91_bickingd@GAR.UNION.EDU ("Bicking, David") (12/04/90)
Date sent: 3-DEC-1990 20:33:06 Lots of stuff deleted :) >> Maybe you want speed? I don't know how fast X would be on the NeXT, but it sure >> couldn't be any slower than the display postscript it comes with. > >Maybe you haven't seen NeXTstep 2.0 running on a 68040 NeXTstation. I should hope it's fast on an '040. Has it been released yet? >> Yes, the SLC is faster than either of the other two. It is also very limited in >> expansion. You cannot make it colour. Max memory is less than either of the >WROONG. SLC is much slower than a 68040-NeXT. NeXTstation is supposed to >be even faster than a SparcStation 1+. I wanna see specs before I believe that, especially considering the slowness of the '030 NeXT (as opposed to the speed of the A3000). >Note: When comparing the A3000UX and the NeXTstation you should remember >that the latter has a 68040 and a motorola DSP. The latter doesn't have >color in it's cheapest configuration, though. When comparing the two you also have to realize the A3000UX can be easilly expanded in MANY ways, especially the CPU. More color, more sound, higher resolution, all can be added easilly. Soon, when C= finishes their device independant libraries, all those expansions will be universally accessible. > > Jouni -- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--=- Dave Bicking Single Tasking????? Just say NO!!!! Union College Box 152 91_bickingd@union.bitnet // Schenectady, NY 12308 91_bickingd@gar.union.edu \X/ Amiga -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
bostrov@storm.UUCP (Vareck Bostrom) (12/07/90)
In article <9012040152.AA09539@ucbvax.Berkeley.EDU> 91_bickingd@GAR.UNION.EDU ("Bicking, David") writes: >>WROONG. SLC is much slower than a 68040-NeXT. NeXTstation is supposed to >>be even faster than a SparcStation 1+. > >I wanna see specs before I believe that, especially considering the slowness of Ok, I am assuming that the SPARCstation-1 and the SLC are the same speed. I have tested a SS-1 but not a SLC, and I have also tested a SS-1+ and an SS-2, as well as 4/490's 4/390's, and 4/330's. In Floating point, the NeXT easily beats the Sparcstation-1, the 1+, and the 4/3xx and 4/4xx. In integer (almost useless, but anyway) the NeXT manages to tie almost exactly the ss-1+. Here are some marks: Machine Dhrystones MFlops NeXT 030/25mhz 7204 Sun 4/60 (ss-1) 19672.1 (max opt) Sun 4/65 (ss1+) 22140.2 (max opt) NeXT 040/25mhz 23661.0 (max opt) Amiga (no sure) 3350.7 (opt???) IBM 3090/200 54894.8 Expected at 800 with vector. Amiga (030/25) 6990.3 Sun 3/60 3790.3 rs6000/520 28222.0 Sun 4/330 25000.0 (max opt) Ardent titan-3 22831.1 HP 9000/845 20833.3 Gould np-1 13730.8 (max opt) symmetry-81k 5696.9 Hp 9000/310 948.0 (max opt -- took an hour to compile) Anyway, those are figures that I (or a friend of mine) have tested ourselves. I am quite sure that a Cube-040 can match a ss-1+ in integer performance, and because of it's cisc nature, I believe that the 040 cube will handle load better than the ss-1+ will also. THESE ARE JUST GUESSES. I havent been on an 040 next much, but I have been on 1+'s, and I will admit that they FEEL fast. The NeXT has to work on that. The only experience I've really had with the ami is games and a bit of programming, but those machines crash more often than...anything. I would love to try an 040/25 amiga running under sys-vr4, but from what I have seen, they seem to buckle under load much more than the NeXT or the ss-1. Of course, the ss-2 will blow away anything except the rios-risc 6000. I hate AIX, though. Note I say (max opt) on a lot of them, on the suns that means -O4 or WHATEVEr we got the best time out of. The SPARCstations would drop to 9000 dhrys with no optimization. Under ami/os, I had major problems getting ANYTHING to compile, and it reminded me much of a macintosh in that respect. Almost all of the ami's I have seen or delt with had poor res screens, major flicker, or other problems. Going from a 1+GX to a 68000 ami is a major headache, even if i go to the ami just to play games. Graphics on the ami do seem