ASUS P5W DH - Problems + Fixes Thread

[Thickbrit]

Here's a few things i have found with my early dh.Bios 1403 has been the best overclocker for me,seems fairly stable apart from occasional loss of video signal at boot/post.
My board will not boot with memory frequency > 499mhz,or fsb much above 400,so i boot with < 399fsb and < 500mhz memory frequency and then memset trd to 7 and my memory will go to 547mhz 4-4-4-4,but not 1mhz more at any volts/timings/devider.Memory stops dead at 547mhz.This board does not seem to like cas 5 at all,with either of the rams in my sig.
My northbridge (fsb limit) goes to 463 fsb (1m) @ 1.9v on normal water,more volts don't help,did about 414 on stock cooler 1.65v,and about 435 on water and 1.65v.
All deviders seem to work fairly well,except the 3/5 (cpu/ram),which seems unreliable,although i haven't looked too hard for the reason yet.
Lower cpu multi's seem to stress the cpu more,at least on the x6800.
Thats about all i've learnt for now,except for 1 last thing,the fact that asus has the 2nd and 3rd primary ram timings totally reversed to all ram software,is totally confusing,annoying,and unnecessary.

[Falkentyne]

Here's a few things i have found with my early dh.Bios 1403 has been the best overclocker for me,seems fairly stable apart from occasional loss of video signal at boot/post.
My board will not boot with memory frequency > 499mhz,or fsb much above 400,so i boot with < 399fsb and < 500mhz memory frequency and then memset trd to 7 and my memory will go to 547mhz 4-4-4-4,but not 1mhz more at any volts/timings/devider.Memory stops dead at 547mhz.This board does not seem to like cas 5 at all,with either of the rams in my sig.
My northbridge (fsb limit) goes to 463 fsb (1m) @ 1.9v on normal water,more volts don't help,did about 414 on stock cooler 1.65v,and about 435 on water and 1.65v.
All deviders seem to work fairly well,except the 3/5 (cpu/ram),which seems unreliable,although i haven't looked too hard for the reason yet.
Lower cpu multi's seem to stress the cpu more,at least on the x6800.
Thats about all i've learnt for now,except for 1 last thing,the fact that asus has the 2nd and 3rd primary ram timings totally reversed to all ram software,is totally confusing,annoying,and unnecessary.

I'm not sure that lower CPU multipliers really stress the 6800 more.
I think what happens is the CPU tries to complete its "power up" routines at the default multiplier (11), before the BIOS even initializes. That's why the X6800 won't post at lower multipliers and very high FSB unless you increase the vcore. The Nemesis already found this out. The vcore increase you need is equal to the vcore needed for posting at the exact same speed you would have if you were using an 11x multi (example: 400 FSBx9=3600 mhz, but would require the CPU to be able to low level POST at 400x11).

That's what I've found out; maybe I'm wrong.

You should be able to test this by using CrystalCPUID to change the multiplier in windows, from your normal stable overclock (let's say you are using 360x10=3600 or 328x11, and then a clockgen or setfsb or whatever to increase the FSB. (goal=400x9?) CPU should still work at the exact same voltage if the mhz remains the same. But it simply won't POST at 400x9 as it tries to post at 400x11.

Multiplier unlocking is done by the BIOS giving a command to the CPU to lower the multiplier, if I recall correctly, not by "writing" a new multiplier into the CPU flash memory or something like that. Otherwise the CPU would keep the lower multplier as default if you switched motherboards, which it does not. (correct me if I'm wrong).

How did I come to this conclusion?

Rather simple, actually.
my X6800 cpu posts and runs fine (dual prime stable) at 360x10=3600, 1.5125v. But put CPU at 370x10=3700, and the CPU won't even POST (which is unheard of for anyone doing a 100 mhz clock higher). If I increase vcore to 1.55v, it posts all the time. 1.575v for stability in windows/games.

That only makes sense if the CPU were actually trying to POST at 4070 mhz (370x11) at 1.5125v, before BIOS takes over and lowers the multiplier.

[Thickbrit]

I'm not sure that lower CPU multipliers really stress the 6800 more.
I think what happens is the CPU tries to complete its "power up" routines at the default multiplier (11), before the BIOS even initializes. That's why the X6800 won't post at lower multipliers and very high FSB unless you increase the vcore. The Nemesis already found this out. The vcore increase you need is equal to the vcore needed for posting at the exact same speed you would have if you were using an 11x multi (example: 400 FSBx9=3600 mhz, but would require the CPU to be able to low level POST at 400x11).

That's what I've found out; maybe I'm wrong.

