9262 with QX9650 and Q9650

Hello.

I have my 9262 configured with a 3.0GHz Quad-Core Q9650 Processor. As an experiment I have swapped the laptop's CPU with my Desktop machine's which is a 3.0GHz Quad-Core Extreme QX9650.

With the Extreme processor that is multiplier unlocked, I can via software overclock it to 3.20GHz with no problems. Past that, 3.40GHz or 3.60GHz the machine becomes unstable to complete most torture benchmark tests. It actually runs fine, even at 3.60GHz but a few minutes into 100% CPU utilization and the benchmark applications tend to terminate abnormally. Its not a thermal issue as I am monitoring that and its stays within specs so i suppose it might not be enough voltage reaching the processor. I don't have an option to change voltages as Clevo/Sager's BIOS of course does not offer such.

At the default speed of 3.0GHz both the Q9650 and the QX9650 perform exactly the same as release exactly the same amount of heat at full load. So my question is, aside from an unlocked multiplier on the Extreme Processor, what else really differs between the QX9650 and the Q9650 and what makes their thermal ratings be so different from each other according to Intel? Is the colossal price difference just for the unlocked multiplier?

I wish I could push the QX9650 in the 9262 further. I know the potential is there. I am using ARCTIC COOLING MX-2 which is the Thermal Compound that gives me the lowest temperatures on load.

Thanks for the input.

 

I would not recommend to continue running the QX9650 since its power specifications are too high to keep the system stable over a long period of time.

The reason why Clevo does not officially support such a CPU is the high TDP that, though may seem stable during use, can cause major issues down the line.

The Q9650 and QX9650 may perform the same on stock, but their power specifications are quite different:
- Q9650 = 95w TDP
- QX9650 = 130w TDP

That huge of a difference will determine a system's future health and stability.

Pretty much you are using it at your own risk... and it will probably void you warranty through Sager if they know you are using an unsupported CPU.

 

Many official European Clevo Vendors sell the 9262 with the QX9650 as a standard option with full warranties.

 

i would guess they were manufactured at the same line, however at the final stage of testing the qx chip was found to be more stable at high clocks/voltage and as such got that trendy X.

as to tdp probably max figure, since the qx can go further it will emit more down the road.

only speculations, no facts stated.

 

That may be true... but they are going against Clevo's specifications.

As long as they are giving a full warranty then that's fine... as long as they will honor it.

As for Sager, they will only warranty any authorized/supported hardware... Sager is an OEM that follows everything that Clevo tells them to the letter... they have to since they are the largest North American importer OEM/distributor. They cant afford to go against the grain to get more buyers with unofficially supported hardware that can cause issues down the line.

 

I suspect exactly the same. They seem to be pretty much the very same chip aside from the locked/unlocked multiplier reason for which at full idle and same clocks speeds, the heat they generate is exactly the same.

The 9262 with the QX9650 @ 3.20GHz is 100% stable and almost fully stable at 3.40GHz and 3.60GHz. I hear lots of noise coming from the external power supply at full CPU load and it gets extremelly hot (even when using stock 3.0GHz speed). I suspect the overclocking limitation is the power reaching the processor. Its a shame because the overclocking potential is all there and let me say it that, at 3.60GHz, the 9262 is one insane machine

 

The qx9650 has TDP 135 W, q9650 has 95 W. If I run qx9650 at 3Ghz is it going to dissipate more heat than q9650? Is it still going to wear off the motherboard?

PL

 

No. It will dissipate exactely the same heat at the same clock speed. TDP 135W for the QX9650 represents the maximum heat dissipation at full load overclocked within the theoretical maximum limits of the processor. At 3.0GHz stock speeds both the performance and heat generated byt the QX9650 or Q9650 are exactly the same.

 

One thing to watch closely is the power supply and the battery for signs of overheating.

Switching power supplies are usually designed to run close to their maximum ratings with 70-80% efficiency. I just took a quick look at my 9262 power supply and the face plate ratings are: input rating is 300VA and the output is 220VA. That implies a 220/300 or 73% efficiency ratings. In other words at full load the PS is an electric heater burning 70 watts

The 40 watt power difference in the chip is almost 20% of the TOTAL rated power supply capacity and at 73% efficiency under full load is burning an additional 30 watts in the PS. While the PS and Battery should FAIL-SAFE we have all seen the videos of on fire laptops or wall warts.

The clevo site doesn't seem to have readily available power consumption ratings. I'll run some direct power measurements on my quad core 9262 tonight but I'd keep the PS away from flammable materials and yourself and don't go out for a drink while you are running benchmarks.

 

ah, ah Very good advice! Like I said, the only thing worrying me on trying to run the 9262 at 3.40Ghz or 3.60GHz is not so much the heat on the laptop itself that seems very close to the readings I get at 3.0GHz but the seamingly increased wheezing coming from the power supply and that gets extremelly hot (even when using stock 3.0GHz speed).

 

Does your PS have the same faceplate ratings as mine and does it have a fan?

Phil Schaadt

 

The QX9650 will disspiate a maximum of 135W at stock (3GHZ), a Q9650 will dissipate a maximum of 95W at stock (3GHz). It has nothing to do with overclocking temperatures I'm affraid. Intel state the TPD at stock speeds so a QX9650 will run hotter than a Q9650, it may not be much but there is a difference.

