elliot@xenna.encore.com (Elliot Mednick) (04/10/91)
Ok, so we have beaten to death the differences between a 386SX and a 386DX. What I would like to know is what to the big attraction of a DX which seems to cost $200 more (without cache). What does the 32-bit interface buy you? Certainly not CPU performance. I did some checking and got these SI (5.0) numbers: CPU SI (5.0) Comments --- -------- -------- 286-16 12.3 My trusty 286 386SX-20 12.8 A friend's system; claims that sometimes it is 15 386DX-20 12.4 C&T, AMI, 2MB 80ns, interleaved, no wait states, no cache 386DX-25 15.5 Not the same system as above, but same coniguration I was under the impression that DX boards (without cache) really scream. I have heard claims of up to 24 SI, w/out cache. What am I missing? Either 1) I've tested out cheap DX boards, or 2) I'm comparing SI 5.0 to SI 4.5. Someone please set me straight before I shell out $$$ for a DX. (No, I am not running Unix, but I am running Windows [because I have to]). __ Elliot Mednick (elliot@encore.com) | This .signature file is undergoing Encore Computer Corp. | remodeling for your convenience. Marlborough, MA. 01752 | Please pardon our appearance.
fdq@athena.mit.edu (Fred D Quintana) (04/10/91)
I just bought a 386 motherboard from ATronics International. Here are the specs: 40Mhz up to 8 megs on board 64k cache AMI Bios (dated 2/1/91) SIS chipset 6 16 bit slots, 2 8 bit slots AMD 386-DXL CPU (Not Intel!) Cost: $825 This board screams! Norton SI (5.0) reports 42.1! I would strongly recommend this board over a 386DX-25. It only costs perhaps $100-200 dollars more and it is fast! Fred -- ------------------------------------------------------------------------------- | Fred D. Quintana \ Never put off till tomorrow what | | Internet: <fdq@athena.mit.edu> \ you can avoid all together. | | UUCP: mit-eddie!mit-athena!fdq \ | -------------------------------------------------------------------------------
david@kessner.denver.co.us (David Kessner) (04/10/91)
There is a common misconception that 386's are faster than 286's, and that 486's are a lot faster than 386's, etc, etc, etc... When running MS-DOS software the 286, 386sx, and 386DX are all about the same speed (for the same clock speed). Cached machines will generally outperform non-cached machines (by 20-30% in 25mhz machines). So why even bother with a 386sx-- why not stick with the 286? Here is the story... If only MS-DOS software is used, a 286 is the best choice. A 386sx will give you better memory managment, as well as multitasking and enhanced-386 in Windows. It will run normal 386 software, but slower than a 386DX. A 386DX is a good chioce because they dont make a 25mhz and 33mhz SX. It will also run 386 software signifigantly faster than an SX at the same clock speed. A 486 will give you floating point performance, and about 30-50% faster overall when compared to a DX of equal clock speed. So. If all you are going to do is run Windows, and you can live with 16-20mhz CPU's, then an SX is perfect. If you are thinking that this performance curve (286=386sx=386dx) is all wrong then you are right! It's not that the CPU's are wrong, but that MS-DOS is wrong-- since it places signifigant limitations on the program running. On the average, a program that is re-compiled to run in 386-protected mode will run twice as fast on the same machine! Running 386-protected mode software, the line-up is more like: A 386dx is two-three times faster than an equivalent 286 program. 386sx is about 30-40% slower than a 386dx. a 486 is TWICE as fast as a 386dx, 3-4 times faster in floating point. The bottom line is that is PERFORMANCE is what you want-- then MS-DOS should not be on your wish list. -- David Kessner - david@kessner.denver.co.us | do { 1135 Fairfax, Denver CO 80220 (303) 377-1801 (p.m.) | . . . If you cant flame MS-DOS, who can you flame? | } while( jones);
c60b-1eq@e260-3e.berkeley.edu (Noam Mendelson) (04/10/91)
