System Benchmarks

When running the 32M benchmark of SPi, we are calculating Pi to 32 million digits and timing the process. Obviously more CPU power helps in this intense calculation, but the memory sub-system also plays an important role, as does the operating system. SPi 32M has been a favorite amongst benchmarks for these very reasons and is admittedly the favorite benchmark of this reviewer.

The reason we still like to show SuperPi performance is that it is a favorite benchmark of the enthusiasts still and has a cult following, but it also shows the strength of the memory subsystem. As already explained, however, our systems don't really have a lot of difference in that memory world with the FSB and memory frequencies being very similar...yet we still see a huge difference in performance. This is the kick off to the benchmarks that show, when all things are equal, CPU frequency still wins every fight.

The latest iteration of the popular system benchmark is PCMark Vantage from the Futuremark crew. The PCMark series has always been a great way to either test specific areas of a system or to get a general over view of how your system is performing. For our results, we simply run the basic benchmark suite which involves a wide range of tests on all of the sub-systems of the computer.

As we just witnessed in SPi 32M, PCMark Vantage running the default test, is nights and days ahead of the default XMP settings. Higher CPU frequency helps with all tasks that PCMark tests for so naturally a substantial increase in performance is there,; over 25% to be exact. With CPU clocks just over %30 higher and performance just over %25 higher, the casual factor for the performance increase is obvious.

Another benchmarking community favorite, Cinebench renders an intense 2D scene relying on all the processing power it can. Cinebench R10 is another 64-bit capable application and is likely the most efficient program tested today at utilizing all cores of a processor. We will be running both the single threaded and multi-threaded benches here today.

Speaking of CPU power, Cinebench is the favorite benchmark for overclockers to measure and compare CPU processing power with. Naturally that makes it the obvious choice for us and this comparison considering the premier system difference between our XMP settings and overclocked settings is CPU frequency. Like the PCMark result, the Cinebench numbers equate to a %30 increase in performance in the single CPU test and a solid %23 gain in the multi-threaded benchmark.

Moving from some of the more 'synthetic' benchmarks, we find ourselves smack dab in the middle of a VOB to DivX encoding task. We will take a VOB rip of the movie Rounders, and convert it into DivX using the default 720P setting of DivX converter v6. This is a real life test of a task that may be routinely seen these days.

As mentioned, we are now entering the 'real world benchmarks' section of today’s program. We start off with the time consuming task of converting a ripped DVD to a useable format on the computer. The home multi-media center is no longer a shelf of DVDs, it is centered around a music server and a huge chunk of digital data and all that data must come from the DVDs you purchase and converted into useable data. The above difference is about an 18% decreases in time required to encode this movie to 720P with the DivX CODEC. That is a savings of over 11 minutes off a 64 minute task...you do the math on your entire DVD collection.

Un-like the DivX conversion we just looked at, Lame Front End is not multi-threaded and only utilizes a single core of a processor. This will obviously limit performance but we should still recognize significant time savings going from the stock settings to the overclocked results. We will be encoding a WAV rip of the Blackalicious album, Blazing Arrow and converting it to MP3 using the VBR 0 quality preset.

As with movies, music too is now primarily stored in a digital medium yet the old business model of selling CDs remains, which results in time required for the end user to convert the data into a digital format that is useable and efficient. Unfortunately the current and long time best encoder of audio is LAME and LAME only uses a single thread. So needless to say, LameFrontEnd, is only going to be running a single thread for the conversion from WAV to MP3. Again, another solid 20% decrease in time needed to complete the task is found; %23 to be exact going from the default XMP profile to the highest clocks we were able to achieve. All of this performance increase is due mainly to the increase in CPU frequency and most noticeable in the real world benchmarks.

Adobe Photoshop CS3 is full x64 compliant and ready and able to use every single CPU cycle our processor has available. Since digital photography is as popular as roller skates were in the 70's, we are going to be timing how long it takes to convert 100 RAW images from a Canon 20D into half size JPG files of maximum quality.

The last of our 'daily digital conversion benchmark' session is the Photoshop conversion of large RAW images to minable JPG files for viewing on the computer or internet. We see only an 18% increase in performance here and this is likely due to a bottleneck in the system that doesn't fully let the CPU flex its processing power. Perhaps memory or hard drive BUSes are being saturated during this task and thus providing less enthusiastic results as the other benchmarks we just looked at.