How much voltage for 3.2GHz/E6400?

kent

2[H]4U
Joined
Apr 8, 2001
Messages
2,603
I was Orthos stable at 1.39/1.37 (idle/vdroop load) for around 13 hours.

The machine completely reset itself, which I attribute to a voltage issue as my PSU (Corsair HX620) is the most impressive PSU I've EVER owned. The voltages don't budge under load (not measured with DMM, FWIW)

I went up to 1.45v in the BIOS which results in 1.41/1.39 (idle/vdroop load) and I'm Orthosing again now.

What voltage have most people required? BTW my Orthos load temps are below 60C measured with TAT/ASUS Probe/CoreTemp/SpeedFan
 
I went about 11 hours running Orthos yesterday at 420 MHz x 8 with the CPU set to 1.45 volts in the bios.

Have you tried adjusting the northbridge voltage yet? I had the NB at 1.45 volts yesterday for the above run but 1.55 volts won't hurt it with the Asus heatpipe cooler.
The Gigabyte DS3 feels like an EasyBake oven compared to the P5B Deluxe. :D

Here's how I've set the memory timings in the bios.

biostimingsvx0.jpg


I set the Write to Precharge Delay to 11 since Asus did that in their 0910 bios so I thought it must be a good idea.
Compared to AMD systems, there's virtually nothing to be gained by adjusting memory timings on the Intel 965 platform.

I think 1.40 volts in the Asus bios should be adequate to get 3200 MHz out of an E6400.
Time for another Orthos run at that setting to find out.

With the bios set at 1.400 volts, CPU-z is showing 1.344 volts with Orthos running. So far so good.

I like CoreTemp and it's reporting 56C Max so far with the Intel heatsink / fan.
 
Seriously, I wish you were my uncle. You don't even know!

I just got back home, so far 3+ hours at 1.39v under load (1.45v in BIOS)

I briefly had my NB at 1.45v, but I've backed it back down to 1.25v (I think that's the first selectable option, after AUTO). If I'm not stable again, I'll bump it back up along with some more vCore.

I'm not trying for anything more than 3.2GHz. I saw your earlier post about 400 vs 401, and if I'm stable at 401 for 24 hours, I'll go back to 400 and see. I'd like the extra GB of memory bandwidth, sure couldn't hurt!

Also, the ONLY memory settings I changed was to 4-4-4-12, I left everything else alone. I saw you changed some things -- how did you figure stability would be achieved?


EDIT: After rereading your post, you mention you may only need 1.34v to reach 3.2GHz. Why do you think I'm going to need more voltage? If I'm not stable at 1.39v (1.45v BIOS), then I plan to keep adding more until stable. Do you think it's a possibility I got a bum processor?
 
Every CPU or chip is different. Not all overclock the same or require the same amount of volts.

It also depends on the rest of your hardware, motherboard, RAM, PSU...

Some people can overclock an E6600 to 3.6ghz with 1.288V and be Orthos stable.

Others cannot even reach 3.6ghz with 1.6V

I can do 4GHZ with 1.52V and I've seen better!
 
8.5 hours Orthos stable now. I estimate I was previously stable around 13-14 hours. I know it was at least 12.

I always go to 24 hours, then I follow up with memtest overnight and then 3DMark overnight as well.


SpeedFan has gotten wacky, reporting 85C on the Core, but TAT and CoreTemp are showing 53-55C. Not sure why, but I'll believe TAT and CoreTemp before SpeedFan/ASUSProbe. It seems whatever probe those programs was reading is the same. I doubt it's an issue.
 
orthos6h43mdw3.png


Testing with the CPU voltage set to 1.400 in the bios is looking good so far. I'll post an update tomorrow morning.

With all your testing kent do you ever have any time left over to actually use your computer? :)

In my early days of overclocking the Asus P5B Deluxe I had most things set to AUTO and was doing pretty good but adding extra CPU voltage wasn't really getting me too far.

The thing that has helped me the most with stability is a little extra northbridge voltage.

On the Asus board I think you could run 1.55 volts for the NB, 24/7 and never have a problem but 1.45 volts for your 400 X 8 goal should be fine.
That's what I'm presently using.
 
Well, it didn't fail Orthos, it just reset again.

I'm fairly sure my memory is stable at 4-4-4-12, seeing as that's what's written into the EPP profile, though the P5B doesn't support EPP.


What do you guys think? I've upped the NB voltage to 1.45 and upped the Core to whatever results in 1.42v under load. Orthosing has begun... again.

Core temps reading in the low 50C (53C currently)
 
My E6400 did 3200 at stock voltage with the stock cooler. But it all depends on the motherboard, the chip, and on your workmanship in putting the system together. And perserverance let's not forget perserverance. ;)
 
orthos19h08mcj8.png


No problems here running my E6400 at 8 X 400 with the CPU set to 1.400 volts in the bios. The northbridge is at 1.45 volts.

