hh2@prism.gatech.EDU (HAAS) (06/02/91)
In article <1991Jun2.041512.29546@leland.Stanford.EDU> fangchin@leland.Stanford.EDU (Chin Fang) writes: >In article <1991Jun2.030215.11584@unixland.natick.ma.us>, bill@unixland.natick.ma.us (Bill Heiser) writes: >|> In article <1991May31.183111.16505@ux1.cso.uiuc.edu> tmkk@uiuc.edu (K. Khan) writes: >|> > >|> >It means that Intel is scared silly of the industry's upcoming move to >|> >the MIPS RISC chips, and is going to milk us for every single cent it >|> >can while it is still able to do so. >|> >|> How does the i486 compare with the 20 - 30 MIPS of the SUN Sparc chips? >|> How about the MIPS chips? >|> > >Hi Bill :-) and everyone, > >I try to stay objective. Let's use i486 33Mhz as baseline for the CISC side, >and IBM POWER archecture as the RISC side baseline. > >Assuming that everything runs in memory, then i486 33 Mhz offers you about 21 >MIPS and less then 2 MegaFLOPS. Now IBM RISC POWERstation 320 20Mhz offers >you 29.5 MIPS and 8.5 MFLOPS. So given enough memory to both, i486 33Mhz ( body of nicely written survey deleted ) >Remember even Intel had to venture into RISC and made this i860? And lot's >people use it as graphic coprocessor or numerical coprocessor, why? RISC, >=> superior floating point performance (> 30 MegaFLOPS), that is. 80 MFLOPS (FP multiply/accumulate in one clock cycle) SO . . . . . Where are the i860 based machines ?!? YEARS ago (it seems) when IBM announced their Wizard adapter (an MCA bus-mastering i860 board) they claimed "AIX support in the next version" Well, the next version has come and gone, and now we can "access" the board from AIX. We can also compile C, C++, Fortran and Pascal code on the Microway i860 card (got one of those: not bad). Where are the machines? Why doesn't anyone port thier PC-based Unix to one of the i860 cards? The only i860 base machine I've heard of is a retro-fit for the SPARC machines. My guess is that the people that CAN make an i860 based machine simply don't want to. Why would IBM port AIX PS/2 to an add-on card?? Would they ever sell an RS/6000 then? Locus probably doesn't because IBM won't let them. What about SCO or interactive? 80 MFLOPS. Anybody got any ideas as to why we haven't seem more from this chip, or rumors of upcoming products? >Chin Fang hh
fangchin@leland.Stanford.EDU (Chin Fang) (06/03/91)
In article <30379@hydra.gatech.EDU>, hh2@prism.gatech.EDU (HAAS) writes: |> In article <1991Jun2.041512.29546@leland.Stanford.EDU> fangchin@leland.Stanford.EDU (Chin Fang) writes: |> >In article <1991Jun2.030215.11584@unixland.natick.ma.us>, bill@unixland.natick.ma.us (Bill Heiser) writes: |> >|> In article <1991May31.183111.16505@ux1.cso.uiuc.edu> tmkk@uiuc.edu (K. Khan) writes: |> >|> > |> >|> >It means that Intel is scared silly of the industry's upcoming move to |> >|> >the MIPS RISC chips, and is going to milk us for every single cent it |> >|> >can while it is still able to do so. |> >|> |> >|> How does the i486 compare with the 20 - 30 MIPS of the SUN Sparc chips? |> >|> How about the MIPS chips? |> >|> |> > |> >Hi Bill :-) and everyone, |> > |> >I try to stay objective. Let's use i486 33Mhz as baseline for the CISC side, |> >and IBM POWER archecture as the RISC side baseline. |> > |> >Assuming that everything runs in memory, then i486 33 Mhz offers you about 21 |> >MIPS and less then 2 MegaFLOPS. Now IBM RISC POWERstation 320 20Mhz offers |> >you 29.5 MIPS and 8.5 MFLOPS. So given enough memory to both, i486 33Mhz |> |> ( body of nicely written survey deleted ) Thank you. |> |> >Remember even Intel had to venture into RISC and made this i860? And lot's |> >people use it as graphic coprocessor or numerical coprocessor, why? RISC, |> >=> superior floating point performance (> 30 MegaFLOPS), that is. |> |> 80 MFLOPS (FP multiply/accumulate in one clock cycle) SO . . . . . Yours uses high end i860 chip. Man! My face is green with envy now :-( I certainly know how to put 80 MFLOPS into good use :-) |> Where are the i860 based machines ?!? YEARS ago (it seems) when IBM [some stuff deleted] Our EECF (electrical engineering computing facility) here has a multi-i860 machine made by FPX (I think) running multiprocessor SYSV UNIX. If you need more info, please email me and I will give you a name for more. |> Would they ever sell an RS/6000 then? Locus probably doesn't because |> IBM won't let them. What about SCO or interactive? |> I like to hear some insight info about this too. I also have been curious for quite a while already. |> 80 MFLOPS. |> Hmmm.... really enticing. A beauty indeed :-) Chin Fang Mechanical Engineering Department Stanford University fangchin@leland.stanford.edu
