p45 (asus p5q pro) / e8400 E0 voltages for 400 FSB?

[bludragon]

Just wondering what chipset and cpu core voltages people would expect to be needed to get an E8400 E0 to a 400 FSB on a P45 (asus p5q pro) board?

So far I've tried:

chipset 1.1v (auto)
cpu 1.2v (measured, set to auto in bios)
fsb 400
multi 7.5 (so cpu not oc'd)

Not stable

chipset 1.2v
cpu 1.2v (measured, set to 1.225 in bios)
fsb 408
multi 7.5 (so cpu not oc'd)

8hrs orthos stable

chipset 1.2v
cpu 1.25v (measured, set to 1.275 in bios)
fsb 408
multi 9 (cpu @3.6Ghz)

Not stable

CPU Temps are ok so far (<57 degC in orthos). Not sure about chipset temp though.

Also, should I be changing any other voltages? Memory and SB are not oc'd, so I'm assuming I can leave them at auto. Just looking at the oc guide, maybe I need to go back and find VTT so I can adjust that...

Seems like this is not quite as easy as people make out!

My goal is to get a rock stable 400fsb (so test at 408) with lowest possible voltages to keep noise levels down. I am quite enjoying stock with speedstep though...

 

[m0moma]

i have C0.. i just changed the fsb to 400 x 9.0 multi and it stayed at 3.6ghz vids at 1.2250v core voltage at 1.288v everything else auto

 

[walk2k]

Runs fine for me on auto vcore. It increases (automatically) to about 1.3x under load at 400x9 = 3.6Ghz. Temps were a bit too hot (76c) for my taste though with stock cooler so I've settled on 3.2Ghz (356x9) for now, though 3.3Ghz (366x9) is about the same really.

Only voltage I increased was vdimm, to 2.1v for DDR2-1066 ram.

At 3.2/3.3Ghz it shows 1.264v under load on auto vcore (1.256 at idle). Temps are a more reasonable 66-67c max (in Orthos, more like 56-57 in real world apps, ie. games). Intel thermal spec is 73.4 or something like that, so keep it under that and you're fine.

I also tested it (briefly) at 4Ghz (444x9) and it ran with no crashes or anything, again with auto vcore, but temps hit 78c in Orthos before I slammed the STOP button and rebooted

To be clear, all I did was set AI Overclock to manual, FSB to 400, then reset DRAM speed to (1066 or 800, whatever you have). Then reboot. That's it, 3.6Ghz easy.

 

[kpo6969]

C0 here
3.6
1.262 MIT, 1.236 bios, 1.216 cpuz, 1.20 cpuz load
Board has bad vdrop (auto sets @ 1.30)

 

[Zapper48]

My CO will run 400 fsb on auto VID is 1.225 but CPUZ shows 1.18 I can't set volts lower than VID in bios.

 

[darthjoe]

So, have you guys been getting stable overclocks with Vcore on Auto, and then backing it down in the bios to the lowest it will go and still boot? I did a simple overclock to 3.6 (400 x 9), and left Vcore on auto, CPU-Z shows the voltage at 1.3V, however, I don't think that the chip needs that much juice. I'm curious as to how much I should try to back the Vcore down to.

 

[datwater]

Originally posted by: darthjoe
So, have you guys been getting stable overclocks with Vcore on Auto, and then backing it down in the bios to the lowest it will go and still boot? I did a simple overclock to 3.6 (400 x 9), and left Vcore on auto, CPU-Z shows the voltage at 1.3V, however, I don't think that the chip needs that much juice. I'm curious as to how much I should try to back the Vcore down to.

I'm in the same boat - on a GA-EP35-D3SP. Auto goes to 1.3v. Been running fine like this for months. Temps are fine too. Should I bother backing down the voltage manually?

 

[bludragon]

I took my cpu voltage off auto because it appeared to bump it up to ~1.35 V at 400 FSB which I didn't think would be necessary.

My main issue is that I've now realised the system is not stable at stock cpu clock and volts at 400 FSB. It would seem that most people are not having that issue, is this a p45 thing, or just my bad luck? I need to do some more testing, but it's a slow process - change a setting, leave running overnight, repeat...

