Coming from NT 3.5X and 4.0, Win 2k pro really was a pleasure to work with. Slim, fast and stable.

arfffff!!!! yes it is and its still pups favorite os with 7 being second and xp being 3rd

@@PhantomWorksStudios i agree with you, windows 98 was my childhood, i got a windows 8 later in 2015 (yes, i had a 98se until 2015) Edit: i then got a windows 10 in 2018 (i still have that one) but i gave it for my little brother bevause i got a new windows 10 in late 2021 (my current pc)

@@CutieFakeKirby I had windows 2000 all through the xp era,vista era, and begining of 7 era

@@PhantomWorksStudios woaw, that's impressive

NT4 Hydra?

Unusual. With the 2001 crash after y2k spending every CIO and their brother obsessed with cheap plastic Dells running XP to replace Unix boxes as for $4000 Dell workstation were a fraction of the cost and could reduce cost by having one integrated platform for everything hence fire the Unix admins for cheap Indian MCSEs in Bangalore. Thankfully the tend of outsourcing stopped as I had to leave IT for several years as employers kept eliminating jobs for h1b1 guys who would screw them up.

The Sun PCii and especially the Sun PCiii were the ones which had adequate levels of performance, having a Socket 370 P3/Celeron at around 600 and 1.4Ghz respectively as opposed to a Super Socket 7 AMD K6-2. Which, for one is a cost reduced K6-3 with less cache and suffers from this massively. In fact the PCiii 1.4Ghz P3 Tualatin is about as fast as a 32bit x86 part could get at the time of its release. And these CPUs were mainly sold to the server market while desktop got first gen P4 (1.6 - 2.0Ghz with.... *DRUMROLL* RAMBUS which were much slower).

@@timothygibney159 Well, what with horseshit like Sun Blade 100/150 with a 550 - 650Mhz out-dated-on-release USPARC-IIe CPU at starting MSRP of 6500USD, the 4000USD Dell with a P3 at around twice the clock speed destroys Sun hardware. Not to mention AMD64 aka Opteron being released very shortly after the introduction of the Blade 150. So, this was sound financial decision making. In fact, they should've stopped buying Suns overpriced bullshit when the U5/U10 came out, which like the fore mentioned, is a half-manufacture cobbled together from chips which have fatal hardware bugs. Sun killed itself by selling bargain basement PC shitware at a 100x markup. AMD64 meant it was game over for Sun's workstations and the UltraSPARC architechture on the desktop. Sun tried to get on the AMD64 and Opteron hypetrain by introducing a new PC workstation brand using AMD64 platform but the design was rushed and the systems suffered from faulty CPU cards which often failed with intermittent faults that caused system crashes. Dunno what the warranty experience was with Sun at the time, it used to be notoriously good, but I suspect they tried to pull some bullshit and wrangle out of honoring their contract. The system used standard PC BIOS for example and AMDs reference platform chipset etc. It was overall on paper actually a half-way decent system in a stylish case that had equitable build quality to contemporary mid-tier Dell, Compaq and IBM systems. But, it didn't sell well. So, their jump turned out to be fall on their face at speed. I can't remember did it come with Solaris x64 on release or was it stuck with the 32bit version or how that story went... It most likely did not and this was one of the reasons nobody gave two shits. Other reason is marketing was neglible. Not Invented Here syndrome at its best. They did later on come out with a competitive line of Intel Core 2 Xeon based servers and I think some workstations(?). The Servers were built into same kind of cases with identical designs as the T1 and later UltraSPARC servers and the very last UltraSPARC workstation the Ultra 45. Which had for example PCIe bus and finally industry mainstream standard memory, that wasn't outdated. They released in an identical case a AMD64 workstation named the Sun Ultra 20. Which didn't sell any better than the short lived brand the first AMD64 Opteron PCs were introduced as.

@@Meton12765the K6-2 had no internal L2 cache, while the K6-3 had 256KB of internal L2 cache. AMD did release the K6-2+ sometime later, although that was essentially a K6-3 with half of the cache disabled.

jurassic park feels

@@juniorsilvabroadcast That was an SGI Irix machine. 

it was a turd

@@thewiirocks i know

@@juniorsilvabroadcast …this! This is UNIX!”

Woah there, spoilers for the plebs who don't have early access.

I could see the appeal of PHB IT management wondering why can't we buy plastic cheap Dell workstations running Windows 2k to replace these? What a pita if you had 100 of them to support

K6? Oh my! \m/ \m/ /m\ /m\ . I suffered through one, and then finally got the cache COAST

@@joeturner7959 K6/2. There was no more COAST with these, i believe. In fact, the SS7 boards all came with at least 256KB of L2, and the K6/2+ and K6/3 also had L2 cache on them. XP on a K6 would be downright torture!

@@andreewert6576 The K6-2 had a K6-2E still only 64 cache. The K6-3 had 64K L1/256K2 and a 20 byte prefetch decode pipeline. CPUworld says that both parts @400Mhz, w/100 Mhz FSB are not diffrent, but they were night and day in performance.

pretty much, it's annoying to even install Windows 2000 with any SP's on this as well. And even if he had gone with a newer Solaris version, it would have created other workarounds jut to get the card working as this SunPCIi card isn't supported past Solaris 9 anyway, so you have to tweak its drivers for Solaris 10 or 11. A slightly better route is getting the next gen of this card in one of these. It has a Celeron 600Mhz, can run windows 10, has an option of using a separate Ethernet from the host Ultra5/10 machine but does support passing audio to the system so you can just have audio through it.

