clong@topaz.rutgers.edu (Chris Long) (06/01/90)
Has anyone else read the article in the June issue of PC World where 33 MHz 386s are compared to 25 MHz 486s? The author conludes that there is no substantial difference in performance in most cases between the two. This shocked me, because I'd expect the 486 to be about twice as fast, on the average. Is it possible that the 8k internal cache of the 486s tested wasn't turned on? Failure to do this could indeed degrade the performance of a 486 by about 50% (see the June issue of Personal Workstation, page 46), which would explain the results reported. Comments? -Chris
orac@chinet.chi.il.us (Paul Giovacchini) (06/04/90)
In article <Jun.1.02.34.49.1990.595@topaz.rutgers.edu> clong@topaz.rutgers.edu (Chris Long) writes: > >Has anyone else read the article in the June issue of PC World where >33 MHz 386s are compared to 25 MHz 486s? The author conludes that >there is no substantial difference in performance in most cases >between the two. This shocked me, because I'd expect the 486 to >be about twice as fast, on the average. Is it possible that the >8k internal cache of the 486s tested wasn't turned on? Failure >to do this could indeed degrade the performance of a 486 by about >50% (see the June issue of Personal Workstation, page 46), which >would explain the results reported. > >Comments? > >-Chris Did the machine tested have an AT bus? It must have. Because the AT bus can not do the "burst mode" of the 486, the performance increase that you will get, as I have read in about 3 different magazines is "negligible". BUT if you have either a MCA or an EISA bus, which can both handle the burst mode, the performance is astronomical. ---- -Paul Giovacchini | " Think you are having a bad day? ( orac\@chinet.chi.il.us ) | I just fried 2 computers and a cat! "
davidsen@sixhub.UUCP (Wm E. Davidsen Jr) (06/04/90)
I have been running benchmarks on 386s and 486s for about a month now,
and the 486 is about 2.6 times as fast as the 386 at the same clock
speed *for 32 bit software*. I have noted that the 486 is actually
slower for some 16 bit arithmetic. If someone has a benchmark which only
measures this, then they will, indeed, get results which make the 486
look bad. Actually if you want to run 16 bit integer programs, the 486
probably is not cost effective.
--
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" -mebeaucham@uxh.cso.uiuc.edu (06/05/90)
In article <1:18 pm Jun 3, 1990 by orac@chinet.chi.il.us in uxh.cso.uiuc.edu:comp.sys.ibm.pc > Paul Giovacchini writes: < Did the machine tested have an AT bus? It must have. Because < the AT bus can not do the "burst mode" of the 486, the performance < increase that you will get, as I have read in about 3 different magazines < is "negligible". BUT if you have either a MCA or an EISA bus, which can < both handle the burst mode, the performance is astronomical. < This conclusion may be contradicted by PC Magazine's study (June 26, 1990, p. 113, "The Bus Wars") where for the ALR Power Casche 4 (with 25 MHz 486 cpu) configured as nearly-identical systems except for the bus, they show that the ISA bus is just as fast as MCA and EISA in 15 different benchmarks. While for most of the tests the results are just about the same, for disk controller operations reading small blocks, ISA and EISA rates were almost identical but surpringly exceeded the MCA rate by approximately 3 times. Also, for each bus, when the ALR 486 was compared to 386 machines of other manufacturers (IBM/MCA, Everex/ISA, Compaq/EISA), the 486 outperformed the 386 on the Dhrystone benchmark by substantial margins. Anyway, PC Magazine's conclusion is that ISA is as good as MCA or EISA for most real single-user DOS applications and is not worth the increased cost. Perhaps there is a benchmark which convincingly illustrates the power of the 486/EISA or 486/MCA over 486/ISA, but I haven't seen it. In terms of number-crunching benchmarks, I am seeing 26-28,000 Dhrystones and 1.0 Mflops for the 486-25 vs. 16-17,000 Dhrys and 0.5 Mflops for most 386-33 machines (under SCO Unix as per Personal Workstation, June, '90, p.63). Having the 387 internal to the 486 is a big plus, I think. In the future, it will probably force a reduction in the cost of floating point hardware. Jim Beauchamp j-beauchamp@uiuc.edu
