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: Scumbria who wrote (36371)	8/26/1998 7:09:00 PM
From: Petz	Read Replies (2) \| Respond to of 1574182

Scumbra, re: K6-3 improved performance You say K6-3 will have minimal (10%) performance improvement because K6-2 systems are already benchmarked with large L2 caches. Moving 256K of the L2 onto the CPU die (K6-3)will have minimal (<10%)impact on performance. This may be true at 300 MHz, but the K6-3 starts at 350 MHz where an L2 cache speedup of 3.5:1 will make a BIG DIFFERENCE. The reason for this becomes clear if you analyze the benchmark results at tomshardware.com The K6-2's Winstone 98 number only increases by 6% when the clock goes from 300 to 350 MHz with the L2 speed kept constant at 100 MHz. This is an increase of 16.7% in the CPU speed. Clearly the L2 cache, and to a lesser extent, main memory, disk and video performance, is inhibiting the performance increase of the K6-2. For the Celeron-A the benchmark improves by 8.3%. For Pentium II, notice that a jump of only 14.3% in CPU clock speed (350 to 400 MHz) allows the benchmark to improve by 8.8% -- the benchmark 'keeps' 62% of the CPU speed increase. CPU.........SpeedIncrease.......WinstoneIncrease......%B/A K6-2........300->350=16.7%......5.8%.........................35% Mendocino 300->350=16.6%......8.3%.........................50% PentiumII 350->400=14.3%......8.8%.........................62% I would expect the K6-3 to be faster than the K6-2 at 300 MHz by about 10%, as you say. But its benchmarks will scale with clock speed much better than the K6-2 with its fixed L2 cache speed, in fact a reasonable estimate is that 56% of the clock speed improvement will apply. This puts the K6-3 benchmark estimates at: K6-3-300 (underclocked chip) 25.7 x 1.1 = 28.3 K6-3-350 28.3(1+0.5650/300) = 30.9 (Pentium II-400 is 30.8) K6-3-400 28.3(1+0.56100/300)= 33.5 (higher than Pentium II-504) Petz