875P-T 5/4 divider help

[MikeyLikesItSI]

Ford: I've tried using the 7.8 refresh once before with no real difference. Around 290 i start getting a pixelated screen. Just every few lines i get a few random pixels. I don't know what that is a sign of. If i run memtest 288 fsb, test 5 locks up after a few minutes. I think that is memory heat related though. I want to pull the heat keepers (spreaders) off my adata and either put sinks on them, or leave them be. This mem seems to error more at 2.5-4-4-8, but i think your nudging me back into the testing mode :-)

 

Too bad i can't lock my CPU at 14x to help rule out CPU error. Since i can't boot into windows for very long at 280 fsb, i think i'm getting CPU instability, but feeding it 1.65vcore doesn't help either. Will post back after some testing tonight.

 

Keep us posted on your results

[fordf250]

Mike did you ever get that ocz memory from Tony? The hynix stuff.

 

Well the 300fsb didnt work but I cant remember ever having this cpu stable at 4200mhz.

I thought I was overheating it so I droped the Vcore to 1.6V and got the System 32 windows corupt screen but after setting defaults it booted back into windows.

 

Superpi and everest:

 

http://client.webshots.com/photo/312249557/315374499xsUCDp

 

I ran the 8m test but lost key board support so could not get a screen shot .The mouse was still working though.Mabye its a driver issue I might not have installed the newer chipset driver on this install.

[The Stilt]

This new 325 bios is excellent

I was able to run SuperPI at 303FSB with 2-2-2-5 setting CPC Addr enabled.

With CPC Addr. disabled 310FSB wasn´t a problem

 

I was using P4 660 cpu and 14x multi.

[MikeyLikesItSI]

Tell me your mouse and/or keyboard are USB. I've noticed that USB goes dead about 290 for me. if i use clockgen, at 292, The lights turn off on my USB hub, and within 10 seconds, it goes blue screen on me.

 

Tony really did try to help me out with the memory. He even offered to trade me some of his personal ram, but i had 2x256, and he had 2x512. We both drew the conclusion that its really just the board that can't take it. But no, i still have that OCZ memory. I was selling it on ebay but the guy that won the auction can't pay by the specified deadline. I might throw it in and play with it some more.

 

I am starting to believe more and more that i'm at the dead end with this CPU/board/ram setup. I just think they are all really close to pushing their max. i just really wanted to get 280 fsb to achieve a stable 4.2ghz.

 

 

EDIT: I think your testing with the 5:4 divider is great, but if you had mem capable of higher speeds, you'd probably find the same results running 1:1. I don't think there is any problem with the divider, i think (be it the northbridge or something else) the board just gets loose at 280+ FSB. Has anybody here ever achieved 24 hour prime stable 280+ FSB? I know stan was close, but i don't know if he was 24 hour 2xprime stable

 

EDIT #2: I also think that this problem will ease off with some higher voltages, thats why i was hunting around XS for a VAGP mod. I've soldered some small stuff before, but this is ridiculous. i'm still trying to work up the nerve to try it.

[fordf250]

I destroyed an NF7 last year and have medium soldering skills so I can understand not wanting to solder on this motherboard. I havent tried the drop mod for the same fear.

 

My mouse is usb and my key board is the old round plug.

i have had this memory stable at 286 on an 865pe board but yes I cant even get it stable at 280 now.

[MikeyLikesItSI]

I've been soldering since i was 12. Even the droop mod on the legs of the ICS chip isn't that bad compared to the controller right under the memory to do the VAGP mod. Maybe if someone offers to do it for me :-). Nah, i'm too much of a do-it-yourself-er. Always have been.

 

Ford, i fear that we have seen the end of the road for the FSB on the 875P-T. I also fear that it cannot be corrected with a bios update. I have a feeling to climb anymore requires more AGP volts (Hey angry, you out there? Can they scale volts higher with a bios?! that would be the cats meow...), or a frozen (or really cold) northbridge

 

I think i'd wind up attempting the pelted NB before soldering that little tiny insy winsy itty bitty leg on that chip. i don't think that leg is more then fifteen thousandths (.015") big.

[fordf250]

Well even if it cant go higher, my 640 and this board at 260-270fsb is the most awsome rig I have ever had.I should get a second 875p-t as I dont think I will go to DDR2 and PCI yet and have a lot of money invested in this one.

[MikeyLikesItSI]

By no means was i trying to say that this board isn't good. 270 isn't anything to laugh at. Especially if running high multi's.

[fordf250]

I didnt think you were saying the board isnt good ,I was just saying how much i like this board even if I cant get it to do what I want at 290fsb.270 is plenty for me.

[The Stilt]

Tell me your mouse and/or keyboard are USB. I've noticed that USB goes dead about 290 for me. if i use clockgen, at 292, The lights turn off on my USB hub, and within 10 seconds, it goes blue screen on me.

 

Tony really did try to help me out with the memory. He even offered to trade me some of his personal ram, but i had 2x256, and he had 2x512. We both drew the conclusion that its really just the board that can't take it. But no, i still have that OCZ memory. I was selling it on ebay but the guy that won the auction can't pay by the specified deadline. I might throw it in and play with it some more.

 

I am starting to believe more and more that i'm at the dead end with this CPU/board/ram setup. I just think they are all really close to pushing their max. i just really wanted to get 280 fsb to achieve a stable 4.2ghz.

 

 

EDIT: I think your testing with the 5:4 divider is great, but if you had mem capable of higher speeds, you'd probably find the same results running 1:1. I don't think there is any problem with the divider, i think (be it the northbridge or something else) the board just gets loose at 280+ FSB. Has anybody here ever achieved 24 hour prime stable 280+ FSB? I know stan was close, but i don't know if he was 24 hour 2xprime stable

 

EDIT #2: I also think that this problem will ease off with some higher voltages, thats why i was hunting around XS for a VAGP mod. I've soldered some small stuff before, but this is ridiculous. i'm still trying to work up the nerve to try it.

 

I actually do have USB keyboard and mouse (Logitech Internet kb and MX1000)

 

I´m running VAGP at 2.02V so, that should help me a bit.

[fordf250]

I,m not so sure its the agp voltage anymore as I loaded the latest chipset driver and droped the agp voltage to 1.6 and upped the Vcore to 1.6V and have been running dual prime and superpi at 32m and have not lost the use of the mouse or keboard yet.

 

I agree that the Vagp needs to be at 1.8V for 1/1 at 277 but I was running memtest free at1.6V this morning. This booting into windows is more affected by Vcore for me.It will not boot into windows at 1.5Vcore and 290fsb even at 1.8Vagp.

[BigStan]

Has anybody here ever achieved 24 hour prime stable 280+ FSB? I know stan was close, but i don't know if he was 24 hour 2xprime stable

I'm running Prime stable at exactly 280fsb with my Hynix based Adata4000. It is quite interesting to me that the mem is stable at that speed because I could never get it faster then 277 on my IC7 Max3. I can boot into Win at 290fsb mem 5:4. It takes 1.65v VCore, 1.8vVAGP. I think with good mem I could be stable at 285. But it would have to be mem that can do 285 1:1. I bid on some 2x512 OCZ 4400 Gold. I'll know later today if I'll be getting it or not. It should do 285. And that should settle a lot of questions for me.

 

The issue of Win getting corruption when getting errors on mem test #7 was when I was running the 14 multi on my 550 D0 proc. I haven't had problems with corruption when running the default multi of either the 550 D0 or this 530J