E8400 is amazing!

smilingcrow · Post by **smilingcrow** » Fri Feb 08, 2008 1:48 am

mcoleg wrote:nope, not here. it's either auto or numeric value.

It may be that Auto = stock voltage when not over-clocked and when you over-clock it automatically over-volts.
The simple way to test this theory is to compare the Vcore at load using CPU-Z at stock FSB and when over-clocked.

mcoleg wrote:what is considered a default vcore for wolfdales? anyone?

It varies from chip to chip.

WR304 · Post by **WR304** » Fri Feb 08, 2008 3:49 am

mcoleg wrote:srsly though, what is considered a default vcore for wolfdales? anyone?

It says 1.225v on the retail packaging.

It varies between chips but that seems to be what the default BIOS setting ought to be.

mcoleg · Post by **mcoleg** » Fri Feb 08, 2008 11:02 pm

WR304, yah, it does say something like that.

smilingcrow - i thought that was the normal behavior, wasn't it? there's been talk about intel implementing voltage regulation on the chip for a while; i think that's what we are seeing. if you look around, all the big m/b manufacturers have suddenly started talking about implementing "smart vcore" of one sort or another. they are not inventing it, they just try to figure out how to use the already existing options that intel gave them.

you can look at it this way - there's a default vcore for all chips of the same class - like 1.225v for e8400. however, the voltage regulation will undervolt the chip according to vid value.

if on some boards bios doesn't work well with that option, that would explain why there's not a whole lot of good overclocks on "default" vcore.

i also think that we have to abandon the notion of a "set" or "static" vcore in favor of "dynamic" vcore. instead of one number we now have a range of numbers the cpu can choose from. that would explain a lot about how these new chips perform.

WR304 · Post by **WR304** » Tue Feb 12, 2008 11:45 am

There's an interesting post about Wolfdale temperatures by unclewebb in this Xtremesystem.com thread:

http://www.xtremesystems.org/forums/sho ... 99&page=51

Measured with an infra red thermometer the temperatures for his E8400 CPU were 10-15c lower than what is reported by Coretemp.

smilingcrow · Post by **smilingcrow** » Tue Feb 12, 2008 12:28 pm

WR304 wrote:There's an interesting post about Wolfdale temperatures by unclewebb in this Xtremesystem.com thread:
http://www.xtremesystems.org/forums/sho ... 99&page=51
Measured with an infra red thermometer the temperatures for his E8400 CPU were 10-15c lower than what is reported by Coretemp.

Iâ€™ve thought for a while that people worry too much about what the absolute Core Temps are and should focus on how much headroom there is before thermal throttling kicks in. If my DTS temps are 20C below throttling and Iâ€™m comfortable with my other temps then Iâ€™m not concerned whether DTS is actually reading 80 or 65C.

If I buy a Wolfdale Iâ€™d like to be able to confirm when throttling is occurring so that I can calibrate the DTS temps that I get. I only say this because Iâ€™m still not clear whether the DTS offset values that CoreTemp reads are accurate for Wolfdale. Has anyone tried using RMClock to see if it recognises throttling with a Wolfdale?

Matija · Post by **Matija** » Tue Feb 12, 2008 12:34 pm

I think it's best to wait for a new revision.

Something's wrong with this one. Temps are all over the place, varying wildly even between chips from the same batch, and Lenovo supposedly returned a giant shipment of mobile Penryns because of DTS issues.

WR304 · Post by **WR304** » Tue Feb 12, 2008 2:32 pm

smilingcrow wrote:Has anyone tried using RMClock to see if it recognises throttling with a Wolfdale?

The official RMClock 2.30.1 doesn't work with my E8500 processor. It can't identify the CPU and doesn't display any voltages or core temperatures.

The more recent RMClock 2.35 (beta 3) appears to work with it though. The CPU appears as unknown but the other information for it appears. The displayed multiplier is still wrong as it ought to be 9.5 for a E8500. BIOS vcore for this CPU is set to 1.275v.

The indicated CPU temperatures are 5c lower than Coretemp 0.96.1 also.

RMClock 2.35 (beta 3) download:
http://forum.rightmark.org/topic.cgi?id=6:1423

RMClock 2.35 Beta E8500 CPU information

RMClock 2.35 Beta and E8500 CPU at 4ghz (Intel Speedstep enabled)

There are a few more relevant posts about temperature over the next few pages of that Xtremesystems thread. They expand a bit on the original post I linked to.

