Asus P5E3 Premium Wi-fi-AP

[Razmatazz]

Ok, I must admit I have been a little afraid about pumping up the nb voltage too much. I had been at 1.45 with the failures and after reading again this page, I tried pushing it up to 1.61 to see if that would make a difference. Superpi now stable at 2M which I had not been able to do before. I am hopeful that the increase in NB voltage will be the answer. Too late to do much more today but tomorrow is another day.


A little more...
Ok, what's wierd is that when I ran the superpi 4M, prime95 was running (forgot to turn it off). It passed. Then I shut of prime95 and tried 8M and the computer crashed immediately. Any throughts?

[soapman72]

Ok, I must admit I have been a little afraid about pumping up the nb voltage too much. I had been at 1.45 with the failures and after reading again this page, I tried pushing it up to 1.61 to see if that would make a difference. Superpi now stable at 2M which I had not been able to do before. I am hopeful that the increase in NB voltage will be the answer. Too late to do much more today but tomorrow is another day.


A little more...
Ok, what's wierd is that when I ran the superpi 4M, prime95 was running (forgot to turn it off). It passed. Then I shut of prime95 and tried 8M and the computer crashed immediately. Any throughts?

Are you running loadline calibration? If you are set this to NORMAL. Bump the NB to 1.61v or even 1.65, and bump the Vcore some. You and I have the same ram, and the same board. You have a CPU that will place more load on the NB though, so 1.61 will be the absolute minimum for you IMO. You can always drop the multiplyer if you are worried about putting too mych Vcore to your CPU. You will still need to bump the NB and once you are stable you will need to bump the NB again when you try a higher multiplyer. Faster CPU speed means more throughput which means more load on NB.

My system is stable and does not Freeze/crash/BSOD I am sure yours can get there as well.

[Razmatazz]

Man, I need to get to bed. It always is (just one more test) you know how it goes... I tried increasing nb voltage to 1.65 and set llc to normal. I then tried running the 4M on superpi and it crashed immediately (this had just worked at a lower nb setting of 1.61 and with superpi running at the same time). Thanks for the input. Any other thoughts? Also, I have not increased the FSB from the default of 400MHz. I can't even think about increasing it further until I can at least get things stable at FSB 400MHz.

[soapman72]

Man, I need to get to bed. It always is (just one more test) you know how it goes... I tried increasing nb voltage to 1.65 and set llc to normal. I then tried running the 4M on superpi and it crashed immediately (this had just worked at a lower nb setting of 1.61 and with superpi running at the same time). Thanks for the input. Any other thoughts? Also, I have not increased the FSB from the default of 400MHz. I can't even think about increasing it further until I can at least get things stable at FSB 400MHz.

What are you running for Vcore (as reported by CPU-Z or everest? What speed is your cpu running at?

[Sectorz]

Soapman72 , can you answer to my last post please ? I have news 2 hours 19 minute before crahs for 3490MHz and Ram 1741 but I have see my cpu voltage in load was 1.41 ... I have change the LLC Auto to normale.

Razmatazz I have the same board same memory and I have a Quad core penry to and a anothers same thing in common , I am not stable to lol. What bios you use ? 400FSB is easy if you tweak clock skew A2 B2 delay 200 for me.

[Razmatazz]

I have BIOS 0505. I have left the Vcore at auto. I had thought that since I am running CPU at default speed, that auto would be fine. Voltage is at 1.216 during prime95. Should I set Vcore manually? If so what would be a good initial setting? I am skeptical that this will make a huge difference but would be willing to give it a go.

Also, should I just not worry so much about Superpi crashing the system if everything else seems to be working well? I would think that this program should be able to run on my system.

[soapman72]

Soapman72 , can you answer to my last post please ? I have news 2 hours 19 minute before crahs for 3490MHz and Ram 1741 but I have see my cpu voltage in load was 1.41 ... I have change the LLC Auto to normale.

Razmatazz I have the same board same memory and I have a Quad core penry to and a anothers same thing in common , I am not stable to lol. What bios you use ? 400FSB is easy if you tweak clock skew A2 B2 delay 200 for me.

I run a dual core (E8400) and bois 0505. I have not needed to bother with clock skews for the RAM and you probably shouldn't need to either. Just make sure that the Vdimm is set to the specs that your ram maker recommends. It sounds like you have snough VCORE, but watch those temps. If you have NB voltage set to 1.61v-1.65v, you should be close.

If you still get a crash after 2 hours, have you tried leaving everything as it is, and setting the clock skews back to AUTO?

