max overclock changes with age? drastically?

[alyarb]

well i have a C1 Q9550. When i first got the chip i benched it for a couple days between 4.10 and 4.25 GHz, and, well i can say i needed about 1.44v and it was running hot as hell. i can't remember the details. after all that i set it to run at 3.4 GHz at 1.2v and it's been happy that way for the past few months. Yesterday i was going to run a bench for someone and wanted to get back up to 4.0 or so, and i couldn't post anywhere above 3.80 with the same voltage config i had been using earlier (GA P45 UD3R F6). on the bright side, i did get 3.75 GHz stable at 1.240 volts. but what gives?

 

[aigomorla]

RAM looks like its giving away..

First to fail = Ram.
Then board
Last is cpu.

 

[alyarb]

oh god you might be right. this ram is from a previous build! it used to do 800 4-4-4-4-12 but now i have to do ~1050 5-5-5-5-15! kept it at 2.1v like the spec card says. it's so funny, because i've been through two PSUs and had to change motherboards to make this 4.2 ghz possible, and poof. i was about to spend 300 on a water upgrade. I should have known i was doomed since i got a C1 in the mail. it was like an omen.

the system will not post, however, even at a measily 3.81 GHz, with memory at 900 Mhz 5-5-5-15. that is what makes me wonder. should i increase vdimm even though the clock is low and timings incredibly lax? i mean, yeah, the memory is almost 3 and a half years old, but i'm at 3.757 GHz right now with the ram at 1066 5-5-5-15. this is why i think it's the cpu. it had no trouble whatsoever at 4.0 ghz out of the box, and now this.

 

[GLeeM]

Thermal paste can dry out after a while. Could that contribute to your problem?

 

[alyarb]

i don't see how i can blame the paste if my temps are reasonable. it doesn't explain how 3.79 ghz is achievable but not 3.81.

 

[OCGuy]

Originally posted by: alyarb
i don't see how i can blame the paste if my temps are reasonable. it doesn't explain how 3.79 ghz is achievable but not 3.81.

Are your NB and VTT/FSB term voltages the same as they were before?

 

[alyarb]

yep i used the exact same settings as before and i'm getting nothing. i tried higher and lower, and nothing.

 

[BTRY B 529th FA BN]

I'd guess the motherboard

 

[taltamir]

running ANY voltage through the cpu, even highly undervolted, causes it to degrade over time. modern cpus are designed and set at a default voltage which they are expected to maintain for at least 3 years by intel. increasing voltage further decreases it faster, much faster. undervolting decreases it slower, but it still decreases

 

[soccerballtux]

No, they warrant the CPU for 3 years, it will run for ~10 on stock, this is common OC knowledge. OC'ing takes this down to 5.
OP: are the voltages measured in CPU-z the same you were seeing a year ago? My Vdroop on this board has increased; had to up my voltage recently to keep the 3.4Ghz.
3.4Ghz is holding strong, it was the motherboard power that was having trouble.

 

[alyarb]

the voltages have been the same. i just can't get any of the FSB speeds i used to. nothing higher than say 450*8.5, but i'm using the slowest memory strap, so it's not that. i dont think its the motherboard, its a UD3R and it wasn't given me any problem when i was OCing earlier.

 

[OCGuy]

Sounds like MOBO or RAM. Your chip isnt going to poop out @ 1.2XV

 

[alyarb]

the VID is 1.25. i'm asking for trouble at 1.24? i didn't think so. but how can it be my RAM? the clocks are abysmal. and i dont want to admit it might be the board.

 

[OCGuy]

Originally posted by: alyarb
the VID is 1.25. i'm asking for trouble at 1.24? i didn't think so. but how can it be my RAM? the clocks are abysmal. and i dont want to admit it might be the board.

You should be able to run that chip @ 1.38v all day. That is why I would bet the farm that it is not the CPU.

Do you have another 775 chip to see if you can break your current FSB wall?

Have you lowered the RAM divider to eliminate it as the problem? What FSB strap are you on for the MOBO if you have that setting? (some mobos have a "FSB hole" meaning they wont POST from 375-400, but 400+ is no problem )

 

[Gillbot]

Personally, I've never had a CPU degrade and i've pushed all of mine pretty far. Usually 1st to go is RAM, 2nd is Mobo then PSU, etc.

 

[alyarb]

while overclocking the fsb i use the slowest FSB:RAM ratio that the board will allow. usually it's 900 MHz 5-5-5-15. sometimes 940 or 960. is that slow enough for you? now that i'm settled in at 442*8.5, I have the ram back at 1066 5-5-5-15. The funny thing is this is all XMS2 that was rated for 800Mhz 4-4-4-12 from the factory. First I had 2x512, and then I added 2x1024, and when I added the newer memory, even though it's rated for 4-4-4-12, they were unstable as a combination. so i have to do 1066 and haven't tried going beyond that. It worked fine on my old 680i and it's doing fine at these speeds on the P45. do you think i should swap the DIMMs around and try them as only pairs?

 

[OCGuy]

4 sticks of RAM and a quad? That is a lot of stress on the Northbridge. How high have you tried the NB voltage?

 

[alyarb]

1.3