The Worst CPUs Ever Made

SmokeRngs said:
The PIIs all had external cache which was usually 512k (if memory serves) but ran at I believe 1/2 or 2/3 speed. The first version of the Celeron had no cache at all but the second version (Celeron A)had on-die cache in the amount of 128k. However, the Celeron cache ran at full speed. In most cases a Celeron A at the same clockspeed as a PII would perform better than the PII as long as what was running was larger than the 512k cache on the PII. The higher speed and lower latency of the cache on the Celeron A was a huge boost compared to the PII.

The Celeron A design basically became defunct in the PIII era as the cache was on-die for the PIII as well as Celerons and the PIIIs had at least twice as much. At that time the Celerons still ran on a 66mhz bus while the PIIIs were 100 and 133. You could still find the occasional PIII based Celery model which was decent because you'd be able to overclock the piss out of it raising the clock speed considerably which left the only downside being the smaller cache.
Click to expand...
Ah, you are correct about that - good memory!
 
schmide said:
It was a regressive step. The 386 dx sx differed in memory interface. The 486 dx sx differed in FPU. Other 486 clones like the AMD dx2 had FPU. It just seem inconsistent at a time when chips needed consistency.

I feel much better about many of the other CPUs on the list.
Click to expand...
Ah, that's right.
The 80386DX had a 32-bit external data bus where as the 80386SX only had a 16-bit data bus.
So basically, it would take a ~33MHz 80386SX to be on par with a ~16MHz 80386DX to process the same integer data.

The Motorola m68k line was very similar with their vanilla, LC, and EC lines as well, depending on the generations.
That makes more sense on why you would think that, though.
 
tayunz said:
If I remember correctly didn't we used to run F@H and Linux on PS3? I felt like that console generation was fun with the PowerPC stuff.
Click to expand...
Yes, it could run Yellow Dog Linux, but if memory serves me correctly, it would only run on and utilize the PPE in the Cell, which was basically an IBM 970 PowerPC CPU with in-order execution, making it quite slow.
It also only had access to the 256MB of RAM on the PS3, and had no access to the GPU acceleration beyond basic 2D or software-3D rendering via the CPU when in Linux or OtherOS.

Yes, it also had Folding@Home on it!
Though, it would only run at about 1500 PPD - even at the time, that was very modest and the Quadro P1000 I have (roughly equivalent to a GTX1050/1050Ti SSF GPU) can get over 100,000 PPD.

You are right, though, it was a fun era with PowerPC, especially with the consoles of that time period!
 
I can think of 2 more to add:

Cannon Lake: the one laptop SKU that exists is slower than Skylake in the same power envelope, and the whole 10nm effort destroyed Intel's process edge.

Llano: AMD got in a lawsuit over it IIRC. It was actually a great little laptop chip, but holy hell were they bad about binning it (mine has like 40% freq overhead with an undervolt) and actually selling the things. Which set a bad tone for future APUs.


There were some troublesome CPUs in early Android phones too, but I can't remember thier names.
 
  • Like
Reactions: N4CR
like this
Even having a whiff of Cell on that list is well out of order. Just flat out wrong. It rescued the ps3 in the end doing half the underpowered gpus work for it from the middle of that consoles lifecycle onward and schooled a generation of developers in multithreading along the way. It was from an alien planet but proved itself in spades by the end.
 
Transmeta. That was supposed to end both intel and amd. lol.
 
I was at ATV when the PS3 was getting ready for release and the early Dev kit was larger than my Corsair 650D, ran like a damn oven and crashed constantly because no one knew how to properly code for the cell... I understand the reason a lot of people love(d) it but it was a nightmare early on.

I also had the pleasure of buying one of those silly P4's with the stupid FSB and RDRAM. So basically I got fucked in every regard - but hey it was supposed to be the latest and greatest and the future of technology! It was replaced by a much nicer A64 shortly after.

Edit: Oh and that dev kit was louder than the floor fan sitting next to me!
 
Shmee said:
Ah yes those damn lazy devs. They just should have created tools for a new type of processor from scratch as well as develop for other platforms! SOOO lazy, or you know, not worth the all the extra effort.
Click to expand...