In article <14534@encore.Encore.COM> elliot@encore.com writes: >Ok, so we have beaten to death the differences between a 386SX and a 386DX. >What I would like to know is what to the big attraction of a DX which seems >to cost $200 more (without cache). What does the 32-bit interface buy you? >Certainly not CPU performance. I did some checking and got these SI (5.0) >numbers: >CPU SI (5.0) Comments >--- -------- -------- >286-16 12.3 My trusty 286 >386SX-20 12.8 A friend's system; claims that sometimes it is 15 >386DX-20 12.4 C&T, AMI, 2MB 80ns, interleaved, no wait states, no cache >386DX-25 15.5 Not the same system as above, but same coniguration First of all--don't post Norton SI results as serious statistics. Norton SI is a very vague measurement of overall system performance, not the CPU's performance. If you're interested, a 20 MHz Model 70 tests 21.5 on Norton 4.0 SI. >I was under the impression that DX boards (without cache) really scream. >I have heard claims of up to 24 SI, w/out cache. What am I missing? >Either 1) I've tested out cheap DX boards, or 2) I'm comparing SI 5.0 >to SI 4.5. Someone please set me straight before I shell out $$$ for a DX. That depends on what you use your system for. But if all you want is raw power, get a 386/25 or 386/33. I would also recommend an internal cache of 64K. +==========================================================================+ | Noam Mendelson ..!agate!ucbvax!web!c60b-1eq | "I haven't lost my mind, | | c60b-1eq@web.Berkeley.EDU | it's backed up on tape | | University of California at Berkeley | somewhere." |
tmurphy%peruvian.utah.edu@cs.utah.edu (Thomas Murphy) (04/10/91)
A dx can but you a couple things....one is speed but this is a relative thing and hard to measure as it depends on memory, disc, and application. I think that for most people and even businesses that an SX will really fill the bill and with the money you save you can buy the math chip or actually have a head start on it....if you are really going to crunch numbers jump to a 486 as it is cheaper than the 386/387 combo and faster too... If you are contemplating going to os-2 or unix you really want 32 bits or just want to toss money but I say an sx for most with plenty of memory and a cookin hard-drive....it is really more the limiter of your system. murph
pjh@mccc.edu (Pete Holsberg) (04/10/91)
In article <14534@encore.Encore.COM> elliot@encore.com writes:
=Ok, so we have beaten to death the differences between a 386SX and a 386DX.
=What I would like to know is what to the big attraction of a DX which seems
=to cost $200 more (without cache). What does the 32-bit interface buy you?
=Certainly not CPU performance. I did some checking and got these SI (5.0)
=numbers:
Using SI to evaluate the performance of a computer is like using gas
mileage to evaluate the performance of a car!
Pete
--
Prof. Peter J. Holsberg Mercer County Community College
Voice: 609-586-4800 Engineering Technology, Computers and Math
UUCP:...!princeton!mccc!pjh 1200 Old Trenton Road, Trenton, NJ 08690
Internet: pjh@mccc.edu Trenton Computer Festival -- 4/20-21/91
jerry@gumby.Altos.COM (Jerry Gardner) (04/11/91)
In article <1991Apr9.235733.11567@kessner.denver.co.us> david@kessner.denver.co.us (David Kessner) writes: >If you cant flame MS-DOS, who can you flame? | } while( jones); ^^^^ People who can't spell??? :) -- Jerry Gardner, NJ6A Altos Computer Systems UUCP: {sun|pyramid|sco|amdahl|uunet}!altos!jerry 2641 Orchard Parkway Internet: jerry@altos.com San Jose, CA 95134 Help stamp out vi in our lifetime. (408) 432-6200
tinyguy@cs.mcgill.ca (Yeo-Hoon BAE) (04/11/91)
In article <1991Apr10.160439.17824@mccc.edu> pjh@mccc.edu (Pete Holsberg) writes: > >Using SI to evaluate the performance of a computer is like using gas >mileage to evaluate the performance of a car! > >Pete >-- What exactly is a Landmark? It's probably another stupid meaningless performance measurer, but it seems to give much favour to 486 machines (relative to 386). -TG
ben@val.com (Ben Thornton) (04/12/91)
c60b-1eq@e260-3e.berkeley.edu (Noam Mendelson) writes: >First of all--don't post Norton SI results as serious statistics. Norton SI >is a very vague measurement of overall system performance, not the CPU's >performance. It is also misleading to consider only raw CPU performance. After all it is the system as a whole that matters when you shop for a machine. Norton's SI is just a number, but is is the only measurement of overall performance that is popular enough to have statistical significance. >+==========================================================================+ >| Noam Mendelson ..!agate!ucbvax!web!c60b-1eq | "I haven't lost my mind, | >| c60b-1eq@web.Berkeley.EDU | it's backed up on tape | >| University of California at Berkeley | somewhere." | -- Ben Thornton packet: wd5hls@wd5hls.ampr.org Video Associates Internet: ben@val.com Austin, TX uucp: ...!cs.utexas.edu!val!ben What's the moral of the story?
