Intel E8400 Wolfdale Living Review

[raysheri]

e8400 on the abitix38quad gt. ballistic 8500 tracers, Zalman 9700, gainward 8800gt gs 1024, abit official 12 bios

been reading all these forums and i've got this setup myself. The temp thing has had me going round scratching my head but i now feel that uguru is posting the right temps.
Core temp is all wrong for this cpu - and other software readings of various descriptions about this system show all different readings which leads me to conclude that the rest of the world has not caught up with the new processor and chipset that is different form previous issues. Currently running at 3376.9 mhz 450.2x7.5, ddr @1080, timings 55515 fsb@1800 without any voltage increases from the defaults (except 2.20 for the ballistics). uguru temps are 30 for sys and 33 for cpu which seems right to me for no load in my environment. Ran orthos and even core temp just sat at 54 which always shows 10 degrees higher than uguru ( and yes the .5 multipliers do work but if i set a whole figure, save and exit, boot into windows xp then reboot into bios it adds the .5 to whatever i had!!!) My intuition is that this is an awsome processor that can go way over the top and the problems are all in the bios and motherboard makers who i don't think have been able to support it properly... YET.

PS just like a new ferrari, this quadGT is still being tested on the track and is not ready for consistant racing (ie Crysis)

[supershanks]

Hi raysheri,
welcome to the forums.

the problems are all in the bios and motherboard makers who i don't think have been able to support it properly... YET.

I'm glad your plesed with your CPU.

Can you perhaps tell us then how Core temp , cpu-z or Bios are wrong. here ??
Post #49 - Reading the CPU Temp Values As you can see this is reading the sensor values directly from the CPU., which tally with the readings from Coretemp.

There is definite evidence of faulty sensors both here and around the web Wolfdale Temp issues

As we have said here in this thread,:-
1) Either you put it down to sensor issues & ignore it focussing on the far more inaccurate analogue Temp (from Guru)
or
2) You believe the sensors, which tends to take the shine of the chip.

but either way you can't blame it on bios or software.

luck

[raysheri]

Thanks for the reply supershanks, my understanding of the program coretemp was wrong due to inexperience and lack of knowledge but thanks to people like yourself and unclewebb it's growing everyday.

"My apologies to everyone on this thread for ever doubting the degrading E8x00 processors issue. The problem is real. Now we just have to find out what a safe maximum voltage should be. Based on the Intel docs I'll be sticking to a maximum of 1.40 volts with my new Wolfie and keeping a close eye on any degrading of performance." " Anyone that lives in fear when their CoreTemps hit 70C don't need to worry. That's nothing! "When I tested my previous E6400 I found that the readings from my IR gun pointed directly at the IHS were exactly equal to what CoreTemp was reporting from about 47C to 85C where it would begin to throttle.

My E8400 is different. I've seen some users reporting maximum CoreTemps of 70C. When CoreTemp reports 70C for my E8400, the IR gun is only showing an actual temperature of 60C. At lower temperatures the difference approaches 15C. I think this will make a lot of sense for anyone that has gone from dual core Conroe to dual core Penryn on the same board.

The other test I did was I ran my E6400 and E8400 at the same MHz and core voltage. With a small hand held high speed fan pointing at the cores, the lowest I could get down to with my E6400 was 47C as reported by CoreTemp and the IR gun. The 45nm E8400 is smaller and more efficient so common sense says it should run cooler. CoreTemp was reporting 57C during this test but the IR gun was only showing 42C. 5C cooler makes logical sense. Penryn running 10C hotter than Conroe at idle makes no sense.

From this I would have to say to take the absolute reported temperatures based on the on chip sensors with a grain of salt. Intel does not document these sensors being used for that purpose. All of the chips may have a similar temperature curve to what I found but there is no guarantee of that. The on chip DTS was designed for thermal throttling and for showing distance to the throttling point which CoreTemp reports correctly. Just keep in mind that 50 units of headroom until throttling does not exactly translate to 50C of headroom.

Better yet, just forget about temperatures. That's the least of ones worries with these chips. There's a huge amount of temperature head room, even when grossly overclocked. "With Penryn, it's voltage headroom that people need to watch."

"I'm 100% sure that TjMax=95C for my e8400 and that value is very likely used across all 45nm E8x00 desktop dual core processors. I ran the exact same test on my Conroe E6400 last summer and it showed 85C on the IR thermometer just as the processor would start throttling. Now for the same test with the E8400 it shows 95C. At 60C actual, CoreTemp shows 35C away from the throttling point, at 70C actual it shows 25C to TjMax and at 80C it shows 15C to TjMax. CoreTemp has always been very accurate at reading the DTS temperature information in MSR 0x19C but not always accurate at guessing what TjMax is. Setting TjMax=95C will result in reported absolute core temperatures exactly equal to the actual core temperature from 60C to 95C.

