daly@ecs.umass.edu (Bryon Daly, ECE dept, UMass, Amherst) (04/22/91)
In article <1991Apr22.155803.8093@tandem.com>, ernest@pegasus.dsg.tandem.com (Ernest Hua) writes: > In today's Wall Street Journal (B6 col 3)... (is this a *@#!! misprint?) > > ... the 486DX [sells for] $581 and up ... > > ... the 486SX is $258 apiece ... Intel has disabled ... the floating point > unit ... > > ... the 487SX ... will list for $799 ... > > <Stuff deleted> > > $258 or $1057 for the CPU > > depending upon your tastes. Who would be stupid enough to do the more complex > design?!!! I saw an add already in the latest either PC Mag or Byte for a 486 SX system from ?ALR? (I can't remember for sure and can't find the ad now) I thought it was some kind of April fool's joke (April issue). My question is Who would buy a White Elephant like a 486SX? The margin between 486's and 386's is disappearing rapidly as it is, and someone would have to be brain dead to pay $200 more for a "487 SX" coprocessor than the actual 486 costs, including the coprocessor. -Bryon daly@ecs.umass.edu
ernest@pegasus.dsg.tandem.com (Ernest Hua) (04/22/91)
In today's Wall Street Journal (B6 col 3)... (is this a *@#!! misprint?)
... the 486DX [sells for] $581 and up ...
... the 486SX is $258 apiece ... Intel has disabled ... the floating point
unit ...
... the 487SX ... will list for $799 ...
Now simple arithmetic shows that a simple design with a 486DX will cost
$581 for the CPU
while a complicated design (allowing for the presence or absence of the 487SX)
will cost
$258 or $1057 for the CPU
depending upon your tastes. Who would be stupid enough to do the more complex
design?!!!
--
Ernest Hua, Associate Design Engineer ernest@tandem.com
Tandem Computers, 19333 Vallco Parkway, Cupertino, CA 95014 408-285-5580mju@mudos.ann-arbor.mi.us (Marc Unangst) (04/24/91)
daly@ecs.umass.edu (Bryon Daly, ECE dept, UMass, Amherst) writes: > Who would buy a White Elephant like a 486SX? The margin between 486's and > 386's is disappearing rapidly as it is, and someone would have to be brain de Intel is targeting the 486SX at people who would otherwise buy a 386 machine. A lot of people have no use for the built-in FPU in the 486DX, since their applications do little or no floating-point math. (Remember that the FPU only helps out with floating-point stuff or complex math like trigonometry; the CPU still does integer math itself.) A 486SX will be MUCH faster than a 386DX running at the same clock speed, since the 486SX has an instruction prefetch cache, and a lot of the instructions have been speeded up (in terms of clock cycles). I know that if I had the choice between a 386DX and a 486SX machine, I'd take the 486SX machine any day. I don't think Intel expects to sell many 487SX coprocessors -- if you need a FPU, you buy the 486DX. Hmm...Since the WSJ article says that the 487SX will supposedly be just a slightly-modified 486DX and will "take over all the functionality" of the existing 486SX, and since the 486SX will be a DX with the FPU circuits disabled, what's to prevent you from pulling your 486SX and just replacing it with a 486DX? Since the 487SX is more expensive than the 486DX, there seems to be no reason why ANYBODY would buy a 487SX. -- Marc Unangst | mju@mudos.ann-arbor.mi.us | "Bus error: passengers dumped" ...!hela!mudos!mju |
rkl@cbnewsh.att.com (kevin.laux) (04/25/91)
In article <HDuV11w164w@mudos.ann-arbor.mi.us>, mju@mudos.ann-arbor.mi.us (Marc Unangst) writes: > I don't think Intel expects to sell many 487SX coprocessors -- if you > need a FPU, you buy the 486DX. > > Hmm...Since the WSJ article says that the 487SX will supposedly be > just a slightly-modified 486DX and will "take over all the > functionality" of the existing 486SX, and since the 486SX will be a DX > with the FPU circuits disabled, what's to prevent you from pulling > your 486SX and just replacing it with a 486DX? Since the 487SX is > more expensive than the 486DX, there seems to be no reason why ANYBODY > would buy a 487SX. Um...Correct me if I'm wrong, but isn't one of the differences between DX and SX that the DX can do 32-bit accesses while the SX can do only 16-bit (like the 386DX vs the 386SX)? This would mean you couldn't just replace the 486SX with the 486DX to get the FPU functionality. Also, what about the speed difference (486SX runs at 20 MHz). And lastly, there is also the difference in power consumption. -- ________________________________________________________________________________ R. Kevin Laux Email: rkl1@hound.att.com AT&T Bell Labs Voice: (908) 949-1160 Holmdel, NJ 07733 Fax: (908) 949-0959
pjh@mccc.edu (Pete Holsberg) (04/26/91)
In article <HDuV11w164w@mudos.ann-arbor.mi.us> mju@mudos.ann-arbor.mi.us (Marc Unangst) writes:
=Intel is targeting the 486SX at people who would otherwise buy a 386
=machine. A lot of people have no use for the built-in FPU in the
=486DX, since their applications do little or no floating-point math.
