2022 2021 2020 2019 2018 2015 Quick Links
Originally encoded on August 15th, 2017 and published the following day
That is what best describes the earliest attempt to record this video, as well as how Windows 98 turned out in the end. Though I'm not sure if the latter is truly a simple "mistake" or malice of the highest degree...
By the time Hardcore Windows 95 and NT were well past being finished and tons more people became aware of this channel all of a sudden, it was only natural that they would want to see a Hardcore Windows 98 series of the same caliber as the previous two. I was already following a pattern where a Hardcore Windows series would debut around 20 years after the release of a version of Windows, but I was undecided on a firm release date for Hardcore Windows 98, for one simple reason: the gold release SUCKED!!!
One early fan suggested meeting in the middle, starting the series at the end of 2018 and ending it at the start of 2019. I found that to be the most ideal option, but that meant there would be a gap of two and a half years between Hardcore Windows NT and 98! To ease the wait, a single-segment Hardcore Windows video spanning 93 minutes began production on July 20th, 2017. This more specifically covered the evolution of Windows 98 prior to its release.
Many people have covered Windows 98 prereleases before, but I wanted to make Hardcore Memphis much more hardware-centric; truth be told, that is the most exciting part about Windows 98, not its batshit insane ideas of making a web browser into an operating system. One channel was already known to test Windows prereleases on real hardware, but Hardcore Memphis is the first of its kind to truly focus on how hardware support in a version of Windows evolved over time.
The established approach was fairly simple: just install a selection of Memphis builds in the most failsafe way possible, copy the setup files to the hard disk and run Setup from there, by order of build number - much like how most enthusiasts prefer to install Windows 95/98. The first "getting hit by a train" was having to adjust to the timebombs Memphis builds are configured with; in general, the safest time to set on your computer is anything before mid 1997.
Rather than using a boot disk generated from a Memphis distribution itself, I used a standard Windows 98 boot floppy. It has FAT32 support, plenty of CD-ROM drivers, no timebomb in IO.SYS, and some extra tools loaded into a temporary 2MB RAM disk. It just works.
Seeing how Memphis evolved in regards to software was more of a side effect of undertaking this project. The earliest leaked build, 1351, is very much reminiscent of Windows 95 with minor changes to the shell, yet already has some significant capabilities like utilizing multiple monitors. According to the beta release notes, only a small selection of video cards supported mutliple monitors in this build, such as the S3 Trio64 and ViRGE. Many other PCI/AGP cards were bound to become supported in due time.
Since I had sufficient equipment to pull it off at this point, I wanted to demonstrate modem functionality as well; I had an internal modem plugged in and connected to an internal phone network powered by some ugly Dell tower from 2005. Getting internal modems to work always tended to be a pain in the ass, especially when using Plug and Play - though that may have had to do with that one BIOS setting nobody understood. Even though Windows 95 and 98 use Plug and Play, it is often ideal to set "PNP OS Installed" to "No" for maximum reliability in getting PnP devices to work! Either way, the modem demonstration idea was scrapped, but wouldn't be entirely forgotten about. All it ended up being used for was downloading an S3 driver I needed over HyperTerminal, anyway.
Things were going very well. Having a multi-monitor setup didn't throw any crashes in this run, and I went through a few builds trying out some of the new things they offered. The "Monitors" tab that originally replaced the Settings tab when multiple monitors are connected did catch me off guard at first, but it's not all that hard to figure out, really. Cutting a full-blown list box of monitors and replacing it with a simple drop-down menu was a wise move, given most desktop computers do not have more than two monitors plugged in.
But as I ran through some of the builds, there was a serious oversight in the recording that I didn't find out until I got down to editing. See, I didn't have two capture cards handy, so I had to use my camcorder to record the secondary monitor. A great spontaneous solution, of course... except, the flicker, the refresh rate!! When a new monitor or video card driver is installed, Windows will often try to set a refresh rate that is ideal for humans working with CRT monitors, but such a refresh rate may not be ideal for most cameras if it is not 60Hz or 120Hz. Good camcorders let the user adjust the shutter speed to align with the monitor's refresh rate, but finding one that can do something as specific as 1/75 or 1/85 might be something to call miraculous.
Such is the case here; Windows gave the second monitor a refresh rate it thought I wanted, and what I ended up getting on the camera was nonstop flickering for the whole time I was using the computer. For it to persist that long, it was safe to say all of the footage was no good; a redo would have to be in order. But so much time was consumed going through those builds, how can all of that just get thrown away?! I ended up making a shitpost out of the otherwise garbage footage to compensate for the time wasted.
Hey, I suppose an extra 48 seconds of Memphis is enough to keep the ravenous crowd occupied.