bill@unixland.natick.ma.us (Bill Heiser) (06/03/91)
In article <30379@hydra.gatech.EDU> hh2@prism.gatech.EDU (HAAS) writes: > >80 MFLOPS (FP multiply/accumulate in one clock cycle) SO . . . . . >Where are the i860 based machines ?!? YEARS ago (it seems) when IBM Isn't the i860 just a co-processor or numeric processor chip, not a cpu chip? I know CSPI has them on their array processors for Harris Nighthawk (and other) computers. -- bill@unixland.natick.ma.us The Think_Tank BBS & Public Access Unix ...!uunet!think!unixland!bill bill@unixland ..!{uunet,bloom-beacon,esegue}!world!unixland!bill 508-655-3848 (2400) 508-651-8723 (9600-HST) 508-651-8733 (9600-PEP-V32)
amichiel@rodan.acs.syr.edu (Allen J Michielsen) (06/03/91)
In article <30379@hydra.gatech.EDU> hh2@prism.gatech.EDU (HAAS) writes: ...a whole bunch of people said stuff.... >>|> How does the i486 compare with the 20 - 30 MIPS of the SUN Sparc chips? >>Assuming that everything runs in memory, then i486 33 Mhz offers you about 21 >>MIPS and less then 2 MegaFLOPS. Now IBM RISC POWERstation 320 20Mhz offers >>you 29.5 MIPS and 8.5 MFLOPS. So given enough memory to both, i486 33Mhz >( body of nicely written survey deleted ) >>Remember even Intel had to venture into RISC and made this i860? ... >80 MFLOPS (FP multiply/accumulate in one clock cycle) SO . . . . . >Where are the i860 based machines ?!? ... >The only i860 base machine I've heard of is a retro-fit for the SPARC >machines. My guess is that the people that CAN make an i860 based machine >simply don't want to. Actually, a whole slew of products are starting to hit the market using the i860 in PostScript aplications. PS accelerators for laserprinters, PS engines for new laserprinters, PS interpreters for output or display applications. While this is a far cry from what you intended, it better than.... My reading of the trade press indicates that intel really pissed a lot of people off. In regards to the i860, they tried to strong arm potential customers into signing contracts in order to support financing for the program, using threats of potential 'chip' shortages, contract supporters get first crack... (While even a no power contract supports intel because they use these contracts to demonstrate the low financial risk, pushing down the bond or financing rate....). On the other side, they told potential developers that they would have no hand or input into the product design. "You will get a little bitty black box that does what we tell you" Nothing that could be potentially used by compteitors was published as part of the specs. these 2 things alone made intel the odd man out in the risc race. Other vendors (such as mips) were very open to design (high level) and responsive to the needs of the customers. Actually, they didn't even have to be responsive, just forth coming with enough info to convince 'the other' engineers that they don't need or want whatever it is they think they want. al -- Al. Michielsen, Mechanical & Aerospace Engineering, Syracuse University InterNet: amichiel@rodan.acs.syr.edu amichiel@sunrise.acs.syr.edu Bitnet: AMICHIEL@SUNRISE
gwoho@nntp-server.caltech.edu (g liu) (06/03/91)
>|> >|> ( body of nicely written survey deleted ) >Thank you. >|> >|> >Remember even Intel had to venture into RISC and made this i860? And lot's >|> >people use it as graphic coprocessor or numerical coprocessor, why? RISC, >|> >=> superior floating point performance (> 30 MegaFLOPS), that is. >|> aclually only 80 mflops when the using pipeline multiply+add instructions out of the registers. often the pipeline cant be used, and only ordinary multiply or add instructions can be used, and not all the data is in the registers, in which case it is about 10 times slower. in real applications, i would imagine that a 40 mhz 860 is about 10 mflops. only hand coded stuff will work at 80mflops. multiplying small matrixes that fit in registers might be 80mflops, but not much else. >|> 80 MFLOPS (FP multiply/accumulate in one clock cycle) SO . . . . . >Yours uses high end i860 chip. Man! My face is green with envy now :-( >I certainly know how to put 80 MFLOPS into good use :-) >|> Where are the i860 based machines ?!? YEARS ago (it seems) when IBM >[some stuff deleted] >Chin Fang >Mechanical Engineering Department >Stanford University >fangchin@leland.stanford.edu
