[rc5] V2 Speed Probs on x86?

Oscar Chang oscarc at divideby0.com
Mon Jun 30 03:33:33 EDT 1997


On Sun, 29 Jun 1997 16:30:55 -0500, Austin T. wrote:

>Hi, I'm new to the list so sorry if this has already been covered, but I've
>just been reading the huge gains from V1 to V2.  Well, my Pentium 133 was
>getting 122 kk/s and now it's getting a max of 154 kk/s, now granted I'm
>happy with the increase, but I'm just wondering why other CPU's such as
>below have jumped so incredibly much?

  The PPro and the Pentium clones (AMD and Cyrix) have bigger CPU caches
and what happens is that instead of going to external memory to get
instructions and memory, it uses those internal caches which runs at the
speed of the CPU (usually around 200mhz depending on the CPU). On the
other hand, our lowly pentiums :) have a smaller cache and probably access
memory more often (only runs at bus speed which is about 66mhz(?).) The
clones also tend to have better branch prediction, and if I'm not mistaken
(don't quote me on this) the AMD RISC-ifies the CISC instructions and then
runs them. That also gives it a boost in speed.
  If any of this is wrong, please correct me. :)

---
oscarc at divideby0.com
You won't be undefined when you Divide By Zero.
http://www.divideby0.com

----
To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in the body.



More information about the rc5 mailing list