• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Abit IP35-e problem

brikis98

Diamond Member
I just put together the following system: Abit IP35-E, Intel C2D E8400, Tuniq Tower HSF, 2x1GB Corsair Ballistics, MSI X1900XTX, Corsair HX520, latest BIOS, latest intel chipset drivers, Windows XP SP3. I wanted to OC this thing, but I decided to check my voltages & temps first, and the following values (all at idle) seem very odd to me:

Speed Fan

CPU: 28C
AUX: 38C
Vcore: 1.13V
+12V: 6.39V
Avcc: 3.25V

HWMonitor

CPU: 28C
CPU Core 0: 43C
CPU Core 1: 43C
ACPI -> THRM: 28C
CPU Vcore: 1.18V
+12V: 6.69V
+5V: 3.74V

I also tried Abit EQ (software that came with the mobo) and HMonitor and neither one showed ANY readings. And even the 2 programs that do get readings are worrying me because:

1. The CPU temp (28C) NEVER changes, even with two instances of Prime95 running a torture test.
2. The CPU core temps seem awfully high for a tuniq tower and also NEVER change.
3. The AUX reading is the only temperature that changes with Prime95 running, jumping up to ~50C after a while.
4. The +12V reading can't possibly be right, or the computer wouldn't boot. Same goes for the +5V reading.


Any ideas what's wrong? What can I do to get accurate readings so I can (safely) try to OC this thing?


update: looks like this is actually an e8400 temperature reading problem to which the abit ip35 series is especially vulnerable. there is a discussion of it here, here and here. It turns out the speedfan 4.34 beta does slightly better with readings. mine is showing:

CPU: 28C
AUX: 46C
Core 0: 38C
Core 1: 38C

These seem reasonable, but unfortunately, they NEVER change, even with two instances of Prime95 torture test! this is quite worrying... and worse yet, goddamn Abit EQ doesn't work at all! this page discusses a SpeedFan config that's supposed to work, but I'm not seeing any good results. Only info that came out of it is that "AUX" in SpeedFan is actually "PWM"... I also tried Core Temp, but it just shows both cores at 43C, a value that never changes, even with two instances of Prime95 torture test running...


Update 2

After switching to the v16 (beta) BIOS, the temp readings are still funky at idle, but work under load. SpeedFan 4.34 beta seems to do the best job of reading the temps. I spent some time OCing this CPU (details through out the thread) and have finally settled on the following:

445x9 (4.05Ghz), 1.38 vcore (1.34 in Windows), 2.2V for RAM, NB/SB/VTT default.

Just finished a ~4 hour Prime95 (two instances) run with no issues. Core temps level out at 55C. Not bad.
 
Originally posted by: brikis98
These seem reasonable, but unfortunately, they NEVER change, even with two instances of Prime95 torture test!
well, if you've been doing your reading as you appear to have been then again you will know that this is common behaviour on some E8xxx CPUs across various mobos from various manufacturers.

 
Originally posted by: Heidfirst
Originally posted by: brikis98
These seem reasonable, but unfortunately, they NEVER change, even with two instances of Prime95 torture test!
well, if you've been doing your reading as you appear to have been then again you will know that this is common behaviour on some E8xxx CPUs across various mobos from various manufacturers.

i have seen this, but some people have also been able to get various programs to work for them, so I was hoping some1 might have a solution for my particular configuration as well. for example, one symptom I have that I haven't seen others experience is that the Abit EQ software that came with the mobo doesn't show ANY readings whatsoever...
 
Update:

for the hell of it, I decided to see how it would behave when OCing and noticed some more strange behavior. I started with moderate settings (as far as an e8400 goes, anyway):

* FSB: 400Mhz (9x400=3.6ghz)
* RAM divider: 1:1 (ddr2 800)
* CPU voltage: 1.38V (def higher than necessary in retrospect)
* RAM voltage: 2.1V

The computer booted up fine and passed Prime95 torture test (2 instances) for 25 min... What's odd is that, after some random time, the various temperature probes finally showed different temps! SpeedFan showed the cores at 55C and Coretemp & HWMonitor at 60C. However, the temps were fluctuating very quickly, so I'm still not sure if they were accurate. They also didn't change for the first ~10 min of the test from their original values (38C), which is a bit suspicous.

