What was the worst computer you ever bought and why?

[MrMaddog]

On 1/22/2022 at 4:43 PM, x=usr(1536) said:

And then there were the early attempts at Winmodem support under Linux.  *shudder*

Ugh, don't remind me...

 

The internal modem card that came with my first PC was not a Winmodem but was "Plug & Play" like my sound card.  Well that worked just fine for Windows 98 but not on Linux (at the time).  The two "Plug & Pray" cards would not work together because of conflicting IRQ settings that I could not manually change so it was one or the other, 

 

I was talking to a Linux guru on an online chatroom and was telling him that I was still getting IRQs from other devices like the serial ports and he said, "Why not just use a serial port for your modem then?"  Next day I went to Staples and bought myself a US Robotics external modem, not only did it work along with my sound card in Linux but gave me the maximum dial-up speed of 52K insteads of the usual 48K with the modem card.  And that's how I discovered an important lession in computers, if you can't solve a software conflict then change the hardware...

 

 

[zzip]

On 1/22/2022 at 1:17 PM, OLD CS1 said:

The other question is if this PCI runs at 33MHz or at the CPU setting.  For the DX4/100 that would be 25MHz.

The DX4/100 is actually a clock-tripled chip, despite the name (marketing strikes again),  so the bus should be running at 33Mhz to get the full performance.

 

On 1/22/2022 at 2:47 PM, Keatah said:

I remember these. Introduced to them via Windows 95/98. Wasn't sure (at first) what to make of them. But when I examined the cards themselves things like "cost cutting" & "cheap" roared into my head from seemingly every direction. Was the computer industry becoming that cutthroat by eliminating one or two chips for savings? Yup.

I had nothing against the concept of a "soft modem".  I remember in my ST days, there was talk that the Atari Falcon DSP could be programmed to act like a modem, and that was a feature!   But my main issue with WinModems was the software was tied to Windows.   I was interested in running multiple OSes, and some of the best TERM programs I had ran on DOS, so I refused to buy a modem that required Windows to use it

 

On 1/22/2022 at 2:53 AM, Keatah said:

I currently have a pile of SB16, AWE32, AWE64 Gold, WaveBlaster daughtercards, and other assorted CL stuffage. I never could afford a Roland or Gravis or any other "top" sounding cards.

I don't think Gravis cards were all that expensive for the time.  I had the Gravis ACE which was an awesome companion to the SB16.  It was just the wavetable and much less expensive than a full sound card.  it sounded much better in games that supported it,  for those that didn't I could fallback on SB16/OPL3 sound

 

On 1/22/2022 at 2:53 AM, Keatah said:

KEYBOARD & MOUSE: Fuck.. whatever I could plug in and get working.

I don't care amount KB/Mouse brand name,  just that they have to feel right.  These days I prefer a mechanical keyboard,  but back then, I went with what ever had a good clicky feel.   I soon learned that PS/2 mice ran smoother than Serial mice, so I switched and never went back.  (It's odd how PS/2 ports made a comeback after the IBM PS/2 was already dead and buried, The ATX redesign made them standard). 

 

I remember the "ergonomic" push for keyboards and mice in the 90s, with a giant lump in the middle of the keyboards I did not find those comfortable.  Same with the mice-- they seemed to have buttons all over the place that I'd keep hitting by accident, causing unwanted things to happen.

On 1/22/2022 at 2:53 AM, Keatah said:

So I found an Evolution board from STB, 1MB, ISA, Cirrus Logic 5422. Man-O-Man! One whole Megabyte! That's what my entire Amiga had! And it did millions of colors! I got it from CompUSA. Still have it of course. I didn't care about Windows' acceleration primitives or hardware cursors or clock speeds. I just wanted trillions of colors onscreen.