c60b-1eq@web-1g.berkeley.edu (Noam Mendelson) (04/12/91)
In article <1991Apr11.204600.15805@val.com> ben@val.com (Ben Thornton) writes: >c60b-1eq@e260-3e.berkeley.edu (Noam Mendelson) writes: >>First of all--don't post Norton SI results as serious statistics. Norton SI >>is a very vague measurement of overall system performance, not the CPU's >>performance. >It is also misleading to consider only raw CPU performance. After >all it is the system as a whole that matters when you shop for a machine. >Norton's SI is just a number, but is is the only measurement of overall >performance that is popular enough to have statistical significance. True, but this thread of messages was concerning CPU's, not complete systems. -- +==========================================================================+ | Noam Mendelson ..!agate!ucbvax!web!c60b-1eq | "I haven't lost my mind, | | c60b-1eq@web.Berkeley.EDU | it's backed up on tape | | University of California at Berkeley | somewhere." |
lbr@holos0.uucp (Len Reed) (04/12/91)
In article <14534@encore.Encore.COM> elliot@encore.com writes: >Ok, so we have beaten to death the differences between a 386SX and a 386DX. >What I would like to know is what to the big attraction of a DX which seems >to cost $200 more (without cache). What does the 32-bit interface buy you? >Certainly not CPU performance. I did some checking and got these SI (5.0) >(No, I am not running Unix, but I am running Windows [because I have to]). So what? You're running *all* 16-bit software and you want to know why having a a 32-bit data path doesn't buy you much? Windows puts the machine into proctected mode, but it doesn't use 32 bit mode. As to the cache, it makes a huge difference even under DOS. I have a 33 Mhz DX with 64K cache. It runs *far* faster than lower power machines. -- Len Reed Holos Software, Inc. Voice: (404) 496-1358 UUCP: ...!gatech!holos0!lbr
andreww@uniwa.uwa.oz (Andrew John Williams) (04/13/91)
elliot@xenna.encore.com (Elliot Mednick) writes: >to cost $200 more (without cache). What does the 32-bit interface buy you? >Certainly not CPU performance. I did some checking and got these SI (5.0) >numbers: >CPU SI (5.0) Comments >--- -------- -------- >286-16 12.3 My trusty 286 >386SX-20 12.8 A friend's system; claims that sometimes it is 15 >386DX-20 12.4 C&T, AMI, 2MB 80ns, interleaved, no wait states, no cache >386DX-25 15.5 Not the same system as above, but same coniguration >I was under the impression that DX boards (without cache) really scream. >I have heard claims of up to 24 SI, w/out cache. What am I missing? >Either 1) I've tested out cheap DX boards, or 2) I'm comparing SI 5.0 >__ >Elliot Mednick (elliot@encore.com) | This .signature file is undergoing >Encore Computer Corp. | remodeling for your convenience. >Marlborough, MA. 01752 | Please pardon our appearance. I have a 386DX/25, with an AMI bios and a C&T chipset, 80ns ram, 0WS. Not sure if its interleaved (dont want to reboot in the middle of reading news!) and SI gives me a Computing Index of 26.3. This is using SI version 4.5. I think older versions of SI give very unrealistic numbers- my system is certainly not 26.3 times as fast as an XT for most things- more like 12 or so. I would guess the 386 numbers you have are using the new version of SI, assuming it has been improved (I have never tried it). Andrew Williams Physics Department, University of Western Australia. andrew@earwax.pd.uwa.oz.au OR andreww@uniwa.uwa.oz.au (note extra w).
