P5B Deluxe: BIOS screens for maximum overclocking success

[ziddey]

It shows PC2-4300 now because you're using 1:1.

When it was stock, you ran 2:3 for ddr800, which was why it showed pc2-6400.

[graysky]

Interesting. Anyone recommend a good memory benchmark program? I've read that you wanna run 1:1 but I'm game to experiment.

[ziddey]

mbench is real quick and lightweight. i'd especially recommend testing the differences in latency at 400fsb and 401.

[Big.Wayne]

mbench is real quick and lightweight

Linky?

[ziddey]

linky no linky

[ziddey]

ok fine http://www.xtremesystems.org/forums/...=memset+mbench

[DL Drafting]

OK Ziddey, suppose your theory on the spread spectrum issue is correct. If so, would it be worth the risk to stability in my system to enable it just to try for some higher clock? Here's where I am now: X6800 3576 mhz 325 x 11 at 4:5 PC-5300 DDR2 800 Crucial Ballistix x 2GB dual channel 4-4-4-12-4-30-10-10-10-10; P5B board and 1004 bios everything is stable. I have been working at this mhertz for weeks now, but would like to see if I can get to 3.8 Ghz on air. I use the Big Typhoon VX, and right now my idle temps are around 30-34c. I am wondering what I can try to increase my X6800 to around 3.8 Ghz on air, or if that is even possible. I have tried 330 x 11, 360 x 10 trying to move past 3.6 Ghz, but I can't seem to get Windows to boot with stability anywhere past 3.59 Ghz. I am really reluctant to jump to say: 400 x 10, just because of the mulitplier on the NB issue as well as the heat. I really can't afford to fry this board. What do you think? Or should I just leave well enough alone. 3.5 is still pretty fast for my drafting work.

[ziddey]

I'm sorry. The spread spectrum theory was disproved. It was just a bunch of events happening at the same time that I pinpointed it on the wrong item. For what it's worth though, it didn't really hurt that much either, but that's not really the point.

That said, try dropping the multiplier even more and working with 1:1 to start. Don't worry about NBCC. Some of your ram subtimings may be too tight as well.

If you want to rule out ram, you can try this first:
5-5-5-15-5-35-3-11-7-13. Or if you're more certain of it, 4-4-4-10-3-25-2-9-4-11.

How many volts are you pushing to the ram? 2.1v?

Stick with 1:1 divider for testing, and turn off static read control for now.

I take it you have a rev1.03g board and don't have vnb adjustments?

How much vcore are you pushing too?

Let's do 8x400 1:1 first to confirm that you can at least scale that high in fsb. Then work multiplier up to 9. You may indeed be needing more vcore.

[graysky]

How did you arrive at those mem. timings? I've been messing w/ mine but honestly, the benchmarks I do don't change:

superpi
x264
lameenc

What application based test would you recommend that actually can use the tighter timings? My current timings on my Ballistix ddr2-800 are: 4-4-4-12-6-42-10-10-10-11 and that's on 1:1 (DRAM:CPU) with the CPU @ 9x333 (Q6600). Oh, DRAM runs @ 2.15V.

Thanks.

[graysky]

I tried your 4-4-4-10-3-25-2-9-4-11 settings and got a ever so slight boost on the video benchmark and a small decrease on the audio while superpi remained the same:

superpi test (1M and 2M) = identical (both gave 17 sec and 41 sec)
lameenc test = about 2 % slower with faster timings
x264 test = about 0.5 % faster on first pass and 0.4 % faster on the 2nd pass.

All these are an average of 2 runs each. I'm actually thinking that the numbers are within error meaning these timings doesn't show a difference in these applications.

[ziddey]

The subtimings don't tend to change performance all too much. However, too low will cause instability, which was why I recommended the weaker set to DL Drafting to try out first.

[DL Drafting]

How many volts are you pushing to the ram? 2.1v? 2.25

Stick with 1:1 divider for testing, and turn off static read control for now. OK

I take it you have a rev1.03g board and don't have vnb adjustments? Where do I look to find out the board revision. Not sure on that one. The bios I am running is 1004. Am running 1.45v on the northbridge.
How much vcore are you pushing too? Vcore I have had as high as 1.45v, which yields an actual 1.39 on CPUID v. 1.38. Most of the time, I just leave it on auto.

Thanks for the input and the advice, Z. I will work on this and let you know how it flies. Also, thanks for the timings. Dan

[ziddey]

You should look into doing the vdroop pencil mod. Very worth it and what you set will be what you get. Be careful about using auto though since it can set extremely high and unnecessary vcores. For instance, 8x400=3200 for me, my board would give me 1.6v. As for the board revision, don't worry about it. That was my mistake. I thought you had the regular p5b.

You wouldn't happen to know the week of your cpu would you?

