Intel Core 2 Extreme QX9650 Review

Admin - 2007-11-21 17:14:31 in CPU's
Category: CPU's
Reviewed by: Admin   
Reviewed on: November 25, 2007
Price: $1,345.00



Do you remember when processors were so big and heavy that you had to lift them with both hands? Well luckily those days are long gone and as technology progresses, processors are slowly shrinking in size while increasing in performance. Over the last two years we have grown familiar with 65 nm processors that actually produce less heat and are more energy efficient than processors of the past . Now there is something new, the 45nm processor which can deliver greater performance and better power efficiency. So what CPU manufacturer is producing such a processor?

The Intel Core 2 Extreme quad-core QX9650 processor is one of Intel’s first 45nm processors to utilize hafnium-infused circuitry to deliver greater performance and more power efficiency. The Intel Core 2 Extreme quad-core QX9650 also runs at 3.0 GHz and has 12mb of L2 cache and a front side bus of 1333 MHz, something that should make any enthusiasts mouth water. Since the introduction of quad-core processors Intel has been a leader in innovative technology but will the latest technology used in the Intel Core 2 Extreme quad-core QX9650 be worthy?


Closer Look:

Our Core 2 Extreme QX9650 was shipped directly from Intel so it was packaged in a black static proof box. The inside of the box is lined with both pink and black foam to protect the processor.







The QX9650 looks no different than any other LGA 775 chip but its what's under the hood that should make the difference.




To install the Intel Core 2 Extreme quad-core processor, you will first need a motherboard that is compatible with the CPU. Whether it be through a bios flash or a motherboard that is designed to support it is of extreme importance. As you can see, I first chose to use the Abit IP35 motherboard which I thought was compatible with the processor, but I was mistaken and later found that at  this time, only the Abit IP35 Pro support it. I then tried my DFI Lanpaty 680i LT, but ended up crushing the resistors on the rear of the board installing the Thermalright Ultra90 heatsink. So finally with a bios flash, I was able to utilize my DFI Lanparty UT P35-T2R.







If your motherboard needs a bios flash in order to run the QX9650, I would suggest flashing the bios prior to installing the processor, all three boards (Abit IP35 not supported) would not even post prior to a bios flash. I had originally chose to use the Thermalright Ultra 90 heatsink on the DFI motherboards because of their preinstalled NB heatsinks. My Zerotherm fan without a mod will not mount unless the fan itself is facing upwards, but since I had already crushed the resistors on one DFI board with the rear plate of the Ultra 90, I decided to use the Zerotherm heatsink facing upward. It comes with an X bracket where the other has an oblong bracket.



Spec Number
CPU Speed
3 GHz
Bus Speed
1333 MHz
Bus/Core Ratio
L2 Cache Size
12 MB
L2 Cache Speed
3 GHz
Package Type

Manufacturing Technology

45 nm
Core Stepping
CPUID String

Thermal Design Power


Thermal Specification


VID Voltage Range

0.85V – 1.3625V




The processor will be tested with our benchmarking suite to show what kind of performance this chip delivers. The benchmarking suite used includes both system tests, as well as gaming benchmarks. To make performance comparisons, the QX9650 will be benchmarked against other processors, including one of the latest CPUs by AMD, as well as other Intels. All chips will be run at default specs to eliminate any variables. All video card settings were left at setup defaults, also to eliminate any variables.


Testing Setup:

Comparison System:


The system tests we will be using are listed below:

We will start with Apophysis. This program is used primarily to render and generate fractal flame images. We will run this benchmark with the following settings:

The measurement used is time to render, in minutes, to complete.


Lower is Better


WinRAR is a tool to archive and compress large files to a manageable size. We will use 10MB, 100MB and 500MB files and test the time needed to compress these files. Time will be measured in seconds.







Specview is a benchmark designed to test OpenGL performance. The tests used for comparison are listed below. The default tests were chosen to be able to compare across platforms. In these tests, higher scores equate to better performance.



Higher is Better


Higher is Better


Higher is Better


Higher is Better


PcMark05 is used to measure complete system performance. We will be running a series of tests to gauge the performance of each individual processor to see which, if any, rises above the others.








Higher is Better



Higher is Better



Higher is Better



Higher is Better


One word, WOW!


Sisoft Sandra is a diagnostic utility and synthetic benchmarking program. Sandra allows you to view your hardware at a higher level to be more helpful. For this benchmark, I will be running a broad spectrum of tests to gauge the performance of key areas of the motherboards.






Processor Arithmetic



Multi-Core Efficiency



Memory Bandwidth



Memory Latency



Cache and Memory



File System


Physical Disks



Power Management Efficiency


Processor Arithmetic,Multi-core Efficiency as well as Cache and Memory are producing values many of us have not seen before. Can you say 12mb L2 cache?



Sciencemark tests real world performance instead of using synthetic benchmarks. For this test, we ran the benchmark suite and will use the overall score for comparison.






Higher is Better!


Cinebench is useful for testing your system, CPU and OpenGL capabilities using the software program CINEMA 4D. We will be using the default tests for this benchmark.


Higher is Better


HD Tune measures disk performance to make comparisons between drives or disk controllers.


Higher is Better



Lower is Better


We are now benchmarking with Cinebench 10, where values are marked higher than Cinebench 9, due to changes in how it reads the multicore CPUs.


Now that the system benchmarks are complete, we will move on to the video benchmarking portion of the review. We will be using an assortment of games to test performance across different processors to look for any performance advantages. The testing will start with resolutions at 800x600 and progressing to 1280x1024 to try and take the video card out of the equation and show the performance at resolutions where processor performance can influence the results.