nbeck@weber.ucsd.edu (Nathaniel Beck) (04/15/91)
My recollection (and it is often wrong) is that SX chips come in 16mh and 20mh varieties. It is hard for me to see why anyone who would settle for that clock speed would demand DX. On the higher end, some friends who were thinking of buying 386 25mh machines (with 387) found that the 486 25mh machines were a much better buy (just a few hundred dollars more for a 50% performance boost). So it may be the case that the only 386 systems that will sell are SX's, or else the price of DX's will drop enough to make the SX irrelevant. I guess Intel would like to make AMD's 386 venture an unpleasant one. Neal -- Neal Beck Dept of Politcal Science, UCSD beck@ucsd.edu Dislaimer: The Regents pay me (a bit!) to distribute my opinions.
ralphs@seattleu.edu (Ralph Sims) (04/15/91)
nbeck@weber.ucsd.edu (Nathaniel Beck) writes: > I guess Intel would like to make AMD's 386 venture an unpleasant > one. Since AMD now has a 386/40mHz chip out, things will really start to heat up. -- halcyon!ralphs@seattleu.edu The 23:00 News and Mail Service - +1 206 292 9048 - Seattle, WA USA +++ A Waffle Iron, Model 1.64 +++
whitney@reed.cs.unlv.edu (Lee Whitney) (04/22/91)
In article <1991Apr12.162615.13529@holos0.uucp>, lbr@holos0.uucp (Len Reed) writes:
)
) So what? You're running *all* 16-bit software and you want to know
) why having a a 32-bit data path doesn't buy you much? Windows puts
) the machine into proctected mode, but it doesn't use 32 bit mode.
)
Your statement is impratical and misleading. First, the 386SX DOES have a 32 bit data path or else it would not be a 386! It's bandwidth is only limited across the bus, and since 90% of 386 buses are 16 bit (ISA), most DX systems would not realize a tremendous advantage over SX's when using 32-bit protected mode software.
A $200 cache may be worth a 25-30% performance increase, but the 5-10% increase of a DX over an SX with a 16 bit bus is probably not.
jones@acsu.buffalo.edu (terry a jones) (04/23/91)
In article <1991Apr21.203004.9909@unlv.edu> whitney@reed.cs.unlv.edu (Lee Whitney) writes: >In article <1991Apr12.162615.13529@holos0.uucp>, lbr@holos0.uucp (Len Reed) writes: >) >) So what? You're running *all* 16-bit software and you want to know >) why having a a 32-bit data path doesn't buy you much? Windows puts >) the machine into proctected mode, but it doesn't use 32 bit mode. >) >Your statement is impratical and misleading. First, the 386SX DOES have >a 32 bit data path or else it would not be a 386! It's bandwidth is only >limited across the bus, and since 90% of 386 buses are 16 bit (ISA), most DX >systems would not realize a tremendous advantage over SX's when using >32-bit protected mode software. > >A $200 cache may be worth a 25-30% performance increase, but the 5-10% >increase of a DX over an SX with a 16 bit bus is probably not. You are also confusing the issue somewhat. The 80386DX system will no doubt have a 32bit wide memory bus. Granted, the ISA bus is only 16 bit, but that is a real good reason not to hang memory out there in the form of an expansion card. Put it all on the motherboard if at all possible. This memory bus is separate and very different than the ISA bus. In the case of the 80386SX, the processor's memory accesses are 16 bits wide, period. So the memory bandwidth of the DX is much greater than that of the SX without even considering the cache benefits. It really is a matter of cost, if you have the money to spend, and are concerned about performance or are going to run Unix, etc., go with the DX. If you are on a budget, go with the SX. The SX system will be cheaper since the cost of the CPU is lower, and the cost of the motherboard will be cheaper since they only have to implement the 16bit wide memory bus. Terry Terry Jones {rutgers,uunet}!acsu.buffalo.edu!jones SUNY at Buffalo ECE Dept. or: rutgers!ub!jones, jones@acsu.buffalo.edu -- Terry Jones {rutgers,uunet}!acsu.buffalo.edu!jones SUNY at Buffalo ECE Dept. or: rutgers!ub!jones, jones@acsu.buffalo.edu
