me this is also the core that has a temperature sensor that gets stuck at 67C away from the throttling point. That is far enough away that it won't cause a problem for me but I was just wondering if anyone else with a stuck sensor has noticed Orthos crapping out on the same core? I think stuck sensors could be a warning sign of a second rate chip that is more likely to degrade and fail even if the core voltage never goes over 1.40 volts.

When I was testing with Orthos, Core1 would be the core to error out, however Core0 is the stuck temp in coretemp, the opposite of yours. My Core0 has never ever moved from 40c to date. Core1 shows 31c right now while typing. Bios says 0 to 2c, go figure.

Larry

[unclewebb]

Big Lar: Thanks for the info. There goes that theory!

[Ron 61]

unclewebb I think you are right about Tj.Max being 95c. Thats what I thought it was all along and your testing seems to confirm it. Thanks for doing the testing. Heat is not an issue with these, all you have to do is look at the skimpy HS that ships with it. Its just going to take time to sort out how much volts these things can take. In 6 months we will know much more.

[scwam]

@Unclewebb, is there a way you could measure the offset between core temp and cpu temp so at least we can have an estimate of the difference between the two?

Personally, I think Coretemp is correct on the Load temps. I've had a 6850 and my E8400 both with a thermistor probe under the exact middle of the PCB/CPU die (touching it). It has read exactly 8 to 11c under coretemps reading at load on both processors. Naturally it will be a few degrees under the actual die reading because the probe I have is touching directly under the bottom of the die at the PCB and not the most hot part (middle of the core).

[laragirl83]

7*550MHz@1.296V, just a little Prime95 Small FFT's test, to make sure it can handle it...

[Capt Proton]

You have to register to download SpeedFan version 4.34b38 but it has no problem reading the on chip digital thermal sensors in the E8x00 processors. If you have the problem where the temperature on one or both of the cores is not changing between idle and full load then the problem is that the sensors in your processor are stuck which is common. Not great quality control by Intel but as long as the DTS sensors start moving before your processor reaches TjMax then they're doing the job that they're designed to do.

This is simply not the case. With Coretemp, both sensors do not go below 37C. Both temps do scale logically above that when I run Prime or OCCT, so, my sensors are not "stuck".
With V4.34 B38, Temp one registers @24C, and only goes up about 2C under load. Temp 2 scales properly. Temp 3 is Bogus, and the last 2 are for the video card.
I know you have to register to download the Betas.
I have found that, with computers, there is little that applies to all, 100%. Once you realize that, life becomes much easier. Seldom is one person 100% correct and anyone that disagrees 100% wrong.

[Damien]

final stable oc and a fun boot at 4.4 at a decent voltage. i think i can get 4.2 stable, matter of messing with settings.

[laragirl83]

funny background image, tough... :s

[unclewebb]

This is simply not the case. With Coretemp, both sensors do not go below 37C.

Most of the stuck sensors get stuck on the way down but when you go to full load, sooner or later they start to work properly. It is possible to have a sticking sensor on both cores. The one on Core0 of my E8400 gets stuck at 38C as reported by CoreTemp.

Your problem sounds unique. If you can PM me two CPUz register dumps, one at idle and one at full load and also send me two side by side screen shots of CoreTemp and SpeedFan at the same idle and full load temperatures then I might be able to figure out what's wrong.

[unclewebb]

I used a Fluke 62 Mini IR Thermometer for my testing and its accuracy is rated at +/- 1C.

As I said before, the DTS is calibrated by Intel and seems to be extremely accurate within a 35C range between 60C and 95C. If your load temps don't get up into this range then even after my TjMax=95C correction, your temperatures based on the DTS are not going to be 100% accurate.

Here's an interesting comparison between my previous E6400 and my new E8400:

I set both CPUs to 1600 MHz and core voltage to 1.080 volts which is the minimum for my P5B Dlx. I'm using a Tuniq and I set the fan to max in a very cold basement to try to get the idle temperatures as low as they can possibly go so I could do a fair, apples to apples, comparison.





The temp sensor in Core0 of my E8400 gets stuck at 38C but the sensor in Core1 works fine. It sat mainly at 35C but would occasionally drop to 34C showing that it wasn't getting stuck like the other one.

Anyone that believes that the DTS is accurate at low idle temperatures and that TjMax=105C for the E8400 needs to somehow explain the 25C difference in idle temperatures between two CPUs that should, in theory, be almost equal.

My E6400 is reporting a core temperature that is about 3C below the ambient temperature of 12C. That's impossible with an air cooler so obviously my E6400 is reporting too low when the idle temp is very low. My E8400 is the opposite. At low idle temperatures it reports too high. I measured an actual core temperature of 35C when it was reporting 40C which is 5C too high. At even lower idle temperatures, you would expect this margin of error to increase further.

To explain a 25C difference we can throw away 1C based on measurement error, room temperature changes, etc. If TjMax is actually 95C, that would explain 10 more degrees which leaves 14 left. Divide that number in two and that means my E6400 might be reading about 7C too low and my E8400 might be reading about 7C too high at idle. That is reasonable based on what I've observed.

I don't believe that Intel designed or calibrated these sensors to be accurate at very low core temperatures. They were calibrated to trigger thermal throttling at 95C. A margin of error that could be as high as +/- 7C at idle is huge so trying to compare idle temperatures is pointless.

[laragirl83]

P5B Deluxe RoxXx!!!

[Big Lar]

I used a Fluke 62 Mini IR Thermometer for my testing and its accuracy is rated at +/- 1C.