I was all into hardware acceleration.   Anything that could offload work from the CPU seemed like a good idea to me.   One think I liked about the Gravis card was that it could mix multiple sound channels in hardware, and play up to 32 digital audio channels independently.   SB16 had two, so the computer had to do a lot of software mixing to get the same effects.   For graphics acceleration,  I knew that the faster the screen updated, the smoother the overall experience was, so I wanted to see them used.

 

The problem was not every custom hardware/coprocessor got put to good use because not every PC had them.   I went for the ASP version of SB16 thinking that the ASP chip would allow cool things,  but almost nothing used it.  Seemed like it only powered "environmental effects", so that your music can sound like it was being played in a concert hall, or a closet (your choice), but all those options seemed to make the music sound worse than without it, so I always turned it off.

 

On 1/22/2022 at 3:33 AM, Keatah said:

I always assumed that PCI had too much overhead to be useful in 486 rigs. Maybe the last Overdrives or the Am5x86 could utilize it. PCI definitely was matched to Pentium, and PII/III/IV.

It had some overhead,  I don't recall there being restrictions on which 486s could handle it though.   Probably it was a case where you'd get more raw speed from VLB, but better stability from PCI.  Both would vastly outperform ISA.

[+OLD CS1]

1 hour ago, zzip said:

The DX4/100 is actually a clock-tripled chip, despite the name (marketing strikes again),  so the bus should be running at 33Mhz to get the full performance.

That explains a lot.  The book for the board indicated 25MHz.  And it displayed DX4/100 in the identification.  Since I cannot go back in time to try again, now I want to pull it out and try again if I can find the CPU.

 

EDIT: hehehehe 27 years later, my buddy says, "Nah, dummy, we ran it at 3x33. It failed when you tried to run it at 4x33 to see what would happen. The 4x25 was my VLB board because it sucked."

[+x=usr(1536)]

18 hours ago, Keatah said:

I think we need to discuss E-Machines!! Who doesn't love to hate those abominations?

I once had a half-dozen of those hunks of shit in my test lab.  We'd deliberately bought them in order to test on the crappiest hardware someone was likely to buy straight off the shelf, and they absolutely exceeded our expectations in that regard.

 

About the only significant thing I can remember of them is that they were responsible for receiving their own test plan which started with the instructions on how to get the OS to boot.  Oh, and they mysteriously seemed to fall on the floor a lot.

[+OLD CS1]

41 minutes ago, x=usr(1536) said:

About the only significant thing I can remember of them is that they were responsible for receiving their own test plan which started with the instructions on how to get the OS to boot.  Oh, and they mysteriously seemed to fall on the floor a lot.

My significant memory is eMachines did not have a 1-800 number for support.  I had a client who bought ten for a survey call center.  The order came in with one dead machine and two dead monitors.  These never got replaced by eMachines and the customer never received a refund.

[Keatah]

2 hours ago, zzip said:

The DX4/100 is actually a clock-tripled chip, despite the name (marketing strikes again),  so the bus should be running at 33Mhz to get the full performance.

Yes. I was always a little confused at the name. I wonder why they did it that way?

 

1- Was it simply because 4 is bigger than 3?

2- Was it because the angular lines of 4 were more coherent with X than 3?

3- Or maybe "four" is easier to pronounce than "three"? No risk of lisping..

4- Was it intended to cause confusion?

5- Since 4 is bigger than 3, it appears to be a faster processor.

6- Maybe it had something to do with the doubling of 2 to 4, less clumsier than 3.

7- 4 is an even number, 3 is odd.

8- 4 looks more futuristic and sharp compared to the bloating curves in 3.

9- Computery stuff tends to happen in powers of 2.

10 - Maybe DX4 rolls of the tongue better than DX3.

 

I truly hate the lying bastards in consumer marketing.

Edited January 24, 2022 by Keatah

[+OLD CS1]

1 hour ago, Keatah said:

8- 4 looks more futuristic and sharp compared to the bloating curves in 3

I think you mean "voluptuous."

