Of course they are..
by
XaXXon
·
· Score: 4, Informative
Of course Intel is going to tell computer manufacturers not to use the other guy's chipset. And the reason? Not because it isn't compatible, or isn't as fast as the Intel alternative, but because Intel will sue them if they do. Because Intel can't/won't put out a chipset that people want to use (only offering RDRAM or old and slow PC133), they sure as hell don't want someone else doing it. With how poorly the P4 is doing, however, you'd think Intel would be happy to have something out there to move more chips..
Intel Just Jealous
by
robbyjo
·
· Score: 5, Informative
It's astonishing how an article could spend that long talking about Intel suing VIA over a chipset which introduces a new DRAM type to an Intel CPU and not mention Intel's PC133 fiasco of two years ago.
For those who don't know, the only reason PC133 exists (as a PC standard DRAM type) is because of VIA. Flashback to early 1999: Intel had the market for chipsets (for Intel processors) almost completely to itself, riding on the enormously successful 440BX chipset, which used PC100. However, P3 speeds were ramping up while memory speeds had been stuck at PC100 for a couple years. The obvious thing to do was to update the BX to support a 133MHz FSB. After all, it was a dead-simple engineering trick (every BX mobo at the time could easily overclock to 133; many were stable up to 150), and the memory makers were already making SDRAM which could safely run at 133 but clocking at 100 because that was the highest official speed.
But instead--and unbeknownst to most of the techie world at that point--Intel had a contract with Rambus which offered them many goodies like the ability to make RDRAM controllers royalty-free (others paid up to 5%) and lots and lots of stock options. However, the contract was contingent on, among other things, Intel agreeing to do everything reasonably in their power to prevent "next-generation DRAM" types other than RDRAM from being paired with Intel processors for the consumer desktop. "Next-generation" was defined as > 1GB/s bandwidth.
PC133 has a bandwidth of 1.066 GB/s.
Moreover, Intel *thought* it was putting the finishing touches on the ill-fated RDRAM-only (at that point) i820 (Camino) chipset, with which they were going to introduce new and badly needed 133MHz FSB P3s. Instead, engineering delays involving the difficulties of getting RDRAM working (eventually they had to settle for only 2 RIMM slots instead of the original 3, a per-channel limitation which remains to this day), and the difficulties of getting a memory translator hub which allowed PC100 to be used on the i820 (a last minute addition when they realized people weren't exactly going to pay $500 for 128MB of RAM) working, pushed the release date back 6 months or so, until November.
Just to reiterate: Intel put off releasing 133MHz FSB P3s, and then when they did release them said that consumers could only use them with a buggy chipset, limited to 2 RAM slots, which offered one's choice of an extra-slow translated implementation of PC100 or of RDRAM which cost 10 times as much per bit as SDRAM. Meanwhile, tests with BX chipsets overclocked to 133 MHz FSB showed that this solution was significantly *faster* than the i820 + RDRAM chipset!
Into this world stepped VIA offering the Apollo133 chipset, the first P3 chipset explicitly designed to use PC133. Nevermind that it was probably *less* stable than an Intel BX overclocked to 133 MHZ FSB. Nevermind that it underperformed the BX@133 as well. And nevermind that then, as now, Intel sued VIA with all their might, among other things requesting injunctions forbidding all VIA products from leaving Taiwan. (The pretext then was that VIA was abusing Intel IP by using the P3 bus with a DRAM type Intel had not sanctioned.)
VIA quickly gained > 50% of the P3 chipset market.
Indeed, the only reason you see ALi, SiS, and soon-to-be nvidia and others getting into the 3rd-party chipset market is because VIA paved the way a couple years ago.
Intel tried every FUD tactic in the book, from suing in multiple jurisdictions to claiming that PC133 SDRAM was not stable (the DRAM itself! And this from the company which had spent the past year patching bugs with RDRAM!). Intel got their ass handed to them in court, and by in the summer of 2000 introduced the i815, essentially the BX@133 product they should have introduced in late 1998.
Intel doesn't like getting humiliated, though, and they've had a seemingly personal vandetta against VIA ever since. In retaliation, they denied VIA the chance to license the P4 bus, as ALi and SiS and (interesting) ATi have done. (This is the basis for the current *threatened* suits. However, it's interesting to note that the P4X266 is currently shipping and no suits have yet been filed, meaning this is probably just a bluff on Intel's part.)
Intel reps were even seen at the recent Comdex show threatening mobo makers who had VIA promotional balloons flying at their booths. All the balloons were taken by the Intel people.
However, Intel's case this time is as flimsy as last. Disregarding potential antitrust concerns, the fact remains that NatSemi, whom VIA recently purchased *did* have a license for the P4 bus, and thus so does VIA.
So does this mean VIA will have similar success as last time? Well, I think they'll easily prevail in court if it comes to that, although it appears that Intel may be playing this one all FUD and no bite: warning mobo manufacturers not to use the P4X266 rather than actually filing any lawsuits. While of course not stated in the article, the well-documented fact is that Intel is telling the mobo makers that if they use the VIA chipset they will have their allocation of Intel's SDRAM (and soon-to-be DDR) P4 chipset, the i845, curtailed or dropped altogether. The result will likely be that only the third-tier mobo makers, who probably wouldn't have gotten a Brookdale allotment anyways, will be using the P4X266.
But another reason VIA won't snap up the P4 chipset market is much more hopeful. SiS' DDR Athlon chipset, the 735, has earned rave reviews, significantly beating every other chipset around. Their upcoming 635 chipset for the P4 will offer all that and more, including support for 333MHz DDR (PC 2700) which is coming down the pipeline now.
