Yes, I know that I am late, but my last two days were filled with meetings and lots of other things, so that I was unable to finish my Monday Blurb on time this week. I hope that this will never happen again.

Times are getting pretty interesting currently. In the next few weeks you will get the chance to find out about Intel's upcoming Pentium III as well as AMD's new K6-3 and at the same time there will be first reviews of NVIDIA's TNT2 and 3Dfx's Voodoo3 as well. Not all of those products are equally exciting, but each of them will raise the limits of PC systems considerably.

Intel's Pentium III Under Pressure

Intel's Pentium III seems to be the most troubled product of those four. The ID No.-issue is still a very hot topic and it could get Intel into serious trouble. Intel's first retreat in this matter, agreeing to leave this ID-No. turned off by default is not satisfying enough for many people. The groups that demand that all Intel CPUs shall come without any ID-No. altogether could possibly still win. This would mean a midsize disaster to Intel indeed, because they would have to recall all Pentium III CPUs that have been produced yet and replace them with Pentium III processors that come without an ID-Number. Whatever the outcome may be, the ID-No. has taken away all focus from the new and shiny 'streaming SIMD instructions' and Intel will have a much harder time marketing this most important new feature of Pentium III.

Who Will Be The First?

There's a lot of chaos going on within Intel anyway. The Pentium III launch was pushed forward to screw up AMD's K-3-launch, but now AMD has pushed up the K6-3-release also. For everyone that's really interested, my sources say that both CPUs will be launched sometimes in the last week of February, but who knows if the two CPU-makers won't change their minds again. Intel's policy in regards to Celeron seems to be so unclear, that even Intel spokesmen don't really know what's going on. First there was recently the rumor of the Celeron 300A being discontinued, not least because it's beautiful overclockability to 450 MHz. Then Intel announced that there won't be any faster Slot1 Celeron than the 366 MHz-version. Now Intel says it's all wrong, all Celeron processors will be available for Slot1 as well as Socket370-systems until the end of 1999. At the same time there seems to be a shortage of Socket370-platforms, which could threaten Intel's Celeron sales in case that the Slot1-Celeron should indeed be discontinued.

7.5 Is Less Than 19 And More Than 14? I Guess Not!

The next thing was Intel's argumentation against my article about their new mobile Pentium II CPUs . I said in this article that Intel is keeping the latest technology of 256 kB on-die L2-cache deliberately from the desktop market, because they can make enough money with the older technology using external L2-cache. Intel told me that after all the die of the new mobile Pentium II processor (code name 'Dixon') represents 19 million transistors, which makes it much more expensive to produce than a normal Pentium II core (7.5 million transistors) plus the external (and pretty cheap) L2-cache modules supplied from RAM makers. Intel claims that Dixon for the desktop would be much too expensive. What they forgot however is that Celeron with its die hosting around 14 million transistors, due to its 128 kB on-die L2-cache, is for some strange reason selling at half the price of a Pentium II. I would be seriously surprised if a Celeron wasn't at least as expensive to produce as a complete Pentium II, the Pentium II might even be cheaper.