 

[badnewcastle]

I've had better luck on my Asus P45 when leaving all the voltages on auto and only changing the FSB, Mem Voltage, Mem timings.

 

[kpo6969]

Originally posted by: darthjoe
So, have you guys been getting stable overclocks with Vcore on Auto, and then backing it down in the bios to the lowest it will go and still boot? I did a simple overclock to 3.6 (400 x 9), and left Vcore on auto, CPU-Z shows the voltage at 1.3V, however, I don't think that the chip needs that much juice. I'm curious as to how much I should try to back the Vcore down to.

I started @ 1.275 and then worked it down.

 

[disports]

It's better not to leave vcore on auto because it usually keeps it higher than it should be. It takes patience to overclock, making sure your computer is stable. I have an E0 E8400 with a VID of 1.25 . I tried setting it to 400 FSB using 1.25 vcore and I don't think my computer posted. I think I needed 1.30 vcore to have it stable at 400 FSB. Set my ram voltage at 1.8, 5-5-5-15 stock timings. Right now, I'm at 1.35 vcore I think with 445 FSB, 1.8v ram voltage, 5-5-5-15 stock timings.

1.336 V idle
1.344 V load

 

[walk2k]

Alright I did some more tests.

E0 E8400, retail box HSF, P5Q-Pro

Vcore = AUTO
RAM is 1066 at 2.1v "stock"

For each test I ran Orthos for 3 minutes each, so max temps are probably not reached but comparable I guess, and vcore doesn't change so...
(At 3.2 I did a longer test because that's what I run it at normally).

----------------------------

3.0Ghz (AI tuner = auto)
334 FSB
RAM 1066 (5:8)

Idle: 1.064 - 1.104v (varies, most typical is 1.072).. = 41C
Load: 1.192v = 57C

---------------------------

3.2Ghz (AI = manual, changed FSB, all else = AUTO)
356 FSB
RAM 1071 (2:3)

Idle: 1.248-1.256v = 41C
Load: 1.264v = 64C (but after 6 minutes = 66C)

---------------------------

3.3Ghz
366 FSB
RAM 975 (3:4)

Idle: 1.256v = 41C
Load: 1.264v = 67C

---------------------------

3.6Ghz
400 FSB
RAM 1066 (3:4)

Idle: 1.304v = 41C
Load: 1.312v = 77C

---------------------------

4.0Ghz
444 FSB
RAM 1067 (5:6)

Idle: 1.328v = 41-42C
Load: 1.336v = 78C after 55 seconds, then I shut it down(!)

So.. I think 3.6Ghz is very doable, at 1.3xx volts anyway. Temps are a bit high for my taste, even though Orthos-level temps are never really reached in games. Suppose I could try backing down vcore, though really I think I'd be better off throwing $15 at a better HSF and maybe even doing 4.0Ghz

 

[nerp]

I think it's nuts that you need that much vcore for 3.6 with an E0. I have a C0 that is 24 hour prime blend stable, 100 pass intel burn test stable at 1.22!

 

[darthjoe]

Did you set those Vcores yourself Walk? My E0 will do 3.6 at 1.3V with the auto setting. 41C Idle and 61C under Orthos load. My original point is that I think 1.3V is more than what's needed for 3.6. Once I get the components into my new case, I'll have to play around with manually dropping the Vcore down.

 

[bludragon]

Interesting, maybe these early E0's are not quite all they're hyped up to be.

Darth, I believe Walk left the vcore on auto. My mobo would also give very similar volts on auto.

Hopefully I'll get some more time for experimentation over the weekend. So far I've gone back to making sure stock is 100% stable.
I think I'm going to start with bumping both the NB and FSB termination voltages up to 1.2V, in the hope that they can let me get away with a lower vcore.

 

[bludragon]

Just tried:

FSB 400
vcore auto = 1.31 load
nb 1.2
fsb termination v 1.2
linx temp 70
orthos temp 60
Error in orthos after 1hr possibly because I decided to try copying some files at that time...

Ho hum. It's getting to the point where I'm going to settle for stock for now.

 

[walk2k]

Yes like I said, vcore was left on auto. 1.312v max at 3.6Ghz. At 4.0Ghz 1.336 max.