You might want to try running your memory at CL5-5-5-15. Not all memory is created equal.
Setting your memory to CL5 will help eliminate one variable that might be holding you back.

Memory voltage is also very important. For testing I would try 2.25 volts and then once you find stability try backing it down a step at a time.

In the SPD settings for my memory CPU-z shows CL3-4-3-9 at 400 MHz but that's wishful thinking.
For stability at those timings I need to feed my D9GMH chips 2.40 volts. No thanks.
 
I got mine to 3200 with 1.35v Bios (1.3v CPU-Z), but it took 1.387v to get another 50mhz. I guess mine is kinda' so-so compared to others here as well.
 
I think it's my memory.

I have the Corsair XMS2-800 (PC2-6400) 4-4-4-12 @ 2.1V (according to newegg and Corsair's own PDF on the stuff)

In my BIOS I have it set for 4-4-4-12 @ 2.15v I figured an extra .5 wouldn't hurt.


I burned a brand new memtest ISO CD and booted it. The memory passed for 2 rounds (all 8 tests) BUT THEN IT RESET ITSELF sometime during the third test.


Good news: I bought it from Newegg, so I can RMA + Refund.

Bad news: I have no other DDR2 to "test" to confirm this.
 
If you're considering an RMA for your memory then don't be afraid to give it some more juice.

I've run my sticks at 2.40 volts and the house didn't catch on fire.

My old Ballistix PC3200 memory would get burning hot but they continue to perform fine. If you're worried just place an extra fan on the memory when you're testing things out.

Dave: What is your CPU voltage set to in the bios? You say you're running stock volts but if it is set to AUTO in the bios then the voltage your CPU is actually getting is not stock. At AUTO voltage varies with FSB speed. It's a good idea when you're overclocking to lock it in place at a fixed value.

As far as I know, Intel rates these C2D processors at 1.325 volts.
 
Well, if the house catches on fire I can order a sweet boutique system with the insurance money! That's one upside.

At any rate, I'm back at school now :)mad: ) and I'll check it out when I get home.


Before I left, I reset everything to stock, save for the memory voltage at 2.15. It's running memtest now and I plan to let it run overnight.

I played HL2 for 30-45 minutes without an error earlier, too so I dunno WTF to think. It's so stable and so fast and shit it passed Orthos for over 12 hours, MOST would call that stable but I like those 24 hour tests. Why do I have to be so picky :D
 
Anyone ever not get their machine Orthos stable and just say fuck it and call it a day?

I played Half Life 2 for like an hour with no problems. Damn thing changes levels so fast it's scary! (I attribute a good chunk of that to my Raptor)


I'm going to see if I can loop 3DMark till I go to bed, then I'll memtest overnight.

BTW: I'm now at 400, not 401. I want MAXIMUM STRESS, lol. CPU Vcore is reading 1.34v under load which is less than the Intel "maximum"
 
Anyone ever not get their machine Orthos stable and just say fuck it and call it a day?

I played Half Life 2 for like an hour with no problems. Damn thing changes levels so fast it's scary! (I attribute a good chunk of that to my Raptor)


I'm going to see if I can loop 3DMark till I go to bed, then I'll memtest overnight.

BTW: I'm now at 400, not 401. I want MAXIMUM STRESS, lol. CPU Vcore is reading 1.34v under load which is less than the Intel "maximum"


get a 2B(or any B actaully, 8B being best but no HB types) pencil and do the vdroop pencil mod. all you have to do is colur in one capacitor. i bet itll be stable all day at same voltage if you do. btw with regards voltage every cpu is different. as long as you arent loading(small ftts) too high above 65C, then add more vcore.
btw have you tried running TAT, another good way to stress the CPU. actualll heats it up more than orthos.

for the pencil mod, just shade in that cap(instead of soldering a VR to it)
p5bvdroop.jpg
 
kent can you read exactly what is written on your DIMMS when you get home.

because if its C4D kit v1.1 then its D9GMH. hence you could up the voltage to 2.3-2.4 and they will scale higher. also recommend a low cfm fan over them(just use a zip-tie to hang it from somewhere)
 
I can't remember what revision my memory is, but I know it's the ProMOS chips. As soon as I got the memory I looked it up on that memlist website (can anyone link me, I forgot it!!!) but it was ProMOS
 
What's a 2B pencil? I've got a #2 pencil lol.

Anyways, what do I shade in? You said capacitor and I don't see any on that picture. Can you edit it in MS Paint for me, please :cool:
 
What's a 2B pencil? I've got a #2 pencil lol.