A Word on ACPI
I suppose I should tell you a bit more about how I originally planned to go about this whole thing while it's still on my mind. My first plan was to install all seven of the builds I selected on one computer I had really been getting a kick out of - a 350MHz Pentium II machine powered by a 440BX chipset. I had two nVidia RIVA TNT2 Model 64 cards plugged in, one AGP and one PCI. After a BIOS update to the Asus P2B-LS, I was able to correct the video device priority so the AGP card, connected to my capture device, would be treated as the primary display.
I ran a test installation of Memphis on that computer, went to the toilet, did some other things I guess, then came back wondering why the fuck my computer shut off, and why I couldn't boot into Memphis after POST. Such are the effects of the timebomb in the Memphis bootloader... no matter, just roll the clock way back on that thing, then try again. Sure enough, wait... what the fuck?! HOLY SHITCOCK, THERE'S A RED SCREEN OF DEATH ON THE SCREEN!!!
ACPI had a rocky start, and didn't really become viable until later in 1999. Nonetheless, Memphis was the testing ground for motherboard manufacturers to try implementing ACPI very early in its development. While the first ACPI specification was released at the end of 1996, around half a year after the last APM revision, it wouldn't be until late 1997 or so that some motherboards like the Asus P2L97 would bring an actual ACPI implementation to the consumer market, and ACPI version 1.0 went through a couple of minor revisions up to February 8th, 1999.
By the time I updated my BIOS, it may have been too far out of reach for Memphis 1351's liking. Maybe one day I'd consider trying it again on the P2L97 I have using its earliest BIOS version... with no switch in early Memphis builds available to disable ACPI in Setup, nor an option available in the P2B-LS's BIOS settings to disable ACPI on that level, I was forced to split the demonstration across three computers, starting with a pure APM dekstop, to which I would begin recording footage on there.
After some demoralization that resulted from all the time wasted, I restarted production on July 31st, and this time, I was prepared. This time, I'd make damn sure to set the second monitor's refresh rate to 60Hz, and check the camera's screen on every boot to confirm it's still set to that. The screwup from before would turn out to be advantageous, as this time I had a better idea of what I needed to do, rather than have to blindly probe for potential hotspots to emphasize. Going through builds 1351, 1400, and 1532 would only end up taking two hours, even when taking the 200MHz K6 CPU into account.
Even so, some things that went smoothly before ended up going wrong in this run, but that was mainly WordPad... right as I was typing up a very important defense against my aggressive boss! Previous attempts had more complete writings like "NO BOSS, I AM WORKING. THIS IS REAL WORK I DO."
Also, regarding that shutdown timer tab in the power management properties from those early builds... turns out that along with the rest of the old control panel carried over into the final release, even though it is not loadable. What for? Maybe some old functions stuck in a DLL for compatibility with some legacy third-party power management software...
Also, that "color highlight" feature was a real eyesore. Wow, a button is suddenly like a hyperlink or something! It can be avoided outright by setting the color highlight option to match the normal text color, but it comes as no surprise that color highlighting on UI elements would end up being disabled later on in Memphis development. Its superior cousin, gradient title bars, first appeared in build 1387, but was very cryptic to activate and required the efforts of a skilled reviewer in order for it to reach public consciousness once again (and prompt the creation of Windows 95D.)
I vaguely recall the hardware installation wizard in one early build of Memphis being able to search for drivers in multiple directories recursively, but find myself having a hard time weaving through all the footage trying to find if that's actually true. If it is, it's yet another one of those good ideas that either takes ages to return (this one did in Windows ME), alongside other ideas that end up getting scrapped forever (Explorer was originally meant to have window tabbing very early in Chicago development).
One question still remains, did Squeezy ever respond to that email I sent him? Well, this is his response...
After I was finished with the first computer, I took it out to the 240P POLITICS broadcasting stadium where the existing Memphis installation would be repurposed to display a screensaver of a very cool 240P POLITICS flag waving in the screen. It's cool!
I really thought everyone would hate 240p Politics when I released the first videos, complaining about it being so so annoying or some shit, but now seeing that it was an instant hit and a reliable news source that delivers only the facts, I figured... why not, let's make another news segment specifically to play back in Hardcore Memphis.
Whilst installing Memphis Beta 2, now known as Windows 98, I win a contest where you make random noises in mic for 4 hrs :
So, this time around I'm using a Dell Inspiron 8000 laptop. It's usable, but it has a broken LCD, so I really only bother with it when I'm using it to record stuff. Defects like these are part of the reason why I don't really bother with collecting old laptops much; the higher modularity of desktops makes them much easier to maintain.
My main goals for this computer were to try to demonstrate ACPI and DVD playback... turned out the former was not in this laptop at all, and as for the latter, I couldn't get Windows 98's own DVD player component to work. What I ended up doing was pulling out an OEM copy of WinDVD and throwing a very rare 240p Politics DVD in the laptop. This computer went by really quickly as well, as I presume I did a test run for that as well.
