Speed of 68040 v. 486

Scott DeMarchi

unread,

Jul 23, 1993, 12:48:52 AM7/23/93

to

Several people have posted to this group telling you folks that you idea that
the mac is faster than the equiv pc clone is FALSE. And I'm even more tired
of cbs...@ccnga.waterloo.edu writing in without cites to claim that a
68040 40mhz is the equiv. of a 486 66mhz.
You folks have three options:
1) check the military archives where they keep benchmarks.
marlin.nosc.mil is a good one. The 486 beats the mac at every chipspeed
on EVERY benchmark (i.e., 486/33>68040/33, etc. -- on diff versions of mips,
specmarks, etc.). If you lame
ass reality deficient mac users would simply ftp files from this site, I
wouldn't have to listen to this shit.
2) buy an issue of the last computer shopper. On real world benchmarks, a
gateway clone for 1/3 the price of the quadra it ran against, beat the
quadra on EVERY REAL WORLD PROGRAM (excel, word, etc.). Get a clue for
only 2.95$.
3) Pc mag has a pentium up against the best quadra. It is more than 300%
faster than the quadra on all tests. Once again, for 3 bucks you folks
could save the world from viewing your stupidity.

Argue about the OS, not about benchmarks you invent.
SD

--
*******
hello
*******

Howard Berkey

unread,

Jul 23, 1993, 2:27:27 AM7/23/93

to

In article <slayer.743402932@gibbs> sla...@gibbs.oit.unc.edu (Scott DeMarchi) writes:
>
>marlin.nosc.mil is a good one. The 486 beats the mac at every chipspeed
>on EVERY benchmark (i.e., 486/33>68040/33, etc. -- on diff versions of mips,
>specmarks, etc.). If you lame

*yawn*
Guess again, dweeb.

I suggest you either check the SPECmark ratings again or learn to
read, barring that.

MIPS are meaningless between two processors.

Look again to verify this, but a 486DX25 is around 8.7 SPECmark89 and a 25
MHz 68040 is around 10.9. A 486DX33 is around 11.1 SPECmark89.

Motorola chips totally blow intel chips of the same clock speed,
especially at floating point operations.

I can pull the exact numbers off comp.arch if you like... or get them
off nosc.mil.

--
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
Howard Berkey how...@netcom.com
Ever notice that most of the people against public sex education are
people that you wouldn't want to have sex with in the first place?
... .. ... ... .. ... ... .. ... ... .. ... ... .. ... ... .. ...

GRUBB

unread,

Jul 23, 1993, 10:36:23 AM7/23/93

to

sla...@gibbs.oit.unc.edu (Scott DeMarchi) writes:

>Several people have posted to this group telling you folks that you idea that
>the mac is faster than the equiv pc clone is FALSE. And I'm even more tired
>of cbs...@ccnga.waterloo.edu writing in without cites to claim that a
>68040 40mhz is the equiv. of a 486 66mhz.

You forgot *me* the author of 'Mac & IBM Info'. I *do* provide
citations, in the sheet and in debates.

>You folks have three options:

>1) check the military archives where they keep benchmarks.

>marlin.nosc.mil is a good one. The 486 beats the mac at every chipspeed
>on EVERY benchmark (i.e., 486/33>68040/33, etc. -- on diff versions of mips,
>specmarks, etc.). If you lame

>ass reality deficient mac users would simply ftp files from this site, I
>wouldn't have to listen to this shit.

There are problems with these benchmarks.
Programmers have found ways to 'trick' most of the benchmarks to producing
higher number than the *average* user would *ever* see. MIPS for example
One other SPecmarks89 and SPECint92 and SPECfp92 has changed *so* much than
the 89 and 92 numbers can't even be comapared. Besides *this* benchmark
needs UNIX, and diffrent UNIX OS produce different results.
Besides the machines of *today* are faster *even though they have the
EXACT same CPU* {Quadra 700 vs Quadra 800 for example. The 800 is *150%*
_faster_ than the 700 and they _both_ use a 68040/33}.
At *best* these types of benchmarks are *ideal* performance which the
*real* world user *rarely if ever* sees.

>2) buy an issue of the last computer shopper. On real world benchmarks, a
>gateway clone for 1/3 the price of the quadra it ran against, beat the
>quadra on EVERY REAL WORLD PROGRAM (excel, word, etc.). Get a clue for
>only 2.95$.

There are problems *here* too. They state that the Centris was not avaliable
which means the study was done *before* Feb 10, 1992. Now consitering that
the average monthly goes to press *three* months before the date on
the cover why did Computer Shopper wait SIX?!? {July 1993 v13 n7 p180(7)).
{Can you say "overly PC-biased test"? I knew you could.}

FileMaker Pro is not that fastest database for the Mac {in fact it is
one of the poorest performers "FileMaker tends to favor reliabliity
over processing speed" (MacUser 1500:105). Do a study using FOXBASE+
next time guys. Photoshop is sensitive to RAM on the Mac side.
Note this in the study.
People have been complaining for *years* that MicroSoft *intentially*
*cripples* their products on the Mac side AND they also have been
*intentially* hiding OS hooks that make programs run faster under
Windows (undocumented Windows). Use WordPerfect instead.
And so it goes. Some 'study" YUH!

The Ingram Laboratories report (IBM System User, Jan 1992 v13 n1 p43(1))
stated the following:
Conclusions:
1) Quadra 900 and 700 outperform all Intel 486 systems tested at <= 50 MHz.
2) Classic II is the fastest machine under $3,000 retail. 41% faster than
a 16 MHz 386SX.
3) Powerbooks 140 and 170 were the fastest notebooks tested. (Since this
was some time ago, the notebooks were 386SX's, not the 486 series.)
4) IIci beat 25 MHz 486 and all 486SX computers.
5) IIsi beat all 386 computers.
6) LC beat every 386SX except one.
7) Classic had competitive performance with 12 MHz 286 costing 2x as much.

1993 comments on above
1) the Quadra 800 would likely beat out all but the Pentium and PowerPC
machines.
2) The Color Classic is the same speed as the Classic II and is $1,389.
3) the 486 series likely are fastest now.
4) The Centris 610 and 650 are faster than the IIci
5) Centris 610 {which is 266% faster} replaces the IIsi
6) The LC was replaced by the LC II which was replaced by the LC III which is
twice the speed of the LC/LC II.
7) both the Classic and 286 machines are gone {except as used}

A *cited* article, an *unbiased* report. So much for the 'PC advantage'.
And *this* was _before_ Apple *really* went 'price cutting happy' {With
*faster* machines even}.

>3) Pc mag has a pentium up against the best quadra. It is more than 300%
>faster than the quadra on all tests. Once again, for 3 bucks you folks
>could save the world from viewing your stupidity.

Big deal. BYTE Aug 93:64 shows that the PowerPC is 1.44 to 4.7 times
_faster_ than the Pentium "Early benchmark tests show the PowerPC
running about 1.5 to nearly five times faster than a Pentium, depending
on the operation" (BYTE Aug 93:6). And *that* is with an *unoptimized*
set of 'transcendental libraries'. Also the PowerPC Mac prototype
shown back in May had a MPC601/80 CPU {apperently a unexpected by product
of the process for MPC601/66}. What's that sound? Oh it is the Pentium going
into riger motis in mid-1994 and resulting collapce of the Pentium
market.

>Argue about the OS, not about benchmarks you invent.

*We* invent? *We* did *not* invent the benchmarks of the Ingram Laboratories
report. *We* did *not* invent the benchmarks of BYTE Aug 93:64.
*We* did *not* invent the *sloppy* and *inaccurate* research of Computer
Shopper. Face it, the PC world's claim 'The PC is always cheaper and
faster than the Mac' is a BIG LIE. As I keep saying the *only*
studies that *support* such a claim are biased and poorly done.