david@kessner.denver.co.us (David Kessner) (04/23/91)
In article <1991Apr21.203004.9909@unlv.edu> whitney@reed.cs.unlv.edu (Lee Whitney) writes: >In article <1991Apr12.162615.13529@holos0.uucp>, lbr@holos0.uucp (Len Reed) writes: >) >) So what? You're running *all* 16-bit software and you want to know >) why having a a 32-bit data path doesn't buy you much? Windows puts >) the machine into proctected mode, but it doesn't use 32 bit mode. >) >Your statement is impratical and misleading. First, the 386SX DOES have a 32 bit data path or else it would not be a 386! It's bandwidth is only limited across the bus, and since 90% of 386 buses are 16 bit (ISA), most DX systems would not realize a tremendous advantage over SX's when using 32-bit protected mode software. > >A $200 cache may be worth a 25-30% performance increase, but the 5-10% increase of a DX over an SX with a 16 bit bus is probably not. Ahhh... No. The 386SX has an INTERNAL 32-bit data path, but a 16 bit external path. It also has 32-bit INTERNAL Address paths, but 24 bit EXTERNAL-- meaning that it cannot address more than 16 meg of RAM... The ISA bus has no baring on the performance of the CPU. On ALL 386 (DX or SX) systems the memory is not accessed over the ISA bus, but rather has it's own 'memory bus'. The 16-bit nature of the ISA bus is just one reason-- the 8mhz standard bus speed is the other. So, since we are talking CPU speed rather than I/O speed, we'll move on... There are two very important issues here: A) The difference between 'MS-DOS Programs' and Protected mode programs, and B) The 16/32 bit EXTERNAL buss's. If all you are doing is running 'normal MS-DOS' Programs then you are correct in saying that a DX is about 10% faster than an SX... But if you were running more 32-bit software, I'd expect closer to 70-90% improvement since the majority of your 'fetches' are 32 bits. So... If you are performance minded-- use a DX and run 32-bit, protected mode software. If you are budget minded, get an SX and run blah 'normal' software. Also, if you notice... Most of the 20mhz DX's are or have been discontinued, so the choice is clearer. 16 or 20mhz machines are SX's, and 25-40mhz are DX's. If all you are running is 'normal' programs, the biggest speed boost will be from clock speed... -- David Kessner - david@kessner.denver.co.us | do { 1135 Fairfax, Denver CO 80220 (303) 377-1801 (p.m.) | . . . If you cant flame MS-DOS, who can you flame? | } while( jones);
davidsen@sixhub.UUCP (Wm E. Davidsen Jr) (04/24/91)
In article <1991Apr23.000348.6644@kessner.denver.co.us> david@kessner.denver.co.us (David Kessner) writes: | The ISA bus has no baring on the performance of the CPU. On ALL 386 (DX or SX) | systems the memory is not accessed over the ISA bus, but rather has it's own | 'memory bus'. There are a large number of SX motherboards which only have 8MB on the board. This is fine for DOS, but falls somewhat short for UNIX. The expansion memory is then put on the ISA bus, with one total w/s at any practical speed. People do run memory on the ISA bus, and cache makes a huge difference. Also, all the early DX boards had a 4MB limit on the motherboard. This system is running 8MB on the ISA bus, and it may be slow but it beats paging! -- bill davidsen - davidsen@sixhub.uucp (uunet!crdgw1!sixhub!davidsen) sysop *IX BBS and Public Access UNIX moderator of comp.binaries.ibm.pc and 80386 mailing list "Stupidity, like virtue, is its own reward" -me
