Author: Tom Kerrigan
Date: 14:59:03 06/13/00
Go up one level in this thread
On June 13, 2000 at 17:27:45, Torstein Hall wrote: >On June 13, 2000 at 17:01:36, Tom Kerrigan wrote: > >>On June 13, 2000 at 15:55:24, Mogens Larsen wrote: >> >>>On June 13, 2000 at 15:41:51, Tom Kerrigan wrote: >>> >>>>No, Thunderbird is a few % faster than an Athlon at same MHz. >>>> >>>>Duron is exactly the same as Tbird, but with 1/4 the L2 cache (64k). >>>> >>>>-Tom >>> >>>I've just purchased an ordinary Athlon. What are the advantages and >>>disadvantages with a smaller cache size. BTW, are the L2 cache of the >>>Thunderbird and the Athlon fullspeed? >>> >>>Best wishes... >>>Mogens >> >>The Tbird and Duron both have on-die "full speed" L2 caches, but they are not >>that much faster than the old off-die caches. Most benchmark scores improve by a >>few percent, but nothing like when Intel moved the PIII cache on-die. >> >>The main benefit of the on-die cache is that it makes the processors much >>cheaper to manufacture. Another, less significant benefit is that the cache >>performance can now scale with the processor performance. >> >>-Tom > >Why was it so much more important for the PIII to get the cach on-die? Primarily because the PIII has a small L1 cache--16k or 32k, I can't remember which. But it relies heavily on the L2 cache, as opposed to the Athlon, which has a 128k L1 cache. Another thing is that Intel redesigned the PIII's L2 cache when they put it on-die, so the bandwidth is higher and the latencies are lower. AMD simply took their cache chips and splatted them on the processor die, so there's no reason for the on-die cache to be much better than the off-die cache. -Tom
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.