On Monday, July 29, 1996 at 2:00:00 AM UTC-5, Tabb wrote:

Igor Obraztsov wrote:

I know, I might be the last person in the world to do it, but today I finally
got on Blizzard's website and looked thru Diablo info/screenshots.
Among the requirements, I saw a line that filled me with dread - Windows95. I've had dealings with win95 before, and that's enough for me. I don't want to have anything to do with that bloated creation of pure evil, contaminating
our poor computers. Is Diablo going to be a Win95-only game???

SNIP

Yep...

Tabb

Lol. Funny going back to see chats in the past. I don't know what the issue all of you people were having. As a 5-year-old at the time, I did not have any issues. I would build P90 machines from parts of broken machines my dad would bring home from work
and hook them up to play Diablo, Starcraft, Doom, Quake, Warcraft II, etc. with very little problems. If there were, I learned how to fix them fairly easily.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Monday, July 29, 1996 at 2:00:00 AM UTC-5, Tabb wrote:

Igor Obraztsov wrote:

I know, I might be the last person in the world to do it, but today I finally
got on Blizzard's website and looked thru Diablo info/screenshots.
Among the requirements, I saw a line that filled me with dread - Windows95. I've had dealings with win95 before, and that's enough for me. I don't want to have anything to do with that bloated creation of pure evil, contaminating
our poor computers. Is Diablo going to be a Win95-only game???

SNIP

Yep...

Tabb

Lol. Funny going back to see chats in the past. I don't know what the issue all of you people were having. As a 5-year-old at the time, I did not have any issues. I would build P90 machines from parts of broken machines my dad would bring home from work
and hook them up to play Diablo, Starcraft, Doom, Quake, Warcraft II, etc. with very little problems. If there were, I learned how to fix them fairly easily.

Win95 was obviously not perfect and had issues, but I think more of this talk has to do with hearsay and bullsh$t, than real issues.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Mon, 15 Oct 2018 23:44:20 -0700 (PDT), devin.moravec@gmail.com
wrote:

On Monday, July 29, 1996 at 2:00:00 AM UTC-5, Tabb wrote:

Igor Obraztsov wrote:

I know, I might be the last person in the world to do it, but today I finally
got on Blizzard's website and looked thru Diablo info/screenshots.
Among the requirements, I saw a line that filled me with dread - Windows95.
I've had dealings with win95 before, and that's enough for me. I don't want
to have anything to do with that bloated creation of pure evil, contaminating
our poor computers. Is Diablo going to be a Win95-only game???

SNIP

Yep...

Lol. Funny going back to see chats in the past. I don't know what the issue all of you
people were having. As a 5-year-old at the time, I did not have any issues. I would
build P90 machines from parts of broken machines my dad would bring home from work and hook them up to play Diablo, Starcraft, Doom, Quake, Warcraft II, etc.
with very little problems. If there were, I learned how to fix them fairly easily.

Win95 was obviously not perfect and had issues, but I think more of this talk has to
do with hearsay and bullsh$t, than real issues.

There were a number of legitimate concerns and criticisms about
Windows95, especially in its early years and especially for gamers.
Here are a few I remember off the top of my head.

Firstly, there were significant performance issues. Machines at the
time usually only had 8 or 16MB of RAM, and Windows95 used a
significant chunk of that just for itself. CPUs were likewise limited
(I think I was still running a 486/33 when Windows95 was released).
For all its other problems, DOS allowed games an exclusive lock on all
the hardware and could maximize performance. The overhead of Windows95
meant a Win95-game running on the same hardware as the DOS version was
always slower. DirectX (and faster CPUs and more memory) would later
mitigate this issue but in the early days of Win9x, this was a
significant problem with the new OS

Secondly, Microsoft was battling against its own reputation. While
Win95 was an improvement, gamers could not help but compare it with
Windows 3, which was just /awful/ for games. Windows 3.x was slow, it
was crash-prone, and its pitiful HAL meant most games could not
utilize the more esoteric features of the hardware to their fullest
potential. Win3 didn't really offer /any/ advantages over playing a
game in DOS. Game selection was also an issue: in DOS, we had games
like Need for Speed and TIE Fighter; meanwhile, the best Win3x had to
offer was Myst and Outpost. Given this background, gamers expected
the same from Win95 (and in the early days of the new OS, those
worries were justified). Any benefits Win9x brought to the table just
weren't enough to counter balance the disadvantages.