david@kessner.denver.co.us (David Kessner) (04/25/91)
In article <3787@sixhub.UUCP> davidsen@sixhub.UUCP (bill davidsen) writes: >In article <1991Apr23.000348.6644@kessner.denver.co.us> david@kessner.denver.co.us (David Kessner) writes: > There are a large number of SX motherboards which only have 8MB on the >board. This is fine for DOS, but falls somewhat short for UNIX. The >expansion memory is then put on the ISA bus, with one total w/s at any >practical speed. > > People do run memory on the ISA bus, and cache makes a huge >difference. > > Also, all the early DX boards had a 4MB limit on the motherboard. This >system is running 8MB on the ISA bus, and it may be slow but it beats >paging! >-- >bill davidsen - davidsen@sixhub.uucp (uunet!crdgw1!sixhub!davidsen) > sysop *IX BBS and Public Access UNIX > moderator of comp.binaries.ibm.pc and 80386 mailing list >"Stupidity, like virtue, is its own reward" -me I would not reccomend running UNIX on an SX-- simply because of the memory constraints. 8 meg is good for most things, but Motif or Openlook (Opengook?) will page. If a _cheap_ UNIX machine is the only priority then an SX is it, but performance will greatly suffer. I know some run RAM off of the ISA bus. If that is the only choice, then so be it-- otherwise it is stupid. Most DX Motherboards have a 'proprietary RAM' slot enabling them to have large amounts of RAM that wouldent ordinarily fit on the motherboard. Current 486 motherboards hold anywhere from 16 meg to 96 meg ON THE MOTHERBOARD. The SX boards are targeted tward the lower-end market, where MS-DOS dominates. In addition, the SX CPU itself is targeted in the same fashon, where 16-bit applicaions are the norm (and 32-bit applications are used, but not common). In this market, 8 meg of RAM is more than enough. -- David Kessner - david@kessner.denver.co.us | do { 1135 Fairfax, Denver CO 80220 (303) 377-1801 (p.m.) | . . . If you cant flame MS-DOS, who can you flame? | } while( jones);
snow@drycas.club.cc.cmu.edu (04/25/91)
In article <1991Apr21.203004.9909@unlv.edu>, whitney@reed.cs.unlv.edu (Lee Whitney) writes: > A $200 cache may be worth a 25-30% performance increase, but the 5-10% > increase of a DX over an SX with a 16 bit bus is probably not. Another issue to consider when deciding between a DX and an SX is the cost of a floating-point coprocessor. When I bought my current motherboard, I could afford an 80386DX only or an 80386SX/80387SX combination. I wasn't sure at the time that I would need the coprocessor, but I'm glad I chose it, since I've made heavy use of it these past few months. Even now, I would not be able to afford an 80387DX (spare cash has been going for memory.) -ken
koch@motcid.UUCP (Clifton Koch) (04/25/91)
From article <1991Apr21.203004.9909@unlv.edu>, by whitney@reed.cs.unlv.edu (Lee Whitney):
-> In article <1991Apr12.162615.13529@holos0.uucp>, lbr@holos0.uucp (Len Reed) writes:
-> )
-> ) So what? You're running *all* 16-bit software and you want to know
-> ) why having a a 32-bit data path doesn't buy you much? Windows puts
-> ) the machine into proctected mode, but it doesn't use 32 bit mode.
-> )
-> Your statement is impratical and misleading. First, the 386SX DOES have a 32 bit data path or else it would not be a 386! It's bandwidth is only limited across the bus, and since 90% of 386 buses are 16 bit (ISA), most DX systems would not realize a tremendous advantage over SX's when using 32-bit protected mode software.