If that doesn't help. but also doesn't make it worse, than you know that playing with that variable is not the answer.

Vrom here, I would begin to increase FSBT and PLL, assuming that you left those on 1.1v and 1.5v respectively up untill this point. I would try 1.28v for Fsbt and 1.58 for PLL.

[soapman72]

I have BIOS 0505. I have left the Vcore at auto. I had thought that since I am running CPU at default speed, that auto would be fine. Voltage is at 1.216 during prime95. Should I set Vcore manually? If so what would be a good initial setting? I am skeptical that this will make a huge difference but would be willing to give it a go.

Also, should I just not worry so much about Superpi crashing the system if everything else seems to be working well? I would think that this program should be able to run on my system.

First, NO Program that other people can run fine should crash your system. Just in case, download a new copy of Super Pi and see if it runs on your computer all the way to 32 million.

I would try a bump in Vcore to about 1.28v-1.30v while under load this cannot hurt you, and may help as I have seen that for a given CPU speed, Vcore requirements may be slightly higher when running a lower multiplyer and higher FSB.

Next, are you still in the 1.61v-1.65v range for the NB voltage? I would try PL 8 if you are currently at PL7. You can do this by setting the Transaction Booster to MANUAL and selecting 8. Leave clock twister on AUTO

Let me know how this works.

[Razmatazz]

Is Superpi Mod v1.5 the right one? This is the one that I have been using.

I tried bumping to PL8 without any effect. NB is up to 1.65. Still crashed.

This seems like a big change from NB 1.41 and PL7 which had run everything stable except crysis and superpi. (which is where I'm still at in terms of stability)
Is it the case that the system would need this much given the fact that I am running default cpu speeds and memory much lower then rated (1600MHz)

I'll give the increase Vcore a shot.

Thanks for the advice.

[Sectorz]

I run a dual core (E8400) and bois 0505. I have not needed to bother with clock skews for the RAM and you probably shouldn't need to either. Just make sure that the Vdimm is set to the specs that your ram maker recommends. It sounds like you have snough VCORE, but watch those temps. If you have NB voltage set to 1.61v-1.65v, you should be close.

If you still get a crash after 2 hours, have you tried leaving everything as it is, and setting the clock skews back to AUTO?

If that doesn't help. but also doesn't make it worse, than you know that playing with that variable is not the answer.

Vrom here, I would begin to increase FSBT and PLL, assuming that you left those on 1.1v and 1.5v respectively up untill this point. I would try 1.28v for Fsbt and 1.58 for PLL.

I have already PLL at 1.58 and 1.30 to 1.34 for the FSBT.If I don’t put this clock skew I just no boot...

Me what I would it a stable 24 hours Prime stable, When you tell that is stable, Razmatazz how many hours of Prime 95 you have run ?This can crash after 6 or 11 hours if you can run it 24 Hours it sure you cannot have problem.

[soapman72]

Sectorz,


what kind of ram do you have? I had many, many problems when I first got this board that were related to OCZ Platinum 1600mhz ram. My board did not like the ram and I went through 3 boards and 3 kits of that ram with lots of weird freezing, BSOD's and crashing games. I got rid of that ram and now have 1800mhz OCZ Reapers, and the board has worked flawlessly since. It took me about 3 sessions running prime95 to find settings that work.


I agree completely with your comments on what truely means stable. Theoretically your system should be able to run prime forever, but I don't have that much time on my hands....LOL And yesa 24 hour prime 85 run with no errors indicates that it is pretty stable, but that is no gaurantee either. Isn't it possible to get an error after hour number 25 or 26? Of course. We could start a whole ne thread regarding what constitutes stable, but that sounds boring.

I will tell you that my system has recently run as long as 6 hours w/prime 95 no errors, 4 hours occt w/no errors and 10 passes on memtest86 with no errors. Could it error in 8 hours? Absolutely, but since I have not had one single BSOD, program freeze or crash, or any indication of OC weirdness whatsoever, I would rather just enjoy my system instead of using it to heat my house. It is summer here in Vegas after all.

I hope you don't reaqlly need to use clock skew adjustment to get your ram to run at it's rated speed. That alone is an indication that your ram will never work well in that board. If you are confident that you have adequate voltage for your NB and VCORE, and have also tested with PLL and FSBT at their lowest (non auto settings) maybe it is time to start RMAing your board or RAM untill you find out the source of the problem.

[Razmatazz]

I ran OCCT for 1 hour with pass. No real long runs of Prime95 yet. I am not overclocked. I can run Superpi Mod v1.5 if Prime95 is also running. When I shut down Prime95 and run Superpi the system will freeze and I will have to restart. Any thought about this one? Weird huh?