It seems as if any core temp below 38C is just not reported... This might explain why, at stock settings, if the Tuniq Tower was keeping the CPU under 38C (even under load), I never saw the temps change at all...

I'm now going to back the voltage down and see if my theory is correct...

EDIT ---> I just noticed something that I'm not a big fan of: I set the CPU Voltage to 1.32V in BIOS. However, in CPU-Z, it shows up as 1.22V idle and 1.208V under load. I know the numer will show up less in Windows, but should it really be this much less?
 
Originally posted by: brikis98
EDIT ---> I just noticed something that I'm not a big fan of: I set the CPU Voltage to 1.32V in BIOS. However, in CPU-Z, it shows up as 1.22V idle and 1.208V under load. I know the numer will show up less in Windows, but should it really be this much less?
Edit/Delete Message
if C1E & EIST are active, yes.

 
Originally posted by: Heidfirst
Originally posted by: brikis98
EDIT ---> I just noticed something that I'm not a big fan of: I set the CPU Voltage to 1.32V in BIOS. However, in CPU-Z, it shows up as 1.22V idle and 1.208V under load. I know the numer will show up less in Windows, but should it really be this much less?
Edit/Delete Message
if C1E & EIST are active, yes.

both are disabled...
 

thx 🙂

just tried the new new beta BIOS and unfortunately, the results aren't any better:

* CPU temp is stuck at 28C
* core temps are stuck at 38C (SpeedFan) and 43C (HWMonitor/CoreTemp)
* Abit EQ still doesn't work
* Only difference I've seen, so far, is that after flashing to the beta BIOS, it allows me to select half multiplier settings (8.5, 9.5, etc) and it defaulted my E8400 to 3.2ghz instead of the default 3.0ghz...
 
One more update:

I tried some OCing of my E8400 with the beta BIOS and the results are a bit more promising:

* I got to 3.6ghz with only a tiny bump in voltage (1.24V) - Prime95 stable 30 min
* I got to 3.8ghz with another small bump in voltage (1.28V) - Prime95 stable 30 min
* I could not get 3.9ghz to be Prime95 stable - computer would totally lock up within 15 seconds of starting the test - even when increasing the CPU voltage all the way to 1.4V... either my CPU/mobo has hit a wall here, or I need to tweak other settings, such as RAM voltage (currently at 2.2V), RAM timings, NB or SB voltage (what are safe ranges for these?), or CPU VTT (what is a safe range for this?).

The biggest difference I'm seeing with this new beta BIOS - and I'm hoping it's not a fluke or my imagination - is less vdroop. It's roughly ~.02V idle and ~0.04 under load. This is a big improvement over the old numbers (.1V and .12V, respectively) and probably responsible for allowing me to get beyond 3.6ghz...
 
another update with the beta 16 BIOS:

* it seems that the mobo does NOT like me to manually set RAM timings - each time I do, I get instability or no POST. so, i left them on auto for now, which works great (pushed my RAM to ddr2 1050 with ease).
* got my CPU up to 445x9 = 4.05Ghz @ 1.38V (1.33V in Windows). Prime95 (two instances of torture test) stable for 30 min, will continue testing/tweaking later.
* At 4.05Ghz, the tuniq tower is keeping the cores at < 55C with the fan speed between 55% (inaudible) and 75% (slightly audible). I suspect that I can keep the same OC with a slightly lower vcore, but i'll have to do lots of stress testing to be sure.

all in all, the results are respectable. i'll push the system to see the max OC i can get, but as I want my system to be quiet, 4.05 will prob be where I stop for 24/7 use, especially if I can get it with less volts.
 
