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: Maxwell who wrote (41567)	11/16/1998 5:38:00 PM
From: Scumbria	Read Replies (2) \| Respond to of 1572605

Maxwell, It is a 5 year old architecture. The branch prediction and decoders of P6 are ancient. You are trying to compare 1993 technology to 1999 technology. All modern microprocessors are starved for data running real world apps. Until dram latencies dramatically improve, the value of one architecture vs. another is limited to the entertainment of CPU architects. Soon we will be running CPU's with a cycle time of 1ns, in systems with average dram latencies exceeding 50ns. This imbalance turns all CPU architects into little more than academics. Unless anyone really cares about synthetic benchmarks which fit in the L1. Scumbria

To: Maxwell who wrote (41567)	11/16/1998 5:46:00 PM
From: Tenchusatsu	Read Replies (1) \| Respond to of 1572605

<So you think the P6 architecture is that good? It is a 5 year old architecture. The branch prediction and decoders of P6 are ancient. You are trying to compare 1993 technology to 1999 technology.> First of all, don't you mean 1996 technology to 1999 technology? After all, I don't remember seeing Pentium Pro machines back in 1993. Second, as for branch prediction, the only thing out there that would make the P6's branch predictor look "ancient" is a hybrid branch predictor. I don't think AMD ever mentioned having a hybrid branch predictor in any of their K7 foils. And third, yes I do think the P6 architecture is that good. Like I said before, from what I've seen in the K7 foils, AMD is pursuing a "Ford Mustang" strategy behind the K7 design. There's nothing particularly new in the K7 that wasn't featured in the P6. AMD is just adding more of the stuff that makes the P6 so great, i.e. more buffers, more cache, more execution units, bigger tables, wider-is-better, etc. And did you take a look at the transistor count? 22 million, compared to the Pentium II's 7.5 million. Geez, that's comparable to the Mustang's oversized 5.0-liter engine! As for the comment regarding "1996 vs. 1999 technology," well, I'll let Paul answer that one. Tenchusatsu