>hello
And GOODBYE. Don't let the door hit you on the way out.

Steve Fuller

unread,

Jul 23, 1993, 2:33:25 PM7/23/93

to

In article <22ot17...@dns1.nmsu.edu> bgr...@dante.nmsu.edu (GRUBB) writes:

>EXACT same CPU* {Quadra 700 vs Quadra 800 for example. The 800 is *150%*
>_faster_ than the 700 and they _both_ use a 68040/33}.

This is wrong. the 700 uses a 040/25. That is why the 800 is 150% faster;
150% clock speed.

Was is really necessary to use 78 '*'s? You post was painful to read.

/s Steve Fuller sfu...@tamu.edu Texas A&M University

cbs...@ccnga.waterloo.edu

unread,

Jul 23, 1993, 3:31:17 PM7/23/93

to

bgrubb, I don't want to steal your thunder but.....

Q700 & Q800 don't use the same CPU.

CPU equivalents in Centris & Quadra's:

Centris 650 = Quadra 700 = Quadra 900 68040/25 MHz
Quadra 800 = Quadra 950 68040/33 MHz
Centris 610 68LC040/20 MHz

--Chris
--
Christopher Smith
cbs...@boomer.uwaterloo.ca
cbs...@undergrad.math.uwaterloo.ca
DC 3527
(519) 885-1211x3581

cbs...@ccnga.waterloo.edu

unread,

Jul 23, 1993, 3:25:49 PM7/23/93

to

Geez. I ftp'd all this stuff, and I *still* can't find a test for
Macintosh machines..... sheesh! The closest thing I could find were
tests of NeXTStation Turbo's running the "can I crawl any faster" NS
3.0. They seemed to have different results for the 33MHz machine that
ranged from "better than any 486-33 and a lot of DX2-66's" to "slower
than an ant". I noticed that the Gateway machine managed to outperform
an RS/6000 550........

Just goes to show that compilers can make *way* too much difference with
C programs. :-)

cbs...@ccnga.waterloo.edu

unread,

Jul 23, 1993, 3:59:07 PM7/23/93

to

sfu...@tamsun.tamu.edu (Steve Fuller) writes:
>In article <22ot17...@dns1.nmsu.edu> bgr...@dante.nmsu.edu (GRUBB) writes:
>
>>EXACT same CPU* {Quadra 700 vs Quadra 800 for example. The 800 is *150%*
>>_faster_ than the 700 and they _both_ use a 68040/33}.
>
>This is wrong. the 700 uses a 040/25. That is why the 800 is 150% faster;
>150% clock speed.
>
>Was is really necessary to use 78 '*'s? You post was painful to read.

ARGH!!!! This is *really* bad.

How do you get 150% x 25MHz = 33MHz??????

I think we have a contagious case of "errors" running around here......

33 is 132% of 25.

Howard Berkey

unread,

Jul 23, 1993, 3:04:52 PM7/23/93

to

In article <CBSMITH.93...@ccnga.waterloo.edu> cbs...@ccnga.waterloo.edu writes:
>
>Just goes to show that compilers can make *way* too much difference with
>C programs. :-)
>

One of the best places to get benchmarks for hardware is, strangely
enough, from the CPU vendors themselves. They are guaranteed to be
using the best performance compilers available to benchmark their
product. The original SPECmark89 values I posted came from a
comp.arch archive and were referenced as being from the vendors (Intel
and Motorola) themselves.

-H-

GRUBB

unread,

Jul 23, 1993, 8:07:07 PM7/23/93

to

cbs...@ccnga.waterloo.edu writes:
>bgrubb, I don't want to steal your thunder but.....
>Q700 & Q800 don't use the same CPU.

Oops.

>CPU equivalents in Centris & Quadra's:

Followed by speeds {times Classic}

>Centris 650 = Quadra 700 = Quadra 900 68040/25 MHz

11 times 10 times 10 times

>Quadra 800 = Quadra 950 68040/33 MHz

15 times 13 times
{Yes, the 800 IS faster than the 950}.

>Centris 610 68LC040/20 MHz
8 times classic speed.

Just shows that while my *example* may have been wrong {and boy was it}
the *conclusion* that use _other_ 68040 benchmarks to demonstrate
Mac performance is DUMB idea unless you caveat the statement
and have something that shows the Mac performance against the
machine whose specs you are citing.
{I do this with my update of the Ingram Laboratories report}

GRUBB

unread,

Jul 23, 1993, 8:21:37 PM7/23/93

to

sfu...@tamsun.tamu.edu (Steve Fuller) write:

I wrote:
>>EXACT same CPU* {Quadra 700 vs Quadra 800 for example. The 800 is *150%*
>>_faster_ than the 700 and they _both_ use a 68040/33}.

>This is wrong. the 700 uses a 040/25. That is why the 800 is 150% faster;
>150% clock speed.

Given. Care the explain the following:

cbs...@ccnga.waterloo.edu writes:
>CPU equivalents in Centris & Quadra's:

Followed by speeds {times Classic}

>Centris 650 = Quadra 700 = Quadra 900 68040/25 MHz
11 times 10 times 10 times

>Quadra 800 = Quadra 950 68040/33 MHz
15 times 13 times
{Yes, the 800 IS faster than the 950}.

>Centris 610 68LC040/20 MHz
8 times classic speed.

Now the 800 is 15/13 times faster than the 950 or 115%. Since the new
SIMM interleaving set up explains 10% of this WHERE is the other 5% coming
from? Same CPU, same SCSI chip {8-bit SCSI-2} both have built-in video,
so what is it?

Gordon Hooker

unread,

Jul 24, 1993, 12:05:05 AM7/24/93

to

In article <22ot17...@dns1.NMSU.Edu> bgr...@dante.nmsu.edu (GRUBB) writes:

>sla...@gibbs.oit.unc.edu (Scott DeMarchi) writes:
>
>Besides the machines of *today* are faster *even though they have the
>EXACT same CPU* {Quadra 700 vs Quadra 800 for example. The 800 is *150%*
>_faster_ than the 700 and they _both_ use a 68040/33}.
>

The Quadra 700, like the Quadra 900, runs at 25 mhz, not 33.

>>hello
>And GOODBYE. Don't let the door hit you on the way out.

Yes, GOODBYE. I'll try not to bust the door off it hinges on my way out.

Gordon Hooker

unread,

Jul 24, 1993, 12:40:56 AM7/24/93

to

In article <CBSMITH.93...@ccnga.waterloo.edu> cbs...@ccnga.waterloo.edu writes:

>sfu...@tamsun.tamu.edu (Steve Fuller) writes:
>>In article <22ot17...@dns1.nmsu.edu> bgr...@dante.nmsu.edu (GRUBB) writes:
>>
>>>EXACT same CPU* {Quadra 700 vs Quadra 800 for example. The 800 is *150%*
>>>_faster_ than the 700 and they _both_ use a 68040/33}.
>>
>>This is wrong. the 700 uses a 040/25. That is why the 800 is 150% faster;
>>150% clock speed.
>>
>>Was is really necessary to use 78 '*'s? You post was painful to read.
>
>ARGH!!!! This is *really* bad.
>
>How do you get 150% x 25MHz = 33MHz??????
>
>I think we have a contagious case of "errors" running around here......
>
>33 is 132% of 25.
>

The difference between the 132% and 150% are explained by the faster memory used by the Q800 over the 950- I'm sure most of you knew that, but at least one didn't.

