Comment by fuzzfactor
2 days ago
>up to a year after release, many gamers still recommended Windows 98. Why? Mostly due to compatibility where things a Voodoo card and a Soundblaster running in MS-DOS were preferable for many titles, and this is something that simply wasn’t on offer with XP.
Actually, mostly since Wxp was slow as a dog compared to W98, because W9x still had direct control of the hardware rather than the sluggishness-inducing Hardware Abstraction Layer (HAL) that NT has always had inserted between the OS and the devices.
W95 was noticeably faster than W98 was too, and both of course move like lightning-speed compared to W10 whose 64bit drags compared to W10-32bit, and W11 is more embarrassing as it continues to further slow with each update (almost every month now rather than only once per year), which makes W10 seem like it was a quite a bit less encumbered than W11.
95 and 98 were roughly the same speed; any differences would likely be due to drivers. The main difference between the 9x and NT lineage is the former is actually a hypervisor for DOS VMs (and the GUI itself can be considered a DPMI application, running in its own VM) while the latter is a "full" OS with a very limited DOS emulator.
Not on low-end kit.
I cut down Win95 to run from a 16MB SSD in 1996, paid for by PC Pro magazine. I knew that OS inside out.
Around the turn of the century my travel laptop was an IBM Thinkpad 701C, the famous "Butterfly". 40MB RAM and a 75MHz 486DX4.
Win95 was great on it, better than OS/2, but the thing is Win95 had a max of 4 IP addresses. In total.
I had a dialup modem (1), an Ethernet card (2), AOL for toll-free dialup (different stack, so 3) and Direct Cable Connection (4).
Add a different modem or Ethernet card and it couldn't bind TCP/IP to it. No more addresses.
I tried NT 4 but it had no power management, no PnP, no FAT32.
I tried Win2K. Not fun in 40MB of nonstandard (and so vastly expensive to upgrade) RAM.
I tried 98SE. Too big, too slow.
So I cut it down as hard as possible with 98Lite.
(Still around, remarkably: https://www.litepc.com/98lite.html )
No IE, no themes, no built in media stuff, no Active Desktop, and it ran reasonably on a 486 in 40MB of RAM.
And it supported more IP addresses!
But it was hard work to get it working, and it was never entirely stable.
No. I reject your statement based on considerable personal experience and benchmark testing.
98 was considerably heavier than 95.
Just look at the ISO files!
95 OSR 2.1 with USB support:
https://winworldpc.com/product/windows-95/osr-21
385MB.
98SE:
https://archive.org/download/windows-98-se-retail
622MB.
98 is a significantly bigger and more complex OS.
Same design, but a lot more stuff piled on top.
I remember that myself.
When people bought a new W98 PC, which was often the first computer for so many consumers, it really did perform quite similarly to earlier-adopters' W95 PC's that were already in action.
The specs on the newer hardware were so much better which made up for it, and progressive sluggishness of Windows was swept under the rug for mainstream consumers, continuing to an extent today. You know, like a snail without a shell ;)
This is why in the '90's when Grove was running Intel and Gates was running Microsoft, professional geeks coined the phrase: "What Andy giveth, Bill taketh away." They didn't wait until WindowsME to say this.
It wasn't really worth it for mainstream apps, but if you had a challenging Office 97 workload, with or without VBA, something like live "real-time" data acquisition, or god forbid any type of ML or simulation, the best improvement you could get was to wipe W98 off the HDD and start fresh with W95. It always has seemed like there was some uncalled-for obstacle to prevent easily installing a previous version of Windows on a new PC though.
Even now this still works to an extent, buy a new mainstream W11 consumer PC, install W10 in a regular ordinary Microsoft dual-boot configuration and see for yourself.
Most people would have so much SSD space left over they could even try a triple boot, how about that W10 ISO from 2015 if you really want to emphasize the difference in how much less sluggish things could have been now. Woo hoo. Plan to stay off the internet when booted to this one, in Device Manager you could even pre-emptively disable the ethernet & wifi.
Of course try it on a HDD if you haven't done that in a while, to see how that feels compared to earlier Windows when you were using nothing but HDDs.
Windows 8.0 is also still fairly installable in new PC's if you want to see what it was like when they had one of their many brilliant engineers taking focused responsibility to achieve faster boot times in particular.
Also the fact that pre-SP2, Windows XP actually crashed (and permanently broke in "interesting" ways) more than Windows 98 in practice, theory be damned. I became so familiar with how to install Windows during this time ...
Yes, SP1 wasn't horrible if you could get it (but who can download something that big on dial-up?), but it still was not great.
I know you can run microbenchmarks to show the increased pointer size of 64 bit Windows can cause a few percentage points of performance difference in certain scenarios but that doesn't jive with the statement "W10 whose 64bit drags compared to W10-32bit".
All I had to do was try them both back-to-back on the same hardware.
If one advertises they drove 2 trims of the same car model to the airport back to back and found cars with 2" smaller wheels are lightning fast because it took 30 minutes longer in the other car then people are, rightfully, going to doubt the test instead of the wheel size. Especially when you're not the only one to have driven cars with different wheel sizes but you are the only one reporting it's the wheel size, specifically, that made the trip significantly longer to take and give the trip as your sole evidence for the claim you know why it was slower.
From my enterprise image/push creation days one example of something I did find different between x64 and x32 was the specific driver bugs/performance. The thing is it went/goes both ways on that, sometimes it's the 32 bit driver that's bugged, sometimes it's the 64 bit driver, sometimes there was a special patch version of the driver but the vendor didn't post both builds. You get the idea. In this case it wouldn't make sense to blame the <x> bit OS variant as inherently being massively slower, but it sure might seem like that with an n=1 test.
2 replies →
Race cars are barren of safety and security features, creature comforts, and even frequently missing windows.
But boy are they sure fast.
But I wouldn’t daily drive one.
Nitpick I know but race cars in well run series actually have quite good safety - just not in the same way because the environment and expectations are different. You don't need/want a reversing camera or parking beeps and boops...
I think part of MS issue is that they keep bundling and pushing "crap useful to some minority" (as well as unwanted ads and features too) by default into ostensibly "your" system and making it hard to focus on what you want it for.
If you want it to focus on gaming performance... well it's more about arcane tweaks rather than having a turn off the shit button.
Maybe the coming Win10 EoL will see a few % points jump to Bazzite or some other linux gaming-focussed distro.
Considering a common use for Windows these days is Steam Launcher, performance is kind of a big deal, actually. Literally the only thing I use my desktop for is to play games, so yes, performance is pretty much the only thing I care about with it.
Race cars have heaps of safety systems not present in road cars. They don't have ABS and traction control because they don't actually increase safety on track with a professional driver. SRS airbags also offer no additional safety when in a 6 point harness and wearing a helmet and neck brace.
Race cars have drastically more safety features than road cars. Your road car doesn't tether your helmet to your headrest to protect your neck and doesn't have a roll cage, for starters.