I disagree entirely about using an Intel chipset. If they use an Intel chipset, why not force AMD to use an AMD chipset? Ooops, wait, AMD is not competent enough to sell chipsets, so we should penalize Intel because they are? What????
Use the best against the best, especially since the same company uses both. This is a processor comparison, and you're trying to measure processor performance. Since both are available platforms, and from the same company, it's a much better comparison. This is particularly so since a lot of the people buying this processor will know enough to buy the highest performing chipset. Certainly no one would buy an Intel motherboard for performance. Reliability, yes. But did they measure that? Well, no, so stick to best versus best and don't penalize Intel because they make a supporting product AMD doesn't.
The problem with using same cost is that it is very easy to skew results. There are plenty of motherboards out there that cost the same, that have widely varying performance. You can pick the worst versus the best for the competitor, and that will make a big difference in results. Just look at the BadAxe performance. They did a test on this motherboard on another site, and it got totally annihilated by the other motherboards. It was dead last in almost every test, and by a lot. So they then use this motherboard to test it against the Athlon 64. It's so obviously a bias, but no one notices it. Intel motherboards are not about performance, they are rock solid. They relax timings for that reason. So, you don't compare performance with them against AMD motherboards, you use comparable motherboards that are made to perform. It's not about price.
With regards to earlier benchmarks against AMD and Intel, I disagree with you. If Intel has a platform that supports better memory or whatever, than you use it. You test best against best. The reason being, if I am out there looking for a new computer or new parts, I can get these parts. I am not going to buy the slowest parts out there when I can get faster ones, so using slow parts in a comparison is kind of pointless, but I'll be choosing between the best for each platform. If I can't buy an AMD platform with, say, RDRAM then that is a liability to that platform that is relevant. I agree though, you have to mention price in the reviews. If the Intel platform costs $1500 and outperforms the AMD platform by 5%, both because of RDRAM, not too many people will find it to be worth it. So, yes, price is important, but in these high-end comparisons, I think they should show best versus best instead of trying to be cute and showing Intel on a sub-ideal platform.
Why wouldn't you touch Nvidia? Because they are the best performing chipset for Intel processors? Or because that is what was used for AMD? Or, is it because you forced AMD to use an AMD chipset, you should force Intel to use an Intel chipset? Which one?
I have no idea why you think the Woodcrest is the same performance as the Opteron. Everything I have seen shows it blowing it out of the water. The results they showed when compared to a Sun based Opteron represented an ugly pounding. However, as you go to more processor units, not cores, you would expect Intel to perform worse and worse because they still use a shared FSB and have one memory controller. So probably with 16 dual cores (meaning 32 processors), the Opteron would show advantages because of the way memory is accessed. Do not hold me to 16, I am just pulling a number out of my head, it may be less or more, but my point is, as you go higher you should see AMD solutions look better. It's got nothing to do with the architecture, but just with the way they each access memory.
I'm not sure this is a big deal for Intel though, since for big iron they have the Itanium 2. It's not really where they want to position x86.
03/27/06 04:45:51 AM]