Problems OCing

[Ruger22C]

E8400, GA-P35-DS3L rev 2.0 with F8a (newest bios except for BETA), 4gB DDR2 800 (2x2) OCZ Reaper HPC, Xigmatek 120mm, RC-590.

I can run 380x9 x2.4 (equals 3.4ghz & 912mhz) stable, but if I lower the multiplier to 2.0 I cannot even get 380x9 x2.0 ( 760mhz ram) stable.
If I increase the CPU volt from 1.15 to 1.156, I get almost immediate BSOD.
RAM @ 2.0x, 2.0v and 2.2v is best, 2.1v gives errors soonest. All three give errors very fast, though.

If I go to the 2.0x (RAM Multiplier), and do a prime 95 test.. I get an error very quickly when doing the test that runs ram the hardest, vs test that runs primarily CPU, an error takes much longer if it ever comes.
What's up with this?

Please help, thanks.

I ran 378x9 x2.4 with STOCK cooling, so with the huge upgrade, I'll be very upset if I can't go any higher at all.

 

[imported_wired247]

loosen up the ram timings and keep it at 1:1 DRAM:FSB until you get the FSB and CPU OC in order.

also, the newest bios isn't always the one you want.

 

[Ruger22C]

I Have to use f8a, to support wolfdale.
I can't set the RAM timings, no where in BIOS to do it unless I'm missing something.

How do do 1:1 dram/fsb?

Thanks.

 

[Ruger22C]

Maybe this will be clearer:

@ 333x9 = 3GHZ, 2.4x multiplier on RAM is 800MHZ.
That's stock.
@ 380x9 = 3.42GHZ, 2.4 multiplier on RAM is 912 MHZ.
That's stable.
@ 382x9 = 3.44GHZ, 2.4 multiplier on RAM, is 916 MHZ.
That's unstable.
@ 380x9 = 3.42GHZ, 2.0 multiplier on RAM, is 760 MHZ.
That's unstable.
@ 400x9 = 3.6GHZ, 2.0 multiplier on RAM, is 800 MHZ.
That's unstable, too.

If I increase V-Core from 1.15v (stock) to 1.156v, I get BSOD.
Anywhere from 2.0 to 2.2v on RAM causes errors on 2.0x multiplier.

WHY DO I KEEP GETTING ERRORS / CRASH?

 

[imported_wired247]

you'll need to talk to someone who knows gigabyte bios a little better. I've never owned a gigabyte board so I would be hard pressed to hold your hand through changing all your settings.


there must be a way to set ram timings.

see if you can boot it at 333FSB and DDR2-667 (what i suppose you are calling a 2.0x multiplier) which is really DRAM:FSB = 1:1

1.156V could be quite low for 400x9.

You could try pulling all but 1 ram stick (if that doesn't work, then try swapping it for your other ram stick) to see if that helps you boot.

To be blunt, it sounds like you have no idea what you're doing, but it also sounds like something fishy is going on, like unsupported FSB to NB strap, or marginally compatible RAM for your mobo.

 

[BonzaiDuck]

Originally posted by: Ruger22C
Maybe this will be clearer:

@ 333x9 = 3GHZ, 2.4x multiplier on RAM is 800MHZ.
That's stock.
@ 380x9 = 3.42GHZ, 2.4 multiplier on RAM is 912 MHZ.
That's stable.
@ 382x9 = 3.44GHZ, 2.4 multiplier on RAM, is 916 MHZ.
That's unstable.
@ 380x9 = 3.42GHZ, 2.0 multiplier on RAM, is 760 MHZ.
That's unstable.
@ 400x9 = 3.6GHZ, 2.0 multiplier on RAM, is 800 MHZ.
That's unstable, too.

If I increase V-Core from 1.15v (stock) to 1.156v, I get BSOD.
Anywhere from 2.0 to 2.2v on RAM causes errors on 2.0x multiplier.

WHY DO I KEEP GETTING ERRORS / CRASH?

Not familiar with that chipset. However, I am familiar with the E8400. I can tell you right away that you have a lot of room there for voltage increase. The retail-box "label" spec is 1.25V "maximum." The "safe" range for CPU_Z readings is 0.85 to 1.36+V. If the values you show here are your "set" values or "settings," I'd say again that you should probably bump up the VCORE to at least 1.25V.