[graysky]

The subtimings don't tend to change performance all too much. However, too low will cause instability, which was why I recommended the weaker set to DL Drafting to try out first.

I tried 3-3-3-10-3-20-2-9-4-11 and found that the results were within error. In fact, all the results were within error of each other suggesting that the timings on my board don't matter.

[ziddey]

do note trcd of 3 is busted. so the lowest you should go is 3-4-3 in that regard.

[DL Drafting]

You should look into doing the vdroop pencil mod. Very worth it and what you set will be what you get. Be careful about using auto though since it can set extremely high and unnecessary vcores. For instance, 8x400=3200 for me, my board would give me 1.6v. As for the board revision, don't worry about it. That was my mistake. I thought you had the regular p5b.

You wouldn't happen to know the week of your cpu would you?

Ziddey,
you might direct me to the explanation of the vdroop pencil mod, because the terminology is losing me.
Here's what I was able to do at 8x400 on 1:1. Yes it did post -- the really goofy post where it goes clear dead, then blips and makes a little sound like it wants to start, but then has second thoughts and shuts off for three more seconds, then goes ahead finally and starts -- 'bout drives me nuts. However, the voltage showing in PC Probe was 1.49v, and the Core Temps returning were 52C and 54C at idle. This vCore is .11v higher than at 11 x 320 on 4:5 (FYI). When I dropped an arithmetical load on the processor from Sandra the temps quickly shot up to 64C and 61C. This seems hot to me, and well above what I am running down at 11 x 320 on 4:5. Do you think I should try to bump that multiplier to 9? I'll give it a try if you still think it worthwhile, but I am concerned about the heat. I'm still auto on everything but the Vdimm, which right now is at 2.25. The timings you sent were good. I used the loosest ones (first string). I am not able to ascertain the date of manufacture, however, this system was purchased from JNCS not too far from you in December last year. I have attached a Sandra shot of the cpu info. Thanks for all the help and good advice. Maybe I will reach 3.8 yet. Dan in Sedalia, Missouri - The Great State Fair City.

[ziddey]

your problem is that you're using auto for too many settings. and exactly like I said, vcore to auto is not a good idea. I'm assuming that asus programmed auto to rise with fsb, and it doesn't factor in multiplier. That's why you can run 11x320=3520 with such lower temps than 8x400=3200. It sounds about right too. When I set 400fsb, I get 1.6v after performing the vdroop mod, so in your case, 1.5v or so may not be all too off.

Find out what vcore is with 11x320, and you know you can go way higher. Your end goal should be 9x400 or better.

http://www.wittekakker.be/projects/p5b/index.html

[DL Drafting]

Yes. I am auto on almost everything, because I really don't know safe voltages to prescribe. I am pretty comfortable with vCore. I have had it to 1.45 as I mentioned earlier. Seems I have to set it at about 1.55 to achieve 1.45. Still experimenting there. Also have had the north bridge to 1.45 with a little extra heat generated. I'm not sure whether and how much to tinker with the other settings, including south bridge. Maybe a little advice there would be very welcome. I am going to try the 9 x 400 to see if we can get it stable. Thanks again. Dan

P5B Deluxe with Wi-Fi Bios 1004/ X6800 11 x 320 4:5/ 2.25vdimm 2GB Crucial Ballistix DDR2 800 4-4-4-12 / Asus Nvidia GeForce 8800 GTX 768mb/ Big Typhoon VX AS-5

[DL Drafting]

Ziddey,
I found Free Cable Guy's settings page and started there for the voltages I was not familiar with, in addition to the great info you already shared with me. I have attached the screenshots, which explain better than I can. You were absolutely correct on the multiplier thinking. When I upped the multi to 9, the temps came down from where they were at 8. Thanks for all your help. I am running smoothly now. Will "stress test" with work this week and report back later if everything stays stable, but right now things look good. After hitting 3760, I throttled back to 3600 (9 x 400) just to work from. I did get a blue screen at 3800. Do you think I should try tightening up my mem timings into the fours?

P5B Deluxe with Wi-Fi Bios Version 1004
Intel Core 2 Extreme x6800 9x415 = 3735 mhz 1660 fsb
2 GB Crucial Ballistix DDR2-800 5-6-6-18-6-42-3-11-7-13 2.25v dimm
Voltage settings (this is what is set, some voltages vary from actual)
vCore: 1.5125; FSB Termination Voltage: 1.45; North Bridge: 1.65; South Bridge: 1.60; ICH: Auto
WD 10K Raptor x 150 GB RAID 1
Seagate 7200K x 320 GB RAID 1
Asus Nvidia GeForce 8800 GTX 768 mb

[DL Drafting]

Here are screen shots of the other results.

