System Review – PC-Club Silencer

General Performance and Encoding

We'll start with the WinStones 2004 results. The real world application of these scores is dubious, as most business/office applications tend to spend most of their time waiting on the user. However, we include them as a general indication of system performance. For more detailed benchmarks, we recommend that you look at our full coverage of Pentium M on the Desktop.

In business performance, the Pentium M 755 does well, helped in large part by the fast 2MB L2 cache. Differences between the tested configurations of the Silencer are, as expected, negligible. Using 2-2-2 RAM improves performance by a few percent, but the change in graphics cards and hard drives has little to no effect on performance. One interesting footnote for the Business WinStones scores is that the Pentium M showed a large amount of variance between runs when we used the onboard graphics. 3% or less is acceptable and with a graphics card, the variance was usually 2% or less. The Silencer with IGP enabled had variance of 11%, in addition to running slightly slower than the discrete graphics configurations. Even with such a large (relative) difference between benchmark runs, we doubt that anyone would notice which system was faster in everyday use.

The Content Creation side of things is even closer in terms of the Silencer performance. Not even the IGP configuration really differs that much in performance. Here, however, we see the handicap of the Pentium M architecture. The Athlon 64 is noticeably faster, topping the highest configuration of the Silencer by just over 20%.

For our encoding test, we use AutoGK to encode a chapter from the “Sum of All Fears” DVD to 75% quality without audio, using both the DivX and Xvid codecs. These are one-pass encoding runs, which don’t get the best conversion quality, but they do provide a look at the video encoding performance of a system. As a more realistic re-encoding benchmark, we'll also include the time required to encode the sample video that we used in our Socket 478 and 745 SFF roundup to a target size of 5 MB, this time including stereo audio and with a fixed video width of 640 pixels. There are 900 frames in the video clip, so you could try to calculate an overall FPS rate for the conversion. We prefer to look at the total time rather than just a single pass, however, as AutoGK 1.60 ends up doing three complete passes plus some miscellaneous work to arrive at the final result.

DivX and Xvid are both very demanding of the memory subsystem as well as floating point/SSE performance. The Pentium M architecture with single channel DDR333 memory is simply not able to compete well with the more powerful desktop chips (although it is still probably faster than the Athlon XP). The SN25P with Athlon 64 3800+ beats the Silencer by 25% to 30% in our AutoGK tests – not even a healthy overclock is going to close the gap.