Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Math Co-Processor and Doom

1,026 views
Skip to first unread message

Jason Weiler

unread,
Jan 18, 1994, 6:00:56 PM1/18/94
to
Chris 'Strunoph' Norman (cbno...@undergrad.math.uwaterloo.ca) wrote:
: Stupid question perhaps, but there's no answer in the FAQ to this one.
: Will a math co-processor speed up Doom? I'm running a 486DLC/40, and
: if I get a 20% increase in speed or more it'll be worth the $$$ to me.

From what I hear, a 386 w/ an FPU will kick ass all over a 486 w/o an FPU
I've seen DOOM run on a 486DX 33 and it was quite smooth and playable. If
you don't have an FPU, I'd strongly recommend getting one, and not just for
DOOM.

Later,
Jason Weiler
wei...@rpi.edu

Message has been deleted

Joseph W. Vigneau

unread,
Jan 18, 1994, 8:22:35 PM1/18/94
to
In article <2hhpn8$9...@usenet.rpi.edu>,

Jason Weiler <wei...@goya.its.rpi.edu> wrote:
>
>From what I hear, a 386 w/ an FPU will kick ass all over a 486 w/o an FPU
>I've seen DOOM run on a 486DX 33 and it was quite smooth and playable. If
>you don't have an FPU, I'd strongly recommend getting one, and not just for
>DOOM.

Um, *all* 486s have an FPU built in, except the SXs... Correct me if I'm
wrong...
--
jo...@wpi.edu - WPI Computer Science |Linux 0.99.14f:Free 32 bit OS for the i386

John C.Kahoon

unread,
Jan 19, 1994, 6:02:36 AM1/19/94
to

The shareware version of doom does NOT use an FPU .The registerd version,
well I'm not sure but, I think the only reason it runs slowly is because
it's ver. 1.1 .I do know that if you have less than 8 meg of ram it won't
run to good. Any way, I'm running a 486 33sx w/8meg (no fpu) and I'm real
happy with running a full screen at low res. It looks like I'm behind the
DOOM cam.

Badger

unread,
Jan 19, 1994, 3:52:15 AM1/19/94
to
I have a 386/40, and just got a math coprocessor. I don't see any
noticable speedup in Doom...what's wrong? Isn't it built to use a
coprocessor? Or did they just use 486-specific code if a 486 is present
(or something)?

I kind of expected to see it run nearly as smooth as a friend's slower
486, due to raw speed. Huh?

Anssi Saari

unread,
Jan 19, 1994, 6:31:02 AM1/19/94
to
In <2hhpn8$9...@usenet.rpi.edu> wei...@goya.its.rpi.edu (Jason Weiler) writes:

:From what I hear, a 386 w/ an FPU will kick ass all over a 486 w/o an FPU


:I've seen DOOM run on a 486DX 33 and it was quite smooth and playable.

It's also quite smooth and playable with my 486SX-33.

:If you don't have an FPU, I'd strongly recommend getting one, and not just for
:DOOM.

What for then? I have no software that uses it, so it'd be wasted money.
Of course, when upgrade time comes, there probably won't be a 486SX2 or SX3,
so then I'll have to waste money on the FPU...

Anssi

--
Anssi Saari s10...@ee.tut.fi
Tampere University of Technology
Finland, Europe

One Lain Meyer

unread,
Jan 19, 1994, 11:31:03 AM1/19/94
to
at...@cleveland.Freenet.Edu (John C.Kahoon) writes:

i'm playing the shareware version 1.1 on a 386 DX 40 with only 4 meg. i'm
using the screen size 2 smaller than full size with minimum detail...
works just wonderful...no choppiness what-so-ever. and yes, i have seen it
played on a 486 DX whatever, and the playability/smoothness doesn't suffer
much at all...

oh and also...you haven't played DOOM until you've played it with a Gravis
pumping out 4 digital channels (plus the music) into a 200 watts/channel amp
driving a pair of speakers with 15 inch ported woofers. crank it all the way up
and you feel DOOM. guess it would be like playing it out in california. =)
check it out...you'll like it.

-OLM

Chris 'Strunoph' Norman

unread,
Jan 19, 1994, 11:54:08 AM1/19/94
to

Some motherboards require you set a jumper to let it know it has a co-pro.
Some let you set it in CMOS. Some autodetect.

If you get any interesting e-mail on this subject, *please* forward it to me.
I'm considering buying a co-pro.

