facebook rss twitter

Review: VIA P4PA-UL P4X266A Motherboard

by Tarinder Sandhu on 10 May 2002, 00:00

Tags: Intel (NASDAQ:INTC), VIA Technologies (TPE:2388)

Quick Link: HEXUS.net/qalf

Add to My Vault: x

Benchmarks

We'll start off as always with Sisoft Sandra's memory benchmark. The Pentium 4 loves bandwidth, so the solution best able to provide it usually excels in practical benchmarks. We'll first have a look at the buffered results.

Even though the memory controller has been enhanced from the transition between the P4X266 and P4X266A chipset, the scores posted by the P4PA are the lowest of our test group. The differences aren't that great, however.

Buffered results only reflect possible memory efficiency when using streaming sources. We know all data isn't like this so we decided to have a look at unbuffered scores by simply switching off the buffering option. This should be reflective of more normal usage.

Here we see the P4PA pull ahead of the I845D by a discernable margin. The SiS 645, in the form of the ABIT SD7-533, running with PC2700 memory, is still the fastest of the DDR chipsets. The RAMBUS equipped I850 continues to lead in this synthetic benchmark. We can attribute the P4PA's relative success to the enhanced memory optimisations present.

Let's now see how the P4PA copes with our first practical benchmark, Pifast. Pifast simply calculates the constant Pi to X million places. It relies on memory throughput as well as sheer CPU power. We've set it to calculate to 10 million places using the fastest method.

Here we see the P4PA beat out the I845D once more, an impressive feat considering the relative immaturity of the P4X266A chipset. As expected, the PC2700 equipped SiS 645 and RAMBUS powered I850 continue to set the pace. The differences, however, are nothing to be worried about.

Next we'll turn our attention to MP3 encoding. We're benchmarking by encoding a 481MB custom WAV file into 128kb/s MP3, using the LAME 3.91 encoder and Razor-Lame front-end.

A tie across the board, corroborating our long-held belief that MP3 encoding revels in pure CPU speed. Different memory subsystems seem to be largely irrelevant with respect to performance.