Thirdly, DOS gamers - and especially those who posted to Usenet - were
familiar with the eccentricities of DOS. Tweaking config.sys and
maximizing lower RAM were arts we had long mastered; playing games on
Win9x would require an entirely new skill-set (and in these early days
of the internet, this sort of information was much harder to get). It
seemed a lot of extra effort for no real advantage

Fourth, many of us had extensive DOS game collections (I still do!),
many games of which ran poorly or not at all in Win9x. Yes, newer
Win9x-native games might run fine, but the whole dual-booting to DOS
thing was really annoying. So if I was going to have dual boot anyway,
why bother with Windows95?


So there were a number of good reasons why people looked askance at
Windows95 and wondered why we couldn't just stay with DOS. After all,
DOS had worked well enough to get us classics like Warcraft, Wing
Commander III, so its not like DOS wasn't a capable gaming platform.
Meanwhile, Win95 was unproven, top-heavy and the only real point of
comparison people had was Windows 3.1. It is no surprise that there
was such an outcry.

(Myself, I was fairly slow to upgrade to Win9x, moving over only in
late '96 or early '97. All the games I was interested in were
DOS-based, and I had customized Win3x to be quite usable. I was less
concerned with the performance issue than I was with the new learning
curve and saw no advantage to the new OS. Even afterwards, I still
maintained a dual-boot, often switching back to DOS and Win3x. While
not overtly hostile to Win95 - I knew eventually I would have to
upgrade - I understood some of the anger espoused by other users. DOS
worked, so why mess around with that?)

Ultimately, Windows would prove itself the better platform but it took
several years and it really wasn't until Windows98 that its victory
was assured. Oddly enough, Windows' biggest advantage to gaming was
one that was almost invisible to the player. It wasn't any performance
gains, or the UI, or the Add/Remove Programs installers: it was
drivers. Although we take Windows drivers for granted now, they were a
radical new idea in 1995.

In DOS, if a game wanted to support the Soundblaster AWE, or 3DFX, or
a Thrustmaster joystick, or a Zoom modem, the game developer would
have to write drivers for each piece of hardware. Especially given the proliferation of new hardware in the mid to late 90s, this was a
daunting task. Win9x simplified this by offering a decent hardware
abstraction layer. Now the hardware manufacturers provided the driver
which alerted the OS - and any game running on Win9x - as to that
hardwares capabilities. It greatly simplified the developer's job,
allowed better scaling of game to computers capabilities and allowed
developers to better make use of more esoteric features of hardware.
Meanwhile, on the end-user side, the gamer could just plug in a game
without worrying if it would work with his joystick or sound-card.
Without this significant advantage, Win9x probably would not have made
the quick inroads to PC gaming as it did.

Anyway, its easy to look back and think "oh, those silly gamers of
1995!" but their concerns were valid and justified at the time.
Windows 95 was unproven, Microsoft had a bad track record when it came
to its operating systems, there were no real indications that things
would improve, and the performance hits were significant. It is no
wonder people were clamoring for DOS versions of new games.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Spalls Hurgenson <spallshurgenson@gmail.com> wrote:

Firstly, there were significant performance issues. Machines at the
time usually only had 8 or 16MB of RAM, and Windows95 used a
significant chunk of that just for itself.

I found that 8MB was actually sweet spot for Windows 95. While the
older version of Windows used significantly less memory, Windows 95 had
a much more advanced disk caching. On Windows 3.1 (and plain MS-DOS)
you were stuck with a fixed size disk cache, but Windows 95 (like modern operating systems) used all of free memory as a disk cache. The size
of disk cache would automatically shrink or grow depending on the memory demands of the programs you were using.

On a 8MB machine running Windows 3.1, you'd set the disk cache to 1MB so
you'd have plenty of free memory for applications. However, not a lot of applications needed that much memory, so a lot of the time a large chunk
of RAM was sitting unused. Because Windows 95 could use that unused RAM
as cache it would often peform better than Windows 3.1 on 8MB machines.

That's one of the reason why I played X-COM Apocalypse under Windows
98 rather than plain MS-DOS. With MS-DOS I was limited to whatever the
maximum SMARTDRV cache was (32MB?) but under Windows 98 it could easly
fit the entire game (~200MB) in its disk cache.