[zzip]

1 hour ago, Keatah said:

Yes. I was always a little confused at the name. I wonder why they did it that way?

 

1- Was it simply because 4 is bigger than 3?

2- Was it because the angular lines of 4 were more coherent with X than 3?

3- Or maybe "four" is easier to pronounce than "three"? No risk of lisping..

4- Was it intended to cause confusion?

5- Since 4 is bigger than 3, it appears to be a faster processor.

6- Maybe it had something to do with the doubling of 2 to 4, less clumsier than 3.

 

I truly hate the lying bastards in consumer marketing.

Who knows?  Maybe they thought the idea of clock-quadrupling was more appealing?   But really which would most consumers rather have:

A) 100mhz processor with 25mhz bus

B) 100mhz processor with 33mhz bus

 

I would think most people would want B,  clock tripling is actually superior to quadrupling,  bigger number isn't always better   Seemed like the real selling point of the chip was 100mhz! (woot!).  Not the number that came after DX--  so I can't figure out what they were going for.

 

I do think they like to confuse people though.   Remember when AMD came out with an actual clock-quadrupled 486?  They named it the 5x86 to make people think they were getting a Pentium!

Edited January 24, 2022 by zzip

[Krebizfan]

I think the X4 was one of Intel's attempts to keep AMD from having a matching model number that ultimately led Intel to invent the Pentium label. Clock doubling/tripling was the best that could be done at the time and the 33 MHz bus was the fastest bus speed that remained stable. The 486DX2-80 and 486DX4-120 and 5x86-150 and 160 were all pushing beyond what the bus could handle. 

 

AMD having similar model numbers to Intel and Intel trying to devise a new model number scheme to keep customers from looking at AMD is one of the more amusing parts of CPU marketing. Now, it is i(odd number) versus R(odd number) while 10 to 15 years ago, AMD had 4 digit Phenoms matching the high end 4 digit Core 2Q and Phenom II had 3 digit numbers that matched the 3 digit numbers of the Core i lineup and after AMD introduced 6 core chips with model numbers just over one thousand, Intel jumped Sandybridge to the two thousands. 

[zzip]

37 minutes ago, Krebizfan said:

I think the X4 was one of Intel's attempts to keep AMD from having a matching model number that ultimately led Intel to invent the Pentium label. Clock doubling/tripling was the best that could be done at the time and the 33 MHz bus was the fastest bus speed that remained stable. The 486DX2-80 and 486DX4-120 and 5x86-150 and 160 were all pushing beyond what the bus could handle. 

But AMD was still making DX4-100 chips, so if that was the plan,  it didn't work.

 

[Greg2600]

Bought my mom a Compaq laptop which wound up being the biggest POS ever over 10 years ago.  It was so incredibly slow right out of the box.  Upgraded the RAM and still total dog, I think mainly because they used an AMD dual-core cpu that was a desktop chip.  Anyway, had to get her a Lenovo a few years later.  I'm currently using the POS as my Xlink Kai server that's about it, ha ha ha.

 

https://support.hp.com/us-en/document/c03015475

[Keatah]

2 hours ago, zzip said:

Who knows?  Maybe they thought the idea of clock-quadrupling was more appealing?   But really which would most consumers rather have:

A) 100mhz processor with 25mhz bus

B) 100mhz processor with 33mhz bus

 

I would think most people would want B,  clock tripling is actually superior to quadrupling,  bigger number isn't always better   Seemed like the real selling point of the chip was 100mhz! (woot!).  Not the number that came after DX--  so I can't figure out what they were going for.

In comparing modern PCs to classic rigs, I believe bus speeds have become less important. Less visible. Less influential as on-chip integration and those multipliers grow. And when I was faced with a choice of either 25Mhz vs 33MHz FSB I had taken the 25. Cost savings was enough that with a little extra pocket change I could either get another HDD. Or MOAR games.