Most of this part was really just running through some products on the Memphis Beta 2 catalog, as well as giving FrontPage Express a spin so as to show where Internet Explorer 4 is at, barely. Every time I glance back at some of this stuff, I'm glad I'm not stuck in the dark age anymore when it comes to making legacy-compatible websites...
The highlight of this build demo was a little drawing outside of it. Enjoy a pirate maroon bass leaping in front of the sun in all his glory... wait, is he holding that sword right?!
The Final Stretch... or a Precursor?
Now comes the most powerful computer in the video, the one that I originally intended to carry the whole way through. It's got the same 350MHz Pentium II and 256MB of RAM, but this time just one AGP RIVA TNT2 Model 64. Multi-monitor support had to get much better at this point, from what I can infer... unless 98 Gold bugged the fuck out on it.
Starting with build 1619, I initially blamed the CD-ROM drive for the installation taking such a long time, but now I'm pretty sure it has a lot more to do with the onboard Adaptec Ultra2 SCSI controller being miserably slow at emulating a primary BIOS hard disk in some failsafe 16-bit compatibility mode. This issue is not prevalent when a 32-bit protected mode driver for the controller is installed. Either way, the installation ended up taking around an hour, which no typical Pentium II configuration should have to drag through.
The next thing I wanted to demonstrate was USB support. Windows 95 OSR2 supported USB devices as well, but few vendors actually ended up writing Windows 95 drivers for their USB devices, and it doesn't even have support for basic HIDs like keyboards, mice, and joysticks. Windows 98 would end up being the first one to really have any capability of working with actual USB devices that would appear on the market shortly after its release.
Much like Windows 98 Gold in its entirety, Memphis 1619 was pretty unstable with USB devices. I plugged in my trusty LaserJet via USB, tried loading the Windows 98 PCL6 driver for it, and... CRASH! Wow, this is just like that embarassing demo Bill Gates joked about with the scanner! There is a higher quality version of that infamous clip if you want to check it out...
Of course, when I upgraded to Beta 3 (Memphis 1650), the ACPI driver was gone, and APM took its place. The reason why this happened was because Windows 98 blacklists BIOS versions prior to December 1999 from using ACPI by default. This can be overridden by running SETUP.EXE /PJ, but it's always more advisable to just update your BIOS version for a more reliable ACPI implementation.
Now, as for that printing of the upcoming JimboVideo logo... I really don't remember what I was supposed to do to get the entire image to fit in a normal landscape-oriented image when printing from Internet Explorer; I was able to fit a similar drawing in one page just fine when I printed from Adobe Illustrator on Windows for Workgroups 3.11. Either way, the HP LaserJet 2200DN is one hell of a compatible printer for 2003, being able to work with that vast of a range of systems!
Most everything I wanted to demonstrate besides the broken USB implementation could've been done in a single build. Going to build 1681, it surely wouldn't be too much longer before Windows 98 would exit beta status. Much like I tried to play GLQuake back in Hardcore Windows 95 with lackluster skills back then, here I figured I'd play Unreal, a game I had just recently started playing then, just to wrap it all up neatly.
Getting Unreal wasn't part of my original plan back when I was planning out the bass lan party video. All I wanted to get, initially, was Unreal Tournament - the original version from 1999, of course. As a result of a sale spanning the entire Unreal series on Steam, I ended up getting the complete set just because I could. Unreal Tournament never ended up making it into that video, partly because I found its demo handling to be sucky, but I figured I'd give Unreal Gold a spin. I wasn't sure what I was in for, but as soon as the music kicked in as I was locked in combat with some LesserBrute bombarding me with rockets, I immediately knew this game was going to be fucking amazing. Somehow, even with all the sorely dated textures, Unreal STILL looks fantastic to this day.
Late 90's 3D games often get mocked for their visuals looking really chunky, but one of my friends would agree that 3D hardware limitations influenced the art direction of a number of these games, making them really stand out in their own way. It's the same reason why R-Type Delta looks so much more vibrant than R-Type Final 2! As soon as the raw graphics actually get good, it seems too many things tend to kick back and lay down so comfy on them...
That trip through Memphis sure was interesting. The final video turned out so long that when I joined VidLii, I made a special version of it optimized for 240p and a hard 10 minute limit. The extreme cramming probably made it harder to follow along with, but hey, while VidLii's gone down the shitter, at least now there's two for everyone!
While Hardcore Memphis did prove to be a serious contender in documenting old beta releases of Windows, it, of course, is only a precursor to what is to follow. What all might that be? I've had over a year to think about it... but I haven't really been one to extensively document prereleases. For more cool Memphis stuff, drop by the Blue OS Museum!