Someone recently asked on the Real World Technologies forum if he should buy an MSI K7T Pro with the 686A Southbridge or the newer K7T Pro 2A and 686B with ATA/100 support. His question was brought about by a recent article on Tom’s Hardware Guide titled ‘Late Vintage: High Degree of Maturity of the KT133’, and specifically this page: ATA/66 vs. ATA/100 – Errors in the New VIA-Southbridge?. Test results had reportedly indicated that the new 686B Southbridge was having a negative affect on graphic performance.
Now to me that just didn’t make sense, but you never know, so I put three VIA KT133 series motherboards to the test. One uses the 686A SB, one uses the 686B SB and one uses the new KT133A with 686B SB.
The following benchmarks were run under Windows 98 SE with the latest chipset drivers installed, 128MB of memory, an ATI Radeon DDR 32MB AGP video card and a WD 200BB ATA/100 7200 RPM HD. Windows was set at 1024×768/64K @ 85Hz. All test were run 3 times and were rejected if not within 3%, best of the 3 was used. Motherboard’s models were chosen because they all come from the same manufacture, BIOS have the same options and all BIOS’s are from the same ‘vintage’. I chose a different video card than used in the test on Tom’s Hardware guide, just in case that was an issue. Being a Pilot run motherboard, the K7VTA Pro may have some room for improvement when a production BIOS becomes available.
Mainboard |
Soyo K7VTA |
Soyo K7VTA-B |
Soyo K7VTA Pro |
CPU |
Athlon 1GHz |
Athlon 1GHz |
Athlon 1GHz |
Chipset |
VIA KT133 / 686A |
VIA KT133 / 686B |
VIA KT133A / 686B |
Memory |
PC133 CAS 2 |
PC133 CAS 2 |
PC133 CAS 2 |
Status of Mainboard |
Production |
Production |
Pilot |
| | | |
Business Winstone 2001 |
41.6 |
42.0 |
40.9 |
Content Creation Winstone 2001 |
39.3 |
39.7 |
40.4 |
| | | |
3D Winbench 2000 | | | |
Score |
95.8 |
95.9 |
96.0 |
CPU |
2.08 |
2.07 |
2.09 |
| | | |
Winbench 99 | | | |
CPU |
91.2 |
91.2 |
91.2 |
FPU |
5460 |
5460 |
5460 |
Business Disk |
6220 |
7640 |
6390 |
High Disk |
18200 |
18400 |
17100 |
Business Graph |
433 |
440 |
424 |
High Graph |
1240 |
1250 |
1240 |
| | | |
Quake III Demo 1 |
126.8 |
127.1 |
127.3 |
| | | |
3D Mark 2000 | | | |
Score |
4760 |
4757 |
4758 |
CPU |
442 |
444 |
433 |
Game 1 |
96.8/65.8/29.6 |
96.7/65.7/29.6 |
96.7/65.6/29.6 |
Game 2 |
94.9/67.4/42.2 |
94.9/67.4/42.2 |
95.0/67.4/42.2 |
Wizmark 3.0 (lower better) |
6.46 |
6.30 |
6.21 |
Tirtanium 1.90 3D Now |
60.2 |
60.2 |
60.1 |
Now, I didn’t run the same test, use the same video card, or the same motherboards as Tom’s Hardware Guide, but I just don’t see any difference at all between the two Southbridge chips.
The difference reported on Tom’s Hardware Guide was very minimal (6~8%). If I’d been doing the test I would not have even given it a second thought, other than to suspect a difference in the BIOS, motherboard, or even a test variance, if not all run at the same time. Based on my results that show no advantage in using the 686A Southbridge, I would suspect there are not any graphics issues with the new VIA 686B Southbridge chip.
Just goes to show how hard it really is to test a product, understand the results and how further testing is sometimes needed before making a conclusion.
Be the first to discuss this article!