=(Remember that the FPU only helps out with floating-point stuff or
=complex math like trigonometry; the CPU still does integer math
=itself.) A 486SX will be MUCH faster than a 386DX running at the same
=clock speed, since the 486SX has an instruction prefetch cache, and a
=lot of the instructions have been speeded up (in terms of clock
=cycles). I know that if I had the choice between a 386DX and a 486SX
=machine, I'd take the 486SX machine any day.
Does this SX have a 32 or 16 bit data bus?
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 -- 4/20-21/91c60b-1eq@e260-1a.berkeley.edu (Noam Mendelson) (04/26/91)
In article <1991Apr25.132715.1945@cbnewsh.att.com> rkl@cbnewsh.att.com (kevin.laux) writes: > Um...Correct me if I'm wrong, but isn't one of the differences >between DX and SX that the DX can do 32-bit accesses while the SX can do >only 16-bit (like the 386DX vs the 386SX)? This would mean you couldn't >just replace the 486SX with the 486DX to get the FPU functionality. Also, >what about the speed difference (486SX runs at 20 MHz). And lastly, there >is also the difference in power consumption. You're wrong. But Intel's naming strategy is completely inconsistent. The 386SX is a 386DX with a 16-bit external data path. The 486SX is a regular 32-bit 486 sans math coprocessor. To be consistent, Intel should've named the 386SX a 388. Or at least they could be consistent with their 'SX' label. -- +==========================================================================+ | 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." |
jgay@digi.lonestar.org (john gay) (04/26/91)
From article <1991Apr26.002911.1793@agate.berkeley.edu>, by c60b-1eq@e260-1a.berkeley.edu (Noam Mendelson): > In article <1991Apr25.132715.1945@cbnewsh.att.com> rkl@cbnewsh.att.com (kevin.laux) writes: >> Um...Correct me if I'm wrong, but isn't one of the differences >>between DX and SX that the DX can do 32-bit accesses while the SX can do >>only 16-bit (like the 386DX vs the 386SX)? This would mean you couldn't >>just replace the 486SX with the 486DX to get the FPU functionality. Also, > You're wrong. But Intel's naming strategy is completely inconsistent. > The 386SX is a 386DX with a 16-bit external data path. The 486SX is > a regular 32-bit 486 sans math coprocessor. > To be consistent, Intel should've named the 386SX a 388. Or at least > they could be consistent with their 'SX' label. I think intel is being consistent with SX in their labeling. You just have to figure out what their labeling means. Internally SX is really Sux so what you have to figure is that a 386SX is a 386 that SuX somehow - with the 386SX they crippled the external bus path. The 486SX SuX in that the math co-processor is either disabled or just not there (conflicting reports on that one). Other than that (from what I have read) Noam is correct, although I also believe part of what kevin says - the part about not being able to directly replace a 486SX with a 486DX - I doubt that they are pin compatible since intel is coming out with a 487SX for the 486SX and they have to talk to each other somehow. john gay.