--
/| (Reverend) Christopher Bruce Norman
\`o.@' Ack! Ack! Thbbbpppt! cbno...@descartes.uwaterloo.ca
=(___)= CS246 Tutor (MC 1014)
U "Strip mining prevents forest fires" -- Smokey the Bear

Christian Holtje

unread,
Jan 19, 1994, 4:28:30 PM1/19/94
to
s10...@ee.tut.fi (Anssi Saari) writes:
>What for then? I have no software that uses it, so it'd be wasted money.
>Of course, when upgrade time comes, there probably won't be a 486SX2 or SX3,
>so then I'll have to waste money on the FPU...

>Anssi

Then when you get a computer with a math-co, you can upgrade your
OS to Linux! Then ALL your math software will use it!

-Waiting for Linux-Doom!
-Doctor What

_________From the computer of --Doctor What--____________________(C. Holtje)___
] doc...@uiuc.edu | God is real -- unless declared integer [
] Mail me for help, a | Yappappaa yappappaa iishanten [
] subscription to my | Hashagu koi wa ike no koi. [
] periodical, or for fun | My other account is a Linux account. [
-------------------------------------------------------------------------------
Thought for the day:

Yet I argue not
Against heaven's hand or will, nor bate a jot
Of heart or hope, but still bear up and steer
Right onward.
-- Milton

B. Scott Bilas

unread,
Jan 19, 1994, 9:57:17 AM1/19/94
to
In article <2hjn87$2...@vixen.cso.uiuc.edu> me...@uxa.cso.uiuc.edu (One Lain Meyer) writes:
>From: me...@uxa.cso.uiuc.edu (One Lain Meyer)
>Subject: Re: Math Co-Processor and Doom
>Date: 19 Jan 1994 16:31:03 GMT

>at...@cleveland.Freenet.Edu (John C.Kahoon) writes:


>>The shareware version of doom does NOT use an FPU .The registerd version,
>>well I'm not sure but, I think the only reason it runs slowly is because
>>it's ver. 1.1 .I do know that if you have less than 8 meg of ram it won't
>>run to good. Any way, I'm running a 486 33sx w/8meg (no fpu) and I'm real
>>happy with running a full screen at low res. It looks like I'm behind the
>>DOOM cam.

I do the same -- full screen low res. And I've got a 33DX. It's mostly
bearable unless I get into some heavy action in multiplayer mode. But as for
the FPU use, I can't imagine why anybody would use floating point graphics
routines. Judging from the speed of Doom I'd bet that it's mostly 32-bit
integer math. Take a look at any version of Fractint for DOS for a good
example of the speed difference. That's pure integer math (up to a certain
resolution). And it's faster than any floating point fractal programs.
However, if Doom truly does use the FPU, I'd love to see it run on a Pentium
system. The pipelined Pentium FPU is supposed to be 4 times faster than a
486-66!

>i'm playing the shareware version 1.1 on a 386 DX 40 with only 4 meg. i'm
>using the screen size 2 smaller than full size with minimum detail...
>works just wonderful...no choppiness what-so-ever. and yes, i have seen it
>played on a 486 DX whatever, and the playability/smoothness doesn't suffer
>much at all...

I don't know, I think there's a big difference. You should see my friend's
DX66 with local bus ATI GUP card compared to my 33 with non-LB Speedstar Pro.
Talk about smooth...

>oh and also...you haven't played DOOM until you've played it with a Gravis
>pumping out 4 digital channels (plus the music) into a 200 watts/channel amp
>driving a pair of speakers with 15 inch ported woofers. crank it all the way up
>and you feel DOOM. guess it would be like playing it out in california. =)
>check it out...you'll like it.

Done! Nothing beats a Gravis (or, I suppose, a Roland + a PAS-16 or SB16).
Most of the time (when the neighbors yell), what works beautifully is plugging
some good monitor headphones into the amp output of the Gravis. Doom isn't
supposed to support 3D sound for the Gravis, but it sure _sounds_ that way.
When an imp throws a fireball at you, you can almost hear the thing whizz by
your head and hit the wall behind you. By the way, try it with a surround
processor. It's weird.

