home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.ibm.pc.hardware
- Path: sparky!uunet!snorkelwacker.mit.edu!thunder.mcrcim.mcgill.edu!homer.cs.mcgill.ca!storm
- From: storm@cs.mcgill.ca (Marc WANDSCHNEIDER)
- Subject: Re: 80X86 Architecture Differences
- Message-ID: <1992Sep15.151734.9373@cs.mcgill.ca>
- Sender: news@cs.mcgill.ca (Netnews Administrator)
- Organization: McGill University, Montreal, Canada
- References: <01GOSKLYJ07S000FY4@DRYCAS.CLUB.CC.CMU.EDU>
- Date: Tue, 15 Sep 1992 15:17:34 GMT
- Lines: 89
-
- In article <01GOSKLYJ07S000FY4@DRYCAS.CLUB.CC.CMU.EDU> MORRO@DRYCAS.CLUB.CC.CMU.EDU ("John K. Duchowski") writes:
- >Hello Everyone,
- >
- > Could someone please explain to me in clear, plain English the
- >differences between architecture of different 80X86 chips ? I have a bet
- >with a friend of mine who says that in the 80486 the math coprocessor
- >is built in ONLY in the chips with clock speeds of 33 MHz and above. This
- >seems odd to me as then what would be the difference between a 25 MHz
- >80386 (not SX) and 80486 ? I would very much appreciate it if the
- >explanation would also include few words on the SX, DX, etc variants.
- >Thank you.
-
- 8086: 16 bit internal registers, 16bit external bus... crappy chip with
- crappy instruction set. Requires 8087 for floating point
-
- 8088: 8 bit external bus, 16bit internal registers, same otherwise -- still
- crappy. Requires 8087 for floating point
-
- 80186: Not quite sure, but it's supposedly got a few more instructions than
- the 8086.
-
- 80286: 16bit chip with full funcitonality of the 8086 + newer 16bit
- protected mode with 24bit addressing for total address space of
- 16MB. Requires 80287 for floating point. Available in 6, 8, 10, 12
- 16, 20, and 25Mhz clock speeds (not exhaustive list)
-
- 80386SX:32bit internal registers, 16bit data bus, 24bit addressing to outside
- world meaning 16MB of addressable memory. Each 32bit word access
- takes two reads. Well suited to 16bit nightmares like DOS.
- Instructions take quite a few cycles to execute, and 2 to be issued.
- Requires 387SX for Floating point. Currently available in 16,
- 20, 25, and 33Mhz versions.
-
- 80386SLC: Intels laptop chip. Much the same as the SX with static registers
- and other 'innovative' power saving features. Don't really know
- much about it.
-
- 80386DX:32bit internal registers, 32bit addressing, 32bit external data bus.
- Otherwise, has the same instruction set as the SX, requies 387DX for
- floating point activity. No caching or buffering really (I believe
- there is some sort of 8byte instruction cache or something like that,
- but nothing to start creaming over.) Available in 16, 20, 25,
- 33 and 40Mhz versions.
-
- 80486SX:A marketing move to kill the 386. It's a 486 chip without the floating
- point unit enabled. Otherwise it's the same as the 486DX. Available
- currently in clock speeds of 16, 20, and 25Mhz
-
- 80486DX:One step up from the 386. Includes floating point unit ('487') built
- in, and has an 8k read cache (NOTE: It is NOT write back, which means
- that systems are still going to need an external cache)
- Instructions take 1 cycle to issue, and typically take a little less
- time than the 386 to execute (although there are exceptions)
- Generally a lot faster. I've heard some rumors of a 5 stage pipeline
- in this machine, but I have never verified it. Currently available
- in 25, 33, and 50Mhz versions.
-
-
- 80486DX2:Basically, this chip has a clock speed with which it communicates
- with the rest of the computer, and an internal clock rate which is
- twice as fast. Although touted as truly innovative by Intel, this
- has been used by Motorola for quite some time now, as well as other
- companies. Ie. a 50Mhz DX2 will have a bus speed of 25Mhz but
- will work internally at 50Mhz.
-
- P5: Intels latest chip coming out this year. Intel claims that it will
- do 100MIPS (ie 1 instruction 1 = NOP) and the list of features told
- by people are wildly variant.
- Most rumors point to it having a (finally) useful register set without
- any stupid dependancies (which the 386/486 have in Protected mode
- only) and some serious integer and floating point hardware, as well
- as certain superscalar features. Not quite sure how realistic all
- this is though.
- The name has also been varying wildly from 586 to "the ULTIMATE CISC
- Chip..."
-
-
- If I have goofed, please don't hesitate to correct me or add something.
-
-
- Toodlepip!
- Marc 'em.
-
-
- --
- storm@cs.mcgill.ca McGill University C program run. Run
- Marc Wandschneider Montreal, CANADA Program run! PLEASE!
- "When 900 years old you reach, look as good you will not." - Yoda
- "Ow!" -J.F.K.~
-