After a year and a half -- IC7G/P4 2.4C problems still

SilverSliver

[H]F Junkie
Joined
Feb 23, 2004
Messages
12,513
I've tried to OC this setup since I've had it, without much luck. So, I'll give my most recent story, and implore some of you experts for help.

I recently upgraded to an X850 XT AGP. As such, I got brave again and made another attempt at stable 24/7 3.0ghz

Bios, manual, 250fsb , 5:4 divider, locked agp/pci bus, PCBstrap800, memory settings at 2.5-3-3-8, AAADD. Pretty routine here. 1.55v. Note that CPUz tells me the voltage when set at 1.55 fluctuates between 1.536 and 1.520. Not sure if this is normal.

At this point, as usual, Prime95 will give me rounding error, .5 expecting less than .4, and simply will not be what anyone would consider "truly stable"


Boot up, get into windows, rock stable. Plenty of benchmarks, plenty of gaming. Temperatures are fine, both GPU and CPU wise. I think to myself, "Wow this is great"

Then it happens. BF2 locks up, crashes tabbed out, GPU stops accepting commands and the whole system locks. Reboot. Posts, gets into windows, black screen. -- This is new....

Reboot into safemode, uninstall the X850 drivers, reboot, reinstall, reboot, gets back into windows, black screen still.

At this point I reboot, reset my cmos jumper, and it gets into windows no problems at all....

Basically I need some help. I've tried to get this thing to OC, I bought a Antec TrueControl 550 a while back thats got solid rails and plenty of juice, and even tried the TCCD G-skill ram a while back that didn't help at all, sold the ram and stuck with my Corsair.

1:1 divider, I can't get past 210 FSB. 5:4 seems stable, but then I get random lock ups, and that scare with the new video card was a bit much.

Im truly stumped. If anyone can help me, I'd really appreciate it.


a1e0187d.jpg
 
What bios revision do you have?

When I first got my IC7-G it had bios revision #16 or something around that on it. I could barely get it to 2.7ghz. I upgraded to the latest bios (24 at the time) and I was albe to go all the way up to 3.4ghz.

Also, try setting your memory settings to Auto - Normal - Auto - Disabled - Disabled. This helped me stabilize at 3.4. Try raising your VCore to 1.575v or 1.6v.
 
I will take a shot. The random lockups are when you hit a memory cell that cant work at the speeds you are pushing it, you gave the proof yourself.

"as usual, Prime95 will give me rounding error"

PC3200 ram = 200 MHz rated = 400 MHz actual

"1:1 divider, I can't get past 210 FSB. 5:4 seems stable"

210 FSB = 420MHz not bad at tight timings. But you are pushing it as you can tell.

when you kick in the divider you are lowing the memory speeds and that allows you to get more FSB before the memory again bites you.


run memtest
loosen your memory timings
try lower?? divider.
If those allow you to get where you want to go, buy faster ram.
 
wwparrish said:
...
loosen your memory timings
...

I forgot about that. Intel systems don't rely on memory latencies as much as AMD systems do. Intel system benefit more from running the ram at a higher FSB. Max out the memory latencies then try.

My memory latencies are at 3-4-4-8, but I can run my ram 1:1 @ 271FSB to get about 6.4gb/s with sisoft sandra. To run at 3.4 I need to use a 5:4 divider but I still get about 6gb/s.
 
I had 522mhz G-skill ram at one point, made absolutely no difference.

I set strap to CPU just now, GAT is ANADD, timings 2.5-3-3-8 -- vcore 1.575

Went 2 minutes into Prime95 torture until I got a rounding error, new record...I might be doing something right, am not sure.

