QX9650 L739A644 + Blitz Extreme

[Aivas47a]

Thanks for taking the time to post all the thorough data eva, you're a mensch!

[Kampfzerstorer]

Awesome data spread!

Honestly I think that is about average for the qx9650.

I can only get mine to 4.1 on a P5e3 deluxe with 1.425 (set in BIOS) ((liquid cooling)) before I have to start giving it too much vcore. I start to get nervous around 1.5v, you're brave to try 1.6v+

[eva2000]

Thanks guys for the feedback.. yeah not sure how high we should go for vcore for these chips... someone who is friends with Intel folks, should ask them hehe

[simonmaltby]

My Qx9650 retail is working well but i'm trying to work out why Prime is so hard to get stable. Not that i rate running long 100% cpu intensive tests ( 10 minues is normally enough for me to feel happy).

Mine is L739A640

My chip does a super 10x400 (4Ghz) at 1.33 bios or 1.31 reported in windows. this is nice and steady and seems prime stable (2 hours before i gave up)

I can also get some very good SuperPi scores by booting to windows at 11x400 (4.4Ghz) and then scaling up to 425FSB using SetFSB. This is done with 1.525v Bios 1.49 in windows. This will run superPi without issues, but most other things crash the sys so it must be limited by the CPU or Mobo.



I can also run all the 3dmark tests at 4.5Ghz with 1.5vcore Bios 1.475 Windows without issue.

But i cant Prime at anthing higher than 4.1 atm.

By all accounts looking at what i have achieved it seems my chip is a good one. Because of this i'm leaning towards the something other than the CPU stopping Prime. If i can bench at 4.5 and super pi at 4.66 i would expect to be able to prime at 4.2 - 4.3?

Any comments or suggestions please?

[eva2000]

Nice even earlier than my L739A644 batch!

maybe prime needs alot more vcore for these chips ???

[simonmaltby]

maybe prime needs alot more vcore for these chips ???

Could do, but i notice that load temps sky rocket as volts increase. Maybe my cooling is just not cutting it... should be though as normal temps are very good its just those full load peaks. Lovley and cool under normal conditions even with high volts.

[eva2000]

Yeah temps sky rocket with voltages on QX9650.. it's normal

[oscar3d]

Totally %100 agree with you. This chip's temps are way so sensitive to VCORE changes...

[simonmaltby]

Well i seem to have found some more performance.

I have managed to...
Prime for 8hrs small FFT at 4.3 (1.45 vcore)
Run SuperPi 1M at 4.7 (1.525 vcore

This is all with fairly basic water cooling doing cpu, 2xgpu's and the chipset.

Looks like my chip is a very good one. It was the motherboard holding me back. More Nbridge volts required and less everything else.

[eva2000]

Well i seem to have found some more performance.

I have managed to...
Prime for 8hrs small FFT at 4.3 (1.45 vcore)
Run SuperPi 1M at 4.7 (1.525 vcore

This is all with fairly basic water cooling doing cpu, 2xgpu's and the chipset.

Looks like my chip is a very good one. It was the motherboard holding me back. More Nbridge volts required and less everything else.

Congrats man.. if only mine could do the same hehe

[safan80]

I just got a 39A as well. I won't be testing till later.

[eva2000]

I just got a 39A as well. I won't be testing till later.

Good luck

Made some interesting findings when I ran Super Pi 32M to the max clock speed I could with my QX9650 L739A644 on water cooling on Asus Blitz Extreme 0904 bios.

Findings:

1. As we all know disabling cores allows cpu to scale a bit higher due to reduced heat etc. Seems my QX9650 scaled alot better with 2 cores disabled. 4 core max = 4330Mhz, 2 core max = 4454Mhz and 1 core well no real sustantial stable gain for 32M over 2 core max.
2. What's interesting is that when 4 cores are enabled my coretemps show core 0/1 fixed at 38C after Dtek Fuzion block was used and under load only core 2/3 moved to higher temps substantially.
3. Now when I disabled 2 cores, coretemp lost all readings even for remaining 2 cores, but everest showed cpu temp as the higher temps - so probably reflective of core 2/3 being the main cores used by Windows XP. If this is true then core 2 or thread 3 in wprime would be the main core that does all the work on most occassions right ??
4. So thread 3 slowness in wprime maybe due to other background processes attached to core 2 (core 0, 1, 2, 3) ?
5. Could be related to how the motherboard bios is arranging the cores ?

4 core max



2 core max



Max CPUZ Validation @4605Mhz



Max Super Pi 1M @4534Mhz
Did do 4567Mhz for 10.078s but system rebooted

[Triggerhappynz]

Just got my qx9650 L739A637 been running prime for over an hour at 416 x 10 @ 1.38v how long are you guys running for before you are happy?

[safan80]

mine is a later batch...

here's 4.4Ghz
http://www.xtremesystems.org/forums/...postcount=4133

[Aivas47a]

Nice. I'm stuck at "only" 4.25ghz (10x425) at 1.45 vcore (actual). L739A

[Kampfzerstorer]

You guys are makin me jealous... lol

Any secrets to makin this thing clock higher?

[safan80]

You guys are makin me jealous... lol

Any secrets to makin this thing clock higher?

yes... voltage

[safan80]

I can also run all the 3dmark tests at 4.5Ghz with 1.5vcore Bios 1.475 Windows without issue.

But i cant Prime at anthing higher than 4.1 atm.

By all accounts looking at what i have achieved it seems my chip is a good one. Because of this i'm leaning towards the something other than the CPU stopping Prime. If i can bench at 4.5 and super pi at 4.66 i would expect to be able to prime at 4.2 - 4.3?

Any comments or suggestions please?

run occt 1.2 beta prime can be stable on the Maximus Formula but the system will still be unstable. the only way to know for sure it to run occt. I can be prime stable with this boardand it means nothing to me.

[oscar3d]

Omg!!

[fadetoblack]

Just got my qx9650 L739A637 been running prime for over an hour at 416 x 10 @ 1.38v how long are you guys running for before you are happy?

what kind of cooling do you have? i ran prime overnight on my qx9650 at 400x10 at 1.3875, it was stable for 6.5 hours BUT when i woke up i looked at the temps and they had reached 77. i looked through the core temp logs and it even hit 80 at certain points so i stopped there right. hopefully i can get it stable at a lower vcore, wasnt successful at 1.36 (prime gave an error after about 1.5 hours)

[oscar3d]

For what you've shown Fade to Black, and based it my experience this is as far as you can take it with the lower VCORE.

I'm with an Asus P5E3 Delux and I'm running mine @ 1.35 10x400. LowerVCORES and it will crash...

Those temps are what most of people are getting on air... Tuniq Tower or TRUE

I'm getting 70-71C with my TRUE... which Coretemp are you using, the latest?

Well 6.5 hours priming... not bad... not bad at all....

[eva2000]

Yeah i was getting around 67-70C on thermalright ultra 120 extreme @4000mhz but at 1.440v room temp around 28+ C

[fadetoblack]

For what you've shown Fade to Black, and based it my experience this is as far as you can take it with the lower VCORE.

I'm with an Asus P5E3 Delux and I'm running mine @ 1.35 10x400. LowerVCORES and it will crash...

Those temps are what most of people are getting on air... Tuniq Tower or TRUE

I'm getting 70-71C with my TRUE... which Coretemp are you using, the latest?

Well 6.5 hours priming... not bad... not bad at all....

im using core temp beta 0.95.4. so you're saying i should keep it at this vcore with those kind of temps or try lowering it a little? btw i swapped out the tuniq and put in a TRUE and temps are a little lower, but for some reason the diff between core 0 and 3 is 10-12 degrees now. so im seeing about 55 on core 3 and 67 on core 0 sometimes.

[joe.denice]

yeah these chips after 1.375 volts seem to get rather toasty, i will try and shoot for 4.2ghz tonight and see how well it goes.

[fadetoblack]

btw, how do you guys have your TRUE's installed? horizontally or vertically? also, what side did you install the fan on, and is it pushing air through the TRUE (out through the back of the case) or pulling ?