Steve Fuller

unread,

Jul 24, 1993, 2:38:19 AM7/24/93

to

In article <22pvah...@dns1.nmsu.edu> bgr...@dante.nmsu.edu (GRUBB) writes:

>sfu...@tamsun.tamu.edu (Steve Fuller) write:
>I wrote:

>>>EXACT same CPU* {Quadra 700 vs Quadra 800 for example. The 800 is *150%*
>>>_faster_ than the 700 and they _both_ use a 68040/33}.
>
>>This is wrong. the 700 uses a 040/25. That is why the 800 is 150% faster;
>>150% clock speed.

>Given. Care the explain the following:

OK, I should have said 'That is one reason the 800 is 150% faster'.
(Are all you nit pickers happy now?)

>cbs...@ccnga.waterloo.edu writes:

[munch]

>Now the 800 is 15/13 times faster than the 950 or 115%. Since the new
>SIMM interleaving set up explains 10% of this WHERE is the other 5% coming
>from? Same CPU, same SCSI chip {8-bit SCSI-2} both have built-in video,
>so what is it?

Well, if it not due to pixie-dust or rounding errors, it could be due to
Dale Adams' Most Excellent Video Controller*. The 800 and 950 do not have the
same video hardware. (950 supports 24bit at some sizes.)

Just a guess.

* Not it's real name.

Timothy Connal Delaney

unread,

Jul 25, 1993, 2:29:44 AM7/25/93

to

bgr...@dante.nmsu.edu (GRUBB) writes:

>Now the 800 is 15/13 times faster than the 950 or 115%. Since the new
>SIMM interleaving set up explains 10% of this WHERE is the other 5% coming
>from? Same CPU, same SCSI chip {8-bit SCSI-2} both have built-in video,
>so what is it?

I'm not entirely sure, but I believe it is because the CPU of the 800 is
a more highly integrated board ... fewer bits (about 3) hanging off it than
on the 950 (about 11 additional boards) if I remember correctly.

_/_/_/_/
_/_| _/_| _/_| _/_/_/ _/_| _/_/_/
_/ _| _/ _| _/ _| _/ _/ _| _/ _/
_/ _|_/ _| _/_/_/_| _/ _/_/_/ _/_/_/_| _/ _/
_/ _| _| _/ _| _/_/_/_/ _/ _| _/_/_/

Tim Delaney u925...@wampyr.cc.uow.edu.au

Unknown

unread,

Jul 23, 1993, 10:08:35 AM7/23/93

to

In article <slayer.743402932@gibbs>, sla...@gibbs.oit.unc.edu (Scott

You (general you, not specific) have many ways to realize how ignorant the
originator of this thread is.
1) realize that it is very easy to make your processor run extremely well
on any benchmark if you cheat to do so when you make the chip. Intel is
very good at making chips that aren't any faster but run the benchmarks
better. This is from a Computer Science major at Princeton, I'm not making
this up.
2) I have that issue of Computer Shopper. These people are idiots. They
said in the article that accessing from disk is faster than from ROM.
Oops! Learn your prefixes. Nano is smaller that milli! They also ignored
many important features of the macintosh. BTW, I don't care HOW fast a
machine is if I can't even USE the damn thing.
3) PC Magizine makes all of their money from (surprisingly) PC'S! If
anyone could afford the Pentium and the bank of fifty fans you need to run
it, they are wasting their time, because their is no software that is
powerful enough to use all that speed. PLUS software has to be recompiled
to take advantage of all the Pentiums features, and it still won't be a
whole heck of a lot faster for all but the most demanding applications.
4) 040's use clock-doubling, too, but Motorola doesn't flaunt it like
intel. A 33 MHz 040 is really cranking out crap at 66 MHz.
5) The mac GUI is so much faster than Windoze that the user doesn't notice
a speed difference. It's true! It may take longer to do an Excel recalc
on a mac, but entering the data, launching the program, setting prefs, and
do anything else is much faster. Mac hardware is much more sophisticated,
and you don't have to install a driver just to use a fucking mouse!

So as you see, the Computer Shopper stuff was NOT real world. The just had
the processor puking information. To actually DO anything useful with a
computer, buy a mac.

-Mark
2)

GRUBB

unread,

Jul 26, 1993, 10:56:21 AM7/26/93

to

Let us try this again.

sla...@gibbs.oit.unc.edu (Scott DeMarchi) writes:

>Several people have posted to this group telling you folks that you idea that
>the mac is faster than the equiv pc clone is FALSE. And I'm even more tired
>of cbs...@ccnga.waterloo.edu writing in without cites to claim that a
>68040 40mhz is the equiv. of a 486 66mhz.

You forgot *me* the author of 'Mac & IBM Info'. I *do* provide
citations, in the sheet and in debates.

>You folks have three options:

Not really

>1) check the military archives where they keep benchmarks.
>marlin.nosc.mil is a good one. The 486 beats the mac at every chipspeed
>on EVERY benchmark (i.e., 486/33>68040/33, etc. -- on diff versions of mips,
>specmarks, etc.). If you lame
>ass reality deficient mac users would simply ftp files from this site, I
>wouldn't have to listen to this shit.

There are problems with these benchmarks.
Programmers have found ways to 'trick' most of the benchmarks to producing

higher number than the average user would *ever* see. MIPS for example.

One other SPecmarks89 and SPECint92 and SPECfp92 has changed *so* much than

the 89 and 92 numbers can't even be compared. Besides this benchmark
needs UNIX, and diffrent UNIX OSes produce different results.
Also none of the benchmarks use a Mac for the 68040 info.
If we can use the NeXT in place of the Mac then we can say that the
BYTE Aug 93:142 shows something is wrong with these benchmarks.
NeXTStep on the Intel 486dx2/66 just *bearly* out performs the
68040/33 not the 50% faster that the SPEC numbers would indicate.

Besides the machines of today are faster even though they have the
EXACT same CPU {Quadra 950 vs Quadra 800 for example. The 800 is *15%*
_faster_ than the 950 and they _both_ use a 68040/33}.
These bechmarks are too old to be of use.

>2) buy an issue of the last computer shopper. On real world benchmarks, a
>gateway clone for 1/3 the price of the quadra it ran against, beat the
>quadra on EVERY REAL WORLD PROGRAM (excel, word, etc.). Get a clue for
>only 2.95$.

There are problems *here* too. They state that the Centris was not avaliable
which means the study was done *before* Feb 10, 1992. Now consitering that

the average monthly goes to press three months before the date on

the cover why did Computer Shopper wait SIX?!? {July 1993 v13 n7 p180(7)).
{Can you say "overly PC-biased test"? I knew you could.}

Also the study was filled with stupidity.
For example they said:
"Beginning at the top level, in the case of our "benchmark of
benchmarks"--FileMaker Pro--the Gateway 2000 486DX/33 was almost 18 percent
faster than the Mac Quadra 950. The 486DX2/66 was 40 percent faster, and
its price is about half that of the Quadra."
Fine, except that is NOT what the data they use {and leave out of the
electronic copy of their sheet} says.

The math they should have used is simple
486DX2/66 time / Quadra 950 time - 1 = % that the 486DX2/66 is faster
than the Quadra 950. The result is *60%* faster for the 486DX2/66.
with FileMaker Pro not the 40% they {somehow} came up with.

If these guys cannot do addition/subtraction and division correctly
then *ALL* the data in this 'study' is suspect since an average involves the
same skills as figuring a percentage {just in a different order}.

Scott DeMarchi

unread,

Jul 26, 1993, 2:47:04 PM7/26/93

to

In <230ral...@dns1.NMSU.Edu> bgr...@dante.nmsu.edu (GRUBB) writes:

>Let us try this again.