 

I always understood that slower buses were more stable. And I used that to briefly console myself when 33MHz became the more popular standard. All in all I wasn't hurting for performance and quickly forgot about it. External bus speeds are always going to be slow anyways. Thanks physics! What are we up to now..? 45x multipliers?

 

[zzip]

4 minutes ago, Keatah said:

In comparing modern PCs to classic rigs, I believe bus speeds have become less important. Less visible. Less influential as on-chip integration and those multipliers grow. And when I was faced with a choice of either 25Mhz vs 33MHz FSB I had taken the 25. Cost savings was enough that with a little extra pocket change I could either get another HDD. Or MOAR games.

 

I always understood that slower buses were more stable. And I used that to briefly console myself when 33MHz became the more popular standard. All in all I wasn't hurting for performance and quickly forgot about it. External bus speeds are always going to be slow anyways. Thanks physics! What are we up to now..? 45x multipliers?

 

When I got my first PC, I had a friend with a 486DX-50 telling me I should get the same because it had a 50MHz bus,  he thought the clock multiplied chips were a scam because you were getting a slower bus.   I thought he knew his stuff so I tried to get one, but by that time DX-50s were hard to come by, everything was DX2 or DX4.  I ended up with a DX4-100.   I had no idea the 486 bus was unstable at 50mhz at the time.   His seemed to work fine.

[Keatah]

2 hours ago, Krebizfan said:

Clock doubling/tripling was the best that could be done at the time and the 33 MHz bus was the fastest bus speed that remained stable. The 486DX2-80 and 486DX4-120 and 5x86-150 and 160 were all pushing beyond what the bus could handle.

Agree. The DX2-80 and DX-50's bus didn't allow for any of those PowerStacker or FastChips or TurboChips that were popular during those heydays.

 

I briefly considered getting a DX-50 EISA system. But salesguy from GW2K actually told me there were specialty issues and the system wouldn't be suitable for me. He was right. And then of course the cost. I'd need to borrow an extra 800.

 

When looking at a 66MHZ FSB Slot-1 board or faster, you'll see all kinds of seemingly pointless zigzags in the Northbridge - Southbridge - CPU - RAM - AGP/PCIe traces. That's to ensure they're all the same length and the signals arrive exactly on-time and in-phase. You don't really see this on 25Mhz or 33MHz boards.

 

I'd rather have all this zigazaggy stuff engineered by chip designer teams whom really know their stuff. Leaving it to chance by the motherboard makers left us with some rotten crap boards.

Edited January 24, 2022 by Keatah

[Keatah]

1 hour ago, zzip said:

When I got my first PC, I had a friend with a 486DX-50 telling me I should get the same because it had a 50MHz bus,  he thought the clock multiplied chips were a scam because you were getting a slower bus.

Ohh dear.. Absolutely not a scam. In hearing that I might suspect your buddy wanted to sound like an expert.

 

No. I was totally blown away when I read the Byte article. The one I'm so fond of. I really need to print it out and frame it. It really was new technology that solved a real-world problem in an elegant way. And it was one of those magical moments. Coming from the 8 & 16 bit world, how in the hell could a CPU operate faster than everything else? It went against everything my infantile feeble mind was founded upon.

 

I was all for the cheaper "scammily price" parts if the most important part could run full-tilt. Besides. Modems and printers had been buffering for years. Hard disks were gaining cache. CPU's already had both internal and external caches. ISA slots never exceed 16 or 20 MHz. So why hold back a CPU with a 1:1 multiplier ratio?

 

Quote

I had no idea the 486 bus was unstable at 50mhz at the time.   His seemed to work fine.

The salesman (of vintage 8-bit knowledge and caliber) at GW2K told me that the problem was in the EISA expansion slots or something. Not all cards would work in this system. Now whether that was bus-speed related or EISA-protocol related, I don't know. He was knowledgeable, but not engineering-degree knowledgeable. And I was still green anyways.

