Socket A DDR Chipset Comparison – Revisited

Pages: 1 2 3 4 5 6

Maturity Makes the Difference

I must say that I am impressed with MSI, even if not the KT266 chipset. After my article comparing the AMD 760, ALi MAGiK1 and VIA KT266 DDR SDRAM Socket A chipsets, I received an email direct from a Senior Bios Engineer at MSI in Taiwan. Concern was expressed that my performance results published for the MSI K7T266 Pro were low, due to an ‘old’ BIOS being used, and an offer was made to send me a new BIOS for enhanced performance. I did express some concern with regards to testing a Beta BIOS, but did offer to try it, so they sent a copy of the 1.0A BIOS file. Soon afterward I noticed MSI had a series of pages up on their site that not only discusses DDR memory, but specific performance ‘tuning’ information on the K7T266 Pro. MSI also posted a ‘release’ version of the BIOS (1.0B), along with a note stating that the BIOS was a performance Beta BIOS and caution was to be used if updating to it. MSI has since followed with the 1.0C BIOS (same note as the 1.0B), and now has a new ‘production’ BIOS (1.1) posted for general use available for download.

So what was I impressed with? That they took an active roll in reacting to somewhat negative reviews done on a number of different sites as well as input from end users, something not seen too often in the industry, especially with Taiwan based manufacturers (perhaps a cultural difference?). Providing an unreleased ‘performance’ Beta BIOS to review sites may show the potential performance, but offering a BIOS especially for performance users (or those disappointed with the original BIOS performance) is commendable. For that MSI deserves to be congratulated. And to have an official site with BIOS tuning information makes it even better. Well done MSI.

So what about the VIA KT266 chipset? Well, with the 1.0B or 10.C Beta BIOS’ I did see an improvement in benchmark scores. Not a lot, but it was there. The 1.0A BIOS was a different story. It resulted in some very good scores, but was not stable at all and a number of tests would not complete. Since it was never publicly available from MSI, that isn’t really a negative, however. With the 1.0B or 1.0C BIOS, the benchmark scores improved and brought the scores closer to those of the AMD 760 chipset, and the K7T266 Pro was stable as a rock.

One option added to the BIOS is the ‘High System Performance’ setting. With the 1.0B BIOS I could not complete all the tests in that mode. With the 1.0C I could, but did have a number of recoverable errors. As long as I set the SDRAM 1T option to Disable, that solved any stability issues. For best performance and stability I used the High Performance option in the BIOS, but set SDRAM T1 to Disable and AGP Fast Write to Enable (Disable was default).

Then came the release version of the BIOS (1.1). I still had problems enabling SDRAM T1, so just for the heck of it I tried an SEC DDR SDRAM module. Now I could run with the High Performance BIOS setting, including SDRAM T1 enabled. I guess it was not a fault of the mainboard, BIOS or chipset. In all fairness I was running with the memory set for CAS 2, along with the most aggressive BIOS memory settings, and the Micron module used was CAS 2.5 and an ‘Engineering Sample’, but the SEC was also rated for CAS 2.5. Interestingly the Micron would run at the most aggressive settings in the ALi MAGiK1 based Soyo K7ADA, but not the SEC. This just goes to show you how hard it is to evaluate components sometimes.

I ran a full set of tests on the MSI K7T266 Pro with BIOS’ 1.0B, 1.0C and 1.1. All were pretty close, and when compared with the previous tests run on the Gigabyte GA-7DX (AMD 761) and Soyo K7ADA (ALi MAGiK1), the AMD and VIA chipsets were now very close. In a few tests the VIA now came in first place (it beat the AMD CPUmark 99 score by one point). However, I noticed that both of the other mainboards also had newer BIOS’s available. I had also updated the VIA 4-in1 Driver Pack to the latest Rev. 4.31 and ALi had a new AGP driver (1.74). So in order to have valid test results I re-tested the Gigabyte and Soyo mainboards also. The results were a bit surprising.

Pages:   1 2 3 4 5 6  Next »

Be the first to discuss this article!