The thing is, with the 5:4 divider I am doing that on purpose, attempting to get my ram "out of the equation" -- Technically the ram is still running at it's stock rated speed, because of the divider.

update-- out of curiosity, bumped to 260 fsb 5:4 -- failed to post
 
My secondary rig is has the exact same setup but with a 9800 pro and Geil pc4000. I was able to run at 3.0 1:1 at those mem timings rock solid stable for a year and a half. I dont think that its a MB or bios issue. I assume that you turned off all thier memory performance tweaks I forget what they call them. Turn off all non essential things in bios like extra pci, firewire, game port parellel port ect. I think I remember people saying that they had issues with the 5:4 divider but I never seemed to. If this is the case with you I would try running it at 1:1 and see if the issue goes way. In real life you want notice the lower clock speed.
How about making a memtest boot disk and checking that out. Run test 5. I bet that might tell you whats wrong.
BF2 is not a good test of stability cause it crashes almost everyones system. I dont think it should have even been released yet. The patch seemed to help but it still has issues.
 
vcore to 2.6, timings to 3-3-3-8, ran prime 95 torture for 5 minutes before error.

same settings, 225 FSB 1:1 failed to post

Same settings, 260 FSB 5:4, checksum error
 
So, any tips on how to get this thing stable past 5 minutes in Prime95 torture test? You'd think I was trying for 300 FSB with the difficulty Im having.
 
10 bucks pay-pal to whomever can help me get this thing to run 250 FSB stable in prime 95, come on, Im desperate
 
Your vcore is going to fluctuate. And it being an IC7 board I'm sure it has the popular voltage problems.

That board should let you do 3 4 4 8 timings on the memory? I would max those values out and start there. Also does the IC7 have the Game Accelerator options? Should be something like Auto, Normal, Turbo, F1 if it does. make sure its set to normal.

Does setting your divider to 3:2 help in anyway?

What voltage are you running your ram at?
 
Mine runs fine at 250 FSB. The only time Ive ever had a problem was when I didnt put electrical tape on the bottom of the mouting plate for my whitewater and that chip near the socket shorted and fried the board.
 
Okay -- updates

5:4 divider, vcore 1.6, vdimm 1.6 and 1.65 respectively, 3-4-4-8 -- GAT ANADD
255 FSB, posts, gets into windows, black screen only

3:2 divider, vcore 1.6, vdimm 1.6 and 1.65 respectively, 3-4-4-8 -- GAT ANADD
250 FSB, boots fine
260 FSB, fails to post

5:4 divider, vcore 1.6, vdimm 1.6, 3-4-4-8 GAT ""
250 FSB -- boots into windows, runs well, benchmarks, etc -- Fails Prime95 in the first minute.

Perhaps my chip is at it's limit? However, when I see basically my same chip by fellow [H] chrisf6969, it makes me wonder. I don't expect anywhere near the same results, I'm just wondering if Im doing something wrong...

screen3666.jpg
 
Could be you just hit the wall with that particular core. But before you give up, if you got a fan handy, take the side panel off and get some serious air blowing on the ram sticks and give it one last try. Humor me. Best of luck.
 
System temp reads 29C
CPU temp under Prime 95 stress load is reading 41 C, but IC7-G's read CPU temps 10C higher than they should.

Prime 95 has been going for 10 minutes now...no errors yet knock on wood.
 
SilverSliver said:
Okay -- updates

5:4 divider, vcore 1.6, vdimm 1.6 and 1.65 respectively, 3-4-4-8 -- GAT ANADD
255 FSB, posts, gets into windows, black screen only

3:2 divider, vcore 1.6, vdimm 1.6 and 1.65 respectively, 3-4-4-8 -- GAT ANADD
250 FSB, boots fine
260 FSB, fails to post

5:4 divider, vcore 1.6, vdimm 1.6, 3-4-4-8 GAT ""
250 FSB -- boots into windows, runs well, benchmarks, etc -- Fails Prime95 in the first minute.

your vdimm appears to be a little off. Did you mean 2.6 ?
 
Fingers crossed ! I am assuming you fired up the fan. If so you know what the issue is now and can deal with it. Have fun :D
 
From what your posting it makes it sound like its your ram thats at fault. If you can keep lowering the divider to gain more cpu clock then that sounds like ram.
 
No fan required. Did 40 minutes and no errors at all in Prime95. Could have done more, but I'm thinking that was sufficient for now.