[soapman72]

Yes, weird.


Did you say that you have loadline calibration set to NORMAL? If it was set to auto, setting it to normal may have resulted in your CPU getting less VCORE, it did in my case at 450mhz fsb. Try bumping the VCore a couple of clicks in the bios and run super pi again. Are you currently at the minimum PLL and FSBT? Don't use auto, as that is not the minimum if you are running higher than 333 FSB, the board will automatically start raising VCORE and several other variables as you increase FSB, and it is better to set these items manually.

[Sectorz]

The I have start a RMA the board left my home tomorrow .

[Razmatazz]

Yes, loadline calibration is set to normal and CPU Pll as well as FSB termination voltage are set to minimum. FSB again is at default of 400MHz. With Vcore set to auto, at load CPUID reads 1.216 I can run Prime95 and Superpi together with Vcore at auto. I have bumped Vcore all the way up to 1.35 with no change. Increasing Vcore doesn't seem to make any difference as the system behaves exactly the same way and Superpi won't run alone. This one has me really stumped. I was hoping that as I was able to "dial in" the clock skew some that this would help stability. Oh well, I'm going to try to further tune the skews and check Prime95 for a longer run. Any other thoughts?

[soapman72]

I dont have he modded Super Pi, but in my experiance, running even a 32m super pi run is not as difficult load as running prime. What I mean by that is a computer can be "stable" enough to complete a 32m super pi run, yet fail prime95 after a relativelu short time, it is not usually the other way around.

I suggest that you run Prime 95 for now, and dont worry about super pi. Try to pay attention to which core fails, and how long it takes. If it is ALWAYS the same core, you may have a weaker core that may require you to play with the GTL for that core. I have been lucky as I have only needed to dial in the Nb Voltage and Vcore on my system.

Also, please make sure that your ram is set to 1.9v

[Razmatazz]

Ok, let me share a little of my latest trials. I have found that clock skew does indeed make a big difference in being able to run my ram with a quad core in terms of stability. I have been trying to dial things in with memtest but have found that this has been very difficult to say the least. I then tried to dial things in via Superpi. If you read my earlier posts you would remember that I had not been able to run Superpi at all without crashing the system. I have tested each ram module seperately. I am stable between 300 and 350 delay skew in one memory slot and between 275 and 350 in the other. I have more testing yet to do. I am confident that this will solve my instability issues and that after I have my ram stable I will finally be able to begin pushing the CPU. To make things even more frusterating to me is that one of the memory modules is not stable at all for Superpi at any of the skew settings regardless of the slot. I already have an RMA in the works. The OCZ guys have again been outstanding to deal with. Will keep you posted.

[WolfpupNX]

Hey all - need some help.

QX9770 running at stock speed (3.2 GHz) on this (P5E3 Premium W-fi AP). Trying to use 1600 MHZ kits of DDR3 (I've tried a few at this point), which should be easy given the QX9770's native 1600 FSB.

So here's the issue...I keep getting an error on Core 0 or 1 after a seemingly random amount of time...usually between a few minutes and 8 hours.

Small FFT test works brilliantly for 24+ hours. So does OCCT (4 hours), SuperPi (32 million calculations), and Memtest86+ (multiple passes - done up to 20 some before).

But I can't get it fully stable on the Blend test for 24 hours, and it's bugging me. Suggestions? I've tried raising vCore just a bit, but haven't tried doing anything with any of the other voltages because I'm not sure what's safe to try them at...

Thanks for any help!

[Pisklink]

I could use some help here on overclocking my new E8400 on the P5E3-Premium mobo.

As far as I understand VID is the standard voltage the chip will run on when at standard speed. I guess my chip's VID was 1.18V @ 3Ghz (Q817A240; pack date 4-6-2008).

I already searched the internet for some basic settings to start from. The following settings got me into Vista64 but when I start orthos it already gives errors after app. 10 seconds. Would pls take a look at my current settings and give some advise:

CPU ratio: 9.0
FSB: 450
FSB strap: 400
PCI-E: 100
DRAM freq.: 1800
Timings: 8-7-7-25-2T @ 1.98V (factory specs are 7-6-6-24-1T @ 1.9V)
CPU volt: 1.425V
FSB temination: 1.3V
NB volt: 1.45V
Temps. idle: 42 and 36 (measured with speedfan v4.34)

Everything else is on auto including, load line calibration, CPU spread spectrum and CPU PLL voltage.

