Daniel,
So, John, what you're saying, I take it, is that the 820 is a well engineered and successful product. Whatever you say, John. A credit to Intel and Rambus both, I'm sure.
I'm glad you agree:-)
Seriously, I would like to know what the exact problem was that delayed the shipping of the 820 systems. Nobody seems to know (who is willing to say).
I have not heard of any problems with the 820 chipset when used in conjunction with RDRAM. Have you?
The MTH is a horrible fallback device that I am sure Intel never really wanted to use. However, one of the articles you quoted questioned whether having it present would slow down Rambus memory to the same speed as PC100. That seems very unlikely and is, I am sure, deliberate "casting of doubt" with no basis in measurement.
Unfortunately, so much rubbish has been written that any real issues are hopelessly clouded. Perhaps that was the intention.
The 820 chipset (and the 840) were designed for Rambus memory. When used with Rambus memory, do you know of any verifiable instances of "flakey" operation?
Incorrect voltage on the power supply to the AGP socket does not count. After all, I would hardly blame the chipset for the fact that I have a Supermicro motherboard which delivers the wrong voltage to the processor. (I am quite happy to name names here as I still have the motherboard in question and can readily prove the matter if necessary.)
John |