>higher number than the average user would *ever* see. MIPS for example.
>One other SPecmarks89 and SPECint92 and SPECfp92 has changed *so* much than
>the 89 and 92 numbers can't even be compared. Besides this benchmark
>needs UNIX, and diffrent UNIX OSes produce different results.
>Also none of the benchmarks use a Mac for the 68040 info.
>If we can use the NeXT in place of the Mac then we can say that the
>BYTE Aug 93:142 shows something is wrong with these benchmarks.
>NeXTStep on the Intel 486dx2/66 just *bearly* out performs the
>68040/33 not the 50% faster that the SPEC numbers would indicate.

Which means it is all OS dependant, which was my point. I was tired of mac
heads telling me how much faster their machines were (which they aren't).
I, for example, have a 486dx50 clone, and a Q800. In every app i own, the
486 beats the socks off the quadra. Conclusion: mac heads citing benchmarks
1) make lots of excuses 2) only look at the ones which support their own
preconceptions 3) ignore the fact that the only thing that matters is
the OS.

>There are problems *here* too. They state that the Centris was not avaliable
>which means the study was done *before* Feb 10, 1992. Now consitering that
>the average monthly goes to press three months before the date on
>the cover why did Computer Shopper wait SIX?!? {July 1993 v13 n7 p180(7)).
>{Can you say "overly PC-biased test"? I knew you could.}

More conspiracy theory from mac heads. APple will be a cement sidewalk
and they'll still be yelling "FOUL"

Do any of you folks have proof on this Microsoft thing? I can see intentional
crippling, but who would notice a small speed differential? Too hard
to believe without any proof.

>The Ingram Laboratories report (IBM System User, Jan 1992 v13 n1 p43(1))
>stated the following:

>A *cited* article, an unbiased report. So much for the 'PC advantage'.
>And this was _before_ Apple really went 'price cutting happy' {With
>faster machines even}.

Come on guys. Everyone knows the Ingram labs reports are foolish. Testing
shitty PC's against macs is not what anyone is advocating. The idiots
didn't use local bus, or good graphics cards -- even a quick perusal of
the PC market shows you can buy cards that are radically diff in performance
SO all the ingram report proves is that rotten pc's are worse than macs.
Wow, it's surely miller time for the mac world.
If you want another real world bench, try MacUser's scroll bar test --
take MSword, and hit the down arrow on the scroll bar. Use a ten page doc
in 10 pt. times. A quadra 800 takes forever, while a 486/50 takes <3 seconds.
Every benchmark is manipulatable...

>Only the dramatic speed increace of Windows 3.1 compared to the one of System
>7.1 keeps the PC world from getting its colective butt kicked speedwise.
>{Windows NT beta runs Windows programs up to 10% _slower_ (PC Week 03/15/93)
>so unless NT gets better the PC will be back down to near Windows 3.0 speeds.}

Proving my point that people need to argue about OS's.

>Also the BYTE Aug 93:142 benchmarks show the 68040/33 {NeXT Turbo} performing
>just *5*% below {if that} a 486dx2/66.

Gee, I wonder if it is the port. Steve Jobs has said that the 486
should run Nexstep FASTER than the 040. Hmmm.

>"For 3
bucks you folks could save the world from viewing your stupidity."
>Yes like the "stupidity" that somehow IDEAL benchmarks like SPECmark and
>corrupted ones like MIPS have *anything* to do with the real world.
>
>Talk about calling the kettle black. Follow your OWN advice.

>>Argue about the OS, not about benchmarks you invent.
>We invent? We did not _invent_ the benchmarks of the Ingram Laboratories
>report. We did not _invent_ the benchmarks of BYTE Aug 93:64 nor the
>ones of BYTE Aug 93:142.

Once again, all benches BUT the Ingram reports are suspect. You guys are
great. ONE report favors the mac, and you treat it like the New Testament.
Bet you've seen PC demons too.

>WE did not _invent_ the *sloppy* and *inaccurate* research of Computer
>Shopper. Face it, the PC world's claim 'The PC is always cheaper and
>faster than the Mac' is a BIG LIE. As I keep saying the *only*
>studies that support such a claim are biased and poorly done.
>{The Computer Shopper one is one of the WORST that I have seen in a _long_
>time. They cannot not even add/subtract or divide correctly.}

>>hello
>And GOODBYE. Don't let the door hit you on the way out.
>

Sigh.

--
*******
hello
*******

Graham K. Johnson

unread,

Jul 26, 1993, 3:37:26 PM7/26/93

to

In article <230ral...@dns1.NMSU.Edu> bgr...@dante.nmsu.edu (GRUBB) writes:
>sla...@gibbs.oit.unc.edu (Scott DeMarchi) writes:

>>1) check the military archives where they keep benchmarks.[marlin.nosc.mil]

>
>There are problems with these benchmarks.
>Programmers have found ways to 'trick' most of the benchmarks to producing
>higher number than the average user would *ever* see. MIPS for example.

So Intel specifically designed their 486's around SPECint92? I don't
think so. This is ridiculous.

>One other SPecmarks89 and SPECint92 and SPECfp92 has changed *so* much than
>the 89 and 92 numbers can't even be compared. Besides this benchmark
>needs UNIX, and diffrent UNIX OSes produce different results.

No, they don't need UNIX; they need some decent way of measuring CPU
time used and a compiler. UNIX provides both of those. Don't blame
the results of the Macs on the lack of a decent UNIX for them.

>Also none of the benchmarks use a Mac for the 68040 info.

WRONG!!!!!:
-------
Towers of Hanoi (hanoi.c) results are included below.

The program (hanoi.c) and latest results (hanoi.tbl) can be obtained
via anonymous ftp from 'marlin.nosc.mil' in directory 'pub/aburto'.
The 'marlin' host number is: 128.49.4.4

Please send new results (new machines, compilers, compiler options) to:
abu...@marlin.nosc.mil. I will keep the results up-dated and post
periodically to 'comp.benchmarks'. Thank you very much.

Towers of Hanoi Puzzle Test Results.

Results as of 19 Jul 1993:
CPU Moves in
System OS, Compiler CPU (MHz) 25 usec ref
--------------------- ----------------------- -------- ------ -------- ---
DEC 3000/500 Alpha OSF/1 T1.3-3, cc -O4 DEC21064 150.0 81.623 47
[highest result]
...
Sun SS 10/30 NOTE 010, SunOS 4.1.3 SSPARC 36.0 39.569 49
[just left this in because I replicated it (38.5xx) under Solaris 2.2]
...
486DX2/66 EISA NOTE 021, OS/2 2.0 486DX2 66.7 22.617 48
80486/50 NOTE 029 80486DX 50.0 19.056 12
Gateway DX2-66 NOTE 030, LINUX 0.99 486DX2 66.7 18.654 29
Macintosh Quadra 950 Sys 7.01, Think C 5.03 68040 33.3 14.034 23
CLUB Falcon 433, ISA NOTE 039, MS DOS 5.0 80486DX 33.3 13.834 36
AMI 80486DX/33 NOTE 042,Linux.99pl7-38 80486DX 33. 13.240 45
Macintosh Quadra 950 A/UX 3.0,cc-DUNIX_Old-O 68040 33.3 13.066 53
Macintosh Quadra 950 NOTE 049, A/UX 3.0 68040 33.3 11.673 53
NeXTstation Mach/NextStep2.2,bin/cc 68040 25.0 11.219 20
Macintosh Quadra 700 NOTE 055,Apple A/UX 3.0 68040 25.0 9.702 41

NOTES:
...
021 Borland C++ V1.0 for OS/2 V1.0. AMI Enterprise Motherboard, EISA
and VL bus slots. Exceptional good result for the 486DX2/66.
[interesting that the best result DIDN'T use UNIX]
...
029 DELL Unix V4 Version 2.2, gcc 2.2.2, gcc -m486 -funroll-loops
-fomit-frame-pointer -I/usr/ucbinclude -lc -lucb
030 gcc 2.2.2, gcc -DUNIX -m486 -O6 -funroll-loops
049 gcc 2.4.5, gcc -DUNIX_Old -O3 -finline-functions
-------
Fhourstones benchmark results are included below.

