Sonny.Shrivastava@f555.n161.z1.FIDONET.ORG (Sonny Shrivastava) (10/13/90)
>KWhets/sec: 882.4 (using FPU) 39.4 (not using FPU) >Dhrys/sec: 2871 FPU, 2717 not >Sieve: 3.70 sec FPU, 3.83 sec >Savage err: 7.988e-10, 2.297e-11 >Savage time: 8.52 sec FPU, 71.43 sec >Savage iterations: 25000 FPU, 5000 > >CPU test: 6.40 (no FPU) >Math: 95.27 FPU, 5.24 >Performance: 5.42 (no FPU) I'd be curious to see how various Macs performed. Here are my results on a Mac IIci, System 6.0.5, Speedometer 2.5 (disregard results in previous message): KWhetstones/sec.: 882.353 Dhrystones/sec.: 5791.506 Sieve (Secs, 100 times): 3.750 Savage Cum. Error: 7.98815e-010 Savage Time: 7.317 sec. Savage Iterations: 25000 CPU Test: 6.71 Math: 149.71 Disk: 3.15 Performance: 7.45 (no FPU) As you can see, the Mac IIci easily, without doubt, blows the doors off the Amiga 3000 with Amax-II. Just compare the Dhrystones/sec. figure. In almost every test, the Mac IIci comes out on top, especiall in the CPU, FPU, and Performance tests, which are the only real-world figures. I think you should give your friend a printout of this message, and let him see the figures for himself. It's easy to blow off a lot of steam about the Amiga when you don't have the facts. Well, here they are! By the way, the numbers in the CPU, Math, Disk, and Performance tests for both the Amiga/Amax-II and the Mac IIci indicate how many times faster they are than a stock Mac SE with a 20 MB hard disk. Therefore, the Amiga-3000/Amax-II combination is 5.42 times as powerful as a stock SE, while the Mac IIci is 7.45 times as powerful as a stock SE. These times are not relative to the Mac II, as you indicated in your message, but to the Mac SE. So, if anybody ever tries to tell you their A-3000/Amax-II combination blows the doors off a Mac IIci or higher, you tell them to buzz off and get a life. You can expect the IIfx to have at least twice the performance of a IIci. Hope this helped! I hunted on a lot of BBSs for Speedometer 2.5, and finally found one after calling about 10 Mac BBSs. If it wasn't for you, I never would have updated my Speedometer! Thanks. -- Sonny Shrivastava - via FidoNet node 1:125/777 UUCP: ...!uunet!hoptoad!fidogate!161!555!Sonny.Shrivastava INTERNET: Sonny.Shrivastava@f555.n161.z1.FIDONET.ORG
chow@batcomputer.tn.cornell.edu (Christopher Chow) (10/20/90)
In article <7878.271B9F30@fidogate.FIDONET.ORG> Sonny.Shrivastava@f555.n161.z1.FIDONET.ORG (Sonny Shrivastava) writes: .Here are my results on a Mac IIci, System 6.0.5, Speedometer 2.5 (disregard .results in previous message): As long as people are posting benchmark result, here is what I get for my Mac II with Dove 030 Marathon board installed. [The Marathon 030 is a board consisting of a 68030 at 32 Mhz w/o any additional cache. MacWorld (August 1980) found it to be approximately same in speed as a IIci without a cache board.] Mac IIci II w/ Mara. II w/Mara & FPU . KWhetstones/sec.: 882.353 75.949 659.341 . Dhrystones/sec.: 5791.506 4966.887 4893.964 . Sieve (Secs, 100 times): 3.750 3.050 3.067 . Savage Cum. Error: 7.98815e-010 2.29745e-011 7.99815e-10 . Savage Time: 7.317 sec. 32.4 9.650 . Savage Iterations: 25000 5000 25000 . . CPU Test: 6.71 6.38 6.38 . Math: 149.71 9.79 134.36 . Disk: 3.15 2.96 2.96 . Performance: 7.45 (no FPU) 6.38 n/a I note in passing that the times posted for the Mac IIci for floating point seem very strange. That math benchmark at 149.71 would be indicative of a testing calling the 68882 coprocessor directly. Note for example, the 149.71 is completely off the wall in comparison to the data given in the "systems comparison" window. Furthermore, the Savage cum. error figures are indicative of the errors which the Motorola chips would generate (SANE exceeds IEEE floating point specs for accuracy at a great expense for speed.) Also, when citing Mac IIci numbers you should explicitly state whether you have a cache, and if so, something about the cache (i.e, name, size, and type (direct, n-way interleave (specify n)). BTW, could someone with a Mac IIci and a cache do some benchmarks and show us what difference a cache makes? Good day. Christopher Chow /---------------------------------------------------------------------------\ | Internet: chow@theory.tn.cornell.edu (128.84.248.35 or 128.84.253.35) | | Usenet: ...{uw-beaver|decvax|vax135}!cornell!batcomputer!chow | | US Mail: 1 Notre Dame Dr - Room 335, Albany, NY 12208 | | Phone: 1-518-426-0687 | \---------------------------------------------------------------------------/ -- Christopher Chow /---------------------------------------------------------------------------\ | Internet: chow@theory.tn.cornell.edu (128.84.248.35 or 128.84.253.35) | | Usenet: ...{uw-beaver|decvax|vax135}!cornell!batcomputer!chow |
wilkins@jarthur.Claremont.EDU (Mark Wilkins) (10/23/90)
Re: recent test results posted.
It seems to me that it should be legitimate to benchmark these systems
with code written especially for 68881/2 math coprocessors. After all, most
math intensive programs these days (Mathematica, Excel, Wingz, to name a
few) are available in versions which take advantage of the coprocessor.
However, that such code was used should be noted explicitly, because
Apple's SANE is slightly more accurate (although not enough to be worth
using it in the vast majority of cases, IMHO)
-- Mark Wilkins
--
******* "Freedom is a road seldom traveled by the multitude!" **********
*-----------------------------------------------------------------------------*
* Mark R. Wilkins wilkins@jarthur.claremont.edu {uunet}!jarthur!wilkins *
****** MARK.WILKINS on AppleLink ****** MWilkins on America Online ******