Individual Posts to save wading though the entire thread:

http://www.xtremesystems.org/forums/sho ... count=1308

http://www.xtremesystems.org/forums/sho ... count=1311

http://www.xtremesystems.org/forums/sho ... count=1328

mcoleg · Post by **mcoleg** » Tue Feb 12, 2008 8:31 pm

good find WR304; gonna try it later on.

Conroy · Post by **Conroy** » Thu Feb 21, 2008 9:01 am

Well, I decided to see if I could get up to 4ghz with this chip, and here are my results:

First, I did a day of memtest86 at 400fsb, 6:5 DDR2 speed to make sure my ram could run >900 Mhz. I had to adjust memory timings from 5-4-4-12 to 5-5-5-18 to get the computer to boot.

Then I did a day of folding+prime95 blend at 6x cpu, 450 fsb, 1:1 DDR2 to make sure the FSB could run >445 Mhz - I needed to bump my DDR2 voltage .1v to get this to run without errors all day. (HWMonitor reports the result of this as 1.92v, which is weird because "normal" ddr2 voltage gets reported as 1.87)

Finally, I tried testing with 9x cpu, 445 fsb, 1:1 DDR2.
I set the CPU voltage to 3.5 because the Intel processorfinder page shows a range of up to 1.3675 and I'm afraid to get too close to that.
Prime95 was only able to run for 6.5 hours before it aborted with an error.
Temperatures at load went up to 40 mb, 48 CPU and 46/50 on my cores.

I guess I'm done with that experiment, unless someone has a theory about how to get my 4ghz oc stable without raising cpu volts anymore.

Regardless, I just wanted to see whether it was possible and I'm not a fan of any overvolting, so I'm probably just going to stick with 3ghz/1.0375v from now on, unless I have a good reason to bump it to 3.6/1.225v for some task, which I don't really foresee. Stock voltage for my processor is 1.225.

mcoleg · Post by **mcoleg** » Sun Feb 24, 2008 8:29 pm

that could still be your ram or the north-bridge. relax memory timings even more and bump voltage a little on both.

Conroy · Post by **Conroy** » Fri Feb 29, 2008 8:34 am

For completeness, I did some more tests to compare power usage at my 24/7 settings of 3.0 Ghz/1.0375v. I don't know if anyone cares, but here it is

70 watts idle
98 watts w/ prime 95 running
154 watts max w/ Folding@Home and World in Conflict playing at the same time (CD not spinning)

I'm not 100% sure about the 154 watt number; it's hard to figure out the max because the wattage jumps all over the place and my kill-a-watt doesn't seem to keep statistics like that.

mcoleg · Post by **mcoleg** » Fri Feb 29, 2008 11:15 pm

there's a trick for that - prime small fft on both cores while running 3dmark (whichever version you have handy). that should push the system close to practical maximum output with the loads not jumping around too much. monitor kill-a-watt for 10-15 min and you'll get your max.

ZircularLogic · Post by **ZircularLogic** » Sat Mar 01, 2008 7:37 am

That's some really fantastic stuff about temperature readings and CoreTemp. Many thanks for posting those links, WR304! I've been fretting about temperatures but delta to Tjunction max is in excess of 45C, which puts actual Tjunction below 50C based on those experiments with the IR thermometer.

Overall I'm really happy with this chip. I found a max stable OC on stock voltage of 405x9 = 3645MHz. CPU-Z reports Vcore as 1.208V loaded/ 1.224V idle. Something close to 4GHz should be a gimme.

wayner · Post by **wayner** » Fri Mar 14, 2008 12:31 pm

For those that have the E8800 series CPUs - do you need to replace the stock Intel CPU fan with a quieter aftermarket fan or is the Intel fan smart enough to run quietly given the low heat levels of the CPU?

ZircularLogic · Post by **ZircularLogic** » Sat Mar 15, 2008 6:56 am

If you are running stock speed and volts (or lower) the stock cooler will provide adequate cooling, but it won't be quiet once that little fan gets moving. Most around here will tell you that the main advantage of an aftermarket cooler is that you can use a 120mm fan at low speed, or no fan at all. (Up to 3.6GHz, the air flow generated by the case and PS fans alone is sufficient for my Ninja to keep the CPU cool at full load.)

mcoleg · Post by **mcoleg** » Sat Mar 15, 2008 6:53 pm

i have not even taken the stock hs out of the box

still, i won't bet on it. it's a smaller (what is it, 70mm?) fan with high top rpm. if you want quiet, you might as well go that extra step and get an after-market hs.