jgay@digi.lonestar.org (john gay) (06/04/91)
From article <1991Jun2.235352.22873@rodan.acs.syr.edu>, by amichiel@rodan.acs.syr.edu (Allen J Michielsen): > In article <30379@hydra.gatech.EDU> hh2@prism.gatech.EDU (HAAS) writes: > ...a whole bunch of people said stuff.... > > Actually, a whole slew of products are starting to hit the market using > the i860 in PostScript aplications. PS accelerators for laserprinters, > PS engines for new laserprinters, PS interpreters for output or display > applications. While this is a far cry from what you intended, it better > than.... Actually, I think (from reading in EE Times) that these are all/most based on the i960. My understanding (also from the EE Times article) is that intel has declared/stated/whatever that the i960 is for sale for embedded products (printers and whatever), the i860 is for co-processor applications (math and graphics), and the i486 is for general purpose cpu apps. Intel is strongly discouraging the use of the i860 as a general purpose cpu because it will cut deeply into the cash cow of i[34]86 chips. BTW, in a column in a recent EE Times someone (I forget who, but he puts out a newsletter about the chip industry) estimated that intel has shipped 9 million 386 chips to date. I think that they have recouped their R&D for the 386 chip by now (and probably for the 486 as well). -- john gay. jgay@digi.lonestar.org
ian@rathe.cs.umn.edu (Ian Hogg) (06/04/91)
In article <1991Jun3.214105.490@digi.lonestar.org> jgay@digi.lonestar.org (john gay) writes: >From article <1991Jun2.235352.22873@rodan.acs.syr.edu>, by amichiel@rodan.acs.syr.edu (Allen J Michielsen): > >Actually, I think (from reading in EE Times) that these are all/most based >on the i960. My understanding (also from the EE Times article) is that >intel has declared/stated/whatever that the i960 is for sale for embedded >products (printers and whatever), the i860 is for co-processor applications >(math and graphics), and the i486 is for general purpose cpu apps. Intel >is strongly discouraging the use of the i860 as a general purpose cpu >because it will cut deeply into the cash cow of i[34]86 chips. > What chips are being used in their HyperCube computer? It was the 386 but not anymore. > >-- > >john gay. >jgay@digi.lonestar.org -- Ian Hogg email: rathe!ian@cs.umn.edu ...!umn-cs!rathe!ian Rathe, Inc ianhogg@cs.umn.edu 366 Jackson Street phone: (612) 225-1401
look@s30.csrd.uiuc.edu (Stephen Look) (06/05/91)
The i860 is used in several multi-processor machines including the Alliant
FX2800 series. I have heard that Intel is planning on an upgrade of the
i860 to make it easier to use in that type of application.
--
Steve Look ka9szw Center for Supercomputing R & D
look@s30.csrd.uiuc.edu 305 Talbot Lab 104 S. Wright
(217) 244-5980 Urbana, IL 61801
"No, we don't have any CRAYs here, they are down the street..."kls30@duts.ccc.amdahl.com (Kent L Shephard) (06/05/91)
In article <1991Jun2.231525.19301@unixland.natick.ma.us> bill@unixland.natick.ma.us (Bill Heiser) writes: >In article <30379@hydra.gatech.EDU> hh2@prism.gatech.EDU (HAAS) writes: >> >>80 MFLOPS (FP multiply/accumulate in one clock cycle) SO . . . . . >>Where are the i860 based machines ?!? YEARS ago (it seems) when IBM > >Isn't the i860 just a co-processor or numeric processor chip, not a ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Tne i860 is a full fledged RISC processor. It has capabilities that make it well suited for vector and graphics functions. It has all the functionality of a regular CPU, nothing like a Weitek or 80x87. >cpu chip? I know CSPI has them on their array processors for Harris >Nighthawk (and other) computers. > >-- >bill@unixland.natick.ma.us The Think_Tank BBS & Public Access Unix > ...!uunet!think!unixland!bill bill@unixland > ..!{uunet,bloom-beacon,esegue}!world!unixland!bill >508-655-3848 (2400) 508-651-8723 (9600-HST) 508-651-8733 (9600-PEP-V32) Kent -- /* -The opinions expressed are my own, not my employers. */ /* For I can only express my own opinions. */ /* */ /* Kent L. Shephard : email - kls30@DUTS.ccc.amdahl.com */