--
l/ // Ross Ridge -- The Great HTMU
[oo][oo] rridge@csclub.uwaterloo.ca
-()-/()/ http://www.csclub.uwaterloo.ca/~rridge/
db //

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Wed, 17 Oct 2018 22:31:57 +0000 (UTC), rridge@csclub.uwaterloo.ca
(Ross Ridge) wrote:

Spalls Hurgenson <spallshurgenson@gmail.com> wrote:

Firstly, there were significant performance issues. Machines at the
time usually only had 8 or 16MB of RAM, and Windows95 used a
significant chunk of that just for itself.

I found that 8MB was actually sweet spot for Windows 95. While the
older version of Windows used significantly less memory, Windows 95 had
a much more advanced disk caching. On Windows 3.1 (and plain MS-DOS)
you were stuck with a fixed size disk cache, but Windows 95 (like modern >operating systems) used all of free memory as a disk cache. The size
of disk cache would automatically shrink or grow depending on the memory >demands of the programs you were using.

My memory says 16MB for Win95 but it's been 23 years so I'll defer to
argue. Besides, memory prices were dropping so rapidly during that era
that the sweet spot probably moved about quite a bit ;-)

Technically it required only 4MB but that was /really/ painful. It
booted, but the OS itself was slow (lots of paging) and things dragged
to a crawl if you had the temerity to actually open an app. I vaguely
recall somebody even hacked the OS - patched out win.com's memory
check and then pared down the rest of the OS - so it ran on 2MB but I
can't imagine that was a very useful build.

32MB was my "sweet spot" for Windows98, with 64MB being outrageous
(and 128MB just excessive).

On a 8MB machine running Windows 3.1, you'd set the disk cache to 1MB so >you'd have plenty of free memory for applications. However, not a lot of >applications needed that much memory, so a lot of the time a large chunk
of RAM was sitting unused. Because Windows 95 could use that unused RAM
as cache it would often peform better than Windows 3.1 on 8MB machines.

My default SMARTDRV cache was 2MB on my 8MB machine during the
DOS/Win3 days, unless I was running big apps that needed more RAM
(like most gamers I of course had a boot menu for every possible
occassion). Win9x, as you mentioned, handled disk caching pretty well
on its own and the only memory I have of fiddling with disk-caching in
later years was to get around some addressing bug that occured if you
had 512+ MB of memory. In fact, this ease-of-use was always something
I sort of regretted, as a I enjoyed tinkering with the disk cache
(yes, you could still modify the vCache settings manually, but doing
so always resulted in inferior performance.. not like SMARTDRV where
you could see immediate and tangible benefits ;-)

That's one of the reason why I played X-COM Apocalypse under Windows
98 rather than plain MS-DOS. With MS-DOS I was limited to whatever the >maximum SMARTDRV cache was (32MB?) but under Windows 98 it could easly
fit the entire game (~200MB) in its disk cache.

By the time Windows98 - and its improved Second Edition upgrade -
rolled around, the Windows9x line had proven itself as a viable gaming platform. Not only was the OS improving - even before Win98 and
Win98SE, Microsoft added significant functionality with DirectX 3 and
5, and various OSR packs - but the app and game developers were
becoming more conversant in creating Win9x executables. Plus, hardware
drivers were becoming optimized, and most new hardware didn't have
drivers for DOS anyway. Hardware in general was becoming faster - the
486 was being phased out, RAM was dropping dramatically in price and
3D accellerators were the new rage - which also dramatically improved
things as well. Except for some oddballs, nobody was asking if Half
Life had a DOS version!

But in the first couple of years of Windows95, it was definitely a
concern, as many DOS games did have better performance than their
Win95 equivalents.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Spalls Hurgenson <spallshurgenson@gmail.com> wrote:

My memory says 16MB for Win95 but it's been 23 years so I'll defer to
argue. Besides, memory prices were dropping so rapidly during that era
that the sweet spot probably moved about quite a bit ;-)

I found that with 16MB you could go either way. For applications that
would work under Windows 3.1 that was an outrageous if not excessive
amount of memory, so you could alllocate a large chunk of memory for the
disk cache. On the other hand this was pretty much the bare minimum if
you wanted to run Windows NT 3.1, but then why would you?

Technically it required only 4MB but that was /really/ painful. It
booted, but the OS itself was slow (lots of paging) and things dragged
to a crawl if you had the temerity to actually open an app. I vaguely
recall somebody even hacked the OS - patched out win.com's memory
check and then pared down the rest of the OS - so it ran on 2MB but I
can't imagine that was a very useful build.