dhinds@portia.Stanford.EDU (David Hinds) (06/05/90)
In article <1990Jun3.181830.2145@chinet.chi.il.us> orac@chinet.chi.il.us (Paul Giovacchini) writes: >In article <Jun.1.02.34.49.1990.595@topaz.rutgers.edu> clong@topaz.rutgers.edu (Chris Long) writes: >> >>Has anyone else read the article in the June issue of PC World where >>33 MHz 386s are compared to 25 MHz 486s? The author conludes that >>there is no substantial difference in performance in most cases >>between the two. This shocked me, because I'd expect the 486 to >>be about twice as fast, on the average. > > Did the machine tested have an AT bus? It must have. Because > the AT bus can not do the "burst mode" of the 486, the performance > increase that you will get, as I have read in about 3 different magazines > is "negligible". BUT if you have either a MCA or an EISA bus, which can > both handle the burst mode, the performance is astronomical. > The bus type isn't directly relevant here, if CPU performance is the question. Memory installed on the system board is not accessed via the I/O bus, so a board can support burst mode accesses to main memory regardless of the I/O bus type. I would expect the overall performance gain for a 25MHz 486 to be only on the order of 5% with burst mode. Remember that this is only improving the performance of cache misses - sure, that helps, but the whole idea is to not go to main memory as often in the first place. The performance improvement will be more significant for 33MHz and 50MHz parts, because the CPU otherwise may have to wait around a lot for main memory. In the latest issue of BYTE, there is a review of the latest Cheetah 486-33 system. The system does not use an external cache, but uses 35ns DRAM's for main memory (!?!) to get near-zero wait states on internal cache misses. And the board has only an ISA bus, though with some tricky performance improvements. I think the important question to ask about the original posting is, for what software? If the 386 and 486 are running DOS code, then they are both taking major performance hits to begin with, and can't take advantage of the 32-bit-wide bus. I would guess that if the CPU is spending all of its time loading and storing data 8 or 16 bits at a time, even if these accesses are largely cached, they will bury differences in instruction execution times. For 32-bit code, execute time will start to become limiting, and the large advantages of the 486 should be much more apparent. -David Hinds dhinds@popserver.stanford.edu
sunesen@iesd.auc.dk (Peter Sunesen) (06/07/90)
I have read some article about, 80486 and that it is running the same speed as a 80386. I can say that the person that is writing this, only not have used a 486 or will say something wrong about it becouse thay not can buy one. I have used a 80486-25, since decenber 1989, and can say that my other Taiwan computer is running about 3 times slower, and it is a 80386-25 with cache. It can be right, that a 80386 is running the same speed as a 80486, but the answer to this is that Harddisk, not is made more quick becouse it is running at a 486. I was having a "verry" quick harddisk 18 ms in acces on the 386, but when I installed it on the 486 it was slow, and was going to wait on the harddisk. So the problem was now that I have speed up the computer but the problem was in fact that I was running more on the harddisk, than on the computer, so the speed rise was not so big, before i put a faster harddisk indsite. But a 486 is not a computer, that the user have to run wordperfect, but to run hevy calculation, and when you do that it is running about 3 times as fast. So juge the 486 on what it is doing and not was the test is saying. 