>-OLM

___________________________________________________________________
Scott Bilas | Beware of Wild Bill and his SPA Nazis!
Terris Engineering Company | sco...@iastate.edu

David G. Kitaguchi

unread,
Jan 19, 1994, 3:58:25 PM1/19/94
to
In article 7589...@ee.tut.fi, s10...@ee.tut.fi (Anssi Saari) writes:
>In <2hhpn8$9...@usenet.rpi.edu> wei...@goya.its.rpi.edu (Jason Weiler) writes:
>
>:From what I hear, a 386 w/ an FPU will kick ass all over a 486 w/o an FPU
>:I've seen DOOM run on a 486DX 33 and it was quite smooth and playable.
>
>It's also quite smooth and playable with my 486SX-33.
>
>:If you don't have an FPU, I'd strongly recommend getting one, and not just for
>:DOOM.
>
>What for then? I have no software that uses it, so it'd be wasted money.
>Of course, when upgrade time comes, there probably won't be a 486SX2 or SX3,
>so then I'll have to waste money on the FPU...
>
>Anssi

Well in actuallity the real speed advantage of the 486 over the 386 is the 1k internal
cache in the 486 chip (I think I got the size correct. Also not to be confused with
the external cache). So a 486sx-33 should easily blow the benchmarks off of 386-40.

----------------------------------------------------------------+-------------
o \ o / _ o __| \ / |__ o _ \ o / o |dav...@gdb.org
/|\ | /\ __\o \o | o/ o/__ /\ | /|\ |(410)614-0639
/ \ / \ | \ /) | ( \ /o\ / ) | (\ / | / \ / \ |I've flipped!
----------------------------------------------------------------+-------------

One Lain Meyer

unread,
Jan 19, 1994, 11:37:23 PM1/19/94
to
dav...@velociraptor.welchdev (David G. Kitaguchi) writes:

>In article 7589...@ee.tut.fi, s10...@ee.tut.fi (Anssi Saari) writes:
>>In <2hhpn8$9...@usenet.rpi.edu> wei...@goya.its.rpi.edu (Jason Weiler) writes:
>>

[SNIP]

>>What for then? I have no software that uses it, so it'd be wasted money.
>>Of course, when upgrade time comes, there probably won't be a 486SX2 or SX3,
>>so then I'll have to waste money on the FPU...
>>
>>Anssi

>Well in actuallity the real speed advantage of the 486 over the 386 is the 1k internal
>cache in the 486 chip (I think I got the size correct. Also not to be confused with
>the external cache). So a 486sx-33 should easily blow the benchmarks off of 386-40.

umm. sorry but you are incorrect...kinda...a properly set up 386 DX 40 will
tromp on a 486 SX 33, at times. especially FPU intensive work...i remember
looking at the benchmarks on my 386 DX 40 compared to a 486 SX 33 at the
shop when i bought it. can't remember the numbers, but the 40 was a decent
way out in front of the 33.

-OLM

Korhonen Tommi

unread,
Jan 20, 1994, 3:22:14 AM1/20/94
to
In <1994Jan19....@news.gdb.org> dav...@velociraptor.welchdev (David G. Kitaguchi) writes:

>Well in actuallity the real speed advantage of the 486 over the 386 is the 1k internal
>cache in the 486 chip (I think I got the size correct. Also not to be confused with
>the external cache). So a 486sx-33 should easily blow the benchmarks off of 386-40.

Well, the cache does benefit the 486, but it doesn't make it so fast.
The speed diference from 33 -> 40 MHz is not so small either.
Most of the 486sx's dont "blow" my 386 40 that easily. I admit they do,
but the difference is minimal, even as small as 10%! (CheckIt)

--
--------------------------------------------- |
Tommi "Ghost "Korhonen k13...@cc.tut.fi | .___o___.
Physics & Something .___o___. x ^ x
--------------------------------------------- x ^ x

Message has been deleted

Brian Tottleben

unread,
Jan 20, 1994, 11:14:24 AM1/20/94
to
>
>Well, the cache does benefit the 486, but it doesn't make it so fast.
>The speed diference from 33 -> 40 MHz is not so small either.
>Most of the 486sx's dont "blow" my 386 40 that easily. I admit they do,
>but the difference is minimal, even as small as 10%! (CheckIt)
>
>


I own a 386DX40 with no coprocessor. I have a friend that owns a 486DX33.
In benchmarks that do _not_ use a coprocessor, 486DX33 will completely
blow away scores that my computer generates (by almost 200%). Some of
this comes from the use of a local bus, but when comparing Doom playability,
(which from everything that I have seen or heard does not use a
coprocessor), the 386DX40 doesn't even compare. Definitely time for an
upgrade!

regards,

---b.s.
--
Bri...@cie-2.uoregon.edu
"A long and painful death is better than no death at all."

