Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: Scumbria who wrote (98677)	3/16/2000 6:05:00 AM
From: ptanner	Read Replies (2) \| Respond to of 1570519

Scumbria -- you got this response over at JC's re: AMD & SSE jc-news.com On SI, Scumbria (who is no Intel-fanatic) is saying that AMD has to implement SSE now, because the X-Box will mean that programmers are going to SSE-optimize all their games. It may be true, and I think I agree that they have to implement SSE. It would be good in terms of using the Intel compilers for SPEC scores, too. So I am wondering what the issues are: Patents: Scumbria thinks there will be patent obstacles. I wonder whether SSE is really that innovative. As far as I recall the issue around MMX was mainly the use of the Trademark MMX, which seems like a minor issue. Time: AMD had time to put the SSE prefetch instructions into Athlon, so they have had the opcode details for quite some time. Seems like SSE could be one of the enhancements in Mustang and Sledgehammer, if AMD wanted it to be. Of course the Xbox debacle only just happened, and they may not have seen the need up until then (though there was always the question of SPEC and the Intel compilers). Opcodes: Scumbria thinks there are clashes, but I checked, and all 3DNow! instructions start with either 0f 0f (SIMD) 0f 0e (FEMMS, fast enter/exit of MMX/fp state) or 0f 0d (Prefetch) which are all free at least up until P-III (SSE). No telling what will happen with SSE2 of course. AMD have been quite willing to add Intel stuff to Athlon. They put in the 64Gbyte memory extensions, the SSE prefetch, etc. So is SSE for AMD the future?

To: Scumbria who wrote (98677)	3/16/2000 9:21:00 AM
From: milo_morai	Read Replies (1) \| Respond to of 1570519

Excellent points on 3dnow and SSE A friend of mine posts "By: NY_Gymkata Reply To: 10548 by TA152H Thursday, 16 Mar 2000 at 3:20 AM EST Post # of 10637 TA: First off I think you underestimate INTCs ability to manipulate the market. INTC needs only to say to the market that SSE optimized apps will be far superior and now any software maker who does not optimize their software will be scorned as having inefficient code. INTC offers incentives to software vendors to optimize, but they do this because they want everyone using their proprietary extensions. AMD developed 3DNow! in conjunction with software vendors and asked for input on what instructions to make. AMD also has agreed to license 3DNow! to any chip maker including INTC in an effort to allow software vendors to develop for a single set of instructions. Admittedly AMD must go this route since they do not have the power to manipulate the market like INTC, but that does not change the fact that AMD is doing what is best for everyone. As for your AltiVec suggestion I fail to see how using PowerPC instructions in an x86 processor will give any benefit to AMD. Perhaps these instructions are better than 3DNow and SSE, but since they are coming from a completely different platform I cannot see how such a maneuver would offer any sort of advantage for MOT or for AMD. As for AMD using SSE that will only happen if 3DNow! is rejected which is not out of the realm of possibility, but also not likely (since the support is still there and growing). If AMD used SSE they would return to being at the mercy of INTC for developing each next generation chip and would be crippled, AMD should continue to develope 3DNow! and enhance it as necessary to offer superior performance. Perhaps Mustang should increase the number of MMX/3DNow! processing units and add new instructions to fuel the need for 3DNow! optimized code. As for critical mass AMD does not yet have it, but they are gaining mass very quickly while INTC is losing ground just as quickly (or quicker with VIA and Transmeta entering the market). Can AMD move the entire industry to 3DNow!: No. Can AMD make 3DNow! attractive enough to have developers continue to use it: Yes, and that is all that AMD needs to continue to use 3DNow!. How many times do you think that software vendors will recompile their code to use INTCs latest sets of instructions if they know that their code will run faster on AMD processors without being recompiled? This is not yet the case, but it may well be the case with Willamette vs Mustang or even Willamette vs T-Bird. We'll have to wait and see. -NY_Gymkata " ragingbull.com I think if Via's new Socket 370 chip Joshua with 3DNow! SIMD does well it will continue support for 3DNow!. Since AMD is sharing it's instructions with everyone will that be a factor? esp. if AMD does get it's 30+% marketshare. Milo

To: Scumbria who wrote (98677)	3/16/2000 1:37:00 PM
From: tejek	Read Replies (2) \| Respond to of 1570519

AMD has never been a good long term investment, and I don't see any reason to believe that has changed. The early arrival of Willy, combined with the X-box disaster, gives me no reason to think otherwise. Scumbria, Your comments of late are an extraordinarily dark view of the current AMD/Intel competition. True there is no question that Intel is a formidable opponent but to believe that the game is forfeit because they have more resources at their disposable makes a mockery of why we all came to be here on this thread. I do not have the tech background to dispute your position re SSE issues but I can argue that the X-Box is a big loss and not a disaster, and the early arrival of Willamette will create problems that are not insurmountable. Your position assumes that Intel will execute smoothly something they have failed to do for the past year, that X-Box will be a huge hit and will set new standards, and that AMD does not have clue as to what's going on and that the Athlon was an incredible fluke and AMD must now give up the ghost. I admit your view of the future is a very real possibility but I have a hunch that there are some other very real possibilities that may prevail instead. ted