I actually wrote code on one, it was well documented on IBM's website how to use the spus. It wasn't remotely difficult, either, but it was a paradigm shift in terms of having to thread everything and write specific spu code to make proper use of them. In the age of dual cores, which were more to allow multiple programs to run together than to thread applications at the time, it required code rewrites for everything. So yes it was "lazy" developers that didn't want to completely rewrite their game engines. Understandable, but it really was the way forward. Of course, the xbox method of adding cores was easier as the old code would work fine and could just have pieces threaded here and there gradually, using mostly the same code methodology rather than the specific spu code required by the cell. But, coding for the spus themselves wasn't remotely difficult, there was just very little code reuse. And they were WAY faster.

My thought was that cfd software would be ideally suited to the cell processor, so I decided to take an open source cfd package and try to port it to the cell. Unfortunately, while not exactly difficult, it quickly became apparent that that wasn't a one man in his free time job and there wasn't a large community working on the cfd software in the first place, let alone a bunch of people willing to buy ps3s and learn how to code the spus. But they were not difficult at all to code for.
 
I almost feel the need to defend the Itanium, Prescot and Cyrix P166
 
I had the Cyrix 166+ processor back in the day. It was supposed to be equivalent to a Pentium 166, but I did some Lightwave 3D renders against my brothers Pentium 100 and the Pentium blew away the Cyrix.
Next day I sold the Cyrix setup to a friend and built a Pentium based workstation, and never used another brand for my setups since, always been Intel.
 
MV75 said:
Transmeta. That was supposed to end both intel and amd. lol.
Click to expand...

I've said it before when this came up..................Efficeon's were actually very impressive for what they were. Paired in the correct system they gave more than adequate performance (sharp had some good designs/systems) and they gave some truly outstanding battery life. They weren't really among the worst because for what they were designed to do you couldn't touch them.
 
Zepher said:
I had the Cyrix 166+ processor back in the day. It was supposed to be equivalent to a Pentium 166, but I did some Lightwave 3D renders against my brothers Pentium 100 and the Pentium blew away the Cyrix.
Next day I sold the Cyrix setup to a friend and built a Pentium based workstation, and never used another brand for my setups since, always been Intel.
Click to expand...
I had a free Cyrix 166, which was beat by my P75.
 
dgz said:
I have 20-somethings on my team who insist Intel has always made the BEST STUFF EVAR heh. They absolutely cannot comprehend things used to be different just ~13 years ago.
Click to expand...
Hey, I’m a 20 something, I take offense to that. Shit......I’m almost a 30 something.

I remember in highschool arguing with a dude over his love for his Penguin D and Nvidia 5800. Same guy, different time periods of PC builds. He refused to hear that there were better options. And he is a 30 something now...
 
I have to throw the Pentium pro into this class.

While the pentium pro was supposed to Intel's whipping horse for new tech, it would up being a clusterfk to those that bought onto the hype (including yours truly). Back in the day these were the 1st processors to hit the 200Mhz speed (66mhz x 3 multiplier). Also having their L2 cache on-chip (but not on-die) clocked to the processor speed they were the granddaddy of the processors that have on-die L2 and L3 caches we use today.

While Intel promised a long upgrade path for this platform it never materialized. The original Pentium Pros ran for 3 years. What Intel delivered for next gen upgrade was a kludge called Pentium II Overdrive which had a pentium chip and l2 cache on a daughterboard that could fit in the Socket 8. Like the Pentium Overdrive that was released for 80x486 owners, it was a mess and did nothing but infuriate the owners of Pentium Pros who saw it as Intel simply paying lip service to their lack of design foresight. What actually followed the Pentium Pro was another dead-end mess called the "Slot 1" which used cartridges that contained the bits of the Pentium II overdrive with an integrated cooler and plastic shell. Pretty much a Nintendo Cartridge containing a CPU. Those were superseded by Intel's return to socketed chips with the Socket 370 Pentium 3 Coppermine

Over the last couple of decades I have saved some of my more memorable chips from the systems I built for myself:

museum.jpg MBP_0357.JPG

That is a quick pic of my 'museum' parts (I have another box full of newer or less memorable chips). The smallest one in the lower right corner is a '286. That box has a few layers the other is of some of my older sip and simm memory. I have a naked pentium 2 Slot 1 cpu somewhere (probably in the middle layer of that box). I killed that one trying to get the dang cartridge & HSF off - learned the trick of that later on) cost $800 at the time. Hardware hacking can be expensive QQ. Anywho - yeah, the Pentium Pro really needs to be on that list.
 
pcgeekesq said:
Shoulda waited for the last train.