very fast for such a cheap machine, though I have seen suns with graphics accellerators that will blow an ami away (admittidly those accellerators are $6k plus). Anyway, the 040 NeXT is as fast as a SS-1+. I expect that an 040 ami will be close, also. >the '030 NeXT (as opposed to the speed of the A3000). > >>Note: When comparing the A3000UX and the NeXTstation you should remember >>that the latter has a 68040 and a motorola DSP. The latter doesn't have >>color in it's cheapest configuration, though. > >When comparing the two you also have to realize the A3000UX can be easilly >expanded in MANY ways, especially the CPU. More color, more sound, higher >resolution, all can be added easilly. Soon, when C= finishes their device >independant libraries, all those expansions will be universally accessible. True. I have seen a VERY expanded amiga, and it was very impressive. It did seem to lack the cpu uumph of the ss-2 or a mips magnum, but it was moving along. The NeXTcube with NeXTdimention and monitor should be impressive also, 68040 CPU, 80860/33 graphics chip (gods, thats 66 MFLOPS (acc. to intel)). I couldn't see actual heavy duty sci work being done on an ami/040, as that is getting close to your maximum cpu power, and you have to expand the graphics by a power of 10 to get workstation quality (IM TALKING res, not color. 4096 isn't bad). Also I haven't found any good optimization compilers for the Ami, and this leaves many of the programs that scientists run dead in the water. (when i say scientists, I mean scientists/engineers, etc.). Also, everywhere I go where there are workstations installed, I see Suns, NeXT's (on occasion), MIPS, DECs, IBM rs/6000's, but NEVER any amigas. I wonder why? Well, as of now, there is no Ami/Unix (though I have heard and seen samples of r4 on an ami) there is no 040 ami (once again, ive seen samples), mid to poor system through put, etc. An 030 machine just isn't a serious workstation, paticulary without unix. The NeXT IS a unix workstation, though up till receintly there has been no 040 for it, all along it has run unix. Anyway, both have their uses. I LOVE the amiga to play games on. It is without a doubt the most devoloped games machine ever. It is fast (for a game machine), and can be used for other things besides games. So...the end. > >> >> Jouni >-- >-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--=- >Dave Bicking Single Tasking????? Just say NO!!!! >Union College Box 152 91_bickingd@union.bitnet // >Schenectady, NY 12308 91_bickingd@gar.union.edu \X/ Amiga >-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Vareck Bostrom Reply to: bostrov@mist.cs.orst.edu Flames to: erica@poux.cs.uoregon.edu
xrtnt@amarna.gsfc.nasa.gov (Nigel Tzeng) (12/08/90)
In article <673@storm.UUCP>, bostrov@storm.UUCP (Vareck Bostrom) writes... ^In article <9012040152.AA09539@ucbvax.Berkeley.EDU> 91_bickingd@GAR.UNION.EDU ("Bicking, David") writes: [Lots'o interesting data removed for bandwith's sake] ^ ^I couldn't see actual heavy duty sci work being done on an ami/040, as ^that is getting close to your maximum cpu power, and you have to expand the ^graphics by a power of 10 to get workstation quality (IM TALKING res, not ^color. 4096 isn't bad). Also I haven't found any good optimization compilers This may be a silly question but wouldn't you tend to want to do your massive number crunching on a mainframe and use the workstation for display pruposes? While workstations are nice not much beats a Cray ;-). Anyway...a lot of the systems (at least those considered "leading edge") around here seem to be a big machine (say a power series Iris) that does the serious data massaging and leaves the display (and secondary massaging like smoothing etc) to workstations (say a personal iris or two). Being on a poorer project our main number cruncher was a Vax and the workstation machine are MacII (yep the originals). Both a kind of slow but get the job done. The only thing the Ami tends to lack is the resolution (which you pointed out ;-). ^for the Ami, and this leaves many of the programs that scientists run ^dead in the water. (when