80486-25 8KB cache ------------------ Norton 4.5 : 40.5 Mips : 10.9 80486-33 128KB cache external ----------------------------- Mips : 14.9 About the bus talk, the MCA is just fore IBM, the EISA will be the future becouse you the can speed up for exsample the display adapter / Harddisk controler. But what good it this if for exsample a TRIDENT VGA 512K display adapter must be speedet down to slow and 8bit if there is a slow SEAGATE SCSI adapter on the computer ????. My conclution is that, the ISA/OLD bus will be on long time more, and the Taiwan computers first will produce EISA when the card to the bus is there. So if you out there is buying a 486 with old bus you is not doing somthing wrong, becouse the 486 is fast in all cases, becouse for exsamble the harddisk is a long time behind the computers i acess time, so when harddisk with 1 ms in acess is ther I will definit by EISA, not before. If some person is on way to buy a 80486, I can give a adress on a company in Taiwan were I have good experiance, just E-mail me. Peter Sunesen sunesen@iesd.auc.dk Tlf: +45 98114268 Fax: +45 98114268
phd_jacquier@gsbacd.uchicago.edu (06/07/90)
(Wm E. Davidsen Jr) writes: > I have been running benchmarks on 386s and 486s for about a month now, >and the 486 is about 2.6 times as fast as the 386 at the same clock >speed *for 32 bit software*. ****** Did you have an ISA or an EISA bus ? (Does it matter anyway? I guess it may with 32 bit software) What 32bit software is available now for DOS ? What 32 bit software is available now for UNIX ? Eric
wozniak@utkux1.utk.edu (Bryon Lape) (06/08/90)
>with 32 bit software) >What 32bit software is available now for DOS ? PC-Magazine just did a comparison with ISA, EISA, and MCA machines. In some cases, the ISA was just as fast and better economic sense. -bryon lape-
davidsen@sixhub.UUCP (Wm E. Davidsen Jr) (06/10/90)
In article <1990Jun7.150857.8104@midway.uchicago.edu> phd_jacquier@gsbacd.uchicago.edu writes: | (Wm E. Davidsen Jr) writes: | > I have been running benchmarks on 386s and 486s for about a month now, | >and the 486 is about 2.6 times as fast as the 386 at the same clock | >speed *for 32 bit software*. | ****** | | Did you have an ISA or an EISA bus ? (Does it matter anyway? I guess it may | with 32 bit software) The answer is that all systems had 32 bit motherboard memory. Bus was not an issue in this case, and for some tests the entire program was resident in cache (where present). | What 32bit software is available now for DOS ? You're asking the wrong person! However, there are C compilers which run with DOS extenders (PharLap?) and use not only 32 bit instructions but protected mode. FRACTINT uses 32 bit instructions in DOS real mode, and it is faster than on a 286 at the same clock. It's even faster on an SX than a 286, and that's comparing 16 bit busses (but 32 bit instructions). | What 32 bit software is available now for UNIX ? Xenix/386, SCO UNIX, SCO Open DeskTop, INteractive Systems UNIX, ESIX, Intel UNIX. I believe within 60 days V.4 will be shipping. Commercial software for these systems uses the 386 (32 bit) instruction set. -- 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
John.Guest@f220.n226.z1.FIDONET.ORG (John Guest) (06/11/90)
There is a PCMag issue (V9N12 or 11) that features just that question - whether the bus - EISA, ISA, MCA. Pretty good and indicates that unless you want to go beyond a 4 user system that ISA is as good as any. This may also be of interest to those of you like myself who were waiting for the EISA to be more widely available. -- John Guest via cmhGate - Net 226 fido<=>uucp gateway Col, OH UUCP: ...!osu-cis!n8emr!cmhgate!220!John.Guest INET: John.Guest@f220.n226.z1.FIDONET.ORG
phd_jacquier@gsbacd.uchicago.edu (06/13/90)
(John Guest) writes... >PCMag features just that question - >whether the bus - EISA, ISA, MCA. [...] They indicate that unless >you want to go beyond a 4 user system that ISA is as good as any. Yes. But they did only DOS stuff. They say that they are going to follow up with a UNIX based similar study though they don't say when!!!. Maybe the results would be very different with UNIX ( I mean ..would favor EISA). Anybody has experience or pointers about this? Eric Jacquier