[And for you kids, that's a song reference.
But it's kind of meant to refer to the risk of trying to stay on the "bleeding edge" of technology.]
Click to expand...

My 23 year old sings that song and Im a Believer. It helps his creativity with underground/freestyle, as do the Doors and other lyricists from your era.
 
cyberguyz said:
I have to throw the Pentium pro into this class.

While the pentium pro was supposed to Intel's whipping horse for new tech, it would up being a clusterfk to those that bought onto the hype (including yours truly). Back in the day these were the 1st processors to hit the 200Mhz speed (66mhz x 3 multiplier). Also having their L2 cache on-chip (but not on-die) clocked to the processor speed they were the granddaddy of the processors that have on-die L2 and L3 caches we use today.

While Intel promised a long upgrade path for this platform it never materialized. The original Pentium Pros ran for 3 years. What Intel delivered for next gen upgrade was a kludge called Pentium II Overdrive which had a pentium chip and l2 cache on a daughterboard that could fit in the Socket 8. Like the Pentium Overdrive that was released for 80x486 owners, it was a mess and did nothing but infuriate the owners of Pentium Pros who saw it as Intel simply paying lip service to their lack of design foresight. What actually followed the Pentium Pro was another dead-end mess called the "Slot 1" which used cartridges that contained the bits of the Pentium II overdrive with an integrated cooler and plastic shell. Pretty much a Nintendo Cartridge containing a CPU. Those were superseded by Intel's return to socketed chips with the Socket 370 Pentium 3 Coppermine

Over the last couple of decades I have saved some of my more memorable chips from the systems I built for myself:

View attachment 94203 View attachment 94206

That is a quick pic of my 'museum' parts (I have another box full of newer or less memorable chips). The smallest one in the lower right corner is a '286. That box has a few layers the other is of some of my older sip and simm memory. I have a naked pentium 2 Slot 1 cpu somewhere (probably in the middle layer of that box). I killed that one trying to get the dang cartridge & HSF off - learned the trick of that later on) cost $800 at the time. Hardware hacking can be expensive QQ. Anywho - yeah, the Pentium Pro really needs to be on that list.
Click to expand...

I remember that now and the first PIIs couldn't beat the pro 200mhz with 512kb cache, the 256kb ones yeah but not the next one up let alone the very rare very expensive 1mb models. You had to wait for the second wave of PIIs to have something like an upgrade. Truly hideous 16bit software performance so it was Win NT or nothing. That point though was where x86 grew up from being only for domestic use to proper serious use, sounding the death knell for high-end risc cpus in servers and workstations like what Silicon Graphics were peddling for 100k a pop.
 
I had a diff experience with P Pro’s, we had P166’s and Pro’s for CAD and the it was perfect for them.
 
People diss the CELL to much go play Final Fantasy 13 thats the full power of the CELL and even today the Graphics and play look and feel better then the newer FF15. The backgrounds are clear and sharp unlike the newer PS4 with blurry backgrounds. It's only clear and sharp in close range but backgrounds are blur because the graphics engine of PS4 and PS4 pro still cannot compete with Nvidia.

It's just sad really. Both the Xbox one and Ps4 today are basicly the same machine. Sony used to be unique in the Gaming world. Even today playing God of War 2 on the PS2 that game sometimes looks better then the current gen stuff. The emotion engine was amazing. Sony has lost it unique systems and just went with what everyone else wanted. Just like Microsoft.
 
I will agree with what a few others have said. No way Cell belongs on this list. The Cell was a game changing chip in a lot of ways. It was ahead of its time... it was collaboratively developed by more then one company. (Sony, IBM, Toshiba)
Cell also helped push a handful of engineers careers forward. Cell was designed to fix todays problems... not the PlayStation 3s.