...

The Fhourstones benchmark (c4 for short) solves positions in connect-4,
...
mix of scalar operations. Fhourstones is a pure 'integer benchmark'.
...
Results as of 22 Jul 1993:

CPU KPos/
System OS, Compiler CPU (MHz) sec REF
---------------------- ------------------------ ---------- ----- ------ ---
DEC 10000/610 NOTE 001, OSF/1 T1.3 DEC 21064 200.0 167.7 42
...
Gateway 486DX2/66 NOTE E03, LINUX 0.99p6 80486DX2 66.7 64.7 27
AMI 80486DX2/66 EISA NOTE B01, MS DOS 5.0 80486DX2 66.7 56.5 23
NeXTstation Mono Turbo NOTE E08, NeXTSTEP 3.0 68040 33.3 39.2 34
CLUB Falcon 433, ISA NOTE B03, MS DOS 5.0 80486DX 33.3 36.7 23
NeXTstation Turbo NOTE 038, Mach 3.0 68040 33.3 29.8 24
Macintosh Quadra 950 NOTE 039, A/UX 3.0 68040 33.3 29.8 45
NeXTstation NOTE 022, NeXTSTEP 2.1 68040 25.0 21.2 15
Macintosh Quadra 950 A/UX3.0,cc-DUNIX_Old-O 68040 33.3 21.2 45

NOTES:
B01 gcc -DUNIX -DSMALL -m486 -O -fexpensive-optimizations
-fomit-frame-pointer
B03 gcc -DUNIX -m486 -O -fexpensive-optimizations -fomit-frame-pointer
E03 gcc -DUNIX -DSMALL -O2 -fexpensive-optimizations -finline-functions
-fomit-frame-pointer -fmemorize-lookups
039 gcc 2.4.5, gcc -DUNIX_Old -O3 -finline-functions
-fexpensive-optimizations
-------
Well, obviously you've been proven wrong.

>If we can use the NeXT in place of the Mac then we can say that the

Look again.

>Besides the machines of today are faster even though they have the
>EXACT same CPU {Quadra 950 vs Quadra 800 for example. The 800 is *15%*
>_faster_ than the 950 and they _both_ use a 68040/33}.

Sorry, that's a general number and includes factors (i.e. video
performance) that would not affect a benchmark.

>These bechmarks are too old to be of use.

Check the date. If you don't like them, send your own in.

>Also the [Computer Shopper] study was filled with stupidity.
>...

>The math they should have used is simple
>486DX2/66 time / Quadra 950 time - 1 = % that the 486DX2/66 is faster
>than the Quadra 950. The result is *60%* faster for the 486DX2/66.
>with FileMaker Pro not the 40% they {somehow} came up with.

That's 66% not 60%, but you're getting much closer now.

--
___
/__ ___ "Uh, you're not gonna joh...@cps.msu.edu
<__/raham / eat that, are you?" Michigan State University
|<. <_/ohnson --Fish Tales East Lansing, Michigan

cbs...@ccnga.waterloo.edu

unread,

Jul 26, 1993, 4:58:25 PM7/26/93

to

I feel like I'm being slighted here, so I thought I'd correct this. I
was commenting on the person's statement that the clock speed of the
Q800 was 150% the clock speed of the Q700, which is just *not* the case.
The faster memory is the single largest contributor to the Q800 speed
improvement over the Q700. Other minimal improvements are: faster I/O
bus and faster graphics (by virtue of using 16-bit over 24-bit).

cbs...@ccnga.waterloo.edu

unread,

Jul 26, 1993, 5:59:02 PM7/26/93

to

sla...@gibbs.oit.unc.edu (Scott DeMarchi) writes:
>In <230ral...@dns1.NMSU.Edu> bgr...@dante.nmsu.edu (GRUBB) writes:
>
>>Let us try this again.
>>higher number than the average user would *ever* see. MIPS for example.
>>One other SPecmarks89 and SPECint92 and SPECfp92 has changed *so* much than
>>the 89 and 92 numbers can't even be compared. Besides this benchmark
>>needs UNIX, and diffrent UNIX OSes produce different results.
>>Also none of the benchmarks use a Mac for the 68040 info.
>>If we can use the NeXT in place of the Mac then we can say that the
>>BYTE Aug 93:142 shows something is wrong with these benchmarks.
>>NeXTStep on the Intel 486dx2/66 just *bearly* out performs the
>>68040/33 not the 50% faster that the SPEC numbers would indicate.
>
>Which means it is all OS dependant, which was my point. I was tired of mac
>heads telling me how much faster their machines were (which they aren't).
>I, for example, have a 486dx50 clone, and a Q800. In every app i own, the
>486 beats the socks off the quadra. Conclusion: mac heads citing benchmarks
>1) make lots of excuses 2) only look at the ones which support their own
>preconceptions 3) ignore the fact that the only thing that matters is
>the OS.

Oh wow! To think we have a PC user saying that it doesn't matter what
your hardware is, the real key thing is the OS? Gee! How many magazines,
etc., have conceded that the MacOS is the greatest thing since sliced
bread?

>>There are problems *here* too. They state that the Centris was not avaliable
>>which means the study was done *before* Feb 10, 1992. Now consitering that
>>the average monthly goes to press three months before the date on
>>the cover why did Computer Shopper wait SIX?!? {July 1993 v13 n7 p180(7)).
>>{Can you say "overly PC-biased test"? I knew you could.}
>More conspiracy theory from mac heads. APple will be a cement sidewalk
>and they'll still be yelling "FOUL"

It's not a consipracy. It's stupidity. Tell me *why* C-Shopper would
select to do benchmark tests on macs, only a month before the Mac line
was going to be majorly revamped (not a secret!), when they planned not
to publish the result for six months? I've heard arguements that
C-Shopper did the benchmark in advance, but only had so much space for
articles (what with all that space for ads :-) and consequently could
not publish until such a late date. Great, if you *know* your reports
are going to take a while to get to press, why not do the studies when a
new line is *just* announced, so that there is a chance they'll be
reasonably up to date when the suckers hit the stands, you won't look
like in irresponsible fool.

I don't think the C-Shopper report was a conspiracy. It was just stupid
and irresponsible. I think that is a fair statement to make, and I think
that it a reasonable reason to discount the entire periodcal as a
reference. (Notice, I did't say "all PC magazines". I said C-Shopper.)

Personally, I question the "intentional crippling" thing myself, but I'd
be willing to bet MS invests a *hell* of a lot more reasources getting
their PC versions to run faster. Why? Because they want their OS to win
out. These are products to augment the value of their OS.

MS has control of Windows, so they can tweak it to make their apps run
faster. They can't do this in the MacOS arena.

>>The Ingram Laboratories report (IBM System User, Jan 1992 v13 n1 p43(1))
>>stated the following:
>>A *cited* article, an unbiased report. So much for the 'PC advantage'.
>>And this was _before_ Apple really went 'price cutting happy' {With
>>faster machines even}.
>
>Come on guys. Everyone knows the Ingram labs reports are foolish. Testing
>shitty PC's against macs is not what anyone is advocating. The idiots
>didn't use local bus, or good graphics cards -- even a quick perusal of
>the PC market shows you can buy cards that are radically diff in performance
>SO all the ingram report proves is that rotten pc's are worse than macs.
>Wow, it's surely miller time for the mac world.
>If you want another real world bench, try MacUser's scroll bar test --
>take MSword, and hit the down arrow on the scroll bar. Use a ten page doc
>in 10 pt. times. A quadra 800 takes forever, while a 486/50 takes <3 seconds.
>Every benchmark is manipulatable...