ntaib@silver.ucs.indiana.edu (Nur Iskandar Taib) (04/28/91)
Speaking of the 487SX... here is something quite weird. The latest IBM products circular mentions a 486SX machine. Forget what its called, but IBM mentions an upgrade to the 487SX. Now comes the weird part. To get the 487SX upgrade, IBM wants the 486SX chip returned!!!! This tells us that the 387SX is no coprocessor, it IS the microprocessor. In which case one would like to know what the difference between the 487SX and the 486DX (?) is! I think someone at IBM goofed big time... -- ------------------------------------------------------------------------------- Iskandar Taib | The only thing worse than Peach ala Internet: NTAIB@AQUA.UCS.INDIANA.EDU | Frog is Frog ala Peach Bitnet: NTAIB@IUBACS ! -------------------------------------------------------------------------------
mju@mudos.ann-arbor.mi.us (Marc Unangst) (04/28/91)
jgay@digi.lonestar.org (john gay) writes: > 386SX they crippled the external bus path. The 486SX SuX in that the > math co-processor is either disabled or just not there (conflicting reports o > that one). Other than that (from what I have read) Noam is correct, although I believe it's disabled, but probably "messed up" in some way. I have been hearing reports from inside Intel that the 486DX is so complex that only about 1 out of 7 chips actually makes it off the assembly line in full working order. I suppose the 486SX is really a clever marketing move by Intel to find something to do with all those 486DX chips with part of the FPU unit messed up. > since intel is coming out with a 487SX for the 486SX and they have to talk > to each other somehow. Not necessarily. If what the WSJ said is right, a 487SX will really just be a real 486DX, and the motherboard circuitry will disable the 486SX if a "487SX" is installed. If this is true, then it's amazing how a 487SX can cost more than a 486DX. -- Marc Unangst | mju@mudos.ann-arbor.mi.us | "Bus error: passengers dumped" ...!hela!mudos!mju |
rcollins@altos86.Altos.COM (Robert Collins) (04/29/91)
In article <HDuV11w164w@mudos.ann-arbor.mi.us> mju@mudos.ann-arbor.mi.us (Marc Unangst) writes: >daly@ecs.umass.edu (Bryon Daly, ECE dept, UMass, Amherst) writes: >itself.) A 486SX will be MUCH faster than a 386DX running at the same >clock speed, since the 486SX has an instruction prefetch cache, and a > But the '486SX is only available at 20Mhz at the moment, while AMD is getting ready to ship 40Mhz '386's. >Hmm...Since the WSJ article says that the 487SX will supposedly be >just a slightly-modified 486DX and will "take over all the >functionality" of the existing 486SX, and since the 486SX will be a DX >with the FPU circuits disabled, what's to prevent you from pulling >your 486SX and just replacing it with a 486DX? Since the 487SX is >more expensive than the 486DX, there seems to be no reason why ANYBODY >would buy a 487SX. The marketing buys at Intel saw to it that the '486SX and '487SX have a slightly different pin out than the '486DX. Therefore dropping in a DX is impossible. It is true that the '486SX is totally disabled by the installation of the '487SX. In fact, you can remove it all together when the '487 is installed. This means you can actually buy the motherboard with nothing but the '487 in it. Through the use of some logic, it is possible have a single socket that can be used for the '486, '486SX, and '487SX. -- "Worship the Lord your God, and serve him only." Mat. 4:10 Robert Collins UUCP: ...!sun!altos86!rcollins HOME: (408) 225-8002 WORK: (408) 432-6200 x4356
ajai@sce.carleton.ca (Ajai Sehgal) (04/30/91)
I got some info from a friend @ Intel on this one, although it seemd a bit strange. The 486SX uses the same mask as the 486DX with the coprocessor connection traces cut. I don't know about the pinout but one would assume that Intel wouldn't make the DX and SX pin compatable. The suggested list price for a 486SX and a 487SX combined is not less expensive than the cost of a 486DX. One has to ask, why would anyone who wants an FPU buy the SX version? You get the same thing with the added overhead of interchip communication.
rcollins@altos86.Altos.COM (Robert Collins) (04/30/91)
In article <1991Apr27.194311.15636@bronze.ucs.indiana.edu> ntaib@silver.ucs.indiana.edu (Nur Iskandar Taib) writes: >Speaking of the 487SX... here is something quite weird. The latest >IBM products circular mentions a 486SX machine. Forget what its >called, but IBM mentions an upgrade to the 487SX. Now comes the >weird part. To get the 487SX upgrade, IBM wants the 486SX chip >returned!!!! This tells us that the 387SX is no coprocessor, it >IS the microprocessor. In which case one would like to know what >the difference between the 487SX and the 486DX (?) is! > >I think someone at IBM goofed big time... > No, Intel goofed. The '487SX is actually a '486DX with a few pins switched position. So, it is possible to remove the '486SX when the '487 is installed. Intel did this so people couldn't plug in a '486DX, and up the clock speed to 33Mhz. They are trying to gouge the public for the cost of the 487 upgrade -- which costs more than a '486DX! When you say that this "tells us that the 387SX is no coprocessor" I think you meant to say '487SX. In which case, you are totally correct, it is no coprocessor... it is the main processor. -- "Worship the Lord your God, and serve him only." Mat. 4:10 Robert Collins UUCP: ...!sun!altos86!rcollins HOME: (408) 225-8002 WORK: (408) 432-6200 x4356
jgay@digi.lonestar.org (john gay) (04/30/91)
From article <1991Apr27.194311.15636@bronze.ucs.indiana.edu>, by ntaib@silver.ucs.indiana.edu (Nur Iskandar Taib): > Speaking of the 487SX... here is something quite weird. The latest > IBM products circular mentions a 486SX machine. Forget what its > called, but IBM mentions an upgrade to the 487SX. Now comes the > weird part. To get the 487SX upgrade, IBM wants the 486SX chip > returned!!!! This tells us that the 387SX is no coprocessor, it > IS the microprocessor. In which case one would like to know what > the difference between the 487SX and the 486DX (?) is! > > I think someone at IBM goofed big time... According to the latest (4/29) EE Times the 487SX IS a 486DX with different pinouts. The 487SX is supposed to disable the 486SX and then become the main processor/co-processor. The 487SX has different pinouts and a higher price (and of course lower speed - 20 MHz) than the 486DX to keep people from just selling 487SXs as processors. john gay.