Yah, that was my experience too. With only 4MB of RAM you definitely
wanted to stick with Windows 3.1. I think I experiemented with a 6MB configuration, but I can't remember what the results were.

Speaking of installing stuff on a 2MB machine, I once tried to install
Linux on a 2MB 16MHz 386SX PC. It was probably as low powered of a PC
as (non-embedded) Linux has ever been used on. We wanted to use it as
dumb terminal, so didn't really need anything better. I booted from an
install floppy that let me do a network install, but the install was going painfully slow. There wasn't enough memory and it was thrashing badly.

Really badly, because it turned out that it was paging the cpio command
(sorta like the Unix version of pkzip but without compression) off of
the floppy. When paging code, Linux (and other operating systems)
won't use the swap file, instead it will just discard the code page
from memory if it needs to page it out and reload it from the executable
when it needs to page it back in. Since cpio command was on the floppy,
the floppy effectively became the swap file.

We managed to find an other 1MB of memory to install and that solved
the problem. I think we actually took the extra memory out once we got
it instaled, but it's been a while.

But in the first couple of years of Windows95, it was definitely a
concern, as many DOS games did have better performance than their
Win95 equivalents.

Honestly, this was more a theoretical problem for me. For the most part
when I started gaming under Windows 95, which was actually when it was in
beta before it was released, I would play MS-DOS games under Windows 95
if could. The ability to ALT-TAB to do something was just too invaluable.
I remember building a Windows 95 machine out of spare parts at work, and
then having a Master of Orion running on for days. During the day I'd
use it to browse the web, and during the evening I'd ALT-TAB to the game.

The real problem for me was games that wouldn't work under Windows 95,
like Ultima 7 and Wing Commander: Privateer because their bizarre (and
stupid) memory managers. Even a game like Master of Magic was a pain
to get working under Windows 95 because of the amount of conventional
memory it needed.

--
l/ // Ross Ridge -- The Great HTMU
[oo][oo] rridge@csclub.uwaterloo.ca
-()-/()/ http://www.csclub.uwaterloo.ca/~rridge/
db //

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Sat, 20 Oct 2018 17:38:02 +0000 (UTC), rridge@csclub.uwaterloo.ca
(Ross Ridge) wrote:

Honestly, this was more a theoretical problem for me. For the most part
when I started gaming under Windows 95, which was actually when it was in beta before it was released, I would play MS-DOS games under Windows 95
if could. The ability to ALT-TAB to do something was just too invaluable.

While alt-tab was a great feature, when it came to games I usually
found it far more problematic than it was worth. Even with DOS games
nominally running in their own virtual machine, too often an alt-tab
would end up with either the game or the computer becoming
unresponsive. So generally, I just didn't use it, which greatly
negated one of the advantages of the new OS.

That said, I have very fond memories of switching between Civilization
for Windows and Word 2.0 back in the Win3 days ;-)

The real problem for me was games that wouldn't work under Windows 95,
like Ultima 7 and Wing Commander: Privateer because their bizarre (and >stupid) memory managers.

In fairness to Ultima 7 (with its Voodoo Memory manager), its extender
was created in an era before other extenders were readily available.
It was actually fairly effective at what it did - it just happened to
be incompatible with all later memory management.

Even a game like Master of Magic was a pain
to get working under Windows 95 because of the amount of conventional
memory it needed.

I never really had that much of a problem with convention memory
management, at least not after 1991. Why 1991? Because that's the year
that Falcon 3.0 came out, and /that/ was the game that was the most
aggressive when it came to its memory demands, both low and expanded*.
Once I got a boot configuration working for Falcon 3.0, all other
games were a snap.

There really was an art to memory management that I almost miss. You
had to figure exactly where where to cram programs in upper memory**,
which often took numerous attempts to maximize the results. I remember
the pride I felt when QEMM and DOS Memmaker started automating the
process and were unable to improve my configurations ;-)

The only time I really had any problems was during the brief interlude
when I played around with STACKER disk compression; its resident
driver was a hog. I eventually got around the issue by putting stuff
like Windows*** and work applications on the Stacker volume. That
freed up enough space for all my games on the uncompressed part of the hard-drive, and whenever I gamed I just booted to a configuration that
didn't initialize Stacker.