Jose Tomines - EENG/F93

unread,
Jan 20, 1994, 3:56:49 PM1/20/94
to
One Lain Meyer (me...@uxa.cso.uiuc.edu) wrote:

mke...@eecs.wsu.edu

unread,
Jan 20, 1994, 7:08:42 PM1/20/94
to
My brother has a 386DX-40 with the Co-Processor and I bumped it two notches with
low detail, but it _does_ got dog slow on Episode One, Level Three, and Ultra-
violence. When open the door where you see the health bonus, it just drops
drastically!In article <2hjn87$2...@vixen.cso.uiuc.edu>, me...@uxa.cso.uiuc.edu (One Lain Meyer) writes:

|> at...@cleveland.Freenet.Edu (John C.Kahoon) writes:
|>
[stuff deleted]


|>
|> i'm playing the shareware version 1.1 on a 386 DX 40 with only 4 meg. i'm
|> using the screen size 2 smaller than full size with minimum detail...
|> works just wonderful...no choppiness what-so-ever. and yes, i have seen it
|> played on a 486 DX whatever, and the playability/smoothness doesn't suffer
|> much at all...
|>

I don't know how come your 386DX-40 doesn't get bogged down!

-Michael

Doom for the Apple //e (a novelty item), how about it ID?

Alan Krueger

unread,
Jan 21, 1994, 12:01:11 PM1/21/94
to
In article <2hi20r$8...@bigboote.WPI.EDU> jo...@dolphin.WPI.EDU (Joseph W. Vigneau) writes:
>In article <2hhpn8$9...@usenet.rpi.edu>,
>Jason Weiler <wei...@goya.its.rpi.edu> wrote:
>>
>>From what I hear, a 386 w/ an FPU will kick ass all over a 486 w/o an FPU
>>I've seen DOOM run on a 486DX 33 and it was quite smooth and playable. If
>>you don't have an FPU, I'd strongly recommend getting one, and not just for
>>DOOM.
>
>Um, *all* 486s have an FPU built in, except the SXs... Correct me if I'm
^^^^^^^^^^^^^^
>wrong...

Bullseye.

(Sort of excludes "*all*", doesn't it?)
--
William Alan Krueger | kru...@cs.umn.edu, krue...@gold.tc.umn.edu
Graduate Student | "I've never been the kind / To close an open mind"
Computer Science Dept. | - "Free Time," Michael Penn
University of Minnesota | "Hellllllooooooooooooooo, Nurse!" - Warner Bros.

Gary Finley

unread,
Jan 21, 1994, 12:16:34 PM1/21/94
to
Joseph W. Vigneau writes

>
> Um, *all* 486s have an FPU built in, except the SXs... Correct me if I'm
> wrong...

You're not wrong, but I doubt that Doom uses the FPU at all. I've just
built a new 486SX-25 (no FPU) and Doom runs just as fast as it does on
a 486DX-25 that has floating point, plenty fast enough for good play.

--------------------------------------------
Gary Finley, Univ. of Alberta Psychology Dept.
gf...@psych.ualberta.ca (NeXTmail welcome!)

Patrick C Leger

unread,
Jan 21, 1994, 3:00:42 PM1/21/94
to
One Lain Meyer (me...@uxa.cso.uiuc.edu) wrote:
: dav...@velociraptor.welchdev (David G. Kitaguchi) writes:

: >In article 7589...@ee.tut.fi, s10...@ee.tut.fi (Anssi Saari) writes:

: >>In <2hhpn8$9...@usenet.rpi.edu> wei...@goya.its.rpi.edu (Jason Weiler) writ\
es:
: >>
: [SNIP]

: >>What for then? I have no software that uses it, so it'd be wasted money.

: >>Of course, when upgrade time comes, there probably won't be a 486SX2 or SX\


3,
: >>so then I'll have to waste money on the FPU...
: >>
: >>Anssi

: >Well in actuallity the real speed advantage of the 486 over the 386 is the \
1k internal

: >cache in the 486 chip (I think I got the size correct. Also not to be conf\
used with
: >the external cache). So a 486sx-33 should easily blow the benchmarks off o\
f 386-40.