daly@ecs.umass.edu (Bryon Daly, ECE dept, UMass, Amherst) (05/01/91)
In article <ajai.673019831@castor.sce.carleton.ca>, ajai@sce.carleton.ca (Ajai Sehgal) writes: > I got some info from a friend @ Intel on this one, although it seemd a bit > strange. The 486SX uses the same mask as the 486DX with the coprocessor > connection traces cut. I don't know about the pinout but one would assume that > Intel wouldn't make the DX and SX pin compatable. The suggested list price From what I've heard, they're not. > for a 486SX and a 487SX combined is not less expensive than the cost of a > 486DX. One has to ask, why would anyone who wants an FPU buy the SX version? They wouldn't, at least not with Intel's current pricing. It's just there for people concerned about adding one later on, but don't need one now. > You get the same thing with the added overhead of interchip communication. Well, there shouldn't be any interchip communication, since the 487SX takes over all duties of the 486SX (which means that a 487SX is basically a 20MHz 486 DX, probably in a different package than the 486DX, for a grossly inflated price.) Rumor says that IBM wants you to return the 486SX if you get the 487SX! (In their 486SX machine) Which brings me back to my original question (sort of): Who is this chip aimed at? The low-end 386 crowd? (16, 20MHz SX; 20, 25 MHz DX) The high-end 386 crowd (33, 40 MHz 386 DX)? The 486DX crowd? At 20 MHz, the 486SX admittedly will probably out-perform a 25MHz 386DX (or less) at integer stuff (lets not drag in the FPU mess). But it does not have much of a chance against a 33 MHz 386 (Oooh! I can feel the flames already!) [My 486-25 (DX) is only a little faster than my 386-33 was, by my own observations, benchmark programs I've run, etc. This is borne out by magazine reviews/benchmarks.] So I'd take a 386DX-33 over a 486SX-20, especially if I wanted to get a coprocessor later. So, is the 486SX being marketed against the 386-33? If so, they're going to need a pretty good price to match the 386-33's price-performance ratio, especially with AMD gearing up production of its clone 386's, which is bound to drive down the 386 prices. I think that Intel has laid another marketing egg with the introduction of the 486SX, and that they won't be very successful with it until it's price comes down (Just as what happened with the 386SX). There's a good article in the Wall Street Journal (April 22?) on the 486SX release that people might be interested in. -Bryon Daly daly@ecs.umass.edu
rkl@cbnewsh.att.com (kevin.laux) (05/01/91)
In article <13471.281da4bf@ecs.umass.edu>, daly@ecs.umass.edu (Bryon Daly, ECE dept, UMass, Amherst) writes: > So, is the 486SX being marketed against the 386-33? If so, they're going to > need a pretty good price to match the 386-33's price-performance ratio, > especially with AMD gearing up production of its clone 386's, which is bound to > drive down the 386 prices. I don't think so specifically; I think the point here is that the 486SX at 20 MHz will outperform a 386DX (at 25 MHz) and that Intel now is *not* the sole supplier (of 386's). The 486SX is being marketed as a cheaper alternative to the 486DX *and* Intel is the only one who makes it (or any 486's, for that matter). They are trying to get the 486(SX) to be the de facto standard of choice for PCs. Lots of bucks for Intel :-). -- ________________________________________________________________________________ R. Kevin Laux Email: rkl1@hound.att.com AT&T Bell Labs Voice: (908) 949-1160 Holmdel, NJ 07733 Fax: (908) 949-0959
tanida@forseti.css.gov (Tom Tanida) (05/02/91)
In article <ajai.673019831@castor.sce.carleton.ca>, ajai@sce (Ajai Sehgal) writes: >I got some info from a friend @ Intel on this one, although it seemd a bit >strange. The 486SX uses the same mask as the 486DX with the coprocessor >connection traces cut. I don't know about the pinout but one would assume that >Intel wouldn't make the DX and SX pin compatable. The suggested list price >for a 486SX and a 487SX combined is not less expensive than the cost of a >486DX. One has to ask, why would anyone who wants an FPU buy the SX version? >You get the same thing with the added overhead of interchip communication. Anybody who does want an FPU would buy the DX version. But the 486SX is (allegedly) faster than a 386 and is going to be released at the same price, so this would provide an alternative to getting a 386. If the 386SX is an indication, Intel's marketing strategy of providing chips for just about every level of PC computing is a good idea. (Then again, I could be biased because of the BusinessWeek article in the most recent issue :-) ). -Tom tanida@esosun.css.gov