Slashdot Mirror
Of course Intel is going to tell computer manufacturers not to use the other guy's chipset. And the reason? Not because it isn't compatible, or isn't as fast as the Intel alternative, but because Intel will sue them if they do. Because Intel can't/won't put out a chipset that people want to use (only offering RDRAM or old and slow PC133), they sure as hell don't want someone else doing it. With how poorly the P4 is doing, however, you'd think Intel would be happy to have something out there to move more chips..
Because they can't outperform Via. Thus... just like their partner, Rambus, counter attack with lawsuit. What a classic. Meanwhile, Via has a very strong case, too.
Some bits here and here at Anandtech, and there is another one at Hardware Central. Then, here and here at Tom's Hardware.
All says: Via Rocks, Intel sucks. I'll leave it to you to judge.
--
Error 500: Internal sig error
It's astonishing how an article could spend that long talking about Intel suing VIA over a chipset which introduces a new DRAM type to an Intel CPU and not mention Intel's PC133 fiasco of two years ago.
For those who don't know, the only reason PC133 exists (as a PC standard DRAM type) is because of VIA. Flashback to early 1999: Intel had the market for chipsets (for Intel processors) almost completely to itself, riding on the enormously successful 440BX chipset, which used PC100. However, P3 speeds were ramping up while memory speeds had been stuck at PC100 for a couple years. The obvious thing to do was to update the BX to support a 133MHz FSB. After all, it was a dead-simple engineering trick (every BX mobo at the time could easily overclock to 133; many were stable up to 150), and the memory makers were already making SDRAM which could safely run at 133 but clocking at 100 because that was the highest official speed.
But instead--and unbeknownst to most of the techie world at that point--Intel had a contract with Rambus which offered them many goodies like the ability to make RDRAM controllers royalty-free (others paid up to 5%) and lots and lots of stock options. However, the contract was contingent on, among other things, Intel agreeing to do everything reasonably in their power to prevent "next-generation DRAM" types other than RDRAM from being paired with Intel processors for the consumer desktop. "Next-generation" was defined as > 1GB/s bandwidth.
PC133 has a bandwidth of 1.066 GB/s.
Moreover, Intel *thought* it was putting the finishing touches on the ill-fated RDRAM-only (at that point) i820 (Camino) chipset, with which they were going to introduce new and badly needed 133MHz FSB P3s. Instead, engineering delays involving the difficulties of getting RDRAM working (eventually they had to settle for only 2 RIMM slots instead of the original 3, a per-channel limitation which remains to this day), and the difficulties of getting a memory translator hub which allowed PC100 to be used on the i820 (a last minute addition when they realized people weren't exactly going to pay $500 for 128MB of RAM) working, pushed the release date back 6 months or so, until November.
Just to reiterate: Intel put off releasing 133MHz FSB P3s, and then when they did release them said that consumers could only use them with a buggy chipset, limited to 2 RAM slots, which offered one's choice of an extra-slow translated implementation of PC100 or of RDRAM which cost 10 times as much per bit as SDRAM. Meanwhile, tests with BX chipsets overclocked to 133 MHz FSB showed that this solution was significantly *faster* than the i820 + RDRAM chipset!
Into this world stepped VIA offering the Apollo133 chipset, the first P3 chipset explicitly designed to use PC133. Nevermind that it was probably *less* stable than an Intel BX overclocked to 133 MHZ FSB. Nevermind that it underperformed the BX@133 as well. And nevermind that then, as now, Intel sued VIA with all their might, among other things requesting injunctions forbidding all VIA products from leaving Taiwan. (The pretext then was that VIA was abusing Intel IP by using the P3 bus with a DRAM type Intel had not sanctioned.)
VIA quickly gained > 50% of the P3 chipset market.
Indeed, the only reason you see ALi, SiS, and soon-to-be nvidia and others getting into the 3rd-party chipset market is because VIA paved the way a couple years ago.
Intel tried every FUD tactic in the book, from suing in multiple jurisdictions to claiming that PC133 SDRAM was not stable (the DRAM itself! And this from the company which had spent the past year patching bugs with RDRAM!). Intel got their ass handed to them in court, and by in the summer of 2000 introduced the i815, essentially the BX@133 product they should have introduced in late 1998.
Intel doesn't like getting humiliated, though, and they've had a seemingly personal vandetta against VIA ever since. In retaliation, they denied VIA the chance to license the P4 bus, as ALi and SiS and (interesting) ATi have done. (This is the basis for the current *threatened* suits. However, it's interesting to note that the P4X266 is currently shipping and no suits have yet been filed, meaning this is probably just a bluff on Intel's part.)
Intel reps were even seen at the recent Comdex show threatening mobo makers who had VIA promotional balloons flying at their booths. All the balloons were taken by the Intel people.
However, Intel's case this time is as flimsy as last. Disregarding potential antitrust concerns, the fact remains that NatSemi, whom VIA recently purchased *did* have a license for the P4 bus, and thus so does VIA.
So does this mean VIA will have similar success as last time? Well, I think they'll easily prevail in court if it comes to that, although it appears that Intel may be playing this one all FUD and no bite: warning mobo manufacturers not to use the P4X266 rather than actually filing any lawsuits. While of course not stated in the article, the well-documented fact is that Intel is telling the mobo makers that if they use the VIA chipset they will have their allocation of Intel's SDRAM (and soon-to-be DDR) P4 chipset, the i845, curtailed or dropped altogether. The result will likely be that only the third-tier mobo makers, who probably wouldn't have gotten a Brookdale allotment anyways, will be using the P4X266.
But another reason VIA won't snap up the P4 chipset market is much more hopeful. SiS' DDR Athlon chipset, the 735, has earned rave reviews, significantly beating every other chipset around. Their upcoming 635 chipset for the P4 will offer all that and more, including support for 333MHz DDR (PC 2700) which is coming down the pipeline now.
And they *do* have a P4 license.