P5B Deluxe with Wi-Fi Bios Version 1004
Intel Core 2 Extreme x6800 9x400 = 3600 mhz 1600 fsb
2 GB Crucial Ballistix DDR2-800 5-5-5-15-5-35-3-11-7-13 2.25v dimm
Voltage settings (this is what is set, some voltages vary from actual)
vCore: 1.5125; FSB Termination Voltage: 1.45; North Bridge: 1.65; South Bridge: 1.60; ICH: Auto
WD 10K Raptor x 150 GB RAID 1
Seagate 7200K x 320 GB RAID 1
Asus Nvidia GeForce 8800 GTX 768 mb

[DL Drafting]

Oh yeah, and I forgot. I had an issue with the PCI going to 1X first time I posted at 9 x 415, so I upped the PCI Express Frequency from 100 to 105 and that brought me back to 16x.

P5B Deluxe with Wi-Fi Bios Version 1004
Intel Core 2 Extreme x6800 9x400 = 3600 mhz 1600 fsb
2 GB Crucial Ballistix DDR2-800 5-5-5-15-5-35-3-11-7-13 2.25v dimm
Voltage settings (this is what is set, some voltages vary from actual)
vCore: 1.5125; FSB Termination Voltage: 1.45; North Bridge: 1.65; South Bridge: 1.60; ICH: Auto
WD 10K Raptor x 150 GB RAID 1
Seagate 7200K x 320 GB RAID 1
Asus Nvidia GeForce 8800 GTX 768 mb

[ziddey]

It should be noted that you're using way too much voltage in some areas. You have a c1 version p965 though like my friend's p5bdlxwifi, so I'm not sure if it'll need more than normal, but generally:

You should be able to keep 1.2vfsb until the high 400's or beyond. vnb is excessively high as well. A lot of people suggest keeping it to 1.55v at a max. Vsb, I've found some instances requiring 1.6v, although at 400fsb, 1.5v should be fine. vich to auto should be fine since I don't think it'll climb, but it wouldn't hurt to fix it to 1.057v just in case.

You can try the tighter timings if you wish, although I'm not sure if you'd possibly need more vdimm. Do changes one at a time so you can narrow down the cause of instability.

Looking good

[DL Drafting]

I started bringing down the voltages this afternoon. As long as I could post and run all my benches OK, I thought it might be a good idea to dial them down just to knock the heat and take some stress off, so I am now at 1.475 on the vcore and 2.15 on the vdimm and 1.55 on the north bridge. Will work the others down after I finish this large POV Ray trace I'm working on for a client, which, incidentally, this machine is whipping right through no prob. Thanks for the suggested voltages too, especially for ICH, because I had no idea what to do there. That helps a bunch. For the record, I went ahead and tightened up everything in the DDR2 to 4-4-4-12-4-30-2-9-11. Totally stable. The bandwith numbers on Sandra jumped to 7345/7346. Compared to stock widths of 5,536/5,528 this is a big increase. They are at least 200 above the runs at 5-5-5-15. You have made my day, Z. Thanks again.


P5B Deluxe with Wi-Fi Bios Version 1004
Intel Core 2 Extreme x6800 9x400 = 3600 mhz 1600 fsb
2 GB Crucial Ballistix DDR2-800 4-4-4-12-4-30-2-9-4-11 2.15v dimm
Voltage settings (this is what is set, some voltages vary from actual)
vCore: 1.475; FSB Termination Voltage: 1.45; North Bridge: 1.55; South Bridge: 1.60; ICH: Auto
WD 10K Raptor x 150 GB RAID 1
Seagate 7200RPM x 320 GB RAID 1
Asus Nvidia GeForce 8800 GTX 768 mb

[ziddey]

nice to hear! good luck

[evil-doer]

ok i need a bit of help here. i just bought my first intel in many many years and this overclocking seems a bit confusing with all the variables. first off heres what i bought.

p5b deluxe, newest bios installed
e6600
2x1gb kingston hyperx (khx6400d2llk2)
evga 8800 gts 640mb
antec sonata iii should be here next week.

without touching any voltages i tried 333x9 and 400x8 with the usual features turned on or off as per the guide here and other sites ive read. windows boots fine and runs for a while. but then a random program will hang, the window stops drawing itself, and if i try to kill the process it doesnt go away. and any other programs i try to start after that will not start. i also cannot shut down windows when its like this and i need to do a hard reset. doing a memtest with this overclock also gave errors, so i figured the ram needed more juice. i upped and upped it until i got to 2.3v and i didnt want to go any higher because that seemed high, and i still got programs freezing, altho after a longer period of time. so i figured i better start overvolting some other things. these settings are estimates because i dont want to reboot right now as im testing how stable it is (its been on overnight and its fine so far). the cpu is 1.3/1.4ish and fsb and nb are both at the first number thats purple.

so first ive gotta ask, that symptom of programs freezing like that, is that because of the memory only? or could it be the cpu as well? was i right in overvolting what i did? im really unsure of what i should do at this point.
thanks for any help.