Anyways, what do I shade in? You said capacitor and I don't see any on that picture. Can you edit it in MS Paint for me, please :cool:


they are pencils with graphite as the lead without the additives that HB pencils have. they range from B up to 8B but it must be a number before the B. look in any art shop or some stationary shops should have them.

picture below showing the capacitor to shade in. if you have a multimeter you can check the resistance but its not really possible to lower it too much with a pencil. also after shading cover it with some tape to keep the graphite in place.

p5bvdroop1.JPG
 
Thank you! I may try out the pencil mod later today.

Get this, I'm running Orthos small FFT + Intel's TAT load simulator + Memtest (for Windows) all at the same time.

So far, memtest has hit 1100% coverage, Orthos hasn't failed, and temps are still below 60C. Stable in this environment for almost 8 hours now!
 
Now almost 2000% coverage on the memory

Everything....still....stable....! :D :D
 
12.5 hours of Orthos stable...

I'm just about to call it a day. I'm headed out to the law library soon and when I get back in about 4-5 hours if it's still running I'm calling it stable.

Memory test is now over 2000%
 
12.5 hours of Orthos stable...

I'm just about to call it a day. I'm headed out to the law library soon and when I get back in about 4-5 hours if it's still running I'm calling it stable.

Memory test is now over 2000%

Real nerds Orthos for 24 hours!!! ;)
 
My e6400 before the current one required 1.34v. Now the new one requires 1.575 in the bios to stay stable at 3.2! Its horrible.
 
I think I have mine set at 1.400 or 1.4125 in the BIOS. Whatever it is, it results in 1.36v when TAT is workloading, and 1.37v during games and Orthos.

It's possible I could go even lower. And I know real nerds Orthos for 24 hours, I may let it continue but I feel like Orthos + TAT workload simulator + MemTest (for Windows) = hella stress.

Something else I wanted to mention, I moved around the plugs on my PSU from 3, 4, 5 to 1, 3, 5 and I changed RAM voltage to 2.25 ("stock" would be 2.1)
 
16.25 hours Orthos stable.

Something I realized around hour 13 was TAT's workload simulator was preventing Orthos from really doing its job. I mean, it was completing tests and such, but hadn't completed one round in 12 hours.

I don't chalk that up to anything terrible, as it was running and I figure TAT's simulator is pretty hardcore as it gets the temps higher than Orthos.

Once memtest hit 2000% on both sets of RAM I was testing I closed those. 20 passes = OK.

For what it's worth, it's never Orthosed this long before without resetting. So either it was the PSU being improperly loaded (I had my PCI-E and all my devices on one rail, I believe) or it was the memory needing 2.25v to be stable. Or my hardware didn't like the 401MHz bus perhaps, I've heard weirder things.

At any rate, we'll see if it makes it to midnight and if so -- it's stable!
 
I've settled on 3360 MHz for my 24 hour setting so now it's time to explore the outer limits and I think I'm pretty close with the Intel heatsink and fan.

I had a look through the [H] overclocking database and discovered that it's going to take over 3500 MHz to be the fastest Intel cooled E6400 so here's my shot at it.

hotc2dmp3.png


It was running at this spec for about 3 hours earlier today but it reset itself and rebooted just like your computer does kent. With 1.5125 volts on the CPU, temps are on the borderline of sanity. If it's still running tomorrow morning, I'll be the first to post. :)
 
I was 30+ hours stable w/ Orthos.

3200MHz (400FSB/400MEM) w/ 1.36v under load.


Very very happy. Not sure if it was the PSU needing it's modular cables switched around or the memory needing more voltage. I'll blame the memory, either way it's Corsair's fault HAHA


3200stable.JPG
 
kent: Better enter that one into the [H]ard OC database. Nice work.

I upped my memory back to 2.25 volts and my computer didn't re-boot while running Orthos like last time.

With Orthos still running this morning I thought for sure that I had done it and would be able to get into the OC database but have a look at this:

hard3520loadev2.jpg


Do you see the problem? I've been screwed by Microsoft! It must have adjusted my clock by 3 seconds during the night so now it looks like my hard earned results are bogus. Bastards! :(

hard3520idle2nv2.jpg


pureevilyt6.jpg


And the wife wonders why I never get a damn thing done during the day while she's hard at work!
 
Sure thing unclewebb, once I get back home and I find my Intel box I can read the LXXX and AXXX numbers off of it. I think the OCDB requires those.

I am going to shoot a line to Corsair and say "HEY, YOUR RAM ISN'T STABLE AT 2.1V" because I don't want any warranty issues with them. At least I have one year to return it to Newegg, and by then I may be wanting a Quad so we'll see.
 
1.3475 (something like that) @ 3.2ghz or orthos fails.

Will have to run orthos for 24 hours but my system soo far runs stable.
 
Back
Top