You should be able to test this by using CrystalCPUID to change the multiplier in windows, from your normal stable overclock (let's say you are using 360x10=3600 or 328x11, and then a clockgen or setfsb or whatever to increase the FSB. (goal=400x9?) CPU should still work at the exact same voltage if the mhz remains the same. But it simply won't POST at 400x9 as it tries to post at 400x11.

Multiplier unlocking is done by the BIOS giving a command to the CPU to lower the multiplier, if I recall correctly, not by "writing" a new multiplier into the CPU flash memory or something like that. Otherwise the CPU would keep the lower multplier as default if you switched motherboards, which it does not. (correct me if I'm wrong).

How did I come to this conclusion?

Rather simple, actually.
my X6800 cpu posts and runs fine (dual prime stable) at 360x10=3600, 1.5125v. But put CPU at 370x10=3700, and the CPU won't even POST (which is unheard of for anyone doing a 100 mhz clock higher). If I increase vcore to 1.55v, it posts all the time. 1.575v for stability in windows/games.

That only makes sense if the CPU were actually trying to POST at 4070 mhz (370x11) at 1.5125v, before BIOS takes over and lowers the multiplier.

Yep,that all seems to make perfect sense to me,but i was saying that i think i noticed higher cpu temps (more stress) running on the lower multi's,than on the higher ones,at the same voltage,probably due to increased fsb,although i thought greater fsb would heat the northy and not the cpu.Nice tip about the pre post multi/vcore thing though,cheers.

[Mykou]

Yep,that all seems to make perfect sense to me,but i was saying that i think i noticed higher cpu temps (more stress) running on the lower multi's,than on the higher ones,at the same voltage,probably due to increased fsb,although i thought greater fsb would heat the northy and not the cpu.Nice tip about the pre post multi/vcore thing though,cheers.

very interrresting !

have to try bios with unlocked multi to compare .

[Falkentyne]

Yep,that all seems to make perfect sense to me,but i was saying that i think i noticed higher cpu temps (more stress) running on the lower multi's,than on the higher ones,at the same voltage,probably due to increased fsb,although i thought greater fsb would heat the northy and not the cpu.Nice tip about the pre post multi/vcore thing though,cheers.

Remember, what I said is just a theory.
I don't know what's really going on. But why would my computer not post at 400x9 @ 1.51v, but posts (and runs 100% stable) at 1.55v?
(BTW 3.6 ghz is dual prime stable at 1.5125v).

[graham_h]

anyone have a 1305 bios link ?

[simonmaltby]

anyone have a 1305 bios link ?

All of them are here
http://wiki.p5wdh.de/doku.php?id=downloads:bios

[jimmsch]

quote from page 56 (I think)

I'm not sure that lower CPU multipliers really stress the 6800 more.
I think what happens is the CPU tries to complete its "power up" routines at the default multiplier (11), before the BIOS even initializes. That's why the X6800 won't post at lower multipliers and very high FSB unless you increase the vcore. The Nemesis already found this out. The vcore increase you need is equal to the vcore needed for posting at the exact same speed you would have if you were using an 11x multi (example: 400 FSBx9=3600 mhz, but would require the CPU to be able to low level POST at 400x11).

That's what I've found out; maybe I'm wrong.

You should be able to test this by using CrystalCPUID to change the multiplier in windows, from your normal stable overclock (let's say you are using 360x10=3600 or 328x11, and then a clockgen or setfsb or whatever to increase the FSB. (goal=400x9?) CPU should still work at the exact same voltage if the mhz remains the same. But it simply won't POST at 400x9 as it tries to post at 400x11.

Multiplier unlocking is done by the BIOS giving a command to the CPU to lower the multiplier, if I recall correctly, not by "writing" a new multiplier into the CPU flash memory or something like that. Otherwise the CPU would keep the lower multplier as default if you switched motherboards, which it does not. (correct me if I'm wrong).

How did I come to this conclusion?

Rather simple, actually.
my X6800 cpu posts and runs fine (dual prime stable) at 360x10=3600, 1.5125v. But put CPU at 370x10=3700, and the CPU won't even POST (which is unheard of for anyone doing a 100 mhz clock higher). If I increase vcore to 1.55v, it posts all the time. 1.575v for stability in windows/games.

That only makes sense if the CPU were actually trying to POST at 4070 mhz (370x11) at 1.5125v, before BIOS takes over and lowers the multiplier.

If this is true than my nice stable OC of 350 x 10 and 1.4000vcore would at least POST if I change the multi to 11 in BIOS, but it doesn't. In fact I have to raise the vcore to 1.4875 in BIOS to POST at 350 x 11.

Like I said 350 x 10 is perfectly stable dual Orthos 12+ hours. RAM settings stay the same as in my sig, but change multi to 11 and no POST, so maybe there is something else going on. I can POST and boot into Windows at 335 x 11 - 1.4000vcore but it is not stable in Orthos.