i say scientists, I mean scientists/engineers, etc.). ^Also, everywhere I go where there are workstations installed, I see Suns, ^NeXT's (on occasion), MIPS, DECs, IBM rs/6000's, but NEVER any amigas. I ^wonder why? Well, as of now, there is no Ami/Unix (though I have heard and No software. We picked the Mac just because NCSA Image was available as PD. It all depends on budget a lot of times. Had the Amiga had NCSA Image and a large selection of science software it might have been chosen instead. The CPU power of the same cost Ami would have been higher. It would have lost something in the resolution battle BUT for a lot of the skymaps the necesary resolution could be measured in millimeters and not pixels ;-). Unix is nice...but a lot of sites here still run VMS. To a certain extent Jobs is correct...the applications matter more than the OS. So long as the data can be transfered easily. ^seen samples of r4 on an ami) there is no 040 ami (once again, ive seen ^samples), mid to poor system through put, etc. An 030 machine just isn't ^a serious workstation, paticulary without unix. The NeXT IS a unix workstation, ^though up till receintly there has been no 040 for it, all along it has ^run unix. The problem that the high end Ami (or Mac for that matter) and mid range NeXTs have is that in many cases procurement of systems is on the breakpoint level. For instance I can run out and pick up a $2500 system at no hassle (low paper work involved and only a couple of signatures). Anything beyond that requires going through the whole procurement hassle (with a breakpoint at around 25k). So long as I stay below that next magic number my paperwork (and chasing superiors down) is the same. Therefore machines in that "class" is the range from $2501 to $25k. A lot of times I need to factor in periperals (I can easily spend 20k in a nice Mac...) so the PC level machines are competetive in certain environments (like office automation...Postscript printer...a nice scanner...hmmmm...I obviously need 8 megs to use MS Word...19inch monitor ;-). On the workstation level Personal Iris, Suns, NeXTs, Mac IIfx and A3000 are all at the same plateau. Needless to say if I'm only buying one system the NeXT, the Mac and the Ami are going to lose (remember...up to 25K it's all the same to me...I can fill in software and other little things with followup procurements in increments of $1000-$2500). Above the 25k level requires more work and more signatures (read as I need to justify why I need such a whiz bang system). These number may only be relevant at Goddard (I don't know the breakpoint levels at other installations) but I suspect they are similar. It's really funny in a lot of ways what you have to do to get the job done. A lot of times you buy a system piecemeal to scrape under an aribitrary limit (like the small procurement thing used to be 1K...). A motherboard here...a disk drive there...etc. Sometimes the question is whether the vendor will piecemeal the system for you so you can get the machine on time (procuements of systems at the 25K level is measured in months...at the $2k level in days or weeks...a big difference sometimes if you didn't plan well enough and find yourself in a crunch). Then it's a game to see if the procurement people catch you trying to sneak a system in that way. If they do then you can expect to have ANY small procurement request nixed until they forgive you ;-). Needless to say people don't try this without extreme need ;-). ^ ^Anyway, both have their uses. I LOVE the amiga to play games on. It is without ^a doubt the most devoloped games machine ever. It is fast (for a game machine), ^and can be used for other things besides games. ^ ^So...the end. ^> ^>> Jouni ^>-- ^>-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--=- ^>Dave Bicking Single Tasking????? Just say NO!!!! ^ ^Vareck Bostrom NT -------------------------------------------------------------------------------- // | Nigel Tzeng - STX Inc - NASA/GSFC COBE Project \X/ | xrtnt@amarna.gsfc.nasa.gov | Amiga | Standard Disclaimer Applies: The opinions expressed are my own.