Ok. Grab a brain. Ingram labs did not test "shitty PC's". They tested
PC's from leading manufacturer's like IBM, Compaq, Dell, ALR, and I
think even the now ledgendary Gateway. When they did the tests, those
were the best machines around. Granted PC's have improved since that
test was made. Guess what? So have Mac's!!! Granted PC's have become
less expensive since that test was made. Guess what? So have Mac's!!!
Granted PC's had/have better graphics cards available for them. Guess
what? So did/do Mac's!!!! Amazing isn't it???

And you are going to tell me that a scroll bar test is the key to
everything, but a generic test done by Ingram labs is a waste of paper?
Come on!!! I don't know about you, but high speed scrolling is hardly
what I'm looking for. That has a *lot* more to do with how your OS
handles objects etc. (which is *always* a tradeoff) than which OS or
machine is faster.

>>Only the dramatic speed increace of Windows 3.1 compared to the one of System
>>7.1 keeps the PC world from getting its colective butt kicked speedwise.
>>{Windows NT beta runs Windows programs up to 10% _slower_ (PC Week 03/15/93)
>>so unless NT gets better the PC will be back down to near Windows 3.0 speeds.}
>
>Proving my point that people need to argue about OS's.

Right. In which case.... I'll cite BYTE magazine's big performance
comparison of "workstations" (using the Quadra 800 & Pentium Compaq I
believe) saying that the Mac, while not as powerfull as the rest, had
the best OS. I'd say that although BYTE's advertising is pretty PC
biased, that they are still a fair source of opinion. They can be wrong
sometimes, and they can also be right. Maybe they were wrong about this,
and maybe they were right, but dammit, it sure looks good to me!!!

>>Also the BYTE Aug 93:142 benchmarks show the 68040/33 {NeXT Turbo} performing
>>just *5*% below {if that} a 486dx2/66.
>
>
>Gee, I wonder if it is the port. Steve Jobs has said that the 486
>should run Nexstep FASTER than the 040. Hmmm.

It should run NeXTStep faster than the 040 'cause they are dropping
features from it. :-)

>>"For 3
>bucks you folks could save the world from viewing your stupidity."
>>Yes like the "stupidity" that somehow IDEAL benchmarks like SPECmark and
>>corrupted ones like MIPS have *anything* to do with the real world.
>>
>>Talk about calling the kettle black. Follow your OWN advice.
>
>>>Argue about the OS, not about benchmarks you invent.
>>We invent? We did not _invent_ the benchmarks of the Ingram Laboratories
>>report. We did not _invent_ the benchmarks of BYTE Aug 93:64 nor the
>>ones of BYTE Aug 93:142.
>
>Once again, all benches BUT the Ingram reports are suspect. You guys are
>great. ONE report favors the mac, and you treat it like the New Testament.
>Bet you've seen PC demons too.

No, but the Ingram labs report does show that the PC user's claim that
"Mac are just sooooo much slower than PC's" is rediculous and clearly
based on biased information. If an independant source concludes that
Mac's are consistently faster than PC's and consistently better
price/performance values, then you can bet that *if* the PC's were
actually faster, it sure isn't by some extreme amount.

>>WE did not _invent_ the *sloppy* and *inaccurate* research of Computer
>>Shopper. Face it, the PC world's claim 'The PC is always cheaper and
>>faster than the Mac' is a BIG LIE. As I keep saying the *only*
>>studies that support such a claim are biased and poorly done.
>>{The Computer Shopper one is one of the WORST that I have seen in a _long_
>>time. They cannot not even add/subtract or divide correctly.}
>
>>>hello
>>And GOODBYE. Don't let the door hit you on the way out.
>>
>Sigh.
>

Sigh.

GRUBB

unread,

Jul 26, 1993, 1:43:04 PM7/26/93

to

(Mark Ashton) writes:
>2) I have that issue of Computer Shopper. These people are idiots. They
>said in the article that accessing from disk is faster than from ROM.
>Oops! Learn your prefixes. Nano is smaller that milli! They also ignored
>many important features of the macintosh. BTW, I don't care HOW fast a
>machine is if I can't even USE the damn thing.

That is not the only example of their stupidity:
Here is one example that shows the flaws in this 'study'.
(Computer Shopper, 07/93 v13 n7 p180(7))

"Beginning at the top level, in the case of our "benchmark of
benchmarks"--FileMaker Pro--the Gateway 2000 486DX/33 was almost 18 percent
faster than the Mac Quadra 950. The 486DX2/66 was 40 percent faster, and
its price is about half that of the Quadra."
Fine, except that is NOT what the data they use {and leave out of the
electronic copy of their sheet} says.

The math they should have used is simple
486DX2/66 time / Quadra 950 time - 1 = % that the 486DX2/66 is faster
than the Quadra 950. The result is *60%* faster for the 486DX2/66.
with FileMaker Pro not the 40% they {somehow} came up with.

If these guys cannot do addition/subtraction and division correctly
then *ALL* the data in this 'study' is suspect since an average involves the
same skills as figuring a percentage {just in a different order}.

More dumb statements:
"Peripheral expansion advantages for the Insight 486SX/25 are significant as
well, with four available ISA slots available, compared to the IIvx's three
NuBus and one accelerator slot."
Now *what* did the PC people say about "nobody compares ISA to NuBus"?
ISA compares with NuBus?!? GIVE ME A BREAK.
Also, since when is the PDS slot an 'accelerator' slot?
PDS can became a NuBus slot {IIsi, Centris 610}, can allow toggeling between
the machine CPU and one on a card {The IIe and upcoming Houdini 486 and 486dx2
cards for example}, and allow 'passthrough' cards {which provide a feature and
allow more PDS cards to be used. There is even a PDS->NuBus card that allows
the PDS to still be used in addition to the NuBus slot.}

"Enhancing the video would take another [NuBus slot],"
Given {they got *something* right}.
"leaving one [NuBus slot] open for a scanner, for example, but not much else."
Sorry Computer Shopper but that is what the *SCSI* port is for.
As I have said on the Internet most of what ISA/EISA is used for in the PC
world is done off of a built-in port {modem/network, printer, drive, ADB,
Sound input/output, SCSI-1, and 8-bit SCSI-2}.

NuBus is limited to 8 {old Macs} through 24-bit video, accelerators {some with
expansion to parallel processing}, CPU-ethernet task sharing, 8 to 16-bit
SCSI-2, DSP, DMA cards {block transfers}, real time video input, and Ethernet.
Due to its speed Ethernet is now becoming a SCSI option for the Macs without
built-in ethernet as an option {Centris and Quadra lines}

"The Insight, on the other hand, could take a LAN card, in modem, and a
scanner, and still have a slot open"
Since modems for most Macs are external, scanners are off the SCSI port and
there is an SCSI-Ethernet expansion, the IIvi {and its successor the Centris
650} would have *3* slots open {1 PDS, 2 NuBus} while the Insight has only one
which is _slower_ by a factor of two to four (Computer Design, 06/01/89 v28
n11 p97(1); I&CS (Instrumentation & Control Systems), 07/92 v65 n7 p23(2)).

Given that Computer Shopper did stupid things like not add, subtract and
divide properly; compared ISA to NuBus; calling the PDS an 'accelorator slot';
ignored the beta machines tested by MacUser in December; waited four to five
months to report their 'findings' the whole study is worthless {It was
worthless as of Feb 10, 1993}.
The up coming macs {August 1993} make it even *more* worthless (MacWeek
07/26/93):

Quadra 840AV {Quadra 800 replacement}: $4,399 - 68040/40 8/230, 3 NuBus, 1
PDS. built-in 16-bit sound input/output, double speed CD-ROM. A built-in
16-bit SCSI-2 port is rumored.

