🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Article View: comp.arch
Article #10741

Re: Any new instructions in a i486?

#10741
From: Wayne Schlitt
Date: Mon, 21 Dec 1992 21:31
49 lines
3012 bytes
In article <BzMEIJ.Anz@pix.com> stripes@pix.com (Josh Osborne) writes:
> In article <PCG.92Dec19162224@decb.aber.ac.uk> pcg@aber.ac.uk (Piercarlo Grandi) writes:
> >No, this cannot be the reason. As a rule the Cyrix 486DLC CPU chip is
> >sold with the Cyrix Fasmath (a 387) coprocessor, the combo being touted
> >quite reasonably as a 486DX equivalent.
> 
> I don't think this is the case (but I would love to be wrong...).  I beleve
> the #clocks each insn takes is more like the 386, then the i486.


to quote an article that was just posted last week to comp.sys.intel:

< Intel 486SX:
< execution unit:  RISC-like execution unit with five stage pipeline. Barrel
<                  shifter. Conditional jump taken/not taken: 3/1 clock cycles.
<                  Instructions that can be executed in 1 clock cycle if the
<                  destination is a register and the source is either a register
<                  or an immediate value:
<                  ADC,ADD,AND,BSWAP,CMP,DEC,INC,MOV,NEG,NOT,OR,POP,PUSH,SBB,
<                  SUB,TEST,XOR
< 
< Cyrix 486DLC
< 
< instruction set: complete Intel 486SX instruction set, including *all*
<                  486 specific instructions: WBINVD (write back and
<                  invalidate data cache), XADD (exchange and add), CMPXCHG
<                  (compare and exchange), BSWAP (Byte Swap), INVLPG
<                  (Invalidate TLB entry), INVD (Invalidate Data Cache)
< 
< execution unit:  RISC-like execution unit with five stage pipeline. Barrel
<                  shifter. 16x16 bit hardware multiplier (16x16 bit multiply:
<                  3 cycles, 32x32 bit multiply: 7 cycles, AAD: 4 cycles).
<                  Conditional jump taken/not taken: 6/1 clock cycles.
<                  Instructions that can be executed in 1 clock cycle
<                  if the destination is a register and the source is
<                  either a register or an immediate value:
<                  ADC,ADD,AND,CDQ,CLC,CLD,CMC,CMP,CWD,DEC,INC,MOV,MOVSX,
<                  NEG, NOT,OR,SBB,SHLD,SHRD,STC,STD,SUB,TEST,XOR

So, for most instructructions that the Intel 486 executes in 1 cycle,
the cyrix 486 does also.  There are some things that take longer,
there are some things that take less time.

I haven't seen any SPEC*92 info on the Cyrix, but it is my
understanding that they are about 10% slower than intel's parts at the
same clock speed.


-wayne
Technical Details
Message-ID: <WAYNE.92Dec21153140@backbone.uucp>
Path: arc1-ca.pugleaf.net!sparky!dsndata!backbone!backbone!wayne
References: <WAYNE.92Dec11164422@backbone.uucp> <9212130000.AA05447@iecc.cambridge.ma.us> <PCG.92Dec19162224@decb.aber.ac.uk> <BzMEIJ.Anz@pix.com>

Show Full Headers
Complete Article Headers
Newsgroups: comp.arch
Path: arc1-ca.pugleaf.net!sparky!dsndata!backbone!backbone!wayne
From: wayne@backbone.uucp (Wayne Schlitt)
Subject: Re: Any new instructions in a i486?
In-Reply-To: stripes@pix.com's message of Mon, 21 Dec 1992 17: 42:18 GMT
Message-ID: <WAYNE.92Dec21153140@backbone.uucp>
Sender: wayne@backbone (Wayne Schlitt)
Organization: The Backbone Cabal
References: <WAYNE.92Dec11164422@backbone.uucp> <9212130000.AA05447@iecc.cambridge.ma.us>
 <PCG.92Dec19162224@decb.aber.ac.uk> <BzMEIJ.Anz@pix.com>
Date: Mon, 21 Dec 1992 21:31:40 GMT