Also, look more closely at how you're setting the RAM speed. For DDR, and 1:1, the memory bus speed would equal the CPU FSB frequency of 400 Mhz, and DDR would equal 800.

 

[imported_wired247]

if this is your ram

http://www.newegg.com/Product/...x?Item=N82E16820227284

then the correct voltage is 1.8V ... do not set it at 2.0 or 2.1 for no reason...

the correct speed for that RAM is ddr2-800, which is FSB 400 (multiplier 2), 2x400 = DDR2-800

+1 on bonzai duck's post


Also, that ram's timing is 5-5-5-18. If your mobo is somehow set to 5-5-5-15 (another common timing) then you could get crashes / no post




But, if THIS is your RAM (I'm guessing here, because you did not specify)

http://www.newegg.com/Product/...x?Item=N82E16820227267

then it does say that the mfg spec is 2.1V. and the default timings are 4-4-4-12 which is pretty "tight" timings. Just because the MFG spec'd the ram at those speeds doesn't GUARANTEE that the RAM will be able to perform at those speeds with your mobo.

However, that is a rather high voltage setting for DDR2-800. Best compatibility is usually achieved at DDR2 voltage around 1.8-1.9V.

So, like I said initially, loosen up your timings, try 6-6-6-18 and 1.8 to 1.9V at first, (consult your mobo manual if you don't know how... there is practically no way in hell the timings are not adjustable)

start at FSB 333 and DDR2-667 and go from there.


In summary, my guess here is that your mobo is only marginally compatible with your RAM which is at what I would call a high voltage specification and very tight default timings. If this turns out to be the case, it is a good lesson for the future as it's usually better to buy RAM that is specified at lower voltages and reasonable/non-exotic timings or speeds stock.
THEN, once you know your RAM is stable at its specified speed you can always ATTEMPT to overclock / overvolt / tighten timings / etc.

 

[Ruger22C]

"Features Patent-pending Reaper HPC Heatsink 2.2V EVP"

At below 2v as I recall, my PC restarts it's self about 6 times, won't even get to boot loader. I'll try it again.
I have the first RAM.

"You could try pulling all but 1 ram stick (if that doesn't work, then try swapping it for your other ram stick) to see if that helps you boot."

I can't boot if I change Vcore volts... When I said 'unstable,' I just meant errors in p95. It still boots fine.

"see if you can boot it at 333FSB and DDR2-667 "

.. I can boot. I just get errors in p95.

"1.156V could be quite low for 400x9."

1.15 works, but with errors in P95. 1.156v is a flat-out BSOD.

"there must be a way to set ram timings."

If someone can show me, I cannot find it in BIOS....!!!!!!

I'll go try lower volts, be back. thanks.

 

[Ruger22C]

At 1.8v, working up to 385FSB & 2.0x multiplier. Can't go higher.
Cannot get 390 FSB with 1.9v... 385 is my limit, 1.8v.
That's an increase of about 10 MHZ from what I was able to get before with 2.1v?
.. I'll try CPU voltage...

 

[Ruger22C]

395FSB @ 1.25v..
Yes, I'm a "noob." It's odd to me, that 1.15 doesn't BSOD and 1.25 doesn't BSOD, but 1.156 gives BSOD.

I'll see if I can find memory timings anywhere...

 

[Ruger22C]

O.K., here it is.. I googled and found out how to get to timings... (It's far from obvious)

What do I change?

CAS in BIOS

 

[BonzaiDuck]

Originally posted by: Ruger22C
At 1.8v, working up to 385FSB & 2.0x multiplier. Can't go higher.
Cannot get 390 FSB with 1.9v... 385 is my limit, 1.8v.
That's an increase of about 10 MHZ from what I was able to get before with 2.1v?
.. I'll try CPU voltage...

Some RAM-makers give you their "warranty" voltage range, although you can start below the bottom of it. I'm guessing that -- as was my experience with OCZ modules of the DDR1 era -- these are max. voltage specs. But EVP gives you margin above that. I forgot what the margin is, but they post it at the OCZ web-site. So it would be safe to set the voltage on those things initially at must below the spec. When you do that, and run them at their spec speed (or below) and latencies, you eliminate the RAM from the equation, allowing you to tweak the VCORE, VTT and other voltages.