garnett@cs.utexas.edu (John William Garnett) (12/08/90)
In article <4128@dftsrv.gsfc.nasa.gov> xrtnt@amarna.gsfc.nasa.gov writes: >In article <673@storm.UUCP>, bostrov@storm.UUCP (Vareck Bostrom) writes... >^In article <9012040152.AA09539@ucbvax.Berkeley.EDU> 91_bickingd@GAR.UNION.EDU ("Bicking, David") writes: > >This may be a silly question but wouldn't you tend to want to do your massive >number crunching on a mainframe and use the workstation for display pruposes? Many of the newer workstations are approaching or surpassing mainframe performance on some types of problems. Also much of the work that used to be done on mainframes is now being done on workstations because CPU time is much easier to obtain on a workstation. A given mainframe may be 10 times faster than some workstation but if you have to share the mainframe cpu with 100 other simultaneous users, the workstation will get the job done first (assuming there are only a few users). It all comes down to price/performance. -- John Garnett University of Texas at Austin garnett@cs.utexas.edu Department of Computer Science Austin, Texas
hunter@phoenix.pub.uu.oz.au (James Gardiner [hunter]) (12/14/90)
In <673@storm.UUCP> bostrov@storm.UUCP (Vareck Bostrom) writes: >>I wanna see specs before I believe that, especially considering the slowness of >Ok, I am assuming that the SPARCstation-1 and the SLC are the same speed. >I have tested a SS-1 but not a SLC, and I have also tested a SS-1+ and an >SS-2, as well as 4/490's 4/390's, and 4/330's. In Floating point, the >NeXT easily beats the Sparcstation-1, the 1+, and the 4/3xx and 4/4xx. >In integer (almost useless, but anyway) the NeXT manages to tie almost >exactly the ss-1+. Here are some marks: >Machine Dhrystones MFlops >NeXT 030/25mhz 7204 >Amiga (030/25) 6990.3 From MY basic knowledge, AmigaDos WOULD HAVE TO BE less of a system overhead then UNIX. This makes your TESTS more like Standardized statistics. >I couldn't see actual heavy duty sci work being done on an ami/040, as >that is getting close to your maximum cpu power, and you have to expand the >graphics by a power of 10 to get workstation quality (IM TALKING res, not >color. 4096 isn't bad). Also I haven't found any good optimization compilers The standard amiga Video quality is OK. Like the Next/Sun etc. there are FAST million colour video boards comming out but, like Next,Sun,etc, They cost an arm, leg, etc. >for the Ami, and this leaves many of the programs that scientists run >dead in the water. (when i say scientists, I mean scientists/engineers, etc.). >Also, everywhere I go where there are workstations installed, I see Suns, >NeXT's (on occasion), MIPS, DECs, IBM rs/6000's, but NEVER any amigas. I >wonder why? Well, as of now, there is no Ami/Unix (though I have heard and >seen samples of r4 on an ami) there is no 040 ami (once again, ive seen >samples), mid to poor system through put, etc. An 030 machine just isn't >a serious workstation, paticulary without unix. The NeXT IS a unix workstation, >though up till receintly there has been no 040 for it, all along it has >run unix. Yes Next is a real unix workstation. Amiga users want MORE then just a Nice UNIX workstation. >Anyway, both have their uses. I LOVE the amiga to play games on. It is without >a doubt the most devoloped games machine ever. It is fast (for a game machine), >and can be used for other things besides games. I find this offensive, "most developed games machine ever" If a Sparc1 was as much as a 500 to buy then it would be a much bigger games machine then an amiga. Just because its inexpensive, bought by young people that like to play games and Software companies supply these games, does not make it a games machine. From what I have read about Next in comp.sys.amiga, I have become interested in looking at one. It may be a nice unix machine, but in either case, there is still enough room for a Amiga 3000UX on my desk as well. Hunter -- James Gardiner [Hunter]. System Admin, Public Access UNIX Melbourne, Australia PubNet: phoenix!hunter | (voice)+613-532-8030 (data)+613-523-9865&+613-532-8029 Internet: hunter@phoenix.pub.uu.oz.au | PO BOX 54 Chadstone Centre UUCP:..!uunet!munnari!labtam!eyrie!phoenix!hunter | Melbourne Austalia 3148