phil@brahms.amd.com (Phil Ngai) (05/02/91)
daly@ecs.umass.edu (Bryon Daly, ECE dept, UMass, Amherst) writes: >I think that Intel has laid another marketing egg with the introduction of the >486SX, and that they won't be very successful with it until it's price comes >down (Just as what happened with the 386SX). Far be it from me to defend intel, but the 386sx was and is very successful and I don't see why the 486sx can't be. Prices can change instantly, and will to maximize the amount of money Intel gets for the 486sx. --
tester@cmcl2.nyu.edu (L Testerville) (05/02/91)
rkl@cbnewsh.att.com (kevin.laux) writes: >In article <13471.281da4bf@ecs.umass.edu>, daly@ecs.umass.edu writes: >> So, is the 486SX being marketed against the 386-33? If so, they're going to >> need a pretty good price to match the 386-33's price-performance ratio, >> especially with AMD gearing up production of its clone 386's,which is bound >> to drive down the 386 prices. > I don't think so specifically; I think the point here is that the >486SX at 20 MHz will outperform a 386DX (at 25 MHz) and that Intel now is >*not* the sole supplier (of 386's). The 486SX is being marketed as a cheaper >alternative to the 486DX *and* Intel is the only one who makes it (or any >486's, for that matter). They are trying to get the 486(SX) to be the >de facto standard of choice for PCs. Lots of bucks for Intel :-). Maybe it's just me, but it seems like Intel is trying to pull a MAJOR marketing scam on us. According to Infoworld, Intel will sell the 486SX-20 to OEMs for $258. Sounds good compared to AMD selling the AMD386DX-40 for $390. But, there's a catch. If you ever want to get a math coprocessor (487SX), you need to shell out an ADDITIONAL _$799_ (which is a technologically brain-damaged approach: the 487SX actually "takes over as [both] the CPU and the math coprocessor"). Now the 486-25 supposedly goes for $588, but for the lame user who opts for a 486SX-20; he/she will have to shell out a total of $1057 (OEM costs, mind you) for what amounts to a 486-_20_! Mind-boggling. Maybe Intel figures nobody will catch on to this plot?
crouse@eng.umd.edu (Gil Crouse) (05/02/91)
In article <1991May1.214432.15762@cmcl2.nyu.edu> tester@cmcl2.nyu.edu (L Testerville) writes: >Maybe it's just me, but it seems like Intel is trying to pull a MAJOR >marketing scam on us. According to Infoworld, Intel will sell the 486SX-20 >to OEMs for $258. Sounds good compared to AMD selling the AMD386DX-40 >for $390. But, there's a catch. If you ever want to get a math >coprocessor (487SX), you need to shell out an ADDITIONAL _$799_ (which >is a technologically brain-damaged approach: the 487SX actually "takes >over as [both] the CPU and the math coprocessor"). Now the 486-25 >supposedly goes for $588, but for the lame user who opts for a 486SX-20; >he/she will have to shell out a total of $1057 (OEM costs, mind you) for >what amounts to a 486-_20_! Mind-boggling. Maybe Intel figures nobody >will catch on to this plot? The big assumption your making is that every one will want to get the 487SX at some point. The math coprocessor market is a fraction of the size of the market for CPUs. For business application the coprocessor really doesn't offer much improvement in speed. Most (or at least many) PC users are perfectly happy with a fast CPU and no coprocessor. If that is the case, buying the 486SX saves the buyer $330. Gil Crouse crouse@eng.umd.edu
goldberg@iastate.edu (Goldberg Adam Michael) (05/02/91)
In <1991May1.233459.1062@eng.umd.edu> crouse@eng.umd.edu (Gil Crouse) writes: >In article <1991May1.214432.15762@cmcl2.nyu.edu> tester@cmcl2.nyu.edu (L Testerville) writes: >>Maybe it's just me, but it seems like Intel is trying to pull a MAJOR >>marketing scam on us. According to Infoworld, Intel will sell the 486SX-20 >>to OEMs for $258. Sounds good compared to AMD selling the AMD386DX-40 >>for $390. But, there's a catch. If you ever want to get a math >>coprocessor (487SX), you need to shell out an ADDITIONAL _$799_ (which >>is a technologically brain-damaged approach: the 487SX actually "takes >>over as [both] the CPU and the math coprocessor"). Now the 486-25 >>supposedly goes for $588, but for the lame user who opts for a 486SX-20; >>he/she will have to shell out a total of $1057 (OEM costs, mind you) for >>what amounts to a 486-_20_! Mind-boggling. Maybe Intel figures nobody >>will catch on to this plot? >The big assumption your