But the focus at this point must be the VCORE. Especially, if you're going to run the E8400 up to 3.6 Ghz at the stock multiplier of 9, There's nothing wrong with setting the voltage initially to 1.25V. If that gives you "readings" of 1.20 or lower, I'd say . . . bump it up again. The fact of the matter: on the E0 E8x00 processors, you could set the voltage to 1.30V and it would show a load value of around 1.24 to 1.25V, if you don't disable the offset in the BIOS. And Anandtech techs explain why it's better to leave it alone.

I had to go up to 1.30V (CPU_Z shows 1.25 under PRIME95 load) -- to get my E8600 stable at 4.1 Ghz. To get it stable at 4.25 Ghz, it looks as though I'll have to set it to 1.3625V, but that only keepst it at 1.304V (CPU_Z) under stress-testing load. If anyone asked my opinion, and it's a conservative opinion, these Wolfdales run coolest with the least strain in voltage with a 20 to 25% over-clock. But you'll see people pushing the 3.33 Ghz E8600 to 4.5 -- with hybrid or water cooling.

So I'd say you shouldn't have any problem running the E8400 up past 3.6, maybe 3.7 -- possibly 3.8 Ghz.

Also -- one more thing, and remember, I'm not familiar with the Intel P45 chipset: when you get close to CPU FSB = 400 Mhz, you need to tweak the CPU_VTT (some BIOS call it CPU_FSB) voltage a little higher -- between 1.3 and 1.4. Especially, you should not adjust it higher than 1.4V, and you're better off if you limit the adjustment to a lower setting.

 

[Ruger22C]

P35

 

[Ruger22C]

I'll just go try 7-7-7-21 and change the vtt a little, see how far I can get.
be back.

 

[Ruger22C]

O.K.
6-7-7-21, +1 on FSB. (or is it 0.1? which ever is more likely, I forgot).

Still can't get 405 FSB, max at 400...

Any other ideas? I'm going to go back to 378 x 2.4 for now

 

[BonzaiDuck]

Originally posted by: Ruger22C
O.K.
6-7-7-21, +1 on FSB. (or is it 0.1? which ever is more likely, I forgot).

Still can't get 405 FSB, max at 400...

Any other ideas? I'm going to go back to 378 x 2.4 for now

400x9= 3.6 Ghz . . . That should not be difficult. I, too, had an E8400, and I was running it on a motherboard that just wouldn't push much beyond 400 or FSB 1,600 Mhz. But I was able to at least get it rock-solid at that speed, and I think I got it up to 405. Looking at the spec for your mobo, that should not be a problem there.

Somebody who is more familiar with your chipset (P35-- I stand corrected) and BIOS --- should chime in here . . . And . . . say . . . did you stress-test the RAM thoroughly at stock settings with MEMTEST86+? Is that RAM listed on the mobo's compatibility "configurator?" Or do these Reapers show the mobo in their own compatibility list?

 

[Ruger22C]

"did you stress-test the RAM thoroughly at stock settings with MEMTEST86+?"

yes

...

didn't check compatibility on either gigabyte or OCZ. It works great @ 378 x 2.4... :\

 

[imported_wired247]

looks like you're getting there. I misunderstood when you said "unstable" originally.

good luck!

you should be able to hit the 400FSB easily... just have some patience and as bonzai said you have plenty of Vcore headroom up until about 1.36V easily


as far as the timings go, 7-7-7-21 is very loose so that should give you some wiggle room... then you can try upping the voltage on the RAM again back to the 2.1V mark where it's rated.

northbridge should be able to handle 400FSB without too much voltage, I think ...

VTT , CPU PLL, southbridge voltage, can likely all be kept on auto


did you loosen the secondary memory timings as well? those can cause instability as well.

ultimately, after your overclock of the FSB/CPU is behaving well, you can try tightening up the ram timings again.