->
-> A $200 cache may be worth a 25-30% performance increase, but the 5-10% increase of a DX over an SX with a 16 bit bus is probably not.
The 386SX has an *internal* 32 bit bus. Externally it's only 16 bit.
You get more than a 5-10% increase. The system memory where all your code you're running does not
reside on the crippled ISA bus. It doesn't make much of difference when you're doing memory intensive
work on cards in the expansion slots (i.e. video). The main reason I see for a DX over an SX is that
the DX's are available at higher clock speeds, especially now that the AMD 40 MHz versions are out.
--
-----------------------------------------------------------------------------
... [uunet | mcdchg | gatech]!motcid!koch
c60b-1eq@e260-1a.berkeley.edu (Noam Mendelson) (04/26/91)
In article <1991Apr24.234848.175@drycas.club.cc.cmu.edu> snow@drycas.club.cc.cmu.edu writes: >In article <1991Apr21.203004.9909@unlv.edu>, whitney@reed.cs.unlv.edu (Lee Whitney) writes: >> A $200 cache may be worth a 25-30% performance increase, but the 5-10% >> increase of a DX over an SX with a 16 bit bus is probably not. >Another issue to consider when deciding between a DX and an SX is the cost >of a floating-point coprocessor. When I bought my current motherboard, >I could afford an 80386DX only or an 80386SX/80387SX combination. I wasn't >sure at the time that I would need the coprocessor, but I'm glad I chose it, >since I've made heavy use of it these past few months. Even now, I would >not be able to afford an 80387DX (spare cash has been going for memory.) It's not practical to buy a 386DX/387 combination. A better buy would be a 486, with its built in math coprocessor. -- +==========================================================================+ | Noam Mendelson ..!ucbvax!web!c60b-1eq | "I haven't lost my mind, | | c60b-1eq@web.Berkeley.EDU | it's backed up on tape | | University of California at Berkeley | somewhere." |
jay@gdx.UUCP (Jay A. Snyder) (05/04/91)
whitney@reed.cs.unlv.edu (Lee Whitney) writes: >Your statement is impratical and misleading. First, the 386SX DOES have a 32 bit data path or else it would not be a 386! It's bandwidth is only limited across the bus, and since 90% of 386 buses are 16 bit (ISA), most DX systems would not realize a tremendous advantage over SX's when using 32-bit protected mode software. > >A $200 cache may be worth a 25-30% performance increase, but the 5-10% increase of a DX over an SX with a 16 bit bus is probably not. > the peripheral bus is not the concern. The memory bus is. The 386sx only has a 16bit path to the memory. ============================================================================== Jay A. Snyder "Let Me Up! I've had enough" wa3wbu!gdx!jay@uunet.uu.net GDX-BBS (717) 737-3249 2400/1200/300bd uunet!wa3wbu!gdx!jay Unix and MSDOS File areas + Xenix executables
burkey@duncan.cs.utk.edu (Michael Burkey) (05/07/91)
In response to the letter to which you responded. A 386SX does NOT have a 32 bit bus to memory, it is 16 bit just like the 286 is. From an external standpoint, this thing is virtually a 286. Internally it is 32 bit and acts like a 386DX. Intel is currently working on plans (AMD also is for that matter) of coming out with a 386SX that is pin compatible with a 286. Can we say good-bye 286?? M Burkey
cyang%peruvian.utah.edu@cs.utah.edu (Cheng Yang) (05/07/91)
