P5Q Pro Turbo & Q9550 A Miracle Combination?

I just sold my E7200 and purchased a Q9550 from Microcenter for $169. I smeared some OCZ Freeze thermal paste on the new CPU and nearly killed myself trying to reattach my OCZ Vendetta 2. Everything i've read about OC'ing a Q9550 hasn't exactly been easy, but when I booted up on stock, I saw that the Q9550 was running at about 1.1v in CPUZ under full load. Is that pretty low or normal?

At any rate, I rebooted into my BIOS on my P5Q Pro Turbo and made a couple minor adjustments including Load Line Calibration, CPU Spread Spectrum, and my memory voltage. I then proceeded to adjust my FSB to 400 from the standard 333. It booted just fine at 3.4Ghz with no other changes. Well, Im feeling adventurous so I reboot and switch the FSB to 450 and made no other changes. It boots without a hiccup to 3.825Ghz .

Well, WTH! I couldn't get my E7200 past 3.55Ghz without 1.43V and other adjustments. So I run PRIME95 for 29 hours and its stable on all four cores!!!!

The only issue I have is one of my cores is running about 10C hotter than the others. The other three hover around 50~55 load while the one runs 60~65. (60 for the first several hours and 65 after 20~hrs) During normal use they are all within reasonable temps, but the one still runs about 10C hotter. I may pull the pain in the #$$ Vendetta 2 off, check the paste,and reattach before I go any higher. But it looks like I may be able to easily hit 8.5 x 500FSB (4.25Ghz) easily with just a couple minor tweaks!

Vcore is AUTO currently ? if yes, whats your Vcore currently ?

For your temps i would no be scared. its normal.

I dont think youll get that high can youll get FSB limited. QUADS are lil hard to take over 480fsb. You can still try


And one question. What CORETEMP is telling about your VID ?

WHy do people keep repeating that the quads are FSB limited? What is the source of all of that?

I went to 4.15 Ghz on air last night, just to brek the DREADED 480FSB BARRIER. Ran Everest Ultimate stress test for 30 mins and stopped it only because the temps were touching 70C. No problems with anything. I am sure if I extend my liquid cooling to the CPU I will go higher than that. I have seen several reviews already done with Q9550 on 4.2 Ghz, which is well beyond 480 FSB. Also, 480 equals 4.08 Ghz! How is that a limit? Most people don't have platforms that support that anyway. You need great motherboard and pretty good cooling to go higher. No AMD can touch 4.0 Ghz anyway, without extreme unpractical solutions.

I got mine from microcenter as well and as far as I am concerned - the sky is the limit!

Its not a UNBRAKABLE limit. Its where it starts be become HARDER.

I dont think anyone can go higher on AUTO ... he reached 450 on AUTO and its already something nice. But if i was him i would MANUAL cause AUTO lean towards OverVolting.

I also reach 4.08 quite EZ with my quad. Temps under 60. Im sure i can go higher.... as you are ...

Everything is auto except the FSB, Memory Voltage, LLC, and Spread Spectrum. Im currently at work and will report more details such as VID and VCORE under load when I get home this evening. I do know that my board is capable of 550FSB or higher so at the very least that will not be a limiting factor.

Why do you think I can't go any higher on AUTO? What is the max recomended voltage for these chips? I know the max safe for the E7200 is about +/- 1.38 ~ 1.41v.

ravaneli said:

WHy do people keep repeating that the quads are FSB limited? What is the source of all of that?

I went to 4.15 Ghz on air last night, just to brek the DREADED 480FSB BARRIER. Ran Everest Ultimate stress test for 30 mins and stopped it only because the temps were touching 70C. No problems with anything. I am sure if I extend my liquid cooling to the CPU I will go higher than that. I have seen several reviews already done with Q9550 on 4.2 Ghz, which is well beyond 480 FSB. Also, 480 equals 4.08 Ghz! How is that a limit? Most people don't have platforms that support that anyway. You need great motherboard and pretty good cooling to go higher. No AMD can touch 4.0 Ghz anyway, without extreme unpractical solutions.

I got mine from microcenter as well and as far as I am concerned - the sky is the limit!

Click to expand...

Try running Prime95 large-fft over 480FSB. You actually have the best chance of getting over 480 b/c you have a GB P45 board, but as said it starts to get very difficult to keep large-fft stable beyond that. Passing 1 stress test for only 30mins doesn't mean you're completely stable.

If there was no FSB limit you'd see people dropping the multi to 8 or 7.5 and running the FSB up to 500-550, but that just won't fly for 24/7 use on a quad. You will have problems.