The problem has always been that using the DTS to report absolute idle temperatures has never been accurate and still isn't. This confuses the issue and TjMax sometimes gets adjusted by programmers to try and make the idle temperatures "look right" but that only screws up the reported full load temperatures. The problem now is that many users full load temperatures never get up to 60C so using DTS based temperature monitoring software will never be 100% accurate for them in the temperature range that they operate at.

Adjusting TjMax to 95C for E8x00 also gives you much more comparable load temperatures to the previous Conroe dual core E6x00 series.

I'm not here to twist anyone's arm so use whatever value you're comfortable with. I'll be using TjMax=95C. Don't wait for Intel to confirm this number because they've never publicly admitted to what TjMax is for any of their desktop processors. They also don't support programs like CoreTemp that try to calculate the core temperature by reading DTS values directly."

[raysheri]

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?"

[supershanks]

That's interesting raysheri , thanks a lot for sharing. There is something definitely strange with the idle temps. my 2nd core locked at 43C during termp monitoring on idle it only shifted when load was applied see E8400 Temp Comparisons

Whilst I think in fairness I've always believed the logic that these chips must run cooler than 65nm, I begrudge paying good money for something that is essentially faulty, albeit a temp sensor. Given Intel's overall market dominance at this point,I struggle to see why they bought these chips to market in this condition, might be different if AMD were breathing down Intel's neck.

Think the one thing I did take from playing with the E8400 is I'm a Quad man now

Luck

[Ace-a-Rue]

Quad MAN!...BUT...most programs, at the moment, can't use more than 2 cores...i know software will eventually be updated/released to use all 4 cores.

[supershanks]

hi ace.

Quad MAN!...BUT...most programs, at the moment, can't use more than 2 cores...i know software will eventually be updated/released to use all 4 cores.

Yes but i think it's more like what I described here Conclusion

Must admit though i think i'm a quad man now It's a bit like comparing a 125cc bike with a kawasaki 1100cc

Whilst the dual core wolfdale is king of the hill superpi wise, Wprime & other benches/operation show the quad to be better. I guess for me it woke me up to want i'd got used to taking for granted.


luck

[raysheri]

thanks supershanks,
I setup my new system in a enermax uberchakra case and with the intention of easily swapping hardware in and out. I got the e8400 because it was $270 au and a $100 cheaper than the Q6600 at the time. I'm waiting till the price hits rock bottom before i get one as i think they are a classic. I've seen that most people with e8400's get 4.0g overclocks on air pretty easily and it seems the key is not to use too much voltage. There are a lot of people with the locked temp read at idle but mine isn't and practically speaking it doesn't seem to make any difference to performance. Ive got a feeling the penryns are really suited to the x48 ddr3 config although i've heard of people getting good results with the 965 chipset!!!
We're coming to the end of an era as intel will be going to a new socket and onchip memory controller in 12mths. and their own video cards, (Nvidia are really in the hotseat for not enabling intel chipset SLI) so i think there rolling out everything they can squeeze out of LGA775 and their existing stock and manufacturing and also to crush the opposition in the market place before they essentially start doing the same thing. In 1 years time we could be looking at Intel new socket, onchip MC, dual/tri/quad GPU configs, and 8 core cpu's. --- "luke, you don't know the power of the dark side"

[supershanks]

Ive got a feeling the penryns are really suited to the x48 ddr3 config

Or X38

Asus P5E3 Deluxe Wifi Living Review

Must admit I'm a bit disappointed to see a max divider of only 2:1
To run my Cellshock at their rated 1800Mhz it means my minimum FSB = 450.
They should be capable of benching to 2,000Hz which means an FSB >= 500, maybe those E8400 & E8500 are looking more attractive.

We're coming to the end of an era as intel will be going to a new socket and onchip memory controller in 12mths. and their own video cards, (Nvidia are really in the hotseat for not enabling intel chipset SLI) so i think there rolling out everything they can squeeze out of LGA775 and their existing stock and manufacturing and also to crush the opposition in the market place before they essentially start doing the same thing. In 1 years time we could be looking at Intel new socket, onchip MC, dual/tri/quad GPU configs, and 8 core cpu's. --- "luke, you don't know the power of the dark side"

There's a lot of truth in that statement mate, especially about the dark side

luck

[Ace-a-Rue]

Originally Posted by supershanks

hi ace.

Yes but i think it's more like what I described here Conclusion Whilst the dual core wolfdale is king of the hill superpi wise, Wprime & other benches/operation show the quad to be better. I guess for me it woke me up to want i'd got used to taking for granted.


luck

SO...you are one of those overkill type guys!