In article <1991May6.170707.27182@cs.utk.edu> burkey@duncan.cs.utk.edu (Michael Burkey) writes: >In response to the letter to which you responded. > >A 386SX does NOT have a 32 bit bus to memory, it is 16 bit just like >the 286 is. From an external standpoint, this thing is virtually a 286. >Internally it is 32 bit and acts like a 386DX. Intel is currently working >on plans (AMD also is for that matter) of coming out with a 386SX that is pin >compatible with a 286. Can we say good-bye 286?? Do you mean it will be possible to just pull out our 286s from motherboards and plug in 386SX? That will be very interesting! It seems that I can save my 286 motherboard! ***************************************************************************** * Ho capito, signor,si!........ * * * *Cheng Yang Home Address: * *Department of Computer Science 808 East, 300 South, #3 * *University of Utah Salt Lake City, UT 84102 * *E-Mail Address:cyang@peruvian.utah.edu Tel: 801-322-1918 * *****************************************************************************
davidsen@sixhub.UUCP (Wm E. Davidsen Jr) (05/08/91)
In article <1991May7.093840.23795@hellgate.utah.edu> cyang%peruvian.utah.edu@cs.utah.edu (Cheng Yang) writes: | Do you mean it will be possible to just pull out our 286s from motherboards | and plug in 386SX? That will be very interesting! It seems that I can save | my 286 motherboard! But do you want to? Unless you have a 16 or 20 HMz 286 you are going to be running a lot slower. BTW: I don't believe this rumor for an instant. If Intel intended to do this they would have done it long ago and killed AMD's 286 business dead. I believe that Intel will use every line they have to make 386 and 486, and adjust the price to sell them. Unless they could sell the chip for as much as a 386 they would lose money, becuase it's not likely to be any cheaper to make. -- bill davidsen - davidsen@sixhub.uucp (uunet!crdgw1!sixhub!davidsen) sysop *IX BBS and Public Access UNIX moderator of comp.binaries.ibm.pc and 80386 mailing list "Stupidity, like virtue, is its own reward" -me
gtephx (Wild Rider) (05/15/91)
sorry, tried email, but (as usual) it bounced. here's the message: Date: Mon, 13 May 91 10:13:39 MDT From: robertsw (Wild Rider) Message-Id: <9105131613.AA06063@gtephx.com> To: gtephx!asuvax!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!uakari.primate.wisc.edu!sdd.hp.com!hplabs!hpfcso!hpfcdj!hpfcjbv!jbv Subject: Re: Why buy a DX over an SX? Newsgroups: comp.sys.ibm.pc.hardware In-Reply-To: <34390002@hpfcjbv.HP.COM> References: <14534@encore.Encore.COM> Organization: AG (formerly GTE) Comm. Sys., Phx, AZ Cc: In article <34390002@hpfcjbv.HP.COM> you write: >/ hpfcjbv:comp.sys.ibm.pc.hardware / fdq@athena.mit.edu (Fred D Quintana) >/ 3:57 pm Apr 9, 1991 / >I just bought a 386 motherboard from ATronics International. Here are the >specs: > > 40Mhz > up to 8 megs on board > 64k cache > AMI Bios (dated 2/1/91) > SIS chipset > 6 16 bit slots, 2 8 bit slots > AMD 386-DXL CPU (Not Intel!) > Cost: $825 > >This board screams! Norton SI (5.0) reports 42.1! > >I would strongly recommend this board over a 386DX-25. It only costs perhaps >$100-200 dollars more and it is fast! > >Fred > ok, where did you get this "mother of all motherboards" :-) from? i mean, post the phone number, address, all that good stuff so others can order it, too. will it work in a standard ibm pc-xt chassis? how much memory did it come with? detail, details, pull-lease! >-- >------------------------------------------------------------------------------- >| Fred D. Quintana \ Never put off till tomorrow what | >| Internet: <fdq@athena.mit.edu> \ you can avoid all together. | >| UUCP: mit-eddie!mit-athena!fdq \ | >------------------------------------------------------------------------------- >---------- cheers, wr (wild rider) -- Wallace Roberts, AG (formerly GTE) Communication Systems, Phoenix, AZ UUCP: ...!{ncar!noao!asuvax | uunet!zardoz!hrc | att}!gtephx!robertsw Internet: gtephx!robertsw@asuvax.eas.asu.edu Bike: '82 GS1100L Suz voice: (602)581-4555 fax: (602)582-7624 Cage: '89 Mustang GT