 

When I was learning about the Pentium II, I was more aware of bus speeds. I learned there was making the whole board clock at 50. RAM, Chipset, cache, the faster glue, the longer traces.. The whole board. The industry needed a few years to get up to speed to make 66/100/133 speeds. The only way to do it at the time was with premium parts being run up against specification.

 

Eventually 100MHz FSB was possible with AGTL+ electrical signaling and new protocols. And especially with the breakthroughs in trace routing software. But of course all this several years after DX-50.

 

Edited January 24, 2022 by Keatah

[Keatah]

7 hours ago, x=usr(1536) said:

I once had a half-dozen of those hunks of shit in my test lab.

Don't you go make'n no funna my e-Machines now boy you hear!

[Keatah]

9 hours ago, zzip said:

I was all into hardware acceleration.   Anything that could offload work from the CPU seemed like a good idea to me.   One think I liked about the Gravis card was that it could mix multiple sound channels in hardware, and play up to 32 digital audio channels independently.   SB16 had two, so the computer had to do a lot of software mixing to get the same effects.   For graphics acceleration,  I knew that the faster the screen updated, the smoother the overall experience was, so I wanted to see them used.

I never paid much attention to anything "Acceleration" until maybe after my first 3D board. Then for graphics it took on a new level of urgency. Thankfully I stopped caring about it around 2006-2010. Still don't to this day.

 

Most everything felt CPU bound anyways except for select 3D games lucky enough to get the 3DFx treatment.

 

And 2D? Ha! Difference between one card and another only seemed to show up in the higher resolutions. Like desktop stuff. Lower resolution in DOS gaming seemed 100% CPU bound.
 

9 hours ago, zzip said:

The problem was not every custom hardware/coprocessor got put to good use because not every PC had them.   I went for the ASP version of SB16 thinking that the ASP chip would allow cool things,  but almost nothing used it.  Seemed like it only powered "environmental effects", so that your music can sound like it was being played in a concert hall, or a closet (your choice), but all those options seemed to make the music sound worse than without it, so I always turned it off.

Yes, so wanted to something pop up that used the ASP chip. But nothing of consequence or wow. I never cared about sound acceleration at all, ever, except for a brief one-year stint with SB-Live! and Cambridge Soundworks FPS2000 4.1 surround speakers.

 

As a complete matter of fact I found it all distracting, concentration breaking even, with sound coming at you from all angles. Thing was if you naturally turned around to see where that helicopter was, well, you'd the rest of mom's basement or last week's gym clothes piling up. The game imagery never turned with you. If sound is all around me I fully expect imagery to be all around me.

 

Wasn't any better then that Quadraphonic hippy shit from the 1970's. The old man had all that Silver & Wood flavoured Hi-Fi equipment which no one could even look at for fear of breakage.

 

And all that EAX and EAX-2 stuff Creative drummed up. All gimmicky. Never found a single game, single piece of music, or sound effect that sounded good. Especially underwater ocean effect, gosh that was lousy! Or even just the traditional concert hall. The sewer pipe one was humorous however. We'd sit around all day with beer making farting sounds with reverb cranked.

 

9 hours ago, zzip said:

It had some overhead,  I don't recall there being restrictions on which 486s could handle it though.   Probably it was a case where you'd get more raw speed from VLB, but better stability from PCI.  Both would vastly outperform ISA.

PCI is a transaction bus. It's own protocol. It's own scheduling and arbitration. Yet I always envisioned the CPU needing extra cycles to prepare and package those transactions, which would overwhelm the small cache of the 486.

Edited January 25, 2022 by Keatah

[The Usotsuki]

AMD did eventually create a clock-quadrupled 486, the 486DX5.

[zzip]

2 hours ago, Keatah said:

Ohh dear.. Absolutely not a scam. In hearing that I might suspect your buddy wanted to sound like an expert.

