Originally posted by: Gagan
Um, lets just take a little refresh down memory lane.
AMD Was on its thunderbird core for GOD knows how long after the P4's came out with the RDRAM.
How long was it until the Athlon XP's came out that DID beat the RDRAM p4's? That was because it was the ****** cores, when coppermine and northwoods came out the comparibly priced did not stand a chance.
Those are when Intel owned the lionshare of the market, 533mhz and 800mhz respectively
Intel has always owned the lionshare of the market, though the P4's with rambus where when they incurred their biggest market share loss until recently with the Athlon 64s.
And coppermine didn't do crap against the top end Athlons or P4s, and got no performance boost from Rambus due to its single channel FSB.
Northwood didn't immediately beat the athlons anyhow, more like caught up in performance, with a model every now and then that would beat AMD, and then AMD would catch up. It wasn't until dual channel PC3200 that Intel had a consistent performance lead. (which was more due to AMD's serious delay of the Athlon 64 than anything else, the Athlon XP was forced to compete for about a year longer than intended)
And I seem to remember Rambus P4s being available from the start, and losing to thunderbirds with SDRAM. Then again, the chipsets at the time barely benefitted from RDRAM or DDRAM until about 1.6ghz and beyond, and it was a while before Intel had DDR chipsets out with comparable performance to their RDRAM chipsets. I'm not sure if that was an actual advantage of RDRAM or not though, but AMD had similar problems with getting DDR performance up to current standards. The P4s at the time sucked, but right from day one they were giving basically the max performance those P4 cores would ever hope to have, it took until I think the i865 to get DDR performance up to par with RDram, and most of Intel's improvements from initial launch were to the core (higher mhz, more cache, hyper threading) whereas AMD stayed in the game by starting out with a great core (comparatively, it didn't see much improvement in mhz and already had a sufficient amount of cache for the designs, the willamette cores were not what Intel had intended the p4 to be), and then just evolving its platforms from complete crap to on par with Intel. I think the nforce 2 was the only athlon xp chipset that showed latency and bandwidth on par with the i865, while the early VIA DDR chipsets showed worse latency than SDR and barely improved bandwidth.
Ok, some kind of summary is probably needed for all that... or I can just make something else up and put it here.
The Rambus platforms for the P4 were years ahead of Intel's DDR platforms, which may or may not be an advantage of Rambus technology.
The old RDRAM was really only well suited to the P4, while SDR was optimal for the aging P3 and G4 designs and DDR optimal for the Athlon, and able to be suited to the P4 with dual channel.
There's nothing wrong with having more competition in the market, but Rambus's focus on high bandwidth is not most beneficial to the PC market, and its price wasn't worth its performance. If Rambus can come out with some new memory that matches or beats DDR2/DDR3 and can have it affordably priced then there's a very good chance they could catch on in the market place. However, they failed previously for some very important reasons, too expensive for mass market, and since a rambus based platform was only top end performance for a very short period (and never cost efficient) they never really caught on in the enthusiast market either. Had it been an Intel only market, we'd probably still see Rambus in some form. It's quite interesting that the memory manufactuers essentially seemed to band together and buck Intel though, you'd think limiting their high end memory (DDR) sales almost exclusively to AMD would have seriously hurt them, but instead it forced Intel to adopt it as well.