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: Jim McMannis who wrote (55718)	4/16/1999 4:04:00 PM
From: Process Boy	Read Replies (2) \| Respond to of 1578938

Jim - Et al. If possible please dig up some detail as to the date and the category of Si (Alpha, Beta, final, etc.), and in addition the geometries, specifically for Coppermine (.25, .18?). If you haven't guessed, I have a problem with the assertions that the test group name(s) imply. PB

To: Jim McMannis who wrote (55718)	4/16/1999 4:13:00 PM
From: RDM	Read Replies (1) \| Respond to of 1578938

My guess for Q2 1999 AMD results: Speed Grade Unit ASP Revenue K6-2 450 800,000 $115.00 $92,000,000 K6-2 400 1,000,000 $80.00 $80,000,000 K6-2 380 500,000 $70.00 $35,000,000 K6-2 366 600,000 $70.00 $42,000,000 K6-2 350 800,000 $60.00 $48,000,000 K6-2 below 350 100,000 $50.00 $5,000,000 K6-III 400 200,000 $160.00 $32,000,000 K6-IIII 450 100,000 $300.00 $30,000,000 K6 Mobile 700,000 $120.00 $84,000,000 K6-2-475 100,000 $150.00 $15,000,000 K6-2-500 50,000 $220.00 $11,000,000 K7-500 10,000 $400.00 $4,000,000 K7-550 10,000 $450.00 $4,500,000 Results 4,970,000 $97.08 $482,500,000

To: Jim McMannis who wrote (55718)	4/16/1999 4:16:00 PM
From: Pravin Kamdar	Read Replies (1) \| Respond to of 1578938

Jim, If cycles refers to the number of cycles to complete the given number of instructions, it looks like the Coppermine kicks the K7's butt in the uncompress and expresso benchmarks, is about even in the fft, and the K7 rules on the alternative. If the K7 can't complete an fft in fewer cycles than a Coppermine (when the K7 is supposed to be far superior in double precision floating point), the K7 is in big trouble. But then again, I'm not really sure how to interpret the numbers you posted. Pravin.

To: Jim McMannis who wrote (55718)	4/16/1999 4:28:00 PM
From: kash johal	Read Replies (1) \| Respond to of 1578938

Jim, Re: K7 benchmarks vs Coppermine These are the log results from a CPU simulator - the MXS program. The simulator allows you to model different CPU architectures and then run common benchmarks. The students can even "design" their own CPU etc. So the results are "normalized" to a given number of clock cycles. The CPU models are pretty simplistic but are meant to show how chip architeture effects the end results. In terms of actual relevance to the "REAL" Coppermine/K7 battle I suspect that these numbers are relatively meaningless. Regards, Kash J.

To: Jim McMannis who wrote (55718)	4/16/1999 4:29:00 PM
From: Tenchusatsu	Read Replies (2) \| Respond to of 1578938

Jim, I don't think those scores that you posted can be valid scores. It's really hard for me to believe that super-confidential performance data from AMD and Intel would appear in a EE 300-level class. Anyway, for the moment, let's assume that the benchmarks are genuine. IPC stands for "Instructions Per Clock". The higher the IPC, the more performance you'll get at the same clock speed. L1 and L2 Miss Ratio is a measure of how well the caches do their job. Larger caches mean lower miss ratios, which mean fewer memory accesses have to go out to a slower memory subsystem. From your post, it seems that the IPC for the K7 and for Coppermine vary widely according to benchmark: IPC for: uncompress expresso fft Alt. Bench. K7: 1.413 1.104 1.143 1.257 CuMine: 2.066 1.758 1.133 0.806 This shows that for equivalent clock speeds, Coppermine has better performance on two benchmarks, equivalent performance on one, and lower performance on the "Alternative Benchmark." The L1 and L2 miss ratios, however, is no surprise. Coppermine consistently shows slightly higher miss ratios in both the L1 and L2 caches. This is because the L1 cache on CuMine (32K or 64K?) is smaller than the oversized 128K L1 cache of K7, and the 256K L2 cache on CuMine is also bound to be smaller than the minimum L2 cache size (512K) for K7. The miss ratios, in general, are very small because it seems that the benchmarks are very small programs themselves and thus can easily fit itself and most of the data it works on in the L1 and L2 caches. So if this document is the real thing, that means Coppermine will beat the K7 in some benchmarks, and fall behind in others. Of course, the veracity of this document depends on whether you trust inside information to come from a Stanford undergrad EE class. ;-) Tenchusatsu

To: Jim McMannis who wrote (55718)	4/16/1999 4:31:00 PM
From: Scumbria	Read Replies (1) \| Respond to of 1578938

Jim, The numbers appear to come from a simulation. The benchmarks they used were SPEC which fit into the smaller L1 cache of Coppermine, allowing for Coppermine to have almost the same cache hit rate as K7. Without knowing any of the details of their simulation it is hard to draw any conclusions. Some of the benchmarks show K7 being faster, and some show Coppermine being faster. I wouldn't try to infer anything from this because it is probably just some junior class project trying to attract attention. I might have to go across the street and whoop on some of those Stanford boys to clean up their act. (Maybe I'll do lunch with Chelsea too.) Scumbria