Linus Torvald, le papa de Linux que l'on ne présente plus, nous montre que lui aussi s'énerve de temps en temps dans les forums! Le sujet portait sur les extensions 64 bits.  
 
"En fait, je suis un peu écoeuré qu'Intel n'ait même pas _mentionné_ AMD dans sa documentation ou ses communiqués, ce qui fait que je serais enclin à renommer (Note: le fil de discussion parlait de la normalisation des noms) la chose en "AMD64" (Note: plutôt que x86-64) juste pour donner mérite à qui mérite est dû. Cependant, cela n'en vaut ni la peine ni la confusion.  
 
A quiconque d'Intel sur la liste: dites à vos managers d'être f*trement honteux d'eux-mêmes. Simplement parce qu'Intel n'a rien à faire de ses clients et a joué avec une autre architecture 64-bit que personne n'a voulu utiliser n'est pas une excuse pour ne pas en donner le mérite à AMD pour ce qu'ils ont fait avec x86-64.  
 
(Je suis vraiment content qu'Intel soit de la partie, mais ce n'est pas joli-joli de ne même pas mentionner AMD dans leur documentation et essayer de la faire de façon à ce qu'on croit que c'était leur idée).  
 
...  
 
Ce que je trouve si irritant c'est que des _heures_ après l'annonce d'Intel, les gens étaient _toujours_ confus sur le fait que la nouvelle puce soit compatible avec celles d'AMD. Pourquoi diable ne sont-ils pas simplement venus le dire, et en parler? Cela a fait que les personnes qui lisent les manuels (qui ne le mentionnent même pas) ont dû convaincre ces gens dans les forums sur l'architecture des processeurs que oui, "ia32e" était réellement le même que "amd64" à part dans les petits détails qui avaient toujours différenciés Intel et AMD.  

So, AMD now will have Cray for large HPC systems, Sun for the departmental and big corp servers, HP for mid sized corp servers, IBM for low-end HPC, Fujitsu for HPC across the board from workstations to small clusters.
 
So Opteron wouldn't sell uh? It looks like Opteron will be filling part of the gap left by SPARC and Alpha plus some bits of Xeon's market.

AMD and the future: the answers at Cebit (AMD)
 
As I appreciate the fact that you all made my trip to Cebit easier (didn't have to think about questions, just ask , Here are already the answers on the questions concerning AMD.
 
1. What are their plans for moving beyond single-core/single-threaded designs? CMP? CMT? SMT?
 
    AMD told me there are no "real" ("looking into it" ) plans for SMT or Hyperthreading, but it was clear to me (although the spokesman didn't confirm) a dual core CPU is coming in the future. The most important platform improvement for Quad/Dual Opteron systems will be a faster HT link between CPUs, which will produce tangible performance increases. Dave Everitt confirmed that the performance increase due to this faster Hypertransport link will be better than just a few percentages.
 
    Other sources say that a "Coherent" and faster clocked Hypertransport should be ready in the beginning of Q3.  
 
2. What's up with the 90nm process?!
 
    The 90 nm process is shaping up very ("fantastic" if you believe other sources) well. No confirmation on when the new 90 nm Opteron/Athlon 64 will be available ("H2 2004" ), but probably at the end of Q3, according to other industry sources.
 
    The 130 nm process is doing very well too (Cg revision reduces power for example), and another speed grade (2.5 or 2.6 GHz?) in 130 nm is very likely (almost certain). AMD had an aircooled (12.5x multiplier) 2.5 GHz Athlon 64 FX (940 pin) running at Cebit.  
 
3. What about DDR-II?
 
    At this point of time, AMD believes (and many others confirm) that DDR-II 400 is more expensive and slower than DDR400, so AMD sees no need at all. When faster DDR-II will be less expensive, AMD will support DDR-II. Current estimate is that the Athlon 64 platform will not change to DDR-II before the end of 2005. The Ace's team opinion: DDR-II tech specs did never really excite us, it is a very conservative design.  
 
3. When will we be able to buy Socket 939 boards?
 
    Socket 939 boards were over all over the place at Cebit. One motherboard manufacturer told us that in May 2004, socket 939 boards will be available.  
 
4. What is the future of Socket 940 and Socket 754?
 
    AMD confirmed very clearly that socket 940 still has a bright and long future ahead. It is, after all, the socket of the Opteron and AMD is commited to keep the Opteron platform stable. The Athlon 64 FX will be socket 939 based from May 2004 on, but faster Opterons will be available for the current socket 940. Socket 754 will see at least one other speed bump (Athlon 3700+?), but is (IMHO) less future proof than socket 940 and 939. The socket 939 "Athlon 64 FX" will have 1 MB L2, and there will be also a Athlon 64 with 512 KB of L2-cache.  
 
5. Do you expect to maintain the current TDP of 89W for 90nm parts?
 
    AMD confirmed that TDP for 90 nm parts will also be very close to 89 W, just like the 130 nm parts. Don't forget that AMD specifies a maximum TDP for a whole CPU line. In other words 89 W is the TDP of the (expected) fastest CPU.