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: JohnD who wrote (100674)	3/29/2000 1:58:00 AM
From: milo_morai	Respond to of 1572839

Tom Kerrigan?s Sledgehammer Preview Thomas.Kerrigan@Colorado.EDU Quick note about the similar AMDZone article The reason I wrote this article is because I was severely disappointed with Alan Smith's Sledgehammer preview at AMDZone. When it was first posted, Alan's article contained a tremendous number of incorrect statements. I sent him a long list of corrections and suggestions, which is why his article is now similar to mine. However, I still think Alan misses the boat on a number of issues. Hopefully, my article addresses these deficits. Contrary to what many people seem to believe, I did not steal my information from Alan. All of my information can be found on presentation slides that are posted on AMD's web site for all to see. What is Sledgehammer? ?Sledgehammer? is the codename for a 64-bit processor that AMD is currently working on. It has the same basic design as the Athlon. AMD is making the chip 64-bit by widening the integer core and tweaking the instruction set. AMD calls the tweaked instruction set x86-64. It?s almost identical to the instruction set that everybody loves to hate, IA-32. That means Sledgehammer can run IA-32 code without a significant speed penalty or complicated new logic. In fact, the x86-64 core is expected to be only 5% bigger than the current Athlon core. So AMD can manufacture an x86-64 chip [almost] as easily and cheaply as a regular Athlon. x86-64 also adds new RISC-like floating point instructions. This is a huge improvement over the hopelessly outdated IA-32 floating point stack instructions. AMD is calling the new instructions TFP, for Technical Floating Point. They will allow Sledgehammer to have floating point performance equal to that of many RISC chips, like the PA, Sparc, and Alpha. According to current rumors, Sledgehammer will be available in Q1 of 2001. It will be manufactured using a 0.15 micron process, have 2MB of on-die cache, and run at a possible 2GHz (although it will probably be introduced at a slower clock speed). AMD has plans to put two x86-64 chips on the same die, so we can look forward to cheap and simple SMP systems. The importance of 64-bit The main benefit of a 64-bit processor is its ability to easily address huge amounts of memory. 64 bits can address several million terabytes, whereas 32 bits can only address 4 gigabytes. With current memory prices, 4GB of RAM only costs a few thousand dollars, so this limitation is not acceptable for many high-performance applications. Current Intel processors have a 36-bit (64GB) address space, but this feature is sort of a hack. Most operating systems don?t make use of it. 64-bit operations do twice as much work as 32-bit operations, and some algorithms can take advantage of this extra work. It?s a lot like the advantage you can get from MMX, but without the awkward assembly language programming, register juggling, and lack of portability. In fact, a lot of code is already written with 64-bit processors in mind, because 64-bit RISC chips have been around for a long time. At least one of the new SPECint benchmarks, Crafty, relies heavily on 64-bit operations. Sledgehammer vs. Itanium Sledgehammer?s obvious competition is Intel?s 64-bit chip, Itanium, which will be available later this year. Itanium is based on a ?post-RISC? design called VLIW. It?s supposedly superior to RISC, but it has its own set of design hurdles (which I won?t get into). Many people believe that RISC chips can run just as fast or faster than VLIW chips. Intel hasn?t released any benchmark data for Itanium, so nobody knows if it will be faster than Sledgehammer. We do know that Sledgehammer will have a big clock speed advantage over Itanium (hundreds of megahertz), so Intel will definitely be fighting an uphill battle. Itanium?s native instruction set is called IA-64, and it?s nothing like IA-32. Itanium will be able to run IA-32 programs, but probably not very fast. This may cause some people to go with Sledgehammer instead. Sledgehammer vs. Willamette If Sledgehammer is positioned as a server processor, it will not be in direct competition with Intel?s next mainstream processor, Willamette. Instead, the 32-bit Athlon will have to compete with Intel?s next-generation chip. However, Sledgehammer?s x86-64 core is not much bigger than the Athlon?s core, so AMD has the opportunity to make x86-64 a standard feature of its entire product line. This would give AMD?s consumer processors the speed advantage of 64-bit integer operations, plus the sweet new TFP instructions. A chip with TFP will be able to whomp Willamette at 3D gaming (which drives the mainstream performance market). I believe AMD should not overlook this option. Conclusion Sledgehammer will definitely be a cool chip, but its success depends on Itanium. Nobody knows how fast Itanium really is, or how much it will cost. These questions should be answered within the next few months. Hopefully AMD will add x86-64 to some of its mainstream chips. This would give strong floating point performance (and 64-bit operations) to the masses. Cross your fingers. . . ucsu.colorado.edu sorry if duplicate.