You also need to define what you call stable. I like to run a full battery of tests to try and find any weaknesses. As I said previously I can run gfx benchmarks all day long at 500FSB (4.25GHz)...and can probably game like that as well. It just doesn't pass my criteria for what I call 24/7 stable b/c it fails certain long-term tests that specifically stress the NB/FSB. If it works for you and you don't mind the occasional BSOD then go for it. I just like to advise people to work toward a completely stable system unless all they are trying to do is get a high score on a benchmark.

If the chip had a higher multi it would go much higher w/ 24/7 stability. As you pointed out 4.08GHz seems like an awfully low limit since it doesn't take a lot of voltage to get there. I can get to 471x8.5 4GHz w/ only 1.256v. So why can't I give it 1.45v and hit 4.2-4.5GHz easily? FSB-wall is why.

If you still don't believe that there is a FSB limit then show me a SS of large-fft running at 550x6. That's only 3.3GHz. Easy, right? Heck , just a CPU-Z validation would be cool.

I'm not telling anyone to stop at 480. I'm just being practical and letting them know they will probably hit a wall. If they don't do thorough testing at these high FSB speeds then they may encounter random BSOD. By all means push it as far as it will go. I spent 4 weeks trying to get 485FSB. I even made it 12hrs on large-fft and then had a core failure at 12-1/2hrs. But aI also had to run the RAM slower and the tRD very loose. It's just not practical to run at the ragged edge of what the FSB can do b/c you end up sacrificing so much bandwidth to get there.

ChaosInMind said:

Everything is auto except the FSB, Memory Voltage, LLC, and Spread Spectrum. Im currently at work and will report more details such as VID and VCORE under load when I get home this evening. I do know that my board is capable of 550FSB or higher so at the very least that will not be a limiting factor.

Why do you think I can't go any higher on AUTO? What is the max recomended voltage for these chips? I know the max safe for the E7200 is about +/- 1.38 ~ 1.41v.

Click to expand...

Be careful w/ a lot of those auto settings. PLL and FSB Term voltages can get dangerously high. Hitting 550FSB on a dual is completely different than hitting it on a quad. I can hit 600FSB w/ a duallie on this board.

Thats the point, its not CPU limited AFAIK. I have some testing to do from jason though in a few weeks. Mine was cherry at 500FSB with P95 Small FFT and blend.. I forgot to mention IBT for a couple hours. I also forgot to mention my Q9450 would go over 480FSB with my P5Q stable, but with the EP45 I have, that was nothing. So I guess in my experience I believe its the board causing that instability that jason is talking about.

Its board limited as there are PLENTY of people on XS that are rock solid stable (w/e that means to them, but being XS, Id say its stable) at 480FSB+.

ravaneli said:

WHy do people keep repeating that the quads are FSB limited? What is the source of all of that?

Click to expand...

The limit is 500FSB and does not pertain to people running p45 motherboards

I picked upa p5q deluxe and a q9300 a year or so ago and had it up to 530

IMO it's partly the board, and partly the CPU. Every chip is a little different and that is also true of the 2 different dies contained in a quad. You will work towards getting one of the dies stable only to make the other die less stable. The electrical routes are not the exact same distance, and the timings can get slightly off causing a failure. The separate GTL settings and more advanced circuitry in the newer chipsets (P45) try to compensate for this and they do an admirable job. It also appears that GB did a very good job in tackling the problem using auto settings. But, at the same time auto settings can make certain voltages dangerously high.

I've seen plenty of SS's of 500-515FSB w/ 45nm quads here and at XS, but never w/ a 12hr+ SS of large-fft running. Large-fft gave me the most problems as I pushed for a max FSB speed. And that was w/ the RAM 1:1 loose, tRD of 13, and CPU multi at 6 to focus squarely on the FSB speed and what was possible.

And even if you can get the FSB up 490-500 you'll have to pump in the voltages so high on VTT, FSB, PLL and loosen up timings so much that you really end up shooting yourself in the foot. Bandwidth goes down, and temps go up.

From my experience the sweet spot is 470-480. Look around at 24/7 settings and see if you don't agree w/ me.

Neur0mancer said:

The limit is 500FSB and does not pertain to people running p45 motherboards

I picked upa p5q deluxe and a q9300 a year or so ago and had it up to 530

Click to expand...