But try telling kids these days. Even DOSBox immediately starts you
off with 632KB low memory... bah!

;-)


--------
* IIRC, for full capabilties, Falcon 3.0 wanted 624 low memory and 2MB expanded. That left only 16kb for DOS and drivers. Microprose games
always were really hard on low memory.
** You could specify exactly which blocks of upper memory to use for
each driver, which often was a trial-n-error process because a)
different computers often used different parts of the upper memory for
their own hardware and disliked (read: crashed) it if you stuck random
programs up there, and b) said programs often had a larger initial
memory footprint than showed up after they finished loading, so you
couldn't assume that just because it took 15KB once you got to the DOS
prompt that you only needed 15KB upper memory.
*** Windows - thanks to all the bitmap resources embedded in its
libraries and executables - compressed really well as I recall

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 2018-10-16, Spalls Hurgenson <spallshurgenson@gmail.com> wrote:

On Mon, 15 Oct 2018 23:44:20 -0700 (PDT), devin.moravec@gmail.com
wrote:

On Monday, July 29, 1996 at 2:00:00 AM UTC-5, Tabb wrote:

Igor Obraztsov wrote:

I know, I might be the last person in the world to do it, but today I finally
got on Blizzard's website and looked thru Diablo info/screenshots.
Among the requirements, I saw a line that filled me with dread - Windows95.
I've had dealings with win95 before, and that's enough for me. I don't want
to have anything to do with that bloated creation of pure evil, contaminating
our poor computers. Is Diablo going to be a Win95-only game???

SNIP

Yep...

Lol. Funny going back to see chats in the past. I don't know what the issue all of you
people were having. As a 5-year-old at the time, I did not have any issues. I would
build P90 machines from parts of broken machines my dad would bring home from
work and hook them up to play Diablo, Starcraft, Doom, Quake, Warcraft II, etc.
with very little problems. If there were, I learned how to fix them fairly easily.

Win95 was obviously not perfect and had issues, but I think more of this talk has to
do with hearsay and bullsh$t, than real issues.

There were a number of legitimate concerns and criticisms about
Windows95, especially in its early years and especially for gamers.
Here are a few I remember off the top of my head.

Firstly, there were significant performance issues. Machines at the
time usually only had 8 or 16MB of RAM, and Windows95 used a
significant chunk of that just for itself. CPUs were likewise limited
(I think I was still running a 486/33 when Windows95 was released).
For all its other problems, DOS allowed games an exclusive lock on all
the hardware and could maximize performance. The overhead of Windows95
meant a Win95-game running on the same hardware as the DOS version was
always slower. DirectX (and faster CPUs and more memory) would later
mitigate this issue but in the early days of Win9x, this was a
significant problem with the new OS

Secondly, Microsoft was battling against its own reputation. While
Win95 was an improvement, gamers could not help but compare it with
Windows 3, which was just /awful/ for games. Windows 3.x was slow, it
was crash-prone, and its pitiful HAL meant most games could not
utilize the more esoteric features of the hardware to their fullest potential. Win3 didn't really offer /any/ advantages over playing a
game in DOS. Game selection was also an issue: in DOS, we had games
like Need for Speed and TIE Fighter; meanwhile, the best Win3x had to
offer was Myst and Outpost. Given this background, gamers expected
the same from Win95 (and in the early days of the new OS, those
worries were justified). Any benefits Win9x brought to the table just
weren't enough to counter balance the disadvantages.

Thirdly, DOS gamers - and especially those who posted to Usenet - were familiar with the eccentricities of DOS. Tweaking config.sys and
maximizing lower RAM were arts we had long mastered; playing games on
Win9x would require an entirely new skill-set (and in these early days
of the internet, this sort of information was much harder to get). It
seemed a lot of extra effort for no real advantage

Fourth, many of us had extensive DOS game collections (I still do!),
many games of which ran poorly or not at all in Win9x. Yes, newer Win9x-native games might run fine, but the whole dual-booting to DOS
thing was really annoying. So if I was going to have dual boot anyway,
why bother with Windows95?

So there were a number of good reasons why people looked askance at
Windows95 and wondered why we couldn't just stay with DOS. After all,
DOS had worked well enough to get us classics like Warcraft, Wing
Commander III, so its not like DOS wasn't a capable gaming platform. Meanwhile, Win95 was unproven, top-heavy and the only real point of comparison people had was Windows 3.1. It is no surprise that there
was such an outcry.