Well, actually it's an 8k cache. But the real speed advantage is that
instructions execute faster on a 486. Many simple instructions
execute in one clock cycle on a 486, as compared to 2 cycles on a 386,
for instance.


: umm. sorry but you are incorrect...kinda...a properly set up 386 DX 40 will
: tromp on a 486 SX 33, at times. especially FPU intensive work...i remember

At what times? Maybe when you've got a larger cache on the 386. I
seriously doubt that a 386 DX 40 will be faster at floating point
stuff than a 486 SX 33, since neither of 'em have an FPU.

: looking at the benchmarks on my 386 DX 40 compared to a 486 SX 33 at the
: shop when i bought it. can't remember the numbers, but the 40 was a decent
: way out in front of the 33.

There's benchmarks, and then there's real programs. I think someone
was just trying to sell you a 386 DX 40.


: -OLM

Chris

--
Chris Leger (bl...@cmu.edu)
Junior, Carnegie Mellon Computer Engineering

"...when people do our songs, it has kind of
a parody effect, like when Bill Murray does a
Beatles song." - Donald Fagen

One Lain Meyer

unread,
Jan 21, 1994, 8:48:18 PM1/21/94
to
Patrick C Leger <pl...@andrew.cmu.edu> writes:

>One Lain Meyer (me...@uxa.cso.uiuc.edu) wrote:
>: dav...@velociraptor.welchdev (David G. Kitaguchi) writes:

>: >In article 7589...@ee.tut.fi, s10...@ee.tut.fi (Anssi Saari) writes:
>: >>In <2hhpn8$9...@usenet.rpi.edu> wei...@goya.its.rpi.edu (Jason Weiler) writ\
>es:
>: >>
>: [SNIP]

[SNIP]

>: umm. sorry but you are incorrect...kinda...a properly set up 386 DX 40 will
>: tromp on a 486 SX 33, at times. especially FPU intensive work...i remember

>At what times? Maybe when you've got a larger cache on the 386. I
>seriously doubt that a 386 DX 40 will be faster at floating point
>stuff than a 486 SX 33, since neither of 'em have an FPU.

umm, excuse me, but then what the hell is that 387 chip sitting on my
motherboard? looks like a FPU to me.

>: looking at the benchmarks on my 386 DX 40 compared to a 486 SX 33 at the
>: shop when i bought it. can't remember the numbers, but the 40 was a decent
>: way out in front of the 33.

>There's benchmarks, and then there's real programs. I think someone
>was just trying to sell you a 386 DX 40.

nope, i bought it because it was cheap, and all i could afford at the time,
and still.

-OLM

Jeff Chan

unread,
Jan 22, 1994, 1:18:49 AM1/22/94
to

I sent email to id about this and they say the Math Coprocessor
is NOT used by Doom. All the more amazing that they did all the
the 3D with integer math.

Jeff Chan
ch...@shell.portal.com
--

Jeff Chan
Internet: ch...@shell.portal.com
UUCP: {apple,claris,pyramid,uunet}!shell.portal.com!chan

Ed Haymore

unread,
Jan 20, 1994, 11:20:41 AM1/20/94
to
B. Scott Bilas (sco...@iastate.edu) wrote:
| I do the same -- full screen low res. And I've got a 33DX. It's mostly
| bearable unless I get into some heavy action in multiplayer mode. But as for
| the FPU use, I can't imagine why anybody would use floating point graphics
| routines. Judging from the speed of Doom I'd bet that it's mostly 32-bit
| integer math. Take a look at any version of Fractint for DOS for a good
| example of the speed difference. That's pure integer math (up to a certain
| resolution). And it's faster than any floating point fractal programs.
| However, if Doom truly does use the FPU, I'd love to see it run on a Pentium
| system. The pipelined Pentium FPU is supposed to be 4 times faster than a
| 486-66!

On my 486DX, there isn't all that much difference in calculation time
between integer versus floating point on Fractint; perhaps a 25% or so
slowdown. If you're doing floating point emulation in software, I agree
you'll see a very significant slowdown -- but having the FP hardware helps
out a lot. (And Intel even admits that the floating point unit on the
486's is "dog meat".)

BTW, an earlier poster in this thread made reference to the size of the
486 internal cache being 1k. It's actually 8k -- although don't the
Cyrix 486's have a smaller cache?

--
Ed Haymore | AA6EJ
e...@byu.edu | Es perillos abocar-se.

0 new messages