Was that just a CPU-Z validation or was it large-fft, small-fft, and Linx stable?

jason4207 said:

I've seen plenty of SS's of 500-515FSB w/ 45nm quads here and at XS, but never w/ a 12hr+ SS of large-fft running. Large-fft gave me the most problems as I pushed for a max FSB speed. And that was w/ the RAM 1:1 loose, tRD of 13, and CPU multi at 6 to focus squarely on the FSB speed and what was possible.

And even if you can get the FSB up 490-500 you'll have to pump in the voltages so high on VTT, FSB, PLL and loosen up timings so much that you really end up shooting yourself in the foot. Bandwidth goes down, and temps go up.

From my experience the sweet spot is 470-480. Look around at 24/7 settings and see if you don't agree w/ me.

Click to expand...

ONe man's stable is another man's not even close. Like I said, Im completely stable AFAIK with 8 hour+ runs of small and blend along with IBT for a bit as mentioned above. If I dont get hangs EVER, Im stable.

For the record, my VTT at 500Mhz is 1.4 (.5 away from intel's documented max). PLL is 1.57 from 1.50 (only option is 1.57, nothing inbetween).

Is that a sweetspot? Sure I will buy that for a dollar. But is that a limit on FSB on the CPU? Hardly. How could that be a limit when Im benching into the 4.6Ghz+ range? I see what you are saying but there is enough data against the specific statement to sway most critics.

Why isnt small fft and blend + IBT/Link enough if it doesnt freeze or hiccup? What is different there that you believe makes the difference? Large FFT only fills up a bit more cache and uses a bit more memory last I remember. Only i7 is large FFT worthwhile. small/blend is where is at for CPU/system. I personally make a custom run for my final stage to use most my ram and the cache.

IMO if it fails any test then it isn't stable for 24/7 use. Trough trial-and-error I found a test that fails pretty quickly when the FSB was being pushed hard, so I use that as a guide to compare my testing to others. I also passed small-fft for over 20hrs at 485FSB, but failed large-fft at just over 12hrs.

When I run benchmarks all I care about is it passing that particular benchmark. It can BSOD as long as I save the SS 1st!

And you are running at 4.05GHz. Why not higher? Why not run 500x8?

And rav is at 3.8. Same question.

If these chips can go so much higher then why not run them there all the time?

I just don't want folks to see some CPU-Z validation or super-pi run at 4.25-4.5GHz and then think they are going to be able to go and get a Q9550 and run those speeds 24/7. I err on the side of practicality.

See, like I said, one man's stable is another man's not at all!

I have no need for higher for 24/7 use. Could I run it at 4.5Ghz stable Small FFT/Blend? Sure I can, and have. I couldnt do that on my P5Q deluxe though at 500Mhz (like I said more board dependent than anything). I have no need for anything more except to bench. I would expect for 24/7 on a EP45 giga board that a Q9550 E0 can run 4Ghz and Q9650 more 24/7. I simply choose not to run it that fast even though I have 500Mhz+ for stability (my stable).

My CPUz validation and Spi runs are in the 4.6Ghz+ range. Note how I have my sig with 4.05Ghz (4.6Ghz+ benchable).

Boards are just as much to blame in OC'ing a chip as teh chip itself. I had my e8400 on my old P5Q-Pro running at 512x8.5 STABLE (and I could prime 512x9, just not stable for 24/7). 512x8.5 would run at I believe 1.425v?? And 500x8 would run at 1.368v.

My motherboard died when my PSU blew and I got a RMA board. Since then, I could never hit 500 FSB, 478 was my wall. I ran 478x9 @ 1.53v, HORRIBLE! 4.3ghz took .1v more with a lower FSB. My chip was a decent was, but my RMA board sucked.

Well, I am home. I got home, rebooted and the only thing I did was set the NB to 1.4v and the FSB to 500. Vista booted up and prime small is running nice so far, I attached a pic.

Am I doing something wrong? Is 1.4vcore on a q9550 dangerious?

EDIT: I manually set the VTT to 1.34v and set the northbridge to 1.4v and re-started prime on small fft. I hope all of these values are not too high.

CPUID and Windows Task Manager seem to be getting in the way.

jason4207 said:

CPUID and Windows Task Manager seem to be getting in the way.

Click to expand...

Looks like someone got distracted

I just manually set the VTT, PLL, and North Bridge to standard voltage and moved my FSB back down to 450. VCORE is only 1.344V under load so I think I'll just leave it at this, Im sure the extra 400Mhz x4 wont be that big of a deal for the extra voltage and heat.

I'd think you'd be able to get it stable at 3.825GHz w/ less vcore. You might have to bump up vtt and pll just a tad to do it, but it would be worth it IMO.