Author: Vincent Diepeveen
Date: 06:50:03 04/07/02
Go up one level in this thread
On April 07, 2002 at 01:00:55, Russell Reagan wrote: >Last night I was cruising the CCC archives reading over discussions on various >data representation approaches in chess programs, and I came across this >statement from 1999 in this post: >http://www.it.ro/ccc_search/ccc.php?art_id=39708 > >"Just wait 5 years and see if you can find a 32 bit machine left." > >Well we are 2 years away from the 5 year deadline. Do you still agree with this? I still at am my viewpoint that if the fastest processor is 64 bits, that i always can convert some 32 bits datastructures within an afternoon to 64 bits if that's faster for the datastructure in question. Bob can afford a bit bigger machines than i can, let's be clear about this. The only real realistic 64 bits machine is the McKinley of course which is supposed to sell start of 2003. It is going to run at 1Ghz and a dual mckinley might be $20000 if you are lucky to get it that cheap. A dual 1Ghz 21264 which is there for $20000 and which is 64 bits, is as you can see at the benchmarks at www.specbench.org at the same speed as a dual 1Ghz 1.33 Ghz MP for crafty. 21264 issues 4 instructions a clock maximum. K7 like 3. Hammer is perhaps available at the market at september 2003, if AMD is lucky, but we must see it to be equally fast for 64 bits operations as it will be for 32 bits. I have no idea what price Hammer will have. It has more built in than other CPUs. It doesn't require a parallel chipset to work in parallel. I am no expert on this terrain of course, but i can follow the idea that it will be a lot cheaper for parallel hardware than todays machines are, but i do not understand why a chip single cpu will be cheaper if it has more stuff on-chip. Nevertheless it most likely won't be a very expensive chip but i have to warn here that intel shouted for like 5 years about the IA64 before it came onto the market and it is a dissappointing chip because it only runs 800Mhz where we can also buy for a part of its price a way faster 1.6Ghz K7 MP. Of course it still is a big victory for intel to have it, because they are not busy with computerchess but they just see the big machines Sun and other manufacturers sell for billions, with just a few processors inside (so pretty cheap in production cost i would guess) and with mckinley they might do a BIG throw into that market. I do not know when 64 bits machines will be the norm, becuase there needs to be competition. Suppose Hammer is 64 bits and outperforming all other 64 bits cpu's by a huge margin AND it can be used for quads and 8 processor machines. Just let's *assume* this. If they sell it for $4000 a processor, they still completely dominate the market then. Now let's be realistic. Sun, Intel, HP, IBM and all those big manufacturers aren't stupid anyhow. If they could make a superb 64 bits processor at 2.5Ghz, they would do it too. And remember, THEIR 64 bits cpu's are very expensive. Intel very realistically already targets Mckinley to start at 1 Ghz, initially it was even 1.2Ghz, that says *something*. the Power4 from IBM at 1.3Ghz is not exactly *cheap* by any standard (power consumption, price etc). It is not realistic to assume Hammer gets on the market soon. A company which produces for the first time in its history a 64 bits processor is going to have the same major problems a major hardware manufacturer like intel had introducing its 64 bits processor. We still didn't mention price. If i can get a 32 bits processor cheap which outperforms for my software a 64 bits processor by 33% in speed, then my choice has been made already. Even if the 64 bits processor is 2x more expensive, if it outperforms the 32 bits processor by just a few %, i will already buy it. There is not a single 64 bits processor on the market which outperforms the best 32 bits processor. I do not see a change here soon. The software world was of course happy to go from 8 bits to 16 to 32 bits. The step to 64 bits is less important. 32 bits already gives a lot of choices, add to that , that the FPU is already 64 bits for a quite a long time. note that fpu is also used for MMX. So going from 32 bits to entire 64 bits processors, that's for the average user not so interesting. And they sure do not ask for it. So when 64 bits processors will dominate all 32 bits processors, i do not see that soon coming, unless AMD manages to make a processor faster and better than any other manufacturer. But remember, they never made a 64 bits processor before, so most likely september 2003 is a bit early, or the 64 bits part of the hammer will be dead slow simply. One of both. >I'm not trying to throw old quotes back at you or anything. I'm really more >curious about how much longer it will be until 64-bit machines are the norm. >I'm also curious if when the time comes that 64-bit machines are the norm, if >they will be on par with the Hz speeds of the 32-bit machines. For example, >right now you can get a 2.1 GHZ Athlon or 2.4 GHz P4 without having to take out >a loan. If you want a 64-bit Itanium, you're looking at $2,500 - $7,000 for a >chip that runs at 733-800 MHz (www.pricewatch.com). So once 64-bit machines are >practical from a price standpoint, will they still be at a third of the speed we >can get from a 32-bit machine? > >If my data from pricewatch isn't entirely accurate please correct me. E.g. if >there are other 64-bit chips that are cheaper and faster than Itanium. Heck, how >much would a good 64-bit system cost today? > >Thanks, >Russell
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.