Hi Charles R; Pretty stunning for a pro Intel review, they note the surprisingly bad results in a lot of the benchmarks, for instance:
Strangely, we found that our overclocked 440BX/SDRAM/133 system managed to best the i820/RDRAM/133 system by 2 to 4% in these benchmark results. Much like the earlier Q3Test results, we have no explanation for these disparities in the face of logic, other than the possibility that the test's control factors being corrupt from a software standpoint or the possibility that in certain areas certain platforms are slightly superior to the i820 (and vice-versa).
Both of the 133MHz systems outperformed the standard 440BX/SDRAM/100 PC by wide margins, achieving an 8 - 12% faster output environment. Even more strangely, the 440BX/SDRAM/100 PC scored highest in our 3D Mark99 MAX scores.
But they still manage to give it a score of 8 out of 10. :)
My personal favorite is:
We were surprised that ZD's Winstone99 benchmark, which heavily tests each of the components in a PC towards their abilities to work together, didn't return a better score with the i820 than the older 440BX/SDRAM/100 PC.
When questioned about this disparity, Intel representatives stated that the Winstone99 test suite, which was designed in the late summer of 1998, does not offer enough of a stress on the i820 to press it into the area where it would produce a statistical advantage versus their previous platform.
The above is so ridiculous. First of all, it can't be correct, if he is trying to imply that the 820 is faster. What he should have said would be does not offer enough of a stress on the 440BX/SDRAM/100. From my career, (admittedly long, long, ago) of evaluating mini supercomputers for Burroughs (yes, Bilow has been around), I find this concept laughable.
No engineer could have put out that garbage. This is total marketing hype, and not even very well done. Even the Intel supplied benchmarks aren't very impressive.
It looks to me like Intel is going to have to take a dive on this one. I've never seen a product from them with this many disadvantages. It's like they all own stock in AMD or something.
Can you add onto this 820 problem list?
(1) Cost of Memory
(2) Latency of Memory
(3) Availability of Memory
(4) Reliability of Memory (and therefore decent launch)
(5) Performance of System
(6) Overheating of Memory
It sure looks to me like INTC is going to give up massive market share, now at the high end, to AMD, and maybe others.
-- Carl |
