A Day in the Life of an AMD Salesman
Vendor (Rod Macpherson, thinking aloud):
"We think that ... <the real thinking is snipped>... Let me ask him just a few tough questions."
Vendor: How many processors can we put on our servers? AMD salesman: We were under impression that you are making mostly personal computers, are not you? I agree, the market for servers is growing as everything else, but not too big to deal with: for every server people would need at least two dozen PCs. I personally would assemble two uniprocessor servers and get exactly twice the throughput than a single two-processor unit with only 40% performance increase at a cost of three.
Vendor: How much more are the mobile versions? AMD salesman: Not too much, really. Especially for you. See this notebook with sales database? It is made by some large (yet Asian) manufacturer and runs K6 at 233 out of standard battery and without a cryogenic cooler. Cool? Your customers are probally ready to pay high bucks for this book.
Vendor: Is the bus speed equal to Pentium II? AMD salesman: Which bus you are talking about? Since what time a CPU was residing on PCI bus? Ahh, you are not sure. I see. By the way, for your safety I would recommend to stay with 66 Mhz system bus for a while, to reduce your cost and for better realability. Maybe later you can migrate to 75 or 83 Mhz, when your printed circuit technology and motherboard designers will be able to handle these speeds reliably. You know, some wires must be traced really accurately to run at 100Mhz to avoid clock skews, reflections, etc. What? Never heard about skews? No problem, we will work it out with you later. You know, our buddy Intel did really smart move with SEC, but I am not sure your company can handle the cost of migration..... Again, the higher speed DRAM would cost you too much. Of course, you always can fool your customers with impressive numbers that buys no performance. Anyway, with our huge internal caches you will not see much performance difference because the cache miss rate is small, and our CPU does effective prefetching when it continues to execute RISC instructions internally. Hello, are you with me, sir? Good. What? Video subsystem? Did you try the Intel 3D tests on their hottest Pentium-Pro? Or Pentium with MMX? Good... Remember what frame rate it was? Correct, typically 2 frames per second, maximum about 6. Right. Do you remember what kids want to play 3D games? 30fps? Right, good job, you would need a 1000Mhz Intel CPU to keep up with this. AGP? How much the AGP would accelerate the screen texture painting? Maximum four times (33->128), but this works for painting only, which is about 30% of the entire 3D engine. Do your math and see - no real gain. Go and buy 3D-righteous or rendition or other accelerator for $200 and enjoy the real game. Need to learn more? Sorry bud, no more time for education, only after you sign the sales contract. OK?
Vendor: Is there an upgradable plug-in? AMD salesman: What? Come on pal, any Intel upgrade costs more than a new MoBo with new (AMD) processor. I really wonder who buys these upgrades today? By the way, who really cares if your PC will pop-up a window in 15 milliseconds faster? What really does matter is the cost of a PC, and with K6 it may be way below $1000. You can put your savings into high-quality 17" monitors. This would make your staff feel better with much higher office productivty. Think about this.
Vendor:
Hmm, I was told by TV that Intel i233 is way better than i200. But now, doing some math (233/200=1.15), I see that the total premiun seems to be too high: mobo redesign, mechanical unrealability, heat problems, non-handlable power requirements. The whole new product line. And all this for nothing. We are tired to keep up with Intel innovations. With 6 months product life cycle, there in no time to make any profit, only Intel's pockets are getting fatter... yehh... We need really to do something. Right now. Where to sign in? |