Overclocking the Intel Core 2 Extreme X9000

by Charles P. Jefferies

Today we are taking a closer look at the new Intel Core 2 Extreme X9000 “Penryn” processor. Last week we witnessed its stock performance in the Sager NP5793 review, and this week we are going to kick it up a notch by overclocking the processor. The X9000 is the fastest dual-core processor available for notebook computers, and has a price to match. Unlike most notebook processors, the mobile Intel Extreme processors can be overclocked from the BIOS (if the notebook manufacturer allows it). In this review, I will take the X9000 from its stock 2.8GHz clock to an incredibly fast 3.2GHz. The last time we saw a notebook processor around 3GHz was during the Pentium IV dark ages.

How does overclocking work?

Before I start, allow me to explain a few technical terms. A multiplier is a ratio of the CPU’s clockspeed (ex: 2.0GHz) to the speed of the front-side bus (FSB). A CPU with a multiplier of 14x will do 14 clock cycles for every cycle of the FSB. The more clock cycles the CPU does per cycle of the FSB, the faster the clockspeed is and the more the CPU can get done in the same amount of time. The FSB is, in basic terms, the data highway between the processor and the memory in an Intel system. Intel posts their numbers in QDR (Quad Data Rate) format – for example, the 800MHz FSB our X9000 actually has a 200MHz clock, but its effective clock is 800MHz. Let’s take a look at the math behind overclocking.

CPU-Z Screenshot:

The X9000 has a 200MHz FSB (800MHz QDR) and a 14x multiplier out of the factory.

14 * 200MHz = 2800MHz clockspeed

Since the Intel Core 2 Extreme has an unlocked multiplier, it can be changed to something higher. Sager allows for the multiplier to be changed to a maximum of 16x. By doing this, the processor runs at a higher speed.

16 * 200MHz = 3200MHz clockspeed

Pretty neat! By changing a simple option in the BIOS, I am able to overclock the processor by 12.5%, which should equate to 12.5% more processing power. How will this increase in clockspeed affect benchmarks? That is what we are going to find out.

The Processor

I promise I will get to the benchmarks right after this section. I need to point out a few of the features and improvements that differentiate of the X9000 from standard Core 2 Duos. For starters, this is a Penryn chip, meaning it is built on a new 45 nanometer process; previous Merom Core 2 Duos were built on a 65 nanometer process. Because they are built on a smaller process, Penryns are able to process more per clock than the older Meroms, all while achieving a higher clockspeed, using less power, and producing less heat. The X9000 is separated from other Penryns not only because of its higher stock clockspeed (2.8GHz versus 2.6GHz for the fastest standard Penryn, the T9500) and unlocked multiplier that I detailed above, but also because of its higher heat dissipation. The TDP of the Core 2 Extreme X9000 is 44W while the standard Penryn’s TDP is 35W. The Extremes can only be used in notebooks that have a cooling system to handle it, and the one in the Sager NP5793 has been designed to do exactly that.

Benchmarking

I benchmarked the Sager NP5793’s processor at 2.8GHz and overclocked to 3.2GHz under identical conditions. The 2.8GHz performance benchmark screenshots will be placed on the left and the overclocked 3.2GHz on the right. For the graphs, the 2.8GHz clock is used as the baseline at 100%, and the 3.2GHz overclock is given as a percentage of it.

Our test unit is configured as follows:

• Screen: 17-inch WUXGA (1920×1200) glossy widescreen
• Processor: Intel Core 2 Extreme X9000 (2.8GHz/6MB L2 cache/800MHz FSB) – 45nm Penryn
• Video card: Nvidia GeForce 8800M-GTX 512MB
• RAM: 4GB DDR2-667
• Hard Drive: 160GB 7200RPM Seagate Momentus 7200.2 SATA
• Optical Drive: 8X DVD Burner
• Operating System: Windows Vista Home Premium 64-bit
• Wireless: Intel Wireless WiFi Link 4956AGN and Internal Bluetooth
• Battery: 8-cell Li-ion

System Performance Benchmarks

Overall System Performance using PCMark05

Results at 2.8GHz. (view large image)

Results overclocked to 3.2GHz. (view large image)

PCMark05 sees a 7% overall improvement, which is not going to be noticeable in everyday activities but could provide some extra muscle while multitasking heavily.

Processor Performance Benchmarks

Windows Experience Index CPU Test

Results at 2.8GHz. (view large image)

Results at 3.2GHz. (view large image)

The overclocked processor barely adds 2% to the performance here, but I refrain from placing too much emphasis on this test as it is quite basic.

wPrime

wPrime is a multi-threaded CPU test – it is similar to SuperPi but has a few more features. It is a much more accurate benchmark for dual-core CPUs. Please see our big comparison thread here.

