Quad core overclocking?

[gorobei]

I'm thinking of swapping out my E4300 (OC'd to 3Ghz right now) for a Q6600 when the price drop comes.

With 4 cores, overclocking is probably unnecessary. And I kinda prefer cool and quiet systems. But what kind of performance increase is possible with the core2 quads?

I do 3D modeling, animation, and rendering; so there is a hefty load on the proc.

 

[Noubourne]

Actually, with 4 cores in gaming, overclocking becomes necessary because they usually clock the 4-core chips lower than the 2-core chips, so MOST games show better performance on the higher-clocked rig.

Lots of people are OCing the quad-cores. I will be one of them come July!

 

[gramboh]

It depends on if your applications are multi-threaded properly or not. Right now not many games take advantage of multiple cores, so the optimal configuration is a higher clocked CPU (e.g. a C2D at 3.6GHz will give more FPS than a C2Q at 3GHz). This will/has to change going forward because the trend is toward more cores rather than higher clock speeds. Your 3D modeling/animation/render apps will likely benefit from more cores and have a large increase in performance (although not always linear other than rendering from my understanding). That being said, with Q6600 priced so low in a few months, why not? I'm going to drop one in my PC toward the end of the year, just for fun (assuming I can handle the heat at 3GHz).

 

[MarcVenice]

qauds run HOT, so you'll need very good air cooling, but even then ya won't get to far with em. 3.6ghz requires water.

 

[aigomorla]

Originally posted by: MarcVenice
qauds run HOT, so you'll need very good air cooling, but even then ya won't get to far with em. 3.6ghz requires water.

uhhh DONT get a quad until you know what your getting. This guy is 100000000000% correct.

They require beefy cooling devices, and most people who overclock them on this forum are running 50+ dollar Air heat sinks, or 300 dollar waterkits.

The people on 300+ dollar waterkits are the ones you see with uber low temps like me. Dont think you'll replicate the temp curve. It takes a fair amount of capital to bring these suckers down to tollerable temps.

My personal best with temps/quad overclocking:
http://i125.photobucket.com/albums/p73/aigomorla/NewTemps.jpg

My chip aint the best quad, and it doesnt compare squat against my E6600. However she will do 3.6@ massive voltages. And load temps peak to near 70C. I load her 24/7 @ 100% on wcg, so having her on those temps 24/7 expecially during the summer is not what i like.

So do your research over again on quads. The chips will put out almost 2x the heat your C2D does. Its 2 C2D's stamped together.

 

[f4phantom2500]

Originally posted by: aigomorla

Originally posted by: MarcVenice
So do your research over again on quads. The chips will put out almost 2x the heat your C2D does. Its 2 C2D's stamped together.

Yeah, but that doesn't necessarily mean it's putting out twice as much heat, because it is still 1 processor; a quad will put out less heat than 2 duals because the 4 cores share a lot of the same things so they don't need twice as much power as a dual, so they wouldn't put out twice as much heat as a dual...but yeah they still do put out a lot of heat.

 

[lopri]

C2D's Tjunction = 85C / C2Q's Tjunction = 100C

Under air-cooling, quad-core overclocking will be bound by the temperature before anything else. (Assuming your motherboard is capable of dealing with quads) At 3.30GHz with very good air-cooling with the system outside the case, load temps soars to 70C and beyond. Gah.. Couldn't imagine what it'd do inside a case so I lowered the frequency to 3.20GHz, which is much better especially after I re-seated the HSF. (about 60C load) And that's with Prime LargeFFT, so with SmallFFT or TAT, the temps will go up even higher.

To be honest, I feel my E6400 @3.60GHz was snappier with Windows XP, than Q6600 @3.20GHz. But with Vista and lots of RAM, I'm liking this quad-core a lot. (especially when I run Virtual PC 2007 or Company of Heroes, or both. ) If anyone thinks of buying one, I suggest to do so immediately after the price drop coming July. With the increased demand on Q6600 and new SKUs, quads (and duals) will probably binned heavily.

Off Topic: What's the Tjunction of E4300? And E2140? Does anyone know? (Can be read from CoreTemp)

 

[Doctahg]

Just ran TAT for 15 minutes..topped out at 69Core 1 67 Core 2. I am playing with 3.3 1200 x11. It was constant for 10 minutes. But the individaul temps in Core temp and the logged ones on TAT stayed high 40's low to mid 50's..why didn't they go up? Supposedly..running the quads at 70C is not a problem.....

 

[gorobei]

Originally posted by: lopri
C2D's Tjunction = 85C / C2Q's Tjunction = 100C


Off Topic: What's the Tjunction of E4300? And E2140? Does anyone know? (Can be read from CoreTemp)

my E4300 Tjunction is 100C as read by cortemp. What kind of difference does this make? I only ask as I'm new to overclocking.
I'm running a noctua 12F, with idle around 25C and load temps around 50C

-----------------------------------------------

So we know that Q6600s run hotter (more cores in apprx same size package), but is the performance increase over stock speeds worth the heat? Also they just released a new revision that can run hotter, right?
I was surprised to hear that general interface responsiveness dropped.

