To: Time Traveler who wrote (34318 ) 7/12/1998 12:49:00 AM From: Maxwell Read Replies (1) | Respond to of 1572094
Time Traveller: It is good to hear from you. I like challenging questions coming from you. It tells me that that are educated people out there that are willing to ask "WHAT IF?", not like the Intelabees who are living in the ivory tower. <<You are perhaps right. This AMD stock has been beaten really badly. What more could it drop? >> Let me remind you that AMD is a "strategic" game piece for Compaq, IBM, HWP, etc. They all want AMD to survive to keep the 800 lbs gorilla Intel from going mad. These box makers love AMD. It allows them to have the "upper hand" in negotating with Intel. They will do anything to keep AMD alive. In the worst case they will be bought out at the verge of bankruptcy. Thus anything below $15 is a good buy on AMD. <<This is not entirely true. Although this is true for the MMX unit, the FPU of the P-II is still superior to K6's. Currently, only if you want to be a hero to bravely put yourself in front of the whole humanity to blast away exotic species after species of weird and very alien warrior creatures, this feature is not very useful.>> You are completely wrong here TT. FPU of K6-2 can take square root and division faster than Intel FPU. K6-2 can take square root in 10 clock cycles whereas PII takes over 100 clock cycles. FPU of K6-2 show up best when doing 3D geometry transformation. This is apparent in games like QUAKEII. A K6-2-333 can do 60 fps at 1024X720 resolutions while a PII-300 can only do 38 fps. These few games are just a few applications that demonstrate the superior of the 3DNow architecture. Just imagine if you write a program that does "many body problems". This is a physics problem that does massive calculations at billions of iteration. The K6-2 will win because it can calculate faster. Future applications will utilize more SIMD architecture of a CPU. This is the whole reason why Intel will come out with Katmai using similar SIMD architecture. Intel is not stupid. They know 3DNow is the future. They are just a little behind and caught by surprised by AMD. <<At this moment, it is probably true, but when the volume ratio becomes 5:1 (P-II:K6), it may not be true any more. Don't we have to look into the future for investment?>> You are wrong again here. AMD just make the silicon chip, put it in a ceramic package, and test then done whereas Intel a) makes the silicon and package in a ceramic package and test b) make a PCB daughter card c) surface mount the resistors, CPU, SRAM d) test the card In manufacturing the more you touch the more likely it fails. Intel requires 4 steps whereas AMD has 1 step. Now you tell me which is more expensive. BTW, if P(a)= probability of a) to pass......0<=P(a)<=1 P(c)=probability of b) to pass.....0<=P(c)<=1 then P(Intel of slot 1 passes)= P(a)*P(c) P(K6 passes) = P(a) <<Time Traveler is really shocked at that statement. Yousef has demonstrated time after time that he really knows what he is talking about. Time Traveler has not seen many serious challenges to his posts including the ones from you. We should be very luck to have him posting here. We have learnt tremendously in the past couple years.>> He only see the grain of each tree in the forest. I see the whole forest. <<Time Traveler did a very primitive error correction to the above message of yours. Do you mean K6-3? Let's assume you do. Don't you think you are talking about a vaporware? Why don't you compare this K6-3 with comparable Intel product in the future?>> K6-2 with integrate 256KB cache is the K6-3. It is not vaporware. I heard AMD has already ship these parts to key customers running well above 350MHz. You will see it in Q4. This chip is faster than a Xeon of equivalent speed. It will be the world fastest socket 7 and will outperform the PII of similar speed or one speed grade above. Its die size will allow AMD to sell it at healthy $250. Consumers will think it a steal compared to any equivalent Intel parts. <<Why does Intel's conservative approach lose out? We are talking about billions and billions of dollars at stake, at least for Intel. Speaking billions and billions, let's have a moment of silence of respect for the late astronomer, Carl Sagan.>> Intel doesn't lose out at all. They have made billions of dollars. Intel has the marketing advantage. They do what maximize their dollar investment. AMD is late and must offer superior and cheaper parts to win customers back. Thus AMD must took the path they took to get the competitive advantage. <<Not quite, DEC does not have a better IC design. Alpha is a younger generation of CPU that does not have to be painfully backwards compatible to the older ones. Naturally, the system architect of Alpha can do so exploring the most efficient way. Intel has to design a CPU that is almost as powerful as the Alpha given the same generation and still has to worry about backwards compatibility because of market demand. In Time Traveler's humble opinion, Intel and AMD x86 designers had a tougher job than the Alpha designers.>> Your point is well taken, but Youse is still wrong about speed and process. <<No, AMD has not achieved great yields. The tell-tale sign is the more units produced the more losses face AMD. Ever since AMD introduced K6, quarter after quarter, AMD has been shipping more and more K6 and losing more and more money. Is this not true?>> What do you mean by great yield? I can tell you one thing AMD produces more good dice on a wafer than Intel can on their PII of 131mm^2. You have to dig deeper into the fact why AMD is losing money. Had AMD made the same money on flash, Vantis, and communication as Q1 they would have made more money than Q1 and less loss. Also AMD was in a quarter of product transistion. They were migrating form 0.35um to 0.25um. Q3 you will get a better picture. <<Intel has been building fabs, too. How do you explain it is not affecting Intel as much as AMD does?>> AMD lives on borrowed money. Intel got billions in the bank and can build a fab anytime at their disposal. <<Have you checked what Intel has in their pipeline? The history of Intel dominating over AMD has repeated itself year after year! Haven't you noticed it?>> Why talk abou Intel here? Intel is expected to have a full pipeline of products to maintain its market share. What is more interesting is how AMD pipeline of products affect Intel future. The fact is that AMD pipeline of good products guarantee AMD's future and its ability to compete and penetrate Intel's market. Intel will have to work hard just to maintain its #1 position. Maxwell PS: I bought some Intel July puts just to insure my shares. I have a feeing that Intel may cough up some bad number.