Lisa Su, director of emerging products at IBM Microeletronics, said, "We started with a clean sheet of paper and sat down and tried to imagine what sort of processor we'd need five years from now."

Cell might not have ended up doing to much outside the Playstation... still the chip design was revolutionary compared to what Intel and AMD where doing at the time. Their design ideas regarding scalebility laid the ground work for things like Threadripper. I would wager to bet we wouldn't have Ryzen at all and tech like infinity fabric without the work done on Cell. Nor would people like Su be where they are now, it makes a lot of sense that Sony has partnered with AMD since Cell, Su and much of the cell team ended up there.
 
ChadD said:
I will agree with what a few others have said. No way Cell belongs on this list. The Cell was a game changing chip in a lot of ways. It was ahead of its time... it was collaboratively developed by more then one company. (Sony, IBM, Toshiba)
Cell also helped push a handful of engineers careers forward. Cell was designed to fix todays problems... not the PlayStation 3s.

Lisa Su, director of emerging products at IBM Microeletronics, said, "We started with a clean sheet of paper and sat down and tried to imagine what sort of processor we'd need five years from now."

Cell might not have ended up doing to much outside the Playstation... still the chip design was revolutionary compared to what Intel and AMD where doing at the time. Their design ideas regarding scalebility laid the ground work for things like Threadripper. I would wager to bet we wouldn't have Ryzen at all and tech like infinity fabric without the work done on Cell. Nor would people like Su be where they are now, it makes a lot of sense that Sony has partnered with AMD since Cell, Su and much of the cell team ended up there.
Click to expand...
Right, most of the early issues with CELL belong to Sony, not the chip itself.
 
I don't know what was worse, was it my Texas Instruments 8086, or my Pentium III that was really a Pentium II?
 
trparky said:
Not all of them were trash, I remember that the first couple of Celerons could be overclocked to hell and back. Some of them were even better than the Pentiums. Sadly, after that Intel neutered them.
Click to expand...

Many of the early Celerons did. I had a Celeron 266 that would break 400MHz easily. Unfortunately, being cacheless made that a moot point as it was still shit. The ones that were faster than the Pentium II's of the day all had on-die full speed L2 cache.

Red Falcon said:
To my knowledge, all of the Pentium II CPUs had external cache.
Even the Pentium II Xeon & Celeron had an external cache, though it ran at the full CPU speed instead of at half like the normal Pentium II.
Click to expand...

Incorrect. All Pentium II's did have external cache. It was half the speed of the CPU. The Celeron on the other hand either had no cache or full speed on-die L2 cache. The cache sizes were smaller than that of their Pentium II brethren, but made up for this by being as fast as the CPU itself. This was the secret to the overclocked Celeron 300A's that hit 450MHz being faster than the Pentium II 450MHz chips. It's also worth noting that early Pentium II's clocked at 200MHz were significantly slower than the Pentium Pro at the same clock speed which had full speed on-die L2 cache.
 
There were tons of bad chips before these happened. The DLC and SDLC 386/486 hybrid chips (turn your 386 into a 486 stuff) were all kinds of awful. TI made some 486 chips back in the low 90's that sucked. But focusing on Pentium forward... The 6x86 chips weren't all bad, they were cheap and fast in regular ops but just sucked at FPU which meant games sucked.

The Itanium should be the #1 and done on this list. Intel sunk a fortune into this tech and it was horrid. We had Itanium clusters at work and they were the biggest POSes we had. Maybe it was ahead of it's time or implemented poorly, or just "misunderstood". But everyone I worked with had universal hatred of the Itanium systems we had.
 
atom said:
I don't know what was worse, was it my Texas Instruments 8086, or my Pentium III that was really a Pentium II?
Click to expand...

Methinks you got a bit confused there :/

Intel is responsible for the 8086 (and it's little brother the 8088) which started the whole WinTel x86 generation of computers. In fact the 8086 is probably the most famous chip out there.