Today's progress:

* 456x9 (4.1Ghz), 1.40V vcore (~1.36V in Windows): Prime95 stable for 30 minutes. Tuniq tower keeps cores at 58C with fan at 75%.
* 467x9 (4.2Ghz), 1.42V vcore (~1.38V in Windows): Fails Prime95 after only a few seconds. Looks like this is where the 9x multiplier maxes out.
* 495x8.5 (4.2Ghz), 1.42V vcore (~1.38V in Windows): Prime95 stable for 30 minutes. Tuniq tower, with fan at 80%, kept the cores at 60C. SuperPI time of 10.813s.
* 525x8 (4.2Ghz), 1.40V vcore (~1.36V in Windows): Boots into Windows fine, runs SuperPI in 10.843s. Unfortunately, fails Prime95 after just a few seconds.
* 506x8.5 (4.3Ghz), 1.44V vcore (~1.40V in Windows): Booted into Windows fine and ran SuperPI in just 10.563 seconds (!). However, Prime95 failed after just a few seconds and Windows behaves oddly. I think this is real close to the chip's limit.

I didn't have time to do long stress tests or to try all the permutations of settings. The highest stable OC I got was 495*8.5 (4.2Ghz), but the core temps got to 60C and the CPU fan was up fairly high, so I don't think I'd want this as my 24/7 configuration. In order to have a quieter/cooler system, I'm now running the following:

445x9 (4.05Ghz), 1.36V Vcore (~1.32V in Windows): SuperPI time of 11.391s. Prime95 stable for 30 minutes. Tuniq Tower keeps the cores at < 55C at just 45% fan speed (inaudible). This is probably the best compromise between speed and heat/noise/energy usage. Would be even better if it was stable with a lower vcore - I'll test that out this weekend...
 
turns out that 445x9 (4.05Ghz) with 1.36 vcore isn't quite stable enough: it fails Prime95 after about an hour. that's a shame, because that was the voltage threshold that kept the temps low enough that my Tuniq Tower's fan could run at just 45%... seeing some of the results online, I was hoping to be able to use an even lower voltage, but with the vdroop, this doesn't seem too realistic now.

i've bumped the voltage to 1.38V and it has made it pass the 1.5hr mark... I'll let it run for ~6 hours, at which point I think it's "good enough"... cores are at 55C, although the CPU fan has to get up to 75% to keep them there... maybe i just need to get a quieter fan for the Tuniq Tower...


EDIT -> forgot to mention that, to be fair, i'm still on a beta BIOS. maybe the full version will be more stable and OC a little better, although I have no clue when it'll be out.
 
Thanks and this has been very helpful. Keep us posted.

My BIOS version is 15, and I too see a .09V discrepancy between BIOS and CPU-Z vcores at idle. I was able to overclock it to 3.6GHz using a 1.325V/1.24V vcore with the stock cooler. Right now I'm awaiting a better CPU cooler as well as the final release of v16 BIOS. It looks promising...
 
Ok, I think I have my final settings, at least for this beta BIOS: 445x9 (4.05Ghz), 1.38 vcore (1.34 in Windows), 2.2V for RAM, NB/SB/VTT default . Just finished a ~4 hour Prime95 (two instances) run with no issues. Core temps level out at 55C. Not bad.
 
Originally posted by: brikis98

Just finished a ~4 hour Prime95 (two instances) run with no issues. Core temps level out at 55C. Not bad.

holy smokes thats a cool running chip!

you could no doubt go higher with some ddr-1000 or ddr 1066. Looks like you'll be held beneath 4.15ghz by your ddr2-800. Nice temps though!. Also, try Lavalyst Everest to read your temps.

Originally posted by: brikis98
which works great (pushed my RAM to ddr2 1050 with ease).

on stock timings? NICE! I wanted ram like that. I was running prime at 4500mhz on air with an e8400 and ip35e on bios 14.b05. I was limited by my supertalent ram at 890mhz and had to resort to 6 6 6 29 timings and clockgen to 4.5ghz. Prime temps went to 86C
 
Back
Top