(Myself, I was fairly slow to upgrade to Win9x, moving over only in
late '96 or early '97. All the games I was interested in were
DOS-based, and I had customized Win3x to be quite usable. I was less concerned with the performance issue than I was with the new learning
curve and saw no advantage to the new OS. Even afterwards, I still
maintained a dual-boot, often switching back to DOS and Win3x. While
not overtly hostile to Win95 - I knew eventually I would have to
upgrade - I understood some of the anger espoused by other users. DOS worked, so why mess around with that?)

Ultimately, Windows would prove itself the better platform but it took several years and it really wasn't until Windows98 that its victory
was assured. Oddly enough, Windows' biggest advantage to gaming was
one that was almost invisible to the player. It wasn't any performance
gains, or the UI, or the Add/Remove Programs installers: it was
drivers. Although we take Windows drivers for granted now, they were a radical new idea in 1995.

In DOS, if a game wanted to support the Soundblaster AWE, or 3DFX, or
a Thrustmaster joystick, or a Zoom modem, the game developer would
have to write drivers for each piece of hardware. Especially given the proliferation of new hardware in the mid to late 90s, this was a
daunting task. Win9x simplified this by offering a decent hardware abstraction layer. Now the hardware manufacturers provided the driver
which alerted the OS - and any game running on Win9x - as to that
hardwares capabilities. It greatly simplified the developer's job,
allowed better scaling of game to computers capabilities and allowed developers to better make use of more esoteric features of hardware. Meanwhile, on the end-user side, the gamer could just plug in a game
without worrying if it would work with his joystick or sound-card.
Without this significant advantage, Win9x probably would not have made
the quick inroads to PC gaming as it did.

Anyway, its easy to look back and think "oh, those silly gamers of
1995!" but their concerns were valid and justified at the time.
Windows 95 was unproven, Microsoft had a bad track record when it came
to its operating systems, there were no real indications that things
would improve, and the performance hits were significant. It is no
wonder people were clamoring for DOS versions of new games.

Well, that was true up to Quake I. Later, with Quake II,
DirectX gave more performance
on gaming than with the pure and raw DOS system calls.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Mon, 13 Nov 2023 10:42:52 -0000 (UTC), Bozo User
<anthk@disroot.org> wrote:

Well, that was true up to Quake I. Later, with Quake II,
DirectX gave more performance
on gaming than with the pure and raw DOS system calls.

I mean, yes and no?

There weren't really any DOS system calls for most hardware; DOS was
extremely bare-boned after all and beyond support for very basic video routines, there wasn't much it could do for games. That's why almost
every game developer who wrote DOS games made their own video routines
to access the video hardware directly, bypassing DOS. And with full
access to the hardware, performance could (and in some cases did)
easily surpass that of even the best DirectX access, just because
there were fewer hoops to jump through.

The downside, though, was that the developer had to code directly to
the hardware, which essentially meant writing their own drivers for
each and every piece of hardware. This was a monumental task that
required a huge amount of effort (so much so that most DOS game
developers just outsourced it), potentially led to bugs or poor
performance, and heaven help the poor gamer who didn't have hardware
on the pre-approved list of compatible video-cards. The big advantage
to DirectX was not so much that it could out-perform bare-metal
programming, but that it was much more convenient, much easier, and
less expensive and time-consuming for the developers. It's performance
was close enough to bare-metal too, that that any discrepancies
between the two were far outweighed by the advantages to consumer and developer.

But you could have get better performance if you coded directly to the hardware... which meant DOS (or a variation of Linux, or a home-coded
OS) and not Windows.

But ultimately, it's all moot. With a huge proliferation in the
variety of hardware, there was no way game developers could keep up
with coding directly to the thousands of different sound and video
chipsets, so they would never have been able to take advantage of that potentially superior performance. Even in 1995, DOS gamers were lucky
if a developer offered a patch that let them take full advantage of
their 3DFX or Virge video card so it could run in accelerated mode. As
CPUs became more powerful, the hit from the hardware abstraction layer
became less and less significant too. And the other advantages of
Windows95 - networking, multi-tasking, etc. - proved too popular
amongst users for DOS to ever win out.

But in 1995 - when most people thought "Windows 3.1" when thinking of
Windows, and 486s remained the most-used CPU - there were real
concerns about the viability of Windows95.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)