AM2

I was reading the Toms Hardware report about AM2 and it had a really short
section on NF5 from Nvidia.

This
looks really exciting. Especially as they have taken the plunge and got
rid of one ATA connector - now you can only connect 2 ata devices! Im
not sure but I seriously suspect Nvidia would have done away with them
entirely except that most DVD drives are still ATA. NF6 will almost
certainly see the death of ATA connectors.

That also means we will be getting 5 SATA slots instead of 4!

To be perfectly honest
thats the most exciting part of the article for me. At the moment AM2
looks fairly generic, ok so it has potential but all AMD have done
(with the exeption of the FX-62) is port all of their current catalogue
to AM2. So we have all of that potential and the last gen technology
just doesnt fill the shoes.

Maybe Im being a little harsh but I
cant help feeling AMD have introduced this new socket way too soon,
unless they can pull out some good FX or A64 chips in the next few
months people are going to start to feel that this is a little bit like
a gimmick, K8L isn't going to be around for a good while yet!

I
see AMD's strategy, getting punters to upgrade to AM2 now so that when
K8L chips appear they will be an 'easy' upgrade but I really dont think
thats going to work. AM2 MB now will not make full useof the potential
of the socket. NF5 is a great example, sure they have dropped one ATA
socket but by this time next year I fully expect ATA to be real legacy
with DVD drives mostly using SATA. Thats especially likely considering
we should have Vista by then
which will bring full native SATA support.

All of this asks an interesting question. What can AMD do to meet the
threat of Conroe. And is Conroe even a threat. People talk about the last 5 or 6 years being AMD's golden era beating Intel of at every post. In a way thats true, AMD did produce some excellent chips for a ,uch lower price than Intel . They also pulled of some technically brilliant feats, but there is another side to that. The lasy few years havn't been so good to AMD. Really they hit a dark era - one that is peaking now with the premature release of AM2. Ok so Athlons, Athlon Xps and Durons were excellent chips and A64 looked set to follow nucely in their footstpes. But then we had the move to S754 (which AMD openly admitted was a transition socket) and S939 - which they claimed to be sticking with for years to come. Unfortunately it seems 939 just did not have enough potential for AMD so now we get another new socket. Thats all very well - we want progression but there has been so many changes in such a short time punters don't know which way to turn. Do they go for chaep 939 chips now when chances are that it will become less and less supported or do they cut their losses and hit AM2, meaning an entire new rig! I sympathise with AMD, it was never going to be an easy transition.

Conroe is going to be AM2's big competition (in a way). So far tests show that Intels new chip is pretty good - much faster than anything currently produced by either company. It's also a great technical acheivment with some new tweaks and enhancements. On the other hand when you consider that Intel have named this a next generation chip the performance margin doesnt looks so good. Compared to the current generation of chips you would expect Conroe's to speed ahead of them (well I would anyway). I just don't understand that. The GHz wars are over but speed is still king and these operating systmes are getting more and more hungry (ahem).

The other thing is that K8L (AMD's next gen chip) is looking much better. Ok so it doesn't bring much technically (except for L3 chche which will be useful and Double Floating Point Operation) but it is a real redesign with potential for expansion. As far as I can see with conroe Intel have sacrificed alot of space for the huge L2 cache at the expense of upgradibility. Conroe will last a good long time but when it comes to the next next generation Intel may have to go back to the drawing board. I hope Im wrong, I wont pretend to fully understand how the Conroe architecture works but I get the feeling it is a case of: "good now but the future; well....".

Again I hope I'm wrong and Conroe doies go from strength to strength, Intel deserve it, they have put a hell of a lot of effort into this chip and it's failure would hurt them badly.

Similarily I hope AMD can deliver in K8L and AM2. If both sides had excellent chips on offer we could see some great leaps in chip design.

One final thing. L3 cache is an interesting and probably important idea, but despite AMD mentioning it in an IET (then IEE) and an IEEE journal interview a year or more ago Intle havn't incorporated it into the Conroe design. Again I see no spac to add it on the designs either - which is very odd!