L2 vs. L1 cache
Badboy5150
instead of expanding the amount of l1 cache? Could someone please
explain why intel chose this design method?
Gregory
Sent via Deja.com http://www.deja.com/
Before you buy.
Badboy5150
Roger D. Clark
educated guess. Tell me what you think, or maybe someone else out there
knows for sure. I would think it went a little like this: In a computer
long ago and far, far away L2 cache was on a seperate chip. Some bright
engineer realized that the closer a chip and it cache were, the faster it
ran. They decided to piggyback the L2 on the CPU for closeness so it could
be faster HOWEVER, your question I assume is since it found its way to
migrate ON the CPU and since L1 also resides there, then why have both?
right? The guesswork comes into play here. I suppose it is easier to vary
the amounts of L2 on a CPU than it is to redesign the whole chip to vary the
amount of L1. The assumption here is that L2 resides closer to the surface
of the chip and L1 is more at the heart of the CPU. Since cache seems to be
the really expensive part of the CPU structure (Pentium vs. Craperon I mean
Celeron is an example of that) I guess they could alter the price of the
chip by altering the amount of cache and since L2 may be easier to alter due
to its placement on the chip, the answer was to keep it seperate..... I
have know earthly idea if I am right, but I bet I could convince a lot of
people that it was so. (I don't bullcrap my way through life though so this
won't happen). What are the thoughts on this out there? L1 by the way is
somewhat faster (not a lot but somewhat) so the idea of why keep L1 at all
and just go with L2 wouldn't be a faster solution either.
Roger D. Clark
"Badboy5150" <bish...@my-deja.com> wrote in message
news:8a9622$6a8$1...@nnrp1.deja.com...
A. G.
Roger D. Clark <rcl...@fuse.net> wrote in message
news:scgc5v...@corp.supernews.com...
designed to use a certain amount of it. My guess is that it is easier to
add cheaper L2 cache to the chip design than to add L1 cache.
--
A. G.
Of making many books there is no end
and much study is a weariness of the flesh.
Ec. 12:12
aplu...@skillmarque.com
Some of the other 'best guess' (!!??) posts in the other reply thread
are sort-of right in some respects - like having the L2 cache nearer to
the CPU makes a difference.
The main reason for there being two caches is sort-of historic but
ultimately, the L1 cache operates at a faster speed (closer to that of
the core of the CPU and sometimes at the same speed), whereas the L2
cache speed is governed by the speed at which the motherboard memory bus
is running (slower than the internal CPU speed).
To illustrate, say you have a '450Mhz' processor: the motherboard clock
speed will probably be 100Mhz, with the processor running its core via a
4.5x multiplier circuit. The Internal (L1) cache will operate from the
internal clock (although not necessarily at the full 450Mhz) whereas the
L2 cache will be tied to the external 100Mhz clock.
In essence, the Internal L1 cache is optimised for data transfers
between the CPU core logic and registers and the L2 cache matches the
timing of the main system RAM.
It would be ideal to have ALL cache memory running at the internal CPU
speed (or part thereof) but to bring this timing out on to the
motherboard raises a number of technical issues.
There's a good article about cache at:
http://cpusite.examedia.nl/docs/cache.html
Regards,
Nigel Kendrick
Ron Young
Pentium Pro and Pentium II the L2 cache resides in the package not the cpu
itself. The Pentium Pro runs its L2 at the same speed as the cpu but the
Pentium II only works at 1/2 the cpu speed so by putting the L2 on board it
could run faster yet continue to be used by the motherboard chipset to tag
the varying system ram.
Badboy5150 <bish...@my-deja.com> wrote in article
<8a95vc$61g$1...@nnrp1.deja.com>...
> I don't understand why intel chose to implement l2 cache on the chip
> instead of expanding the amount of l1 cache? Could someone please
> explain why intel chose this design method?
>
> Gregory
>
>