I switched strap to CPU, brought the vdimm back to stock, and set the mem to AGP init.

Somehow...this worked. Stable 250 FSB 5:4 divider, 3-4-4-8 timings. Hopefully it stays stable with my extra gig of ram coming in next week.

And yah I meant 2.6 on the vdimm. My mistake :p

Im stumped though on how anything I did new really effected it stability wise...seems like it just randomly decided to be stable. Don't think I'll touch it past this, as you are correct, I think my RAM is holding me back. No biggie, this is what I wanted :D
 
i agree with the above comment. your ram might be whats limiting you. try heatspreaders or active cooling your ram and see what happens. getting 250mhz out of that ram that runs at 200 stock is pretty ambitious. i have a similar setup to yours and the highest i can overclock my cpu with my ram running synchronously is 220mhz x 13 = 2.86gHz.
gluck!
 
i had the same issues with my ic7 and ic7-g except i had pc4400, the motherboard chipset is usualy the problem, ironically the later ic7s are worse for overclocking then the old ones try the cpu and ram out in a friends max3 and i bet you can get higher clock easy(max3's have th best of the best parts that abit could get, i belive the rest were speed binned to the ic7-g max2, ic7-g, and ic7 )

for all i know it could be just two bad cpus, but ahwell thats in the past, see if you can find a friend who has a max3 tho jsut to rule out the cpu and memory
 
John Gault said:
BF2 is not a good test of stability cause it crashes almost everyones system. I dont think it should have even been released yet. The patch seemed to help but it still has issues.
runs fine on my system, my friends athlon-xp, my other friends dothan laptop, his p4c as well as my p4e, never knew anyone had crashing issues
 
ryuji said:
i had the same issues with my ic7 and ic7-g except i had pc4400, the motherboard chipset is usualy the problem, ironically the later ic7s are worse for overclocking then the old ones try the cpu and ram out in a friends max3 and i bet you can get higher clock easy(max3's have th best of the best parts that abit could get, i belive the rest were speed binned to the ic7-g max2, ic7-g, and ic7 )

for all i know it could be just two bad cpus, but ahwell thats in the past, see if you can find a friend who has a max3 tho jsut to rule out the cpu and memory

Suddenly I don't feel quite so bad owning a rev 1.0 IC7-G
 
acascianelli said:
Suddenly I don't feel quite so bad owning a rev 1.0 IC7-G
i found a system for testing for the issue, both the ic7 and the ic7-g had high nb voltage i think it was somthing like 1.65v compared to say the max3's 1.49v(googled for it), i found that out when i went to volt mod both of them, the max3 is lower in voltage, proven by the fact that the nb heat sink is much cooler then it used to be
 
Right now I have my IC7 and 2.4C at 3.3 ever since October 2003. Running Perfect.

This really was the best combo back in the day
 
Well as I thought, my ram has been holding me back the entire time. Im rock stable now on this divider at 250 and 1.57vcore, but, anything over 200 FSB on the ram, ie, 204+, I get prime 95 errors within the first minute. This will satisfy me until my next build, now just to fix my NB fan.
 
Never tried it that high, never even gone to 2.7 in fear of frying this touchy corsair
 
I've had similar problems with my 2.4C (SL6Z3, I believe?) and P4C800-E Deluxe. My RAM is BH-5 PC-3500, it shouldn't hold anything back, especially when using 5:4 ratio because then when you set it at 250, it's only running at 200. But this isn't stable. I've tried all the way up to 2.65 volts and it still wouldn't be stable. Highest load temps I see are 42-43C.

I tried a DDR Booster on my RAM, taking it's voltage slowly up to 3.3volts. Got some decent timings out of it, but it still seems my processor doesn't want to do 3.0GHz stable.
 
I have OCZ PC4200, this ram is pretty tempermental with voltage. It will run 271 1:1 @2.8v but when I have it running at 227FSB (284FSB 5:4) it will ONLY run with 2.7v, nothing less nothing more.
 
Back
Top