Texas Instruments had an 8080, some 486, and the TMS9900 which had a really unique architecture. It was one of the very 1st 16-bit processors put there having only a 3 16-bit registers on-chip (Instruction Pointer {IP}, Status {flags} register and a Workspace pointer {WP}). 16 x 16-bit General purpose registers were stored in ram. It was particularly efficient with subroutines which only required changing the WP to do a context switch and point to a whole new set of registers - including the IP and Flags. This was the processor I cut my teeth on when programming Assembly language. Unfortunately because of the sheer unorthodox architecture, it never really caught on in much beyond the TI-99/4A home computer/game console. That's a real shame because it was a real hoot to code on. While not the most popular of chips the TMS9900 and Motorola 68000 holds special places for me.

Of the Intel line of processors I would say the 8080 and PIII were a couple of my favorites. Pentium, PPro, PII and Celeron for me were kindof ... meh!. The Ppro had huge promise, but it never really went any further in that form than a single generation.


Burticus said:
The Itanium should be the #1 and done on this list. Intel sunk a fortune into this tech and it was horrid. We had Itanium clusters at work and they were the biggest POSes we had. Maybe it was ahead of it's time or implemented poorly, or just "misunderstood". But everyone I worked with had universal hatred of the Itanium systems we had.
Click to expand...

There's a reason we all referred to them as the Itanic cpus
 
Last edited:
Zepher said:
I had the Cyrix 166+ processor back in the day. It was supposed to be equivalent to a Pentium 166, but I did some Lightwave 3D renders against my brothers Pentium 100 and the Pentium blew away the Cyrix.
Next day I sold the Cyrix setup to a friend and built a Pentium based workstation, and never used another brand for my setups since, always been Intel.
Click to expand...


one other chip that i was kinda surprised didnt make this list was the WinChip. Man.. was that a slow turd like the cyrix.
 
katanaD said:
one other chip that i was kinda surprised didnt make this list was the WinChip. Man.. was that a slow turd like the cyrix.
Click to expand...

Did any of the Via chips make it on the list? I don't have the time to read it right now lol

edit:.. nevermind, just read the first paragraph.

To make it on to this list, a CPU needed to be fundamentally broken, as opposed to simply being poorly positioned or slower than expected. The annals of history are already stuffed with mediocre products that didn’t quite meet expectations but weren’t truly bad.
Click to expand...

edit2: decided to read it quick. Via wasn't included, but Cyrix was twice, so I'll take it lol
 
Last edited:
atom said:
Yes I am very confused. Sorry for that. Was thinking of a 486 clone. Windows 95 era AST brand computer, although I can't find any mention of them using anything but an Intel CPU.

Edit: I think it was the 486 DX2
https://cpumuseum.jimdo.com/museum/texas-instruments/80486-d-slc/
Click to expand...

I worked at AST in the low 90's. The business class machines were all Intel, but the "home" systems (ie - Advantage, etc) of the 486 variety could have been Intel, Cyrix, AMD, etc... depending on model and whatever they got bulk deals on. There were tons of Cyrix and AMD DX2 (50-66mhz) models.

"https://cpumuseum.jimdo.com/museum/texas-instruments/80486-d-slc"

Not that one. Those are DLC cpu's... basically a 386 replacement chip that had some internal 486 functions in it. They were marketed as "just as good as a 486"... and they most certainly were not. BUT, they did make for cheap 386 socket upgrades. I listed these in my earlier post about awful CPUs. I had a couple of these in cheapo systems because they were dirt cheap. OMG they were so slow, and had no FPU to speak of. But they were super cheap, so they made good test bench systems.

And when I say cheap, and I'm talking about the low 1990's here, I think these things were like $25 compared to a 486 DX-33 for $200. I bought some that were surface mounted (not socketed) for like $30, motherboard and all. I remember trying to play X-Wing on one of these and it was pure agony. DOOM was OK though.
 
Last edited:
  • Like
Reactions: atom
like this
Why is Bulldozer on the list? It was competitive with Sandy Bridge and Intel's single thread performance advantage... lets just say it melted down to a spectre of its former glory.

I still have Bulldozer and Piledriver machines active today.
 
Why is Prescott on the list instead of Williamette on s423?? Both the socket AND the architecture had to be replaced within a year of release. This was a way bigger failure than Prescott was.
 
You must log in or register to reply here.
Back
Top