As I said before, the DTS is calibrated by Intel and seems to be extremely accurate within a 35C range between 60C and 95C. If your load temps don't get up into this range then even after my TjMax=95C correction, your temperatures based on the DTS are not going to be 100% accurate.

Here's an interesting comparison between my previous E6400 and my new E8400:

I set both CPUs to 1600 MHz and core voltage to 1.080 volts which is the minimum for my P5B Dlx. I'm using a Tuniq and I set the fan to max in a very cold basement to try to get the idle temperatures as low as they can possibly go so I could do a fair, apples to apples, comparison.





The temp sensor in Core0 of my E8400 gets stuck at 38C but the sensor in Core1 works fine. It sat mainly at 35C but would occasionally drop to 34C showing that it wasn't getting stuck like the other one.

Anyone that believes that the DTS is accurate at low idle temperatures and that TjMax=105C for the E8400 needs to somehow explain the 25C difference in idle temperatures between two CPUs that should, in theory, be almost equal.

My E6400 is reporting a core temperature that is about 3C below the ambient temperature of 12C. That's impossible with an air cooler so obviously my E6400 is reporting too low when the idle temp is very low. My E8400 is the opposite. At low idle temperatures it reports too high. I measured an actual core temperature of 35C when it was reporting 40C which is 5C too high. At even lower idle temperatures, you would expect this margin of error to increase further.

To explain a 25C difference we can throw away 1C based on measurement error, room temperature changes, etc. If TjMax is actually 95C, that would explain 10 more degrees which leaves 14 left. Divide that number in two and that means my E6400 might be reading about 7C too low and my E8400 might be reading about 7C too high at idle. That is reasonable based on what I've observed.

I don't believe that Intel designed or calibrated these sensors to be accurate at very low core temperatures. They were calibrated to trigger thermal throttling at 95C. A margin of error that could be as high as +/- 7C at idle is huge so trying to compare idle temperatures is pointless.

Exactly how/where are you taking the temps with your IR? I have a Raytek here that is pretty accurate, I use it for taking readings of the board/sinks,etc..

Larry

[THE JEW (RaVeN)]

P5B Deluxe RoxXx!!!

Nice results. Yeah, this article seems to suggest that the 965 might be pretty decent for the E8*** chips:

http://www.madshrimps.be/?action=get...85&articID=672

This reminds me of the early Conroe motherboards as we had to wait a few months to have high FSB frequencies as the first chipset that was capable of running high FSB's was the 965. The latest chipset are not ready to take the 45nm chips beyond 600+ FSB and most of the boards will be stuck at 550MHz FSB or less on air cooling. The P35/X38's all have the same issues, so the only solution is a newer chipset or better support for the Wolfdale and Yorkfields. However, something strange occured when reading about this issue and talking to people who tried to bypass these problems. It seems that one of the older chipsets, the 965, have no problem at all reaching 620MHz FSB on air cooling! So in order to max out your CPU, you'll have to find a P5B or Commando. On air, however, every motherboard will do well.

Anybody else experience this? If one could hit above fsb 450, I don't think the strap issue would be an issue on the Asus boards (as per Tony's thread).

Input appreciated.

[unclewebb]

You can't get closer to the true core temperature unless you pull off the IHS.

I realize that when a processor is at full load that there are temperature gradients across the processor where parts of the processor are hotter than other parts. By doing my testing at idle and letting the cores slowly come up to temperature I believe the readings with the gun on the IHS are very close to the actual core temperatures. Likely within the +/- 1C accuracy of the Fluke 62 that I'm using.

I covered the E8400 with some thin masking tape to reduce the shine which improves the accuracy and repeatability of the readings. After the computer sits, unplugged for 12 hours, the measured temperature of the covered IHS is exactly equal to the ambient temperature near the computer so I think this test method is valid.

THE JEW (RaVeN): I'm using a P5B Deluxe with the P965 chipset and it works well enough. The practical 24/7 limit for most of these processors is somewhere around 450x9 so it works perfect. I think someone here was running one of these at a FSB of 590 or 600 MHz.

[Big Lar]

Well, I guess you get the title of the biggest "caijones" then

Nice work there

[unclewebb]

Well, I guess you get the title of the biggest "caijones" then

Either that or village idiot! Amazing it still runs great after that abuse.

[Big Lar]

Too funny!

Well, it does in fact seem to get the job done very well tho.

Back on topic tho, I sure wish this darn chip would read temps correctly, but I guess it is doing fine. I run my water at 50 to 53f, and I imagine the temp prob may be due to to low of a reading. I used to get 22c or so idle with this block on my Opty at 3 gigs. Taking that into effect I am not too worried about overheating.

In my ramblings of reading, I came across someone who stated that under 27c the temp was afoul. May have been here that I read it, just lost track of where, however that could add to the equation in my respect.

Larry

[jcniest5]

Well, earlier today I just went and exchanged my Q748A223 for a Q746A483. Also, while there, I also exchanged the Dual Core Celeron E1200 for the E2160 for about $11.00 dollars more. Anyway, about the E8400 exchange, I'm not sure if I did the right thing or not.

Edit: I also saw a few Q746A518 over there. Should I have exchanged for that batch instead? I think the CS people might already be too tired of seeing me so often lately.

[Grnfinger]

Hey that's good to know, all the data people report is interesting.. How are you're PLL and FSB Term set?

CPU PLL 1.50
FSBT 1.40