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

Wyodiver said:
I don't know, I loved my AMD Athlon 2500+ Barton, paired with a Radeon 9500 Pro. I really enjoyed that slice of time. And I have high hopes for AMD now.
Click to expand...

While the CPUs are fine I'm sure, I loathed the entire Socket 462 era. Heaps and heaps of flaky motherboards running terrible drivers, SATA controllers, LAN adaptors and chipsets. Data corruption, blue screens, memory issues, slow USB..... Brand didn't matter much. Asus, MSI, Gigabyte, etc. were all plagued by these issues. For every stable AMD rig I'd encounter, I would have to deal with a dozen awful ones.

I am sure its not AMD's fault though. I blame SiS, nVidia, and VIA. But dealing with it all at work made paying the Intel "tax" for an Intel motherboard chipset at home worthwhile.
 
While I agree with having the K5 in the piece is warranted, your commentary on it is off. The K5 released only a few months after Pentium Pro, so it is a bit unfair to say it took ideas from the Pro. The K5 was actually an AMD RISC 29050 processor which had been released nearly 6 years earlier, grafted to a x86 translating front end.

K5's issue was that 1) it suffered several delays in development, and 2) AMD wasn't able to scale its frequency as hoped, with it initially topping out at 100MHz at a time Intel was releasing 200Mhz Pentiums. Had it scaled correctly, its FPU performance would have been comparable, and its integer performance would have been beastly.

The end result was the same, but it is unfair to characterise it as a poorly implemented Pentium Pro clone.
 
  • Like
Reactions: xrror
like this
DracoDan said:
Missing CPUs that *I* think belong on this list:
  • Intel P4 Willamette - Very hot, power hungry, slower than PIII, socket 423 was replaced after barely a year (no upgrade path!), Initially required crazy expensive and effectively slow RAMBUS, support for old PC133 added later while AMD was already on DDR.
  • AMD Am5x86 - You thought the K5 was bad? Cute... AMD tried to pass this off as a Pentium competitor, it was no coincidence that it used a 486 chipset and socket. We had one of these that ran at 133mhz, AMD was honest-ish by giving it a PR score BELOW its actual MHz speed! The 133mhz part was rated at "PR75". In reality it was about half the speed of a P75 for any floating point, which more and more programs had started to use. Also, since it was 486 based almost all compatible motherboards were limited to ISA (no PCI).
  • Transmeta Crusoe - most people have never even heard of these, they were ultra-low power CPUs used in a handful of laptops (like the tiny Sony VAIO PCG series. Unfortunately they were also ultra-low performance. It wasn't an actual x86 CPU, but instead used "code morphing software" to translate x86 instructions to it's own VLIW architecture. The designers intended to be able to "upgrade" the CPU by updating the firmware (aka code morphing software) but this never came to be. It was great in theory, but it just couldn't keep up with even the lowest end AMD and Intel CPUs of the time.
  • Anything VIA (applies to their CPUs, