The Top 5 Worst CPUs of All Time @ [H]

Great list.

How about the P4 Williamettes? When they launched a couple hundred mhz faster than top tier Pentium 3's, they were still slower in overall performance. They also launched with RAMBUS which was a stillborn product idea that was quickly replaced with DDR.
 
CaptNumbNutz said:
Great list.

How about the P4 Williamettes? When they launched a couple hundred mhz faster than top tier Pentium 3's, they were still slower in overall performance. They also launched with RAMBUS which was a stillborn product idea that was quickly replaced with DDR.
Click to expand...

Just came in to say this- those socket 423 with rambus-uggh
 
No love (or hate, I suppose?) for the Transmeta CPU's? Advertised as full x86 compatibility with superior performance and power savings. Really had okayish x86 compatibility, horrid performance, and only marginally better than competing products on power. A total failure of a chip, and the followup was even worse, as it maintained their goofy code morphing system, still ran dog slow, and lost the negligible power savings.

And a second on the craptastic Willamettes. Worse speed than their predecessor and expensive memory tech with no benefits. Also honorable mention for Tejas, the Pentium 4 replacement that got canned before it ever made it to market. Double the power consumption of the already-ridiculous Prescots (PresHots) and apparently worse performance.
 
Paladin21 said:
No love (or hate, I suppose?) for the Transmeta CPU's? Advertised as full x86 compatibility with superior performance and power savings. Really had okayish x86 compatibility, horrid performance, and only marginally better than competing products on power. A total failure of a chip, and the followup was even worse, as it maintained their goofy code morphing system, still ran dog slow, and lost the negligible power savings.

And a second on the craptastic Willamettes. Worse speed than their predecessor and expensive memory tech with no benefits. Also honorable mention for Tejas, the Pentium 4 replacement that got canned before it ever made it to market. Double the power consumption of the already-ridiculous Prescots (PresHots) and apparently worse performance.
Click to expand...

Transmeta Crusoe and Efficeon where revolutionary products. Intels chips got much more efficient around that time because they straight up cribbed transmetas work (they where well on the way to loosing the patent suits... and instead bought their way out with licence deals, Transmeta needed the $). Neither of the transmeta chips where ever marketed as being faster then x86 chips... they where marketed as sipping power which is what they did. Transmetas Longrun tech is licenced and used to this day by AMD, Intel, Nvidia just to name the majors.... if a chip dynamically changes its voltage and frequency chances are its doing so with the tech developed for the Crusoe, Transmetas Longrun involves designs that help in the fab process allowing lower voltages. Anyway bottom line not bad chips... just a brand new company that got Fd over by Intel and where trying to flog a chip design that was a good 5-10 years ahead of its time. If they had come along just a few years later, and found a way to make a bit more operating cash for the first few years we could easily all be running transmeta chips in our mobiles instead of arm.

The worst processor ever though... has to be the first P4s. Intel plain lost their minds... a stupid long pipe, and trying to force rambus on the world. Really what was intel thinking. A processor with a really long pipe that induces a ton of cache misses... what can we do to make it suck even more, I know lets tie it to memory that also has terrible latency, costs 30-50% more to manufacture, designed by a company that charges 30-50% more in licencing fees. Its a can't miss product. Its what happens when marketing decisions collide with engineering... they wanted a memory the other guys didn't have for the spec sheets, what they got was a terrible product that in my mind is the worst chip/system ever made. But they where able to advertise 1.4ghz machines running 800mhz ram... ohhh sounds so fast. :)
 
Last edited:
  • Like
Reactions: dugn
like this
Itanium! What a disgraceful thing. I remember having a chip and not being able to do crap with it.

If you have time in your hands, and this hasn't been done yet, here are some other ideas for articles:
- CPUs/GPUs (best, worst)
-- By year
-- By socket
-- X versus Y (AMD/Intel, Nvidia/ATI)
-- Highlight of the year (Best PC equipment each year, or 5 years, decade)
-- Insert another crazy nostalgic thing that makes us remember the good and bad old times

Just some suggestions. The K5s were bad, but it made me remember the AMD K6 II 450Mhz, which was my beast at that time.
 
When I was a kid I spend all my savings on a Pentium 60 system... sold my Amiga500 to afford it... Traumatized for life.

I wanted to use it with 3D Studio, but even simple shapes would sporatically be corrupted when moving the camera. The P60 was only partly to blame for the trauma as it was a shit system generally, e.g it only exhibited some semplance of stability when it had 'warmed' up for about 30min.

The 90's where awful times to be a PC enthusiast imo.
 
I love these best/worst articles. Part of it is nostalgia and part of it is just refreshing my memory on history. No matter what they are interesting and informative and the discussions are usually informative and interesting to read as well.

I was hoping for the PIII 1.13Ghz honorable mention and wasn't disappointed.
 
However, you have to admit, the P!!! 1.13 in Slot 1 was a pretty menacing looking beast with it's huge aluminum cooler and twin cooling fans.

I remember having a 1Ghz P!!! S370 in the day, the highest it would overclock to was 1007Mhz....
 
BulletDust said:
However, you have to admit, the P!!! 1.13 in Slot 1 was a pretty menacing looking beast with it's huge aluminum cooler and twin cooling fans.

I remember having a 1Ghz P!!! S370 in the day, the highest it would overclock to was 1007Mhz....
Click to expand...

That was the big problem with the 1.13Ghz, it was nothing more than an overvolted and overclocked 1Ghz. They simply weren't stable because 1Ghz was about the max for that revision. The Tualatins were too far out and Intel was getting its ass beat in the Mhz war by AMD so they tried to slide that out and it backfired.
 
CaptNumbNutz said:
Great list.

How about the P4 Williamettes? When they launched a couple hundred mhz faster than top tier Pentium 3's, they were still slower in overall performance. They also launched with RAMBUS which was a stillborn product idea that was quickly replaced with DDR.
Click to expand...
Exactly, I was expecting them to be no.1, not the P5, that was only on market for a very short period and more of a meh rather than an actual FAIL.

And I was pretty satisfied with my K5PR133. In non floating point performance it often matched the P166 from Intel. Yes floating point was weak even compared to the Pentium 133, but the significantly cheaper price meant I was pretty happy with it. I always felt that I got my money's worth and more, when I smugly showed benchmark results to my friend where my cpu matched or beat his P166 that cost more than twice.
 
Love these lists.

I recall throngs of my college friends getting Cyrix MII-based laptops, then getting upset when RTSs e.g. Panzer General ran like a slideshow. There was a reason they were heavily discounted compared to Intel-based machines at the time.

Would add the original Celery processors to the list, but I suppose I understand why they're not there: they did what they were meant to and allowed the masses to access the "world wide web" (via Netscape, of course).
 
SmokeRngs said:
That was the big problem with the 1.13Ghz, it was nothing more than an overvolted and overclocked 1Ghz. They simply weren't stable because 1Ghz was about the max for that revision. The Tualatins were too far out and Intel was getting its ass beat in the Mhz war by AMD so they tried to slide that out and it backfired.
Click to expand...

Yeah, I remember. It was a bitter disappointment. The Slot 1 1.13 also had a cooler that was a big hunk of copper:

1LPQya3h.jpg


I've still got two Slot 1 P!!!'s as well as two 1.2Ghz Tualatin-S chips.
 
I loved my P54CS 166 overclocked to 200Mhz with a Voodoo 3 PCI - Used to play Unreal Tournament pretty damn well.
 
Very cool read. I didn't start building PC's til 08 so most of those I'd never heard of. It's always cool to learn a little history about this thing of ours.
 
Man, I have some good ones for you.

VXWorks based systems, we had a few hundred systems deployed with the NEC R4310 and the processor was mis-printed and mis-clocked when shipped from the manufacturing plant.

Itanium, this could have been great, but support blew badly.

PowerPC, can't get into specifics, but some of their AP-101's that were early gave us some fits, and I mean major fits!

The SuperSPARC-I & SuperSPARC-II during the transition from Cray to SGI to Sun, not the processor, but the politics!

Maybe the 386SX, without the math co-processor these were dogs.

Oh well. that is my list, I have others that have caused a ton of time and thermal problems over the years. One of the biggest headaches was the Intel Pentium's with the FDIV bug, and having to swap out hundreds of processors.
 
A nice trip done memory lane, how far we have come indeed! Lists like this sure make me appreciate where we are now. It has been an amazing year for CPUs; Ryzen, Threadripper, Skylake-X, Kaby Lake-X and soon Coffee Lake. Now if only I could pick one....
 
I feel like I got clickbaited after reading that article. Thanks [F]laccid crew.



nah it was ok.
 
ChadD said:
Transmeta Crusoe and Efficeon where revolutionary products. Intels chips got much more efficient around that time because they straight up cribbed transmetas work (they where well on the way to loosing the patent suits... and instead bought their way out with licence deals, Transmeta needed the $). Neither of the transmeta chips where ever marketed as being faster then x86 chips... they where marketed as sipping power which is what they did. Transmetas Longrun tech is licenced and used to this day by AMD, Intel, Nvidia just to name the majors.... if a chip dynamically changes its voltage and frequency chances are its doing so with the tech developed for the Crusoe, Transmetas Longrun involves designs that help in the fab process allowing lower voltages. Anyway bottom line not bad chips... just a brand new company that got Fd over by Intel and where trying to flog a chip design that was a good 5-10 years ahead of its time. If they had come along just a few years later, and found a way to make a bit more operating cash for the first few years we could easily all be running transmeta chips in our mobiles instead of arm. :)
Click to expand...

I'm not claiming that their underlying research was poor, they had a great deal of interesting technologies that they used. What I'm saying is that Crusoe, as a processor, was shit. It was roughly half the speed of an Intel CPU of the time, and didn't get anywhere near twice the battery life. They did indeed claim that their chips were actually faster, based on their microarchitecture being more efficient and that this didn't show up in benchmarks because they needed to be run multiple times to get cached and let the instructions be optimized. It never got remotely near the same speed, let alone better. Some programs would just never run, period, and others were glacially slow. Transmeta had good ideas, but their implementation was bad. IEEE gave them an award because of the research behind the chip. If you want to judge on that basis, the Wilametes were fine too because NetBurst was a successful architecture (for a while), and it was just that implementation that was poor.
 
Love these lists.

Sojoe said:
The 90's where awful times to be a PC enthusiast imo.
Click to expand...
I don't agree. It was much more dynamic, multiple companies for each hardware segment, constant and big improvements each generation...


Btw wrong Discussion link from the article.
 
K5s were even worse, I harmed/killed 2 trying to overclock them.
The first one died after only raising its speed a bit.
I figured it might just have been a duff one so tried again.
The 2nd one didnt die but would only run at reduced speed after.
I didnt let anyone have one after that!

I dodged a bullet with the P4s. Stupidly expensive with Rambus, not good performers and very hot.
Lucky AMD had the Athlon XPs and Barton mobile for us nutters. Then the Athlon 64.
I bought loads of them, fun times.
 
Paladin21 said:
Transmeta had good ideas, but their implementation was bad. IEEE gave them an award because of the research behind the chip. If you want to judge on that basis, the Wilametes were fine too because NetBurst was a successful architecture (for a while), and it was just that implementation that was poor.
Click to expand...

Fair enough... I am not sure enough TM chips got into the wild to ever really judge them. I take your point though they where not what many (including their designers) hoped. I did run some tests on them back in the day though and on battery life they where no doubt 40-50% easier on battery then the intel chips of the day. I think what did them in there was that the few products that actually shipped with their chips where low end to say the least and OEMs shipped them with smaller batteries. By the time their second processor hit Intel really had flat out cribbed their tech, Transmeta didn't have the cash to see the court case through. In the end Intel threw them a bone $$ and they took it. They where just to early. I do believe if they had come to the party a few years later they could have been one of the largest players today. A few years later and they could of set their sights on tablets and phones, before ARM became the go to. As it was they where trying to compete for the early laptop market, and consumers didn't even care much about battery life on those beasts back then (I remember I used to sell them) they wanted to plug it in at a hotel they didn't care if it worked on the plane. :)