Results at 2.8GHz. (view large image)

Results at 3.2GHz. (view large image)

The overclocked processor really shines in the wPrime multithreaded benchmark, where we see gains from 14% -17%. This benchmark is entirely CPU-focused.

Super Pi

Results at 2.8GHz.

Results at 3.2GHz.

Like wPrime, the CPU-only Super Pi benchmark sees a great improvement from the overclocking. Mathematical calculation programs will see a nice benefit from the overclock based on what we have seen thus far in the CPU-only benchmarks. Note that Super Pi is only single threaded so it only utilizes one of the cores.

SiSandra Arithmetic

Results at 2.8GHz. (view large image)

Results at 3.2GHz. (view large image)

SiSandra Multimedia

Results at 2.8GHz. (view large image)

Results at 3.2GHz. (view large image)

We see nice gains across the board in SiSandra, with gains ranging from 14% – 21%. Most impressive, considering the CPU was only overclocked 12.5%.

Rendering Performance using Cinebench 10

Cinebench Release 10 is the latest version of Maxon’s rendering benchmark, based on the Maxon CINEMA 4D animation software. It is designed to measure the performance of the processor and graphics card under real world circumstances. More information can be found at www.maxon.net/pages/download/cinebench_e.html

Results at 2.8GHz. (view large image)

Results at 3.2GHz. (view large image)

Cinebench shows the significant performance boost given by the extra clockspeed of the overclocked CPU. The gains are almost the same as the CPU overclock percentage (12.5%) in every section of the benchmark, showing that every megahertz counts.

Gaming Benchmark

Synthetic Gaming Performance using 3DMark06

Results at 2.8GHz. (view large image)

Results at 3.2GHz. (view large image)

While the increase in 3DMark06’s CPU score parallels the percentage overclock of the CPU we are seeing in other CPU tests, the overall score which includes the actual 3Dbenchmarks done by the graphics card does show that the benchmark is limited by the graphics card and not the CPU.

I have omitted the actual real-life game benchmarks from this article because in my tests, I found less than a 1% difference between the framerates produced on the 2.8GHz and 3.2GHz setup, which could be attributed to error. This indicates that the CPU is not the bottleneck in games.

Overall CPU Benchmark Performance Increases

Averaging all of the CPU-focused benchmarks, we can see the 3.2GHz processors averages a 13% performance increase over the 2.8GHz, which is slightly more than the percentage the CPU was overclocked (12.5%). This is exactly what should be expected of a CPU overclock for CPU-dependent tests (generally speaking) – eventually, overclocking the CPU will lead to less and less gains for every megahertz added, but that is only when the RAM and other components come into play, and we cannot reach those clockspeeds in a notebook at current. Here, the RAM and FSB are not bottlenecking the processor and we get a linear performance increase based on the increase in clockspeed. I could not ask for better results from this machine.

Conclusion

The Intel Extreme Editions are coveted chips by computer enthusiasts chiefly because most cannot afford to put chip priced at close to $1,000 in their computer. The Extremes come with bragging rights but more importantly an unlocked multiplier, which allows for painless overclocking to an extent. Thanks to Sager providing the BIOS options to overclock the CPU, the X9000 can be put to work and the extra megahertz really pay off for CPU-dependent benchmarks. Note that Sager is one of the few companies to offer a notebook with an overclockable processor. Some manufacturers offer an Intel Core 2 Extreme processor yet no option to overclock it, which defeats the purpose of paying the extra money for the chip in the first place.

So, who would benefit from having an Intel Core 2 Extreme processor? I have observed that most people believe a faster processor will help improve gaming performance, however in this day and age of dual-core processors it is rarely the bottleneck for games. Gaming is not helped by the overclocked CPU in general because the video card is the bottleneck, not the CPU. For those primarily using their computer for gaming, the Extreme Edition is simply not worth the extra cash – even paired with the beastly 8800M-GTX 512MB video card, the GPU is still the bottleneck. However, for demanding users that perform CPU-intensive applications on a daily basis such as math simulations, encoding, and rendering, the Extreme Edition warrant the extra money because it can cut down the time spent running those applications significantly. Note that the fastest mobile Core 2 Duo processor next to the Extreme X9000 is the T9500, which runs at a slower 2.6GHz. Given the up to 21% performance improvement going from 2.8GHz to 3.2GHz, imagine how much of an increase we would see going to 3.2GHz from 2.6GHz.

Before I close, I would like to tip my hat to the Sager NP5793 – it held up wonderfully throughout all of the benchmarks, and never showed signs of instability while overclocked. Heat did not increase and the machine stayed incredibly cool. It is a standout machine in the high-end notebook market and should be high on the list of anyone looking for the ultimate in mobile power.