Athlon 64 Motherboards and AMD
This content was originally featured on Amdmb.com and has been converted to PC Perspective’s website. Some color changes and flaws may appear.
Chaintech provided us with some interesting roadmaps of their different motherboard lines: Zenith (high-end), Apogee (mid) and Summit (low-end). The Zenith motherboard that is currently planned for the Hammer platform is the 8VJS. It features the K8T400M chipset with the addition of the VT8235 south bridge. Current specs include 2 DIMM slots, AGP 8x, gigabit LAN and serial ATA RAID. USB 2.0, Firewire and SPDIF audio are included as well. The schedule for mass production was set for early April.
The currently shown Apogee branded motherboard is called the 8NJL and uses the NVIDIA Crush K8 chipset. It also includes 2 DIMM slots, AGP 8x, 10/100 LAN, USB 2.0, Firewire and SPDIF audio. It was set for a mass production date of early April.
The original reason for my attending of the event was that both of these motherboards would be ready for sampling and the press would be able to take a working motherboard home with them to poke and prod at. With AMD’s recent announcement that the Athlon 64 processors would be pushed back until September of this year, not only did we not get the samples that we were expecting, the entire line of Chaintech motherboards that were planned for the Athlon 64 are going under a re-evaluation. Why would Chaintech want to go back and change the specifications of their already completed motherboards? The problems lie in the south bridges from the various chipset manufacturers – all the major players have brand new versions and revisions of their south bridges coming out between now and September. And, because the Athlon 64 platform depends mainly on that south bridge for its ability to separate motherboards from competitors, all manufacturers are going to want the latest and greatest that VIA, NVIDIA and SiS have to offer when the Athlon 64 finally reaches the consumers and media. Thus all manufacturers, not just Chaintech, are going to be frustrated during these next six months as they are forced to basically do their research and timing all over again.
Needless to say, in this respect, the Chaintech Evolution Event was very disappointing. Chaintech, however, could not do anything about this as the decision was completely on AMD’s shoulders.
What AMD Had to Say…
With all the controversy of the AMD Athlon 64 delay among the crowd at the event, here is what AMD presented to us at the official briefing as to the reason AMD chose to make this move of releasing the Athlon 64 in September:
1. This would allow a better alignment with AMD’s partners
2. September is a better time for a processor launch
3. Integrated graphics would have a good increase in availability on Hammer boards at this time
4. It will align the Athlon 64 with the mobile Hammer platform launch.
If you are careful, you will notice what we did here in Spain – that there are really no new answers in here. Much of it is vague and easily debatable. What many presume, but we have no evidence to support, is that AMD is still not happy with their yields (good yields are necessary to have a cost-competitive desktop processors) or they were not yet happy with the frequencies they were reaching. It is rumored that the Opteron is going to be launched at speeds of 1.4 GHz – 1.8 GHz and at those speeds a desktop launch may not have been feasible.
The AMD presentation did offer a couple of other interesting notes including information that DDR2 support would be ready for the Hammer processors in 2004 – but he would not say how soon in 2004 and how this DDR2 implementation would be accomplished.
If you were going to be looking for a 2500+ Barton processor, expect to be on a long waiting list, as it will have lower availability than the rest of the processor family at launch time.
Also, I learned that the original planned speed of the 3000+ processor was to be set at 2.25 GHz and then was dropped to the 2.17 GHz when they found that the performance improvements provided by the additional 256 K cache were better than expected.