The game tests that we use are as follows:

First up we have Far Cry. This game makes extensive use of pixel shaders and features Polybump Normal Mapping technology to increase character details.

We will be using the Hardware OC Benchmarking Utility version 1.8 with the following settings.

In older games such as Far Cry, lower resolutions are where the CPU is a factor as you can see at the resolution of 800 x 600 the overclocked frame rates almost peaked 400 FPS. With the release of newer DirectX 10 games which are made for multi-core processors we should start to see higher frame rates in higher resolutions with the CPU giving the video card a boost. In the past as you crept into the higher resolutions the CPU was taken out of the equation leaving most of the work to the video card.


F.E.A.R. is a newer game that includes its own benchmarking utility. We will be using this test to benchmark the game. This game introduces a new AI model that emulates real squad behavior. It has the ability to counteract the moves you make rather than having a predictable routine.




The settings we will use are below:







Quake 4 is next up for testing. We will be using the Hardware OC Quake 4 Benchmark Utility version 1.5 to complete the testing with this game. You will need to update to the most current version for the latest time demo and bug fixes. Average FPS (frames per second) will be the measure used.


The settings we will use are listed below:







Call of Duty 2 is a WWII first-person shooter game that is dated, but still maintains a tremendous online following. This test will consist of a timed run on the Stalingrad multi-player map, measured by Average FPS (frames per second).




The settings used are listed below:







Benchmark: BioShock

BioShock is an interesting new first person shooter that has recently hit the market. This title poses quite a challenge for video cards.


After looking at the last four benchmarks you will notice that even overclocked at 1024x768 when the CPU becomes less of a factor, the FPS are virtually the same.


Need For Speed: Most Wanted. For this test, we will time each race and record the average FPS (Frames Per Second) achieved.


The settings we will use are listed below.









3DMark06 is one of the benchmarks that always comes up when a bragging contest begins. 3DMark06 presents a severe test for many of today's hardware components. Let's see how this set of processors perform. We will only be comparing the E6700, E6750 and AMD 6000+ for this series of benchmarks. The settings used are listed below.








Because this review is all about the processor, we will be showing how the CPU scored in each of the resolutions we tested.



Here is a list of the technologies built into the Intel Core 2 Extreme quad-core QX9650 and some interesting facts you may want to know.





I just might not ever use another DFI board again. After spending days overclocking testing for stability and making sure I had the perfect settings my DFI Lanparty UT P35-T2R decided it wasn’t going to work at all. I began getting continuous reboot cycles and when I finally would boot was constantly get bios errors or have the board not recognize my mouse or keyboard. (PS2 or USB) Not only that the constant boot cycles at times more than 30 in a row corrupted my hard drive. Is it a Bios problem? Is there an incompatibility with the Yorkfield chip? Right now your guess is as good as mine, I even tried my Q6600 again and still no dice. By the way I’m still waiting to hear back from DFI tech support after leaving about 10 messages on their “No one is available, well call you back if you leave your name and number” machine.

Luckily I was able to get a decent overclock before all this happened, that was 3.825 Ghz, FSB at 425 x 9 and DDR2 9200 Ram timings of 5-5-4-12 with 333/1066 for the divider. Other settings including voltages were CPU at 1.3500V, NB 1.37V, RAM 2.23V, Thermal management; disabled and C1; disabled.

By now you have read the results with this overclock, and can see that even with an overclock of just 800MHZ the performance increases tremendously not just in one benchmark application but all.

I would have hoped to at least have the screenshot of the overclock with CPU-Z but even on another system my data in not accessible. So when I get back the DFI motherboards from RMA or get another board that supports the Yorkfield processor, I will go ahead and update this page with not only a screenshot but possibly an even higher overclock. Thank you for your understanding.


After completing all the benchmarks in the suite, I can only arrive to one conclusion, the Intel Core 2 quad-core QX9650 performs. To date it has produced the highest scores out of any processor I have used or reviewed. I like the fact that even overclocked and on air cooling, my CPU temperatures did not go above 49 degrees Celsius at 100% load and that’s even with ambient room temperatures of 78 to 80 Fahrenheit or 23 to 26 degrees Celsius!

As I mentioned earlier, at this time most gaming benchmarks used do not utilize the full potential of a multi-core processor, as OCC tests them further, we will start bringing in new gaming benchmarks that do utilize processors to their fullest potential and start testing in higher resolutions. Luckily new games are slowly being released and we may finally see how well our expensive video cards stand up also. Since I know it will be asked, yes I did game in higher resolutions, and found that games ran smoother and even appeared to be more realistic than other processors. Grass and other shadings were more clear than I have ever noticed with an Intel processor.

Ok getting back on topic, programs like Photoshop, Microsoft Word, opened easily usually in half the time I’m use to, but my major concern with these programs is when I either have multiple pictures in high resolutions open or I’m saving a large document in word. Normally no matter how much RAM I have in a system I always get lag out of Photoshop just performing easy tasks like auto level, or when saving my word document and being able to go get a drink out of the fridge before it’s done. Ok I might save bigger files than most and have way too many high resolution pictures open but it is what I do and the QX9650 performed way better than my expectations, I didn’t have to wait at all for the pictures to auto color or level nor could I get up and get a drink while saving a word file.

The only drawback of this processor may be its price. It might not be affordable to everyone at this time, but along with the newest technology comes a higher price and depending on where your priorities are, it can be a wise investment... You may not have to upgrade for a couple of years.