 

[BonzaiDuck]

Originally posted by: Ruger22C
"did you stress-test the RAM thoroughly at stock settings with MEMTEST86+?"

yes

...

didn't check compatibility on either gigabyte or OCZ. It works great @ 378 x 2.4... :\

Wait a minute . . . . somebody else might know better here . . . with my nVidia 780i chipset, I don't contend with the factor I'm going to suggest.

There is a feature known as the Northbridge "strap." It must be adjusted for different FSB ranges. I think I recall reading that 400 Mhz was a threshold.

And this "strap" feature is implemented in the Intel chipsets -- I'm pretty sure.

Somebody who's already posted here might know about this.

 

[Ruger22C]

6-7-7-21 didn't help.
Secondary memory timings? What should I do?

Bonzai, hopefully someone can shed light on that.

 

[geno]

Don't hold your breath for a 400Mhz+ FSB on that board, they're notorious for hitting 400 and stopping there. The most I was able to get, stably, from my P35-DS3L was 400, anything more and I was lucky to boot into windows.

 

[elconejito]

I was able to get a P35-DS3R (how similar is this to your DS3L?) with an e8400 to 425x9 (3.8Ghz) on my friends computer with VERY little tweaking. If I recall correctly, I set the Vcore to 1.25v, and added +0.1 to the FSB and that was it. He has OCZ Reapers as well (with the funny heatsink on top) and they are running right around their spec 1066 speed (don't recall exactly the frequency), all RAM timings on auto.

 

[BonzaiDuck]

Originally posted by: elconejito
I was able to get a P35-DS3R (how similar is this to your DS3L?) with an e8400 to 425x9 (3.8Ghz) on my friends computer with VERY little tweaking. If I recall correctly, I set the Vcore to 1.25v, and added +0.1 to the FSB and that was it. He has OCZ Reapers as well (with the funny heatsink on top) and they are running right around their spec 1066 speed (don't recall exactly the frequency), all RAM timings on auto.

I think the DS3R and DS3L vary over RAID implementation, can't be totally sure unless I check, though. The GA-EP45-UD3P and UD3R are similar boards, with the "P" version fitted for two PCIe x16 slots -- and only one x16 for the "R" version.

 

[drewp29]

Long time reader, first post . . .

I have an e8400 as well, but with a Gigabyte x38 MB. Currently it is clocked, and has been for 11 months now, to 3.92GHz with the 9.0x multi and 2.00B ram divider. My VCore is set at 1.39375, and it is rock stable in Prime95. I tried for a long time to get 4.0GHz out of it, but I won't go over 1.4V VCore, and 1.4 would not get it stable.

Sounds to me like you need to give the little buddy more Juice. Don't be afraid to bump it up above 1.36, its not going to kill the chip. Or the MB is the issue - my x38 is known to be a good overclocker, which is why I chose it.

Just my 2 bottlecaps!

 

[BonzaiDuck]

Originally posted by: drewp29
Long time reader, first post . . .

I have an e8400 as well, but with a Gigabyte x38 MB. Currently it is clocked, and has been for 11 months now, to 3.92GHz with the 9.0x multi and 2.00B ram divider. My VCore is set at 1.39375, and it is rock stable in Prime95. I tried for a long time to get 4.0GHz out of it, but I won't go over 1.4V VCore, and 1.4 would not get it stable.

Sounds to me like you need to give the little buddy more Juice. Don't be afraid to bump it up above 1.36, its not going to kill the chip. Or the MB is the issue - my x38 is known to be a good overclocker, which is why I chose it.

Just my 2 bottlecaps!

Yeah -- I started with my current build/project last summer with an E8400, but I couldn't get it to 3.8 Ghz with the 680i motherboard i was using. Then I had to RMA the board -- turned out to be a bad BIOS PLCC chip which I could've replaced myself. But by that time, I just went ahead and bought a different motherboard, and picked up an E8600 in the same purchase.

Somewhere in these threads, I posted today with my cautious approach to volting the Wolfdales -- mindful of the Intel specs, and factors like vDroop and vOffset. My wild guess: I could push my E8600 another 50 Mhz to 4.3 Ghz by edging up the voltage some more, but "stimulus" won't buy me a replacement CPU if I need one, and I prefer splurging for "want" and not "need."