It was hot, but stable. I didn't test it for hours, but I'm sure it would run indefinitely at 3.6.. (not so sure about 4.0). Never got any errors in Orthos.

I haven't had time to start tweaking vcore to see what's stable. Also I learned the hard way to disconnect the HDD during such tests.. I corrupted my Windows install once (on another computer).

 

[walk2k]

Originally posted by: bludragon
Just tried:

FSB 400
vcore auto = 1.31 load
nb 1.2
fsb termination v 1.2
linx temp 70
orthos temp 60
Error in orthos after 1hr possibly because I decided to try copying some files at that time...

Ho hum. It's getting to the point where I'm going to settle for stock for now.

I also got an error in Orthos once when I was messing around in SpeedFan or CPU-Z at the same time (not during the above tests). I don't know much about Orthos but it might not be a big deal to get an error if you try to multi-task at the same time.

The real test is, can you play your games or whatever real-world apps for hours at a time with no crashes or anything.

 

[disports]

lol well nerp, sometimes you get a good chip and sometimes you get a bad chip!

 

[walk2k]

Did some tests with manual vcore:

3.6Ghz / 400fsb
vcore set to 1.2500v, measures (cpu-z) 1.224v idle, 1.240v load
temps: 41c idle, 66-67c Orthos after 5-6 minutes.

Quite a difference from auto vcore = 1.312v and 77c!

 

[bludragon]

Originally posted by: walk2k
Did some tests with manual vcore:

3.6Ghz / 400fsb
vcore set to 1.2500v, measures (cpu-z) 1.224v idle, 1.240v load
temps: 41c idle, 66-67c Orthos after 5-6 minutes.

Quite a difference from auto vcore = 1.312v and 77c!

Nice! Wanna swap chips

Although

Originally posted by: walk2k
I don't know much about Orthos but it might not be a big deal to get an error if you try to multi-task at the same time.

Just to make sure I was starting from solid ground, I've had mine back at default clocks and running orthos the entire weekend (49 hrs and counting...) and no amount of multitasking has caused any errors yet. I guess it depends what you use the machine for, but I would be just too paranoid about some random file corruption, or other error, if I were to run at a speed that would error in orthos.

 

[walk2k]

Going lower

3.6Ghz
Vcore set to 1.2500 in bios, measures 1.208 in bios and 1.200v in CPU-Z (idle), 1.208-1.216v load = 64C max.

Games = 53C average.

3.6Ghz cool with retail HSF = pwnd.

As for the error, it only happened once and I haven't seen it again. Normally I agree you don't want to see errors ever, but with programs like Speedfan and CPU-Z that bang on the hardware directly you never know. Haven't seen any crashes or even hints of crashes in playing tons of games and stuff.

 

[Termie]

Walk - looks like you found the secret - manually setting vcore. I've been running my C0 at 3.6/400 for a while, setting vcore manually in the bios at something like 1.2125, with CPU-Z at idle indicating 1.19. Sounds like others have similar results, and yours is pretty similar too. I'm surprised the E0 stepping isn't better, but your isn't much worse, at least. Your temps, however, are indeed very cool. I game at 55C with an aftermarket Nirvana heatsink, so you're doing very well there. Maybe that's the big advantage of E0.

 

[darthjoe]

I was able to play around with my E0 8400 yesterday. 9 x 400, and manually lowered the vcore to 1.250V. Per Realtemp, I'm running around 32-33 C idle and 48-49C under Prime 95 load.

Maybe I'll shoot for 4.0ghz this week

 

[walk2k]

Played a bunch of games, some like HL2 barely tip 50C on 1 core, the other running at more likc 45C.... lol. ETQW runs both cores at about 53-54C when running a server with 16 bots...

It's possible I could push vcore lower too, I may try again if I feel adventurous... Then again 1.21v is pretty close to what it runs at stock (1.19v) so...

Did have 1 crash in ETQW, software error, didn't bring the whole system down but I'll keep an eye on it. Normally a pretty stable game, but nothing is 100%. HL2, both Eps, and Bioshock all ran perfectly for 30-40mins at a time (all I had time for yesterday).