<!–#set var="article_header" value="VIA Apollo KT266 Revisited –
Much Ado About Nothing” –>
Introduction
Sometimes the truth hurts and all wishes for the better turn out to be futile. That’s what VIA had to experience last week when the first respectable online reviews of VIA’s Apollo KT266 hit the web, showing lackluster performance compared to AMD760 chipset. Both chipsets support AMD’s SocketA Athlon processor and DDR-SDRAM, and therefore a combination that is expected to yield excellent performance. Unfortunately even AMD’s 760 chipset was never able to show a speed advance over PC133-systems (typically equipped with VIA’s Apollo KT133A chipset) of more than 10%. VIA’s rather late DDR-supporter Apollo KT266 could hardly reach 5% more speed than its PC133-brother KT133A, which could only be seen as a major disappointment. DDR is still a buzzword, but more due to marketing hype than due to actual major improvements over the cheaper and reliable performing PC133 SDRAM.
My demand is simple. I want to see at least as much of a performance benefit from DDR-SDRAM as what AMD760 is just about able to provide. A DDR-chipset that cannot even provide 10% performance improvement over PC133 has simply failed the class goal. ALi’s MaGiK1 chipset has already disappointed the majority of us and now VIA’s Apollo KT266 had failed to beat or at least live up to AMD760 just as much.
A New BIOS – The Magical Cure?
My judgment over Apollo KT266 was clear. Its performance was not good enough to justify its very existence. This was also not changed by the fact that its price is actually lower than AMD’s counterpart, because if anyone really wants to save money, he should go for the even cheaper Apollo KT133A chipset plus PC133 memory and do completely without DDR. A ‘Team DDR’-sticker on your PC doesn’t make it better, faster or more impressive. The performance has to live up to the expectations.
After the posting of our first review last week there was a minor VIA-friendly website that did not want to believe the truth, foolhardy attacking us as well as Anand, who had come to a similar conclusion. A new BIOS was supposed to change everything and KT266 would all over a sudden perform really well. We were finally equipped with all the tools to flash this new BIOS onto our MSI test board and since we take our job rather seriously, we re-tested our KT266 platform with this new BIOS to see if VIA’s wishes would finally come true.
Test Setup – Updated
We obviously used the same setup as in the initial article and simply added the new results to the benchmark scores we had tested initially. Our platform was still MSI’s K7T266 Pro motherboard, the BIOS used was “a6380vsp.10a”. Our board was from the very start equipped with the mysterious 1kΩ R126 resistor and not with the ‘R127’ resistor.
There had also been discussions in our forum about the fact that I had used different memory in the AMD760 and KT266 motherboards last week. Let me first commend those readers for their vigilance. Of course this is an unusual situation. However, there is a very good explanation for it. The AMD760 motherboard MSI K7 Master-S runs best with the Micron PC2100 DDR-memory, while this memory does not provide the best results in MSI’s KT266 motherboard K7T266 Pro. I tried to be as fair as possible to both and therefore used with each board the memory that makes it perform best. I am not able to give you an explanation for the strange behavior. Theoretically, the Infineon DDR-memory has the fastest specs and should therefore run best in both motherboards, but the K7 Master-S with its rather old BIOS seems to prefer the older Micron memory for the time being. I hope that MSI will finally provide a BIOS-upgrade for this board, because it might well be that it could make AMD760 look even better.
System | VIA Apollo KT266 | AMD 760 | VIA Apollo KT133A |
Processor | Athlon ‘C’ 1.33 GHz, overclocked to 1.466 GHz |
||
Motherboard | MSI MS-6380 “K7T266 Pro”, rev. 1.0 Pre-Release BIOS |
MSI MS-6341 “K7 Master-S”, rev. 0D Pre-Release BIOS |
Asus A7V133, rev. 1.01 BIOS 1004 |
Memory | 256 MB Infineon PC2100 DDR-SDRAM, CL2, Setting CL 2, 4-way Interleave | 256 MB Micron/Crucial PC2100 DDR-SDRAM, CL2, Setting 8-8-5-2-2-2-2 | 256 MB Micron/Crucial PC133 SDRAM, CL2, Setting 2-2-2 |
Graphics Card | NVIDIA GeForce 2 Ultra Reference Card, 64 MB, Driver 6.67 (Win98/Win2k) |
||
Hard Drive | IBM DTLA-307075, 75 GB, 7200 RPM, ATA100, FAT32 Win98, NTFS Win2k |
||
DirectX | 8.0a |
||
Desktop Resolution for BAPCo’s Sysmark 2000 and Webmark2001 | 1024x768x16x85 |
||
Quake 3 Arena | Retail Version no sound |
||
Unreal Tournament | Version 4.28 (patched) no sound |
||
Evolva | Rolling Demo v1.2 Build 944 Standard command line = -benchmark Bump Mapped command line = -benchmark -dotbump |
||
3D Mark 2000 | Build 335, Default Benchmark |
||
SiSoft Sandra Standard | Version 2001 |
||
Desktop Resolution for SPECviewperf 6.1.2 | 1280x1024x32x85 |
||
FlasK Settings | Video Codec: DivX 3.11 alpha, Fast-Motion, keyframe every 10 seconds, compression 100, data rate 910 kbps Audio Codec: audio not processed Video Resolution: 720×480, 29.97 fps, interlaced Resizing: Nearest Neighbor |
Memory Performance
You can see that the new BIOS was indeed able to improve the memory performance of KT266. However, the VIA DDR-chipset is still lagging behind AMD760. You can imagine what this means for the rest of the benchmarks.
Office Application Performance
The performance has been improved, but the class goal has still not been reached. AMD760 remains the winner.
3D Games
I won’t comment on each result anymore, because the tendency is obvious.
MPEG4 Encoding
FlasK MPEG is very CPU and memory dependent and so you can see another example for the improved performance of KT266 due to the new BIOS. However, AMD760 remains untouched.
SPECviewperf 6.1.2
Finally in SPECviewperf 6.2.1 the situation remains the same. The new BIOS is indeed able to improve KT266’s performance, but AMD760 is still the winner.
Final Conclusion
We took the criticism seriously and made the effort to retest VIA’s Apollo KT266 chipset with the latest BIOS. However, the situation remains the same. AMD760 is the only viable solution for Athlon plus DDR-memory. Apollo KT266 and DDR-SDRAM are simply not providing enough of a performance advantage over KT133A plus PC133 SDRAM to justify its additional cost. The bottom line remains the same. If you want Athlon plus DDR go AMD760!
I’d also like to mention a few other important issues. We are actually testing our DDR-platforms with PC2100 DDR-memory that has a CAS-latency of only 2 clocks. The majority of DDR-memory modules available come with a CAS-latency of 2.5, which makes those DDR-platforms perform even worse than what you can see reflected in our benchmark results. You can imagine that with CL 2.5 DDR-memory, the performance advantage of KT266 over KT133A is getting close to zero, if it is not even slower. Keep this important issue in mind when making a choice between KT133A and a DDR-platform. PC133-SDRAM with CL2 is common nowadays, while PC2100 DDR-SDRAM with CL2 is extremely hard to get and also rather expensive.
Please don’t make the mistake to judge badly about DDR-SDRAM only because it doesn’t offer much of a performance improvement in Athlon or Pentium III systems. Those two processors are simply not designed to really benefit from the virtues of DDR-SDRAM. I still claim that Pentium 4 is currently the one processor that would show the biggest improvement with DDR-memory. Unfortunately we are unable to prove that, since Intel would not supply a DDR-platform for Pentium 4 anytime soon. I am sure that Rambus is very relieved to hear that.