He read a bunch of PC magazines and knew a lot of stuff.   But it could also be that he had already invested a decent amount of money into his DX50 when the DX2's started to show up, so he needed to justify to himself that he made the right purchase decision.   And he did do a lot of research before buying it.   

 

I wonder how well-publicized the 486 50mz bus problems were before the DX2's showed up?  It could very well be it wasn't talked about, and a 50mhz 486 was top of the line.   ("Buy our best CPU!  It's unstable!"  said Intel never)

 

But from the times I got to use it, I never noticed any problems with it.  And I don't recall him complaining about trouble.

 

1 hour ago, Keatah said:

I never cared about sound acceleration at all, ever, except for a brief one-year stint with SB-Live! and Cambridge Soundworks FPS2000 4.1 surround speakers.

That was one thing I did care about.   As a test, I could play the same mod/s3m file (since they required a lot of realtime mixing) on my SB16 and Gravis Ultrasound ACE.   when played on the SB16 they might consume 25% of the CPU, while on the ACE it would consume less than 1% of the CPU.   Pretty big difference!   So I reasoned that when playing a game that required a lot of sound mixing, the SB16 would require more CPU resources that could be used to enhance performance in other areas.

 

1 hour ago, Keatah said:

PCI is a transaction bus. It's own protocol. It's own scheduling and arbitration. Yet I always envisioned the CPU needing extra cycles to prepare and package those transactions, which would overwhelm the small cache of the 486.

If the overhead was as bad as all that, people would have avoided 486 PCI motherboards like the plague.  In real world terms it probably just took a little bit of a hit.

[Keatah]

37 minutes ago, zzip said:

I wonder how well-publicized the 486 50mz bus problems were before the DX2's showed up?  It could very well be it wasn't talked about, and a 50mhz 486 was top of the line.   ("Buy our best CPU!  It's unstable!"  said Intel never)

Pretty sure the CPU and nearby electronics were fine. Provided they were built exactly as needed. And it may have been motherboard manufacturers complaining about it, getting something to be stable at those speeds.

 

The Gateway guy said something about the expansion bus slots not working with all cards. So maybe it was just that. It was so long ago and I didn't dwell on it since cost was rising quickly, into the $3200 range for their EISA system.

 

From the consumer's point of view there weren't many fuck-up CPUs. The amount of validation and testing that goes into them is enormous. Read the Pentium Chronicles for just a glimpse of what goes on.

 

Some other fails I was aware of was the infamous FDIV bug in the P60 and P66. The lack of L2 cache in the first Celeron (Covington IIRC), though that was intentional and "fixed" in Mendocino. And then there was some big stability problem with the Pentium III at 1.13GHz. A microcode thing at certain temperatures or something. But it was a stop ship on it.

 

I didn't bother after that. CPUs are too complex. And each modern one has a list a mile long. As long as they're characterized and microcode adjusted to the best of the designers' abilities it's all good. Every BIOS contains microcode to D'L into the CPU at power-on. Every time.

 

37 minutes ago, zzip said:

So I reasoned that when playing a game that required a lot of sound mixing, the SB16 would require more CPU resources that could be used to enhance performance in other areas.

I always liked the CPU to orchestrate everything when it comes to sound. I liked the granular control it exerted on a less powerful card even if it meant losing a couple of FPS. Always felt the sound to be more exact with timing. When something explodes, the game software generate a more timely sound, less lag. Same when having a sound stop precisely when it came time to. Less decoupling. Less hanging notes.

 

Seems the more CPU-hungry a card, the more "right" it seemed.

 

If the CPU sends a simple command rather than a whole series of commands, and the soundcard is making the sound on its own (so to speak), it feels out of sync. Sounds can happen a little earlier or later that what the developers intended. The soundcard may take a few ms to prepare itself and get the sound out. And each card could be quicker or slower in doing that.

 

The Apple II was king in this manner. Gotta have the least lag of anything. Who cares about the screech!