making is that every one will want to get the >487SX at some point. The math coprocessor market is a fraction of the >size of the market for CPUs. For business application the coprocessor >really doesn't offer much improvement in speed. Most (or at least >many) PC users are perfectly happy with a fast CPU and no coprocessor. >If that is the case, buying the 486SX saves the buyer $330. Of course, the whole 486SX thing is a big scam for Intel. The difference between a 486SX & DX? The SX has the internal math processor (all 486DXs have the math co-processor built-in) disabled. So the 487SX is actually a 486DX in disguise. What it amounts to is Intel trying to scam us into believing that we're better off getting a 486SX for the same prices as an AMD 386-40. Personally I'd rather buy an inferior chip from AMD than a better one from Intel. Increased competition -> lower prices -> more competition. Just say No (to Intel). Adam --
rcollins@altos86.Altos.COM (Robert Collins) (05/02/91)
In article <ajai.673019831@castor.sce.carleton.ca> ajai@sce.carleton.ca (Ajai Sehgal) writes: >I got some info from a friend @ Intel on this one, although it seemd a bit >strange. The 486SX uses the same mask as the 486DX with the coprocessor >connection traces cut. I don't know about the pinout but one would assume that >Intel wouldn't make the DX and SX pin compatable. The suggested list price >for a 486SX and a 487SX combined is not less expensive than the cost of a >486DX. One has to ask, why would anyone who wants an FPU buy the SX version? >You get the same thing with the added overhead of interchip communication. This is only the third time I will post this, so sorry if I sound a bit frustrated. At the present time, the '486SX, '487SX, and '486DX are all from the same mask. In the '486SX, the FPU is disabled. Both the 486SX and 487SX have different pin outs with respect to each other, and the DX. So none of them are pin compatible with each other. Intel has distributed schematics for a motherboard based on the 486SX/487DX combination. When the 487SX is installed in the system, it is connected in a way that quite literally disables the 486SX. In other words, you have an expensive chunk of silicon doing nothings. In fact, when you have the '487SX installed you can remove the '486SX. In the future, Intel plans to reduce the mask to remove the FPU from the '486SX to get the cost down. Why would Intel do this? Why would the 486SX/487SX combination be more expensive that the DX? Why would the make them all pin incompatible with each other? They have $$$ in their eyes. The market is supposed to perceive the '487SX as a coprocessor. A coprocessor can obtain a much higher profit margin than a CPU. So, Intel is trying to push the '487SX as a coprocessor to the '486SX, when it really isn't a coprocessor. It's simply a '486DX in a different skin. But as a coprocessor, it can afford a higher market price than a CPU. (Remember, the '487SX electronically disables the '486SX when both are installed.) Why are they all pin incompatible? To keep the whiz kids from putting in a DX, upping the crystal, and having a 33Mhz '486 at a much cheaper cost. $$$ $$$ and more $$$. This is all a brainchild of marketing people. Even many of the engineers at Intel think it is a stupid idea, and deplore the deceptive marketing technique. The same holds true for many of the field representatives, they think it is a sleazy marketing practice. But, rest assured, if you have a '487SX, you have all of the benefits of having the FPU on the same mask of silicon as the CPU, because it is nothing more than a repinned '486DX. -- "Worship the Lord your God, and serve him only." Mat. 4:10 Robert Collins UUCP: ...!sun!altos86!rcollins HOME: (408) 225-8002 WORK: (408) 432-6200 x4356
tester@cmcl2.nyu.edu (L Testerville) (05/03/91)
goldberg@iastate.edu (Goldberg Adam Michael) writes: >In <1991May1.233459.1062@eng.umd.edu> crouse@eng.umd.edu (Gil Crouse) writes: >Of course, the whole 486SX thing is a big scam for Intel. The difference >between a 486SX & DX? The SX has the internal math processor (all 486DXs >have the math co-processor built-in) disabled. So the 487SX is actually >a 486DX in disguise. What it amounts to is Intel trying to scam us into >believing that we're better off getting a 486SX for the same prices as >an AMD 386-40. >Personally I'd rather buy an inferior chip from AMD than a better one from >Intel. Increased competition -> lower prices -> more competition. >Just say No (to Intel). I'm glad SOMEone has gotten to the point here...