FYI, 3d apps do take advantage of multiple procs during rendering and have done so for a while now (years) with linear scaling (double and quadruple). The really sad part is they're only barely starting to take advantage of the GPU in simulation and effect calculation.

 

[yacoub]

E4400 is 100C in CoreTemp95, 85C in CoreTemp94. There's plenty of debate as to which one to go by. There is exactly 15C difference in the temps they report on both cores as well. Everything from the BIOS to SpeedFan to CoreTemp94 report 15C lower than CoreTemp95. CoreTemp95 is the only program that claims temps are 15C higher than everything else. Does this mean CT95 is right and everything else is wrong, or the other way around? No one has been able to say for sure.

My E4400 (@ 3.0GHz) idles around 29C aka 44C, Loads in Orthos at 53C aka 68C.

 

[terentenet]

Quad Cores overclock just fine now. There were some problems for the 680i users at first, fixed now with new rev. boards.
When cooled properly, quad cores can reach allmost 5GHz. 4.8 is highest I've seen.

3.6GHz is what I run. I can hit 4GHz easily, but wouldn't be too good for the Vapo LS unit that being used to cool the CPU.
3.6GHz 100% stable at 1.475v. 4.05GHz 100% stable at 1.525v

 

[Doctahg]

I haven't pushed it yet, as I'm still playing with the cooling, but I can't go higher than 3.3 right now. I haven't pushed the voltage up higher than 1.34 yet though. I tried 3.49?ish..got BSOD's.

 

[aigomorla]

Originally posted by: terentenet
Quad Cores overclock just fine now. There were some problems for the 680i users at first, fixed now with new rev. boards.
When cooled properly, quad cores can reach allmost 5GHz. 4.8 is highest I've seen.

3.6GHz is what I run. I can hit 4GHz easily, but wouldn't be too good for the Vapo LS unit that being used to cool the CPU.
3.6GHz 100% stable at 1.475v. 4.05GHz 100% stable at 1.525v

and you HAVE A ES chip.

Why not go out and buy a Q6600. Trust me, ES chips are a completely different story, and ive been working hard at getting one.

So NO ES chips please. There not fair.

It would be like me bringing a SLR to a SL500 convention. Thats seriously not fair to drop ES's in the comparisons.

 

[lopri]

I think it's more to do with his phase-change cooling? I wouldn't doubt that many of current retail Q6600 will do 3.60GHz with a vaporchill. Even I can run a few benches @3.60GHz and mine is a retail 640F that is known to be a sh*tty batch.

 

[terentenet]

Yes, my overclocking results are 90% determined by the phase change cooling. ES.... last 2 CPUs, ES is all I could find. ES X6800, now ES QX6700. It's because I got those from a guy with connections in Taiwan, before being released to the public.

My computer is also NOT tuned for high overclocks. All build is inside a Lian Li V2000B Plus II case, including the Vapo unit. That compressor dumps heat inside the case. Lots of heat. Check the picture in my signature to see what it looks like.
It's a hybrid. Phase change for the CPU, watercooled all others with 2 loops.

 

[aigomorla]

Originally posted by: terentenet
Yes, my overclocking results are 90% determined by the phase change cooling. ES.... last 2 CPUs, ES is all I could find. ES X6800, now ES QX6700. It's because I got those from a guy with connections in Taiwan, before being released to the public.

My computer is also NOT tuned for high overclocks. All build is inside a Lian Li V2000B Plus II case, including the Vapo unit. That compressor dumps heat inside the case. Lots of heat. Check the picture in my signature to see what it looks like.
It's a hybrid. Phase change for the CPU, watercooled all others with 2 loops.

*sigh*

no offense, im just jealous at your ES chip....

But ES chips arent fair.

Dammit wheres yoxxy, im sure he would have something relavent to contribute to this thread. Since he has to be the one person who i think has the most ES chips on this forum.

But yeah, ES chips arent fair. :\

 

[terentenet]

Aigomorla, why do you complain about your chip? It ain't bad. 1.275v for 3.375GHz is not bad at all. At 1.4v max you should be able to keep 3.6GHz stable. Have you tried it yet?

 

[Kromis]

Originally posted by: yacoub
E4400 is 100C in CoreTemp95, 85C in CoreTemp94. There's plenty of debate as to which one to go by. There is exactly 15C difference in the temps they report on both cores as well. Everything from the BIOS to SpeedFan to CoreTemp94 report 15C lower than CoreTemp95. CoreTemp95 is the only program that claims temps are 15C higher than everything else. Does this mean CT95 is right and everything else is wrong, or the other way around? No one has been able to say for sure.

My E4400 (@ 3.0GHz) idles around 29C aka 44C, Loads in Orthos at 53C aka 68C.

Holy crap! My E4300 is at 2.4GHz and it idles around 50C! I haven't tried the load yet...

WTF?!?! I also have a Arctic Cooling Freezer Pro 7!