In my opinion these settings are way to high as compared to other E8400 users here on the board. However lowering one of these settings results in BSOD or no POST. Thanks in advance for the advise which is much appreciated!!!!

[pumbertot]

I could use some help here on overclocking my new E8400 on the P5E3-Premium mobo.

As far as I understand VID is the standard voltage the chip will run on when at standard speed. I guess my chip's VID was 1.18V @ 3Ghz (Q817A240; pack date 4-6-2008).

I already searched the internet for some basic settings to start from. The following settings got me into Vista64 but when I start orthos it already gives errors after app. 10 seconds. Would pls take a look at my current settings and give some advise:

CPU ratio: 9.0
FSB: 450
FSB strap: 400
PCI-E: 100
DRAM freq.: 1800
Timings: 8-7-7-25-2T @ 1.98V (factory specs are 7-6-6-24-1T @ 1.9V)
CPU volt: 1.425V
FSB temination: 1.3V
NB volt: 1.45V
Temps. idle: 42 and 36 (measured with speedfan v4.34)

Everything else is on auto including, load line calibration, CPU spread spectrum and CPU PLL voltage.

In my opinion these settings are way to high as compared to other E8400 users here on the board. However lowering one of these settings results in BSOD or no POST. Thanks in advance for the advise which is much appreciated!!!!

is your cpu even capable of 4050MHz at those volts? drop FSB back to 400, you cant just jump in and go for such a high o/c right away.

[Pisklink]

Ok, I think I know what the problem was because I set load line calibration to "performance" which differs tremendously in Vdroop. At this moment OCCT reports a flat and steady Vcore but without load line calibration (llc) enabled Vcore would drop sometimes with 0.03 - 0.04V. Hence setting the Vcore as mentioned to 1.425V without llc this results in an effective Vcore of app. 1.39V. At this moment I still have the same Vcore but the system runs at 450 FSB.

[Sectorz]

Soapman72, I have news

I have receive my board from RMA and now I have do a fast test and FSB 430MHz 3440MHz ram 2*2GB 1720MHz stable 4:20 hours for the moment (I will keep it in load for the night to have a 24 hours stable for have a base to push more after) with only NB 1.53v CPU 1.25V RAM 1.92V clock skwe 175 A2 B2 3/4 off setting at auto.I hope to crush minimum 450 of FSB this time I know my cpu can do it and the ram to.

[soapman72]

Soapman72, I have news

I have receive my board from RMA and now I have do a fast test and FSB 430MHz 3440MHz ram 2*2GB 1720MHz stable 4:20 hours for the moment (I will keep it in load for the night to have a 24 hours stable for have a base to push more after) with only NB 1.53v CPU 1.25V RAM 1.92V clock skwe 175 A2 B2 3/4 off setting at auto.I hope to crush minimum 450 of FSB this time I know my cpu can do it and the ram to.

That is great! It is a shame about all the time that you spent trying to get your system stable though. Hopefully it will go better from now on. Keep us posted.

Your settings sound about right. I was at 3.8ghz and 1800mhz ram which required 1.57v nb to be completely stable. 4.0 ghz requires 1.61v with the ram at 1800mhz 2N. If I run 1N it looks like I need 1.65v NB I am treading close to the limit when it comes to NB voltage. I may just run the command rate at 2N to save the NB.

[Sectorz]

The test have pass 14 Hours now.I will try 435 of FSB, before go to job always with 1/1 ratio.But for the moment I am in performance level 8 ram in 2T yhis is not good, but with my others board I have never successfully bench table more then 3490MHz and 1160MHz for memory and now 3440MHz and 1720MHz for memory.

Edit:This do 5:30 hours at FSB 435 3490MHz Ram 1740MHz same voltage of FSB 430MHz.

Someone can tell to me if if normale to have 50 degrée max on is 4 core with a penryn 3.5GHz and 1.23V reel with a watercooling 720
radiator ?

Edit 2: Error after 3:30 hours FSB 440MHz 3525MHz Ram 1760MHz 2*2GB always with NB 1.53V.

Edit 3: before go to sleep I start anothers test this time FSB 440MHz 3525MHz Ram 1760MHz 2*2GB but NB 1.55 and AI clock strong perflvl 7

Edit 4: 1 Error on core 3 after 7 jours 51 minute someone have idea what can I do for have this 24 hours stable ?I have shutdown the test after 12 hours stable on the 3 others core.

[pumbertot]

doe sthe NB get hot with 1.6V on it? do you need aftermarket cooling on it or is stock solution adequate? about to buy one of these boards.