phil@brahms.amd.com (Phil Ngai) (05/03/91)
goldberg@iastate.edu (Goldberg Adam Michael) writes: >The SX has the internal math processor (all 486DXs >have the math co-processor built-in) disabled. So the 487SX is actually >a 486DX in disguise. What it amounts to is Intel trying to scam us into >believing that we're better off getting a 486SX for the same prices as >an AMD 386-40. >Personally I'd rather buy an inferior chip from AMD than a better one from >Intel. Increased competition -> lower prices -> more competition. The current Infoworld has a comparison of 486sx and 386dx-40 systems and the results were that the 486sx is not much faster than the 386dx-40. Depending on the machines and the benchmark, the 386dx-40 is sometimes faster than the 486sx. You do have a good point. If you don't buy the Am386, you may not have a chance to buy the Am486. Bringing products like this to market takes an enormous amount of work and money and AMD has taken a big gamble on the Am386. If Intel succeeds in killing it with the 486sx, AMD may not have enough resources left to develop the Am486. (These are my opinions only, not that of the company, but of course I have a vested interest.) --
john@jwt.UUCP (John Temples) (05/03/91)
In article <280@altos86.Altos.COM> rcollins@altos86.UUCP (Robert Collins) writes: >The '487SX is actually a '486DX with a few pins switched position. Does this mean someone could make a simple adapter to allow using a DX in place of an 487? -- John W. Temples -- john@jwt.UUCP (uunet!jwt!john)
daly@ecs.umass.edu (Bryon Daly, ECE dept, UMass, Amherst) (05/03/91)
In article <1991May2.181631.13603@amd.com>, phil@brahms.amd.com (Phil Ngai) writes: > goldberg@iastate.edu (Goldberg Adam Michael) writes: >>The SX has the internal math processor (all 486DXs >>have the math co-processor built-in) disabled. So the 487SX is actually >>a 486DX in disguise. What it amounts to is Intel trying to scam us into >>believing that we're better off getting a 486SX for the same prices as >>an AMD 386-40. > >>Personally I'd rather buy an inferior chip from AMD than a better one from >>Intel. Increased competition -> lower prices -> more competition. > > The current Infoworld has a comparison of 486sx and 386dx-40 systems > and the results were that the 486sx is not much faster than the > 386dx-40. Depending on the machines and the benchmark, the 386dx-40 > is sometimes faster than the 486sx. I don't buy the 486SX being faster than the 386DX-40. IBM's 486SX marketing for their system claims the 486SX is CAN BE AS FAST AS a 386-33. This implies that it is nearly or is as fast as the 386-33 for most things, but is not faster. I can't see it coming close to a 386-40. Let's look at some math: Some magazine's real-world benchmarks on a bunch of 386-33 and 486-25 systems showed that the 486-25 was, on average, 10-15% faster than a 386-33. Now, reduce the speed of the 486 by 20% to make it a 486SX-20. You get 5-10% SLOWER than a 386-33. The Am386-40 has twice the clock rate of a 486SX, and it is a pretty weak claim to say that the 486 is twice as fast as a 386 at the same clock rate. Now, if adding a coprocessor is considered, the 486SX is even more of a loser, since it costs so much more to add the 487SX "co-processor" than it would to get add a 387. What is a 486SX? It's just a souped up 386DX, that Intel has slowed down to 20MHz, to be in direct competition with existing products (386-33, 386-40). What is gained by customers? Nothing. And if it hurts AMD's competition, we lose, too. Intel is using this marketing scam to try and squash it's competitor. I agree with an earlier poster who has said: JUST SAY NO (to Intel and the 486SX)! > > You do have a good point. If you don't buy the Am386, you may not have > a chance to buy the Am486. Bringing products like this to market takes > an enormous amount of work and money and AMD has taken a big gamble on > the Am386. If Intel succeeds in killing it with the 486sx, AMD may not > have enough resources left to develop the Am486. > > (These are my opinions only, not that of the company, but of course I > have a vested interest.) > > -- I have no vested interests in either AMD or Intel, except as a consumer wanting more innovation and less marketing BS. -Bryon Daly daly@ecs.umass.edu
plim@hpsgwp.sgp.hp.com (Peter Lim) (05/06/91)