This is why both (for example) a Q9450, Q9650 are both rated as 95W TPD - it's the maximum they will reach under full load at stock speeds. It doesn't mean they definately will hit 95W though. This is why Clevo don't support the QX9650.

 

this sound odd, the op said he saw no measurable differences.
while you say it should dissipate 40% more. while i dont doubt the intel numbers, marketing is a tricky business.

 

It sounds strange. I will try tomorrow to find something on the intel website about this.

PL

 

See here for thermal specs of Intel Quad Core CPUs:

http://download.intel.com/design/processor/datashts/318726.pdf

If you go down to "Processor Thermal Specification" all the details are there - table 5-1 compares the Processor Thermal Specifications, 5-3 is the QX9650 and 5-4 covers the Q9*** series CPUs.

The key word here is "maximum" TPD - I'm not saying that a QX9650 dissipates 40W more heat at stock speeds but it can as a maximum under full load.

 

It's hard to get a 100% load on a quadcore CPU in normal usage. In terms of casual measurements, 3Dmark06 loads my Xeon x3360 to about 40% in my OPC built Clevo 901c. I doubt the faster chip is loaded much differently for computing power and hence electrical draw and heat dissipation.

Something like Fritz Chess Benchmark can be used to load all 4 cores to measure the difference.

Phil

 

Perhaps 40% in the Graphics tests but 3DMark06 loads all 4 cores at 100% during the CPU tests so it will push the CPU a fair bit.

 

new programs and games are being developed to be multi-threaded.

up to now, only a few games (like Unreal Tournament 3) are fully multi-threaded to use all cores at full load.

But if the power draw is too much for the system, I expect to see power supply failure and motherboard damage.

 

I understand about the CPU test but I don't see it fully loaded test for both the CPU and the GPU at the same time, is there a setting I missed?

Phil

 

In 3DMark,
the CPU test only tests the CPU (fully multithreaded).
the GPU tests taxes the videocards

If you want to tax the CPU and GPU simultaneously, do a Unreal Tournament 3 benchmark... since its fully multi-threaded to handle up to 64 core... if you got it.

 

I've just made some direct power measurements and it looks to me that the PS can provide enough juice to fry the laptop but its really close. The PS might go first but I don't have any knowledge of the PS internal circuits.

In either case I sure wouldn't put a 135 (thermal) watt processor in my Clevo based machines.

Phil Schaadt

 

I have my own systems development based benchmarks that can crank both the CPU and GPU. I ran those and that is how I concluded that the Power Supply has just enough juice to fry the laptop with the heavier draw (135 thermal watts) processor.

Of course you might get lucky and have both go up in flames at the same time.

Phil

 

Totally agree.

Especially since its not a cheap notebook to throw away after ruining it from experimenting.

 

I'll also concur about no 135W TPD CPUs in the D901C.

Gophn's right on the tests - it won't stress the GPU and CPU 100% simultaneously. Part of our burn in process for OC'd desktops is to loop 3DMark06 (but don't run the CPU tests or Deep Freeze) and at the same time run 4 primes. This will give you a constant 100% CPU useage and consistant GPU useage. Running 3DMark06 at 1920 x 1200 will actually make the GPUs run hotter than at 1280 x 1024 so you can try this too. Or you can run the 1024 wPrime test which runs for about 600 seconds with 2 runs of 3DMark06. Of course, as already mentioned, it's probably not representative of every day usage but if not tested properly you can get a system that seems fine but will occasionally crash in games etc.

 

UT3 benchmark is definitely the test I will recommend for stressing the CPU (single and multi-core) and GPU (single or dual) at the same time.

I just did some stressing with my FX-60 and old 7800GTX... GPU and both core ran at 100% load.... temps went to peak (but safe with fans at max).

 

Thanks for the benchmarking tips I'll have to try UT3. I run 3Dmark06 at 1920 x 1200 and a cpu loading test of my own for the four cores to 100%. No problem on my OPC built Clevo machine.

A few more power measurements make it clear that with a fully charged battery there is enough power capacity available with the Power Supply to fry your machine if you want to do something dumb.

Interesting thread and thanks again for the tips.

Phil Schaadt

 

so.... what was your end result from your testing?

 

2 weeks intensive care in the hospital for 3rd degree burns

 

Rotflmao!!!

 

Well, he made this post on 9/14, and on 9/18 he made a post about putting in a new Q9650. By the first of October, he's seen shopping for a new laptop.

Hmmm....

 

nothing to do with you getting a new laptop. just talking about your lien on another issue is all. but you carry on and have a nice day!
no wonder we never seen any pictures.

had to retract my last statement
you mean this place right.
http://www.mycizmo.com/?main_page=product_info&cPath=65_67&products_id=181&action=configure

 

GTA 4 for PC (hardware):

good news: the minimal settings look better than the 360 version (55-70FPS)

bad news: maximum quality (draw distance, objects, textures, shadows) at 1280*1024 on a ATI 4870 can't run smoothly. 15-25fps

there are pc gta4 hardware vidoes.

 

what about medium settings at 1920x1200...do you think those would look decent?