I had this setup too but much later and not as my main computer. Maybe in 2007. I think I was running Win98 on the card some years later just to test it. I think I was mostly running OpenBSD, not Solaris, hence I couldn't even utilize the fancy card.

It is a standard Socket 7, but considering how slow it already is, I don't think we can spare the performance.

@@NCommander Yes the Cyrix is *very* slow. About the only worthwhile upgrade would be the K6-2+ 570ACZ with unlocked L2 cache. But even then, the SunPCI BIOS doesn't enable the L2 cache of the 2+ processors and it has to be done manually with software.

Maybe a Pentium overdrive would do the job?

@@NCommander The CPU isn't the problem. A K6-2 400 can run all of that stuff with no issue. The problem is with Sun's dreadful implementation of getting a PC-on-a-card to work with their workstation. Apple did better with their versions, using much weaker CPUs that that K6-2. This is an example of the reasons why Sun started to go downhill about that time and no longer exists today. If Apple could make a PC-on-a-card work with their janky classic Mac OS, then doing it with a then-modern OS like Solaris should have been doable. Do you know if these can be used by themselves, as ordinary PC systems-on-a-card outside of a Sun workstation?

@@johncate9541 I have these cards, tried configuring them in a different PC under Linux. But the logic of the PCI-to-PCI bridge chip and the hardware bringup process used on the Sun requires some intensive reverse engineering to make work outside the Sun ecosystem. Not something to be taken lightly.

For video compatibility problems, my guess is that SunPCi software emulates the entire VGA card. I don't think Sun would've put actual VGA capture hardware inside the card, as it would increase the price of the card. When you pass '-vga' to software, it probably tells card to use actual VGA chip instead of software emulation. And for storage, they might be using some kind of proprietary protocol between driver(and BIOS) and host. This would explain why DOS applications work just fine, while booting Windows NT as-is just crash during boot: Windows NT would try to find the boot device through IDE, but in reality there's no actual IDE devices. I think the lack of mouse support during WinNT installation is going to be similar situation as well.

Fight-club programmers edition.

Sounds interesting. Well, I got my hands on Solaris only in 2005 when 10 was out. And I was amazed with how much documentation SUN had for the software. At that time I was managing a couple of FreeBSD gateways for sharing Internet between offices and always was amazed with its Handbook, but then Solaris appeared to be the bible of all what I had previously seen. 🙂

Ghost 8.0 for DOS is my favorite DOS program.

That's due to VFAT mapping being patented by Microsoft I suspect. The Windows CD however has all files in 8.3 format, and I did try multiple things like copying them across as well.

@@NCommander To a small extent, some of the 8.3 names can be controlled within certain limitations inside of various Windows versions. Outside Windows, though, you can generate some short names that wouldn't ordinarily be legal inside of Windows. I've seen scenarios where *NIX will embed colons and other characters which Windows NT will interpret as the token to separate the file name from the alternative data stream, for instance. In a nutshell, it can get really messy when multiple OSes touch the filenames.

at least my blade 150 runs quake III super fast at 1600x1200 using the xvr600 pci

Around 2001 I was running a lab full of Ultra 10s and 5s for CFD plus my own desktop PC (P3 running Linux.) When the HDDs started failing in the SUNs I installed Linux on a few of them and got a very decent performance improvement for our workload. Mind you, my PC was still much faster than any of them despite being a fraction of the price!

Jumpstart here was just because I could, and I didn't want to deal with the hassle of doing a lot of disk swapping. Plus I often reset my machines to clear setups, so this makes it easy to do in the future with other suns.

If you use Solaris 9, you need to change some links so the SunPCI kernel driver can be loaded

@@NCommander If it's not too much to ask, would you be able to upload the SunPCI CD as a ISO to the Internet Archive? Those CDs and disk contents are starting to fade off the internet. SUNWspci_13 can be found, but not the original disk. Kevin Hooke has some great blog posts too if you ever want to try this process on Solaris 10. Loved the video!

Which turned into the rotten pumpkin in 2 years 🤐 But it was a right decision as it was possible to upgrade to XP and drag its butt up until AthlonX2/Core2Duo

What we actually had back in the day was BOCHS. That was how one ran Windows as slow as possible from whatever machine we wanted. It was unusable slow, but it was just so amazing to see a complete Windows desktop *emulated* under another OS. Fun fact: The ReactOS project started up at this time. They still haven't released a working OS some 24+ years later.

my work here is done :)

I'm glad I never bothered to play with my SunPCI, since it looks so painful here. It never interested me because I had much faster PC's around the house, I'd already played with a similar concept in the DOS card in the Apple Mac Quadra, an MS Windows install in a window is nothing new considering how much I've been using VMs for the past 20 years and my usage of UltraSPARC machines was almost always with OpenBSD.

Nope, doesn't see the hard drive. There are ways to do it with an NFS root from the Solaris host.

@@NCommander good god the performance must be horrifying

@@NCommander do it do it do it do it

@@NCommander Alternative option is to connect a HDD over IDE directly to the card, but that might be less fun I think.