 SU.HARDW.PC.MOTHERBOARD (2:5020/299)  SU.HARDW.PC.MOTHERBOARD 
 From : Aleksandr Konosevich                2:5004/9.7      Tue 27 Sep 94 12:51
 Subj : .  Cx486DX40? (.  #$%)


⠪, ᭮  Cx486DX/DX2 :

1) ⨬p ᮯpp p⨢ ⥫  ⮩  ⠪⮢
( 10%-25% p業⮢ p)

2) p 㬭⥫ (BYTE/WORD/DWORD MUL/IMUL  3/3/7 ⠪⮢

3) , 䨣p騩  write-back  write-through (, 
᪠  write-back  "饫"  ﭨ p CCR2). p⨢
RESET', p,   ... :(

4) 祭/몫祭 write burst mode  p p⭮⥩

5) 楫 ⥫ Green PC Features ( ⮬᪮ 몫祭 ᮯp
p ⢨  p  饩 p䬥⨪  Advanced SMM).
 ᫥ p Cyrix  ᮡ, .. 8  p権 
- ᯮ p p 稢襣 :( RES4 (LOADALL 486) ...

---
 * Origin:  aleks@sibkom.omsk.su  (2:5004/9.7)
 SU.HARDW.PC.MOTHERBOARD (2:5020/299)  SU.HARDW.PC.MOTHERBOARD 
 From : Aleksandr Konosevich                2:5004/9        Thu 25 May 95 13:15
 Subj : thomposon (cyrix) 486dx2-50


 05 May 1995 18:41, egor nicolaev wrote in a message to All:

 en>     .  祬 ⠪ 襢?

⠪,  p  Cyrix Cx486DX/DX2 (   業 ⠬ SGS Thompson)

p⢠ p i486/Am486 :

1) p  (8 )   ᪮䨣pp   write-back
2) 祭  ⢮ write buffers (4 -> 8)
3) p 㬭⥫  楫᫥ p䬥⨪ (byte*byte, word*word ->
   3 clock count, dword*dword -> 7 clock count)
4) pp p⠥  㤢 ᥣ ( ⫨稥  i486, ⪫饣 2x
   p cache misses)
5)  p p ᮯpp
6) pp p⠥ p쭮    襬 ᫥ , 祬  i486
7) ⮬᪮ ⪫祭 ⠭  ᮯp p ⢨ p饭  FPU
8) p p SMM ( ⥫ - ⮫쪮  SL-p)
9)  ᮢ⨬  p᪨ software

H⪨ :

1) p 믮 p権  設⢥  ᮢ  
    i486/Am486.

2)  "㣠"  SysInfo (. p) --+
                                                  |
                                                  V
------------------------------------------------------------------------------

     Dear Aleksander,

     The Norton benchmark is very inaccurate.  YOu will find that the
     figure of 113.5 is correct for the Cyrix CPU.  This of course does NOT
     mean that our chip is 20% slower.  The Norton benchmark uses only a
     very small number of instructions in a very tight loop of code.  Since
     our CPU was designed from scratch, 35% of instructions run about the
     same speed as Intel, 35% run faster and 30% slower than Intel.  It
     just so happens that the benchmark for Norton uses a mix of
     instructions that we are slower on.

     This is well demonstrated by the fact that if you use Norton SI
     version 6 with a 66MHz Cyrix CPU it says it is an Intel running at
     76MHz - it computes the clock speed by timing a known loop of
     instructions and then working out the equivalent clock speed.  This
     loop of code is predominantly made up of instructions which the Cyrix
     CPU runs faster, since the program doesn't specifically recognise the
     Cyrix CPU, it presumes it is an Intel running at what it thinks is
     76MHz!

     Hope this explains the situation,

     Regards


     Peter Farrow.

-----------------------------------------------------------------------------

3) 襢 ⮨  p  i486/Am486 ... |)))))

 : ᨦ  Cyrix   ᮦ.    Cx486DX2-V-80 or V-100 MHz ...

                          With best wishes, Aleksandr

P.S. , 襢 -  ᪮쪮 ? ;)
---
 * Origin:  eshslabs@univer.omsk.su  (2:5004/9)