Edited January 25, 2022 by Keatah

[zzip]

8 minutes ago, Keatah said:

I always liked the CPU to orchestrate everything when it comes to sound. I liked the granular control it exerted on a less powerful card even if it meant losing a couple of FPS. Always felt the sound to be more exact with timing. When something explodes, the game software generate a more timely sound, less lag. Same when having a sound stop precisely when it came time to. Less decoupling. Less hanging notes.

 

Seems the more CPU-hungry a card, the more "right" it seemed.

 

If the CPU sends a simple command rather than a whole series of commands, and the soundcard is making the sound on its own (so to speak), it feels out of sync. Sounds can happen a little earlier or later that what the developers intended. The soundcard may take a few ms to prepare itself and get the sound out. And each card could be quicker or slower in doing that.

It's actually the opposite, when you have a card with multiple channels and DMA, all the CPU has to do is say is "play this sample", on channel 1 "and this sample" on channel 2, and this on channel 3, etc.  The results are instantaneous.

 

When you have something like Sound Blaster which only has a single stereo channel, and you need to play multiple samples simultaneously,  then the CPU needs to first mix all those samples together into a single sample, which will inevitably involve some overhead and latency, then finally play it.   

[nadir]

On 1/24/2022 at 5:30 PM, zzip said:

When I got my first PC, I had a friend with a 486DX-50 telling me I should get the same because it had a 50MHz bus,  he thought the clock multiplied chips were a scam because you were getting a slower bus.   I thought he knew his stuff so I tried to get one, but by that time DX-50s were hard to come by, everything was DX2 or DX4.  I ended up with a DX4-100.   I had no idea the 486 bus was unstable at 50mhz at the time.   His seemed to work fine.

Huh, I never heard anything about the DX-50 being unstable before now.  I had one.  DX-50 with EISA bus.  I was even able to overclock the EISA bus to 10 mhz with 3 EISA cards in it.  With just 1 EISA card I could get 12.5 mhz stable.  Eventually I dropped in one of those "586" upgrade chips.  Set at clock tripling mode it was stable for 16-bit gaming (including 32-bit extenders), but would flake out with true 32-bit protected mode software.

 

I had friends with the DX2-50 and vs the original DX-50  the DX2 was noticeably slower even in purely 16-bit software.

[nadir]

Worst computer I ever owned was an AMD K6-2.  Second worst was an AMD Opteron.  Both ran really hot and crashed all the time.  Third worst was my Apple PowerBook G4.  It was so poorly designed that if you picked it up wrong it would flex so much that the battery would fall out...

[DragonGrafx-16]

A Compaq laptop in 2008 where the hard drive crashed not even a month later... my parents bought it for me (I would have been 17 at the time) from Best Buy and the store had to replace it with a different brand of laptop because they unsurprisingly stopped selling the Compaq one. I didn't buy it though.

 

Worst PC I bought myself was my Gateway 2000 I bought around 5 years ago when I first got into DOS gaming on real hardware. Sometimes it would crash when playing Doom. Maybe something was messed up with the RAM IDK. But the Dallas RTC died and I tried to have a socket put in and a new one installed but that apparently bricked the hardware so I replaced it with an AT tower with a K6 that I bought off of the VCF forum for $20.

[DragonGrafx-16]

10 minutes ago, nadir said:

Worst computer I ever owned was an AMD K6-2.  Second worst was an AMD Opteron.  Both ran really hot and crashed all the time.  Third worst was my Apple PowerBook G4.  It was so poorly designed that if you picked it up wrong it would flex so much that the battery would fall out...

The K6 line were great CPUs... the problem you must have had is that the system builder probably didn't give the case enough cooling. Perhaps they didn't even install a fan on it. My K6 (200MHz) in my DOS PC needs one... a heatsink isn't enough. The K6 system I have is beast for DOS gaming.

Edited March 29, 2022 by DragonGrafx-16