Centris 660AV {Centris 610 replacement}: 68040/25. $2,139 - 8/80, built-in 16-
bit sound input/output; $2,799 - 8/230, double speed CD-ROM.

If this is an example of the 'proof' that the PC world uses to 'show'
that Mac are more expensive and slower than PC then the PC world must
be running scared.

Bill Coleman

unread,

Jul 26, 1993, 6:42:12 PM7/26/93

to

In article <slayer.743402932@gibbs>, sla...@gibbs.oit.unc.edu (Scott DeMarchi) writes:
> Several people have posted to this group telling you folks that you idea that
> the mac is faster than the equiv pc clone is FALSE.

Gee. I've found Macs to be faster.

> You folks have three options:
> 1) check the military archives where they keep benchmarks.

> The 486 beats the mac at every chipspeed on EVERY benchmark

^^^^^^^^^

Could be. I haven't studied every benchmark. Although, I don't spend a
significant amount of my computing time running benchmarks....

> 2) buy an issue of the last computer shopper. On real world benchmarks, a

^^^^^^^^^

> gateway clone for 1/3 the price of the quadra it ran against, beat the
> quadra on EVERY REAL WORLD PROGRAM (excel, word, etc.).

The Computer Shopper "tests" have already become infamous here in the
Mac advocacy group. Not what I would call a "scientific" experiment.

Regardless of the Benchmarks, I say again that I don't spend much time
running benchmarks.

> 3) Pc mag has a pentium up against the best quadra. It is more than 300%
> faster than the quadra on all tests.

This doesn't surprise me. The Pentium should be faster than a Quadra.

> Argue about the OS, not about benchmarks you invent.

^^^^^^^^^^

So far, every "benchmark" you have quoted talks about processor speed. No
matter how fast you make the processor, you've still got to interact with
a user. Running DOS on a Pentium is like having a 600 HP engine in a Yugo.
Sure, theoretically it is fast, but it won't perform well in the real world.

The processor is just one part of the computer, and the computer is just
one part of the human / computer system.

The bottom line is, people are more effective using Macs, with less training
time, than with other types of computers. Macs sport a consistant, easy to
use interface to all applications. Macs are sufficiently fast to get the
job done. People enjoy using Macs. Most people don't enjoy using PCs.

That, my friend, is the most important benchmark of all.

--
Bill Coleman, AA4LR ! CIS: 76067,2327 AppleLink: D1958
Principal Software Engineer ! Packet Radio: AA4LR @ W4QO
Hayes Microcomputer Products, Inc. ! UUCP: uunet!hayes!bcoleman
POB 105203 Atlanta, GA 30348 USA ! Internet: bcoleman%ha...@uunet.uu.net
Disclaimer: "My employer doesn't pay me to have opinions."
Quote: "The same light shines on vineyards that makes deserts." -Steve Hackett.

Russ Taylor

unread,

Jul 26, 1993, 7:15:30 PM7/26/93

to

Actually, the IIvx PDS slot IS properly called an accelerator slot. It lacks
the hooks for performing like a full PDS, according to apple. (this from
MacUser reviews on the P600 and IIvx). I'm quite happy with my IIvx 12/80 CD,
I plan to use the slots for ethernet and a 2-page B&W display.

--
*---------------------------------------------*
* From the often troubled mind of Russ Taylor *
* rta...@ursula.uoregon.edu @cie.uoregon.edu *
*---------------------------------------------*

Jerome Chan

unread,

Jul 26, 1993, 8:56:57 PM7/26/93

to

Someone wrote:

> > Several people have posted to this group telling you folks that you idea that
> > the mac is faster than the equiv pc clone is FALSE.
>
> Gee. I've found Macs to be faster.
>

Do you mean for equally priced systems, the PC is faster at CPU intensive
application?

---
The Evil Tofu

GRUBB

unread,

Jul 26, 1993, 11:50:13 PM7/26/93

to

joh...@cps.msu.edu (Graham K. Johnson) writes:
>In article <230ral...@dns1.NMSU.Edu> bgr...@dante.nmsu.edu (GRUBB) writes:
>>sla...@gibbs.oit.unc.edu (Scott DeMarchi) writes:
>>>1) check the military archives where they keep benchmarks.[marlin.nosc.mil]
>>There are problems with these benchmarks.
>>Programmers have found ways to 'trick' most of the benchmarks to producing

>>higher number than the average user would *ever* see. MIPS for example.

>So Intel specifically designed their 486's around SPECint92? I don't
>think so. This is ridiculous.
It is ridiculous because that is NOT what I said.
What I said was that "*Programmers* have found ways to 'trick' most of the
benchmarks to producing higher number than the average user would *ever* see."
Also note that is use *MIPS* as an example of this *not* SPECint92.
The SPECmark set are *far* less suspeptable to these trick in software than
other benchmarks.

For example. Speedometer use several benchmarks including dyestones.
I ran it and it said that my IIsi was running a Quadra speeds.
What had happened was that I had hid MS Mord wich was still running and
interfered with Speedometer producting high results.

Also on the Mac 24-bit color is a real memory and speed eater and so
a good number of 24-bit NuBus video cards also have accelorators for
quickDraw or the CPU on them. An accelorator will mess up *any* benchmark.
{This was a infamous trick in the early days of benchmarks. There are
some rumors that it goes on today.}

Benchmarks are at best what is called "junk" data. They look impresive
but like junk food they do not bear examination.

Real world results are really subject to *all* kind of monkeying. The
choice of OS, programs, and machine can all effect the results.
This is because benchmarks measure one thing or a combinations of things.
.
Compare the SPECint92 and SPECfp92 of the Pentium and PPC. Then look
at BYTE AUg 93:64. Not what one would expect from the SPECmarks is it?
Both are benchmarks but they show *very* different results. Why?
Because they look at different things.

Howard Berkey

unread,

Jul 27, 1993, 12:38:46 AM7/27/93

to

Grubb. Graham. Put it to rest. Neither of you are convincing
anyone.

Grubb,your info sheet (a good idea except for the massive crossposts
and the fact that 90% of the people on the groups you send it to don't
give a shit) never fails to ignite a big flamewar. While I am all for
a big flamefest, a monotonous one bores the hell out of me. Maybe
next time throw in some juicy lies or something when you post it to
the .advocacy groups so people will get REALLY riled. Maybe claim to
be Bill Gates illegitimate child or something. Or use John_-_Winston
as a source or quote A. Abian or something. Something not involving lots
'o caps and asterisks, please.

Graham, while you started based in fact it seems you're losing it. Or
maybe I have that backwards. Anyhow, It's fun to blast really
ignorant PC freaks, but ones who have 1/2 a deck are merely annoying.
Either be an idiot or buy a mac or leave this group or face the fact
that most mac user's image of PC-KKKlone-abUser is a no-life dweeb
"jacking into cyberspace" at 300 baud. Only joking. The point is,
going in 1/2 cocked only encourages the continuation of dumb boring
flames over BGRUBB's info-sheet. Please, flame whole-heardedly, or
leave. :-)

Really, people. This is an advocacy group, and while huge threads of
the same old flame are definantly in its charter, this one is going
too far.

C'ya,

-H-

--
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
Howard Berkey how...@netcom.com
Love the music... it will love you back.