/ phil@brahms.amd.com (Phil Ngai) / 2:16 am May 3, 1991 / writes: $ You do have a good point. If you don't buy the Am386, you may not have $ a chance to buy the Am486. Bringing products like this to market takes $ an enormous amount of work and money and AMD has taken a big gamble on $ the Am386. If Intel succeeds in killing it with the 486sx, AMD may not $ have enough resources left to develop the Am486. $ It might help if AMD lower the price and fulfil its perceived mission of freeing the world from Intel monopoly (which translate to CPU at more reasonable price). Of course, this might start a price war which AMD might not be able to live through (pure speculation here). But, on the other hand if Am386 didn't really offer any price performance advantage, then the world might not be really keen to buy from AMD and AMD might not sell enough Am386 to recoup R&D cost etc. etc. (again pure speculation here). So, which path is AMD going to choose ? $ (These are my opinions only, not that of the company, but of course I $ have a vested interest.) $ These are also my very own opinion. And I think so far, AMD doesn't have a sure fire formula. We are waiting......... Regards, ___o``\________________________________________________ ___ __ _ _ Peter Lim. V````\ @ @ . .. ... .- -> 76 MIPS at under US$20K !! --- -- - - /.------------------------------------------------ === == = = >--_// . .. ... .- -> 57 MIPS at under US$12K !! `' . If you guessed SUN, IBM or DEC, you are wrong ! E-mail: plim@hpsgwg.HP.COM Snail-mail: Hewlett Packard Singapore, Tel: (065)-279-2289 (ICDS, ICS) Telnet: 520-2289 1150 Depot Road, Singapore 0410. #include <standard_disclaimer.hpp>
burkey@duncan.cs.utk.edu (Michael Burkey) (05/07/91)
I hope that some of the motherboard manufacturers out there may read this. -- MAKE NO PROVISIONS ON YOUR BOARDS for the STUPID 487SX!!! -- design your boards to take either 486DX (25/33) or 486SX chips -- if the board is not designed for a brain-damaged 487SX, all you have to do is swap in a real 486 when you want the co-processor (for $200 cheaper) and you could easily build boards to be compatible at 20 or 25 Mhz. -- since the 486SX would be disabled anyway there is no reason a REAL 486 could not be installed in another socket!!! Mike Burkey (KILL INTEL!!!!) (C&T, AMD, & NexGen shall TRIUMPH!) (especially considering that C&T and AMD are both working on 486 clones)
whitney@lil-ed.cs.unlv.edu (Lee Whitney) (05/07/91)
) >>Maybe it's just me, but it seems like Intel is trying to pull a MAJOR ) >>marketing scam on us. ) ) ) Of course, the whole 486SX thing is a big scam for Intel. The difference ) between a 486SX & DX? The SX has the internal math processor (all 486DXs ) have the math co-processor built-in) disabled. So the 487SX is actually ) a 486DX in disguise. What it amounts to is Intel trying to scam us into ) believing that we're better off getting a 486SX for the same prices as ) an AMD 386-40. ) ) Personally I'd rather buy an inferior chip from AMD than a better one from ) Intel. Increased competition -> lower prices -> more competition. ) ) Just say No (to Intel). ) Hell Yes! Hopefully Intel will get stung hard and the public will reject the fact that they are purposely retarding technology and putting their own greed ahead of customers. I will now buy and recommend AMD over intel in any case where two chips are similar in price/performance. Just say 486SuX!
rcollins@altos86.Altos.COM (Robert Collins) (05/07/91)
In article <-WB+FNC@jwt.UUCP> john@jwt.UUCP (John Temples) writes: >In article <280@altos86.Altos.COM> rcollins@altos86.UUCP (Robert Collins) writes: >>The '487SX is actually a '486DX with a few pins switched position. > >Does this mean someone could make a simple adapter to allow using a DX >in place of an 487? YES! And that is exactly how we have our ICE-486 (In Circuit Emulator) attached to the SX socket. -- "Worship the Lord your God, and serve him only." Mat. 4:10 Robert Collins UUCP: ...!sun!altos86!rcollins HOME: (408) 225-8002 WORK: (408) 432-6200 x4356
phil@brahms.amd.com (Phil Ngai) (05/08/91)
plim@hpsgwp.sgp.hp.com (Peter Lim) writes: >It might help if AMD lower the price and fulfil its perceived mission >of freeing the world from Intel monopoly (which translate to CPU at What makes you think AMD is not doing this? How long do you think AMD can lose money before you don't have to worry about whether or not there is an Intel monopoly? (my opinions only) -- The enemy of my enemy is my friend.