Netburst I don't think anyone could really call a successful architecture from a performance stand point. Sales perhaps sure... they where a MHZ marketing coup which was always the plan. I envision the planning stage for the p4 sounded sort of like this;
Sales Dept Slime ball in a over priced blue suit, "What we really need is Mhz lots and lots and lots of Mhz."
Engineer dude, "well Mhz doesn't really = performance and we have these ideas for the P pro cache system we call it CORE"
Sales dude, "but it won't run any faster right it will still be what 1ghz or so.... I can't sell that"
Engineer dude 2, "well we did have this crazy idea back in collage for a cool execution trace cache.... perhaps that could make up for doubling the pipeline depth so we can double the instruction pipe and push the mhz... it could perhaps run about even with our new core design in performance but run at double the frequency"
Sales dude, "yes yes that sounds perfect put that core design in the archives and lets double things.... yes double that's a word I can sell"

lol I'm sure it didn't go down exactly that way.... but in fact I think it sort of did. Intel released Netburst in 2001 and killed it off in 2006. They replaced it with Core which was an evolutionary update of the P6 which they first used in 1995. P4 is responsible for Intel spinning their wheels for at least 5 years performance wise... so they could kill the competition off with a BS GHZ marketing campaign. The first core chips could have been released 5 years earlier, for that alone p4 is a fail in my book. :) (edit: about the only useful thing I think came from Netburst was the idea for a uop cache with out a 20 stage pipe to worry about I am not sure they would have thought to bother storing things that way)
 
Last edited:
I missed out on the K5, my first amd system was a K6, it stilled suck at floating point math tho, kind of regretted that purchase.
 
My P4 2.4C @ 3.0Ghz did everything I wanted it to in the day.

I've still got it in it's associated motherboard somewhere, it's the second motherboard as the first one caught fire when one of the VRM's failed in a blaze of glory - Nothin' wakes you up faster than your PC on fire!
 
Paladin21 said:
I'm not claiming that their underlying research was poor, they had a great deal of interesting technologies that they used. What I'm saying is that Crusoe, as a processor, was shit. It was roughly half the speed of an Intel CPU of the time, and didn't get anywhere near twice the battery life. They did indeed claim that their chips were actually faster, based on their microarchitecture being more efficient and that this didn't show up in benchmarks because they needed to be run multiple times to get cached and let the instructions be optimized. It never got remotely near the same speed, let alone better. Some programs would just never run, period, and others were glacially slow. Transmeta had good ideas, but their implementation was bad. IEEE gave them an award because of the research behind the chip. If you want to judge on that basis, the Wilametes were fine too because NetBurst was a successful architecture (for a while), and it was just that implementation that was poor.
Click to expand...

And the 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.
 
Yeah Bulldozer wasn't that bad, Atoms definitely should have a place in the list. Something that ought to be mentioned (but it's not that bad too) is AMD's "triple cripple" 3 core CPUs. Hahaha I've never seen a CPU with so much hate.
 
Paul_Johnson said:
And the 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.
Click to expand...

The only Efficeon system I had experience with used one of the first models out that was a direct replacement for the Crusoe systems. It was faster...sortof...but still grossly slower than the competing Intel and AMD systems we had. As I recall, it was in the same ballpark as the crappy VIA systems that we had bought as replacements for the Crusoe-powered portables we had previously ditched. The majority of my experience (and comments) was directed at Crusoe. It was a terrible, terrible chip to actually use in a system. Interesting technology that was ahead of its times in many ways, yes. A good CPU, no. I guess if your goal is absolute power efficiency it was somewhat better than the competitors at the time (though all the systems I had direct experience with used really small form factors and batteries, and so you didn't get any better battery life), but performance on the damn thing was so poor you could make Windows stutter just dragging a window around the desktop and office apps would constantly hang and pause. I hated every minute of using and supporting them.

Also, while thinking back down memory lane, I'd give an honorable mention to the DX4 family from Intel. Did anyone else have to service them with those damn, goofy adapters to get them to bolt into some motherboards? Also a real issue at the time trying to explain to customers why a DX4-100 was a worse buy than a Pentium 75. It's bigger numbers! Bleh.
 
Last edited:
Pentium 4's on XDR memory...
Guh those were fun trying to sell them take back because of lack of upgrade options for memory!
 
Agree with others on the Willamette P4s. Somewhat surprised to not see Prescott on the list - they were garbage compared to the competition.

Content-wise, it's a nice article, but the grammatical errors are distracting. Plural nouns don't have apostrophes. "Its" is possessive and "it's" is a contraction of "it is."
 
I still have some old procs in a box somewhere.
procs.jpg
 
  • Like
Reactions: Meeho
like this
You must log in or register to reply here.
Back
Top