Howard Berkey

unread,

Jul 27, 1993, 2:42:08 AM7/27/93

to

In article <232gp4$j...@msuinfo.cl.msu.edu> joh...@cps.msu.edu (Graham K. Johnson) writes:

>In article <howardCA...@netcom.com> how...@netcom.com (Howard Berkey) writes:
>>Graham, while you started based in fact it seems you're losing it. Or
>>maybe I have that backwards
>

>No, I think I started in fact and stayed in fact and it frustrates you.
>

Oh yes Graham. I'm so frustrated. Let me grunt in anger that you
can't recognize humor. Ungh Ungh. There, all better now. :-)

>>Anyhow, It's fun to blast really
>>ignorant PC freaks, but ones who have 1/2 a deck are merely annoying.
>

>I am not a "PC freak".
>

Prove it. :-)

>>going in 1/2 cocked only encourages the continuation of dumb boring
>>flames over BGRUBB's info-sheet. Please, flame whole-heardedly, or
>>leave. :-)
>

>Sorry for trying to be somewhat objective/not REALLY HATING
>Macintoshes. I'll try to do better. Maybe I'll go try to install
>PhotoShop on a Mac II and compare it to a 486 and post my results,
>concluding that Macintoshes suck. You can't flame what I say, so you
>flame the fact that I'm not saying anything you can flame. Interesting.
>

Boy, when some people miss the point, they REALLY MISS THE POINT.
Look Graham, if I really wanted to flame you, I would. This is not a
flame. THIS IS AN APPEAL FOR MORE CREATIVE FLAMING, DAMN IT !!!

I could care less if you spend your free (non-self-flagellating)
timeinstalling photoshop on 1990-era macs. I just wanted to se a
non-brain-dead flame, thanks.

>I just think it's really scary that GRUBB's "Info-sheet" is in the same
>directory as the comp.sys.mac.hardware FAQ. People might actually be
>influenced by it. It actually still says that "the Mac tends to be
>cheaper than an equivalent equipped IBM machine." What a joke.
>It would be easy to flame the hell out of cbsmith or Steve Kanefsky every
>day, but GRUBB is the more dangerous one. He's the one claiming
>objectivity and making errors left and right.

I make no apologies for BGRUBB. On the other hand, Steve and Chris
seem to be much more objective. What do they say that makes you so
defensive? :-)

GRUBB

unread,

Jul 27, 1993, 10:02:09 AM7/27/93

to

how...@netcom.com (Howard Berkey) writes:
>In article <232gp4$j...@msuinfo.cl.msu.edu> joh...@cps.msu.edu (Graham K. >

Johnson writes:
>>I just think it's really scary that GRUBB's "Info-sheet" is in the same
>>directory as the comp.sys.mac.hardware FAQ. People might actually be
>>influenced by it. It actually still says that "the Mac tends to be
>>cheaper than an equivalent equipped IBM machine." What a joke.

Is a joke how you took it of context.
In general, when one adds all the standard Mac hardware features to an IBM
{built-in input/output sound support, SCSI, PDS, built-in monitor support,
built-in networking, standard interface, and NuBus 90 equivalent in higher
machines} the Mac tends to be cheaper then an equivalent equipted IBM machine
{Especially since some IBM monitors can be used with Macs which cuts some more
of the Mac's cost (MacUser Aug 1992:158-176)}."

In short if you *add* PC stuff ISA/EISA cards and have equvalent bus
{PDS ~=VL-bus} {NuBus ~= EISA/MCA/EISA-2} to get the same built-in features
the mac will *In general* be cheaper. There will always be exceptions
mail order, used machines and the like.

>>It would be easy to flame the hell out of cbsmith or Steve Kanefsky every
>>day, but GRUBB is the more dangerous one. He's the one claiming
>>objectivity and making errors left and right.

Mind telling me how *your* *misciting* my sheet is evidence that
I am "making errors left and right"?
Remeber the preamble:
"The reason for this general data sheet is to consolidate and condense the
information out there, so that people in both camps can be clear and accurate
about what they are saying about their machines. Since computer technology
is always changing there are always going to be points in which the sheet will
be lacking or incorrect on information. So, please just don't say the sheet
is incomplete or incorrect but also give me clear and concise information to
make the needed corrections. All prices are in US dollars.
To keep this data sheet accurate please provide, if possible, article
citations for the information provided or corrected and keep opinions to
a minimum. As this is a general data sheet, keep the info provided short and
simple. Finally, keep the information relevant to the section corrected.
Thank you."

If there are *still* errors there *is* a way to correct them. All it requires
is a little research to get the citations that show that particular piece of
info is wrong. In some cases where good evidence is provided I forgo the
citation requirement {For example, the "5.3 MB/s max" reference on ISA}.

Don't blame *me* for you slothness. If I *am* "so dangerous" then do
*not* ignore the sheet or take things out of context. Show *all* of us
*what* is in error in the sheet. Show us that is blatently wrong.
For the cited material provide citations to show that the older stuff is in
error. Even the Computer Shopper article got used:
and "A user pumping up a Windows machine past 64Mb (or even 16Mb in some
cases) can encounter some nasty conflicts." (Computer Shopper, 07/93 v13 n7
p180(7)).
Futher more the later on state that the Mac expands to these and higher
MB level with no problem. SO the cite above is in the context of which
it was in the article.

>I make no apologies for BGRUBB. On the other hand, Steve and Chris
>seem to be much more objective. What do they say that makes you so
>defensive?

I seem to be the one that is making him nervous. Steve and Chris are
just mixing it up.
*I* just provide citation to back up my statements.

:-)n) writes:

Bailey Brown

unread,

Aug 3, 1993, 2:35:47 PM8/3/93

to

j...@world.std.com (Jeffrey B Kane) writes:

>I have a 486DX2 and Quadra 33MHz both on my desk. I think a large
>factor in people assuming that the 486 if faster is the perceived
>difference in the GUI. Apple learned a long time ago that certain
>GUI functions need to be throttled down, so that humans could use
>them (scrolling a list is a good example). The Mac lets you read
>the items as they go by by creating an intentional maximum speed.
>Hypercard also has simmilar "throttles". As best I can tell Windows
>does not. By being less sophisticated, the GUI looks faster, but
>what it is, is less refined. Things like window zooming (with
>the Mac's graphics effects of a zooming window) are other examples
>of intentional speed changes.

> Jeffrey

This confirms my long-held suspicion that Mac users, on average, are
willing to put up with a slow computer while PC users are not. If
what you say is true, then this sort of expectation has been
codified in the MacOS.

I want the machine to do everything as fast ast it can. I like
the way many scrollable Windows windows move the text in realtime
lock-step with the scroll slider. This way *I* can decide how fast
I can skim the text, and a 66MHz machine with a fast hardware blit function
on the graphics card makes this silky smooth.

------------
Bailey Brown "Above all else, confusion reigns."
Intergraph Corporation
bbr...@casca.b11.ingr.com Procol Harum

Ishir Bhan

unread,

Aug 3, 1993, 6:00:49 PM8/3/93

to

In <1993Aug3.1...@infonode.ingr.com> bbr...@infonode.ingr.com (Bailey Brown) writes:

>I want the machine to do everything as fast ast it can. I like
>the way many scrollable Windows windows move the text in realtime
>lock-step with the scroll slider. This way *I* can decide how fast
>I can skim the text, and a 66MHz machine with a fast hardware blit function
>on the graphics card makes this silky smooth.

There is a 9k or so CDEV that does this called "Scrolling". I think it's
on ftp.apple.com. It's one of the best and smallest control panels I've
picked up this year, and it's free.
--
-----------------------------------------------------------------------------
Ishir Bhan | Apple Macintosh PowerBook Duo...
ib...@husc.harvard.edu | Finally, a REAL computer!
-----------------------------------------------------------------------------