 

[aigomorla]

Originally posted by: terentenet
Aigomorla, why do you complain about your chip? It ain't bad. 1.275v for 3.375GHz is not bad at all. At 1.4v max you should be able to keep 3.6GHz stable. Have you tried it yet?

no it takes 1.475 in bios to keep her at 3375. To pull 3.6 i require 1.57V.

That pushes my load temps kinda high. upper 60's lower 70's depending on ambients on full load.

My watercooling setup is just that good to keep her load temps that low.


i want a Quadcore that can do 3.6ghz @ 1.5V in bios. Those are kinda hard to find.

So i am still looking for a B batch, Quad, or a ES quad.

 

[kallekill]

When was it Intel will introduce the smaller 0.45 chips? Will the July Quads at 266 $ be 0.45-chips?

 

[terentenet]

Originally posted by: aigomorla

Originally posted by: terentenet
Aigomorla, why do you complain about your chip? It ain't bad. 1.275v for 3.375GHz is not bad at all. At 1.4v max you should be able to keep 3.6GHz stable. Have you tried it yet?

no it takes 1.475 in bios to keep her at 3375. To pull 3.6 i require 1.57V.

That pushes my load temps kinda high. upper 60's lower 70's depending on ambients on full load.

My watercooling setup is just that good to keep her load temps that low.


i want a Quadcore that can do 3.6ghz @ 1.5V in bios. Those are kinda hard to find.

So i am still looking for a B batch, Quad, or a ES quad.

Damn! I am not good at overclocking, I just began my OC ventures when I got the phase change setup, 1 month ago.
The need for less volts can be due to the low temps I get with phase change, so less electron migration and such. A friend who overcloks frequently told me that.
I see you have an EVGA A1 board. So do I. BIOS P28? Try my voltages please:
CPU : 1.475v
FSB : 1.4v
SPP : 1.5v
MCP : 1.55v
SPP<->MCP : 1.55v
I run FSB@450MHz, 8 multiplier (3.6GHz) or 450x9 for 4.05GHz.
In BIOS - CPU Features disable everything (leave the 4 cores though).
Memory:
SLI Memory: Disabled
Memory unlinked, but running 1:1 (1800QDR), memory 900MHz 4-4-4-12 2T @ 2.25v.
If I run Linked and Synced, my computer will lock from time to time.
You might be able to squeeze more, I have 4Gb (4x1Gb sticks) and playing with those is tricky.

In BIOS, I've tried to stop everything I'm not using. Stopped IEEE 1394, Stopped the IDE controlled (I only have 6 SATA drives), stopped HD audio, Serial Port and FDC controller.
That should free up more IRQs and put less strain on the chipset.

As far as I've noticed, Vcore is not everything there's needed for a stable overclock. For example, if I lower the SPP voltage, I will crash. The tricky bit is that I can get around this by either uping the SPP again or up the Vcore more! So, your chip might overclock well with not many volts, on the Vcore, you might have the voltages messed in BIOS.

 

[cmdrdredd]

Originally posted by: gorobei
I'm thinking of swapping out my E4300 (OC'd to 3Ghz right now) for a Q6600 when the price drop comes.

With 4 cores, overclocking is probably unnecessary. And I kinda prefer cool and quiet systems. But what kind of performance increase is possible with the core2 quads?

I do 3D modeling, animation, and rendering; so there is a hefty load on the proc.

Just a quick note...almost every Quad I've seen will run 3.2Ghz no problem at all. People say Quads suck for overclocking say that while looking at heat. They get hotter that's all, but I've built a system for a guy at work with a Q6600 and it did 3.2Ghz no problems at all. It was stable in every test, but needed more voltage than a dual core CPU.

So yes, Quads overclock well but will be hotter. That's it.

 

[Samus]

My experience with my quad core hasn't been cpu temps, but northbridge temps of the 975x chipset. seems to run 70c+ all the time, and when it hits 78c+ the system becomes unstable. I have active cooling and a huge heatsink on it.

When I ran 800mhz fsb cpu's like my old 6300 dual core, the northbridge never went above 60c. above 1066fsb really hammers it.

DFI Infinity 975x

pics:

http://www.imgplace.com/getcode.php?img=directory/dir3413/1180379812_1459.jpg]

[/url]

http://www.imgplace.com/getcode.php?img=directory/dir3413/1180379814_4906.jpg]

[/url]

http://www.imgplace.com/getcode.php?img=directory/dir3413/1180379816_9744.jpg]

[/url]

http://www.imgplace.com/getcode.php?img=directory/dir3413/1180379819_5379.jpg]

[/url]

333mhz from 266mhz is totally stable. 350mhz requires more northbridge voltage, but then I run into heat issues after awhile and it wont pass prime after 3 hours consistantly.

too bad, because 350fsb is 2.8ghz even, which is a great one-up from the 2.13ghz stock. and as said, games love the raw speed, not the number of cores: my quad core at stock 2.13 was slower in battlefield 2142 than my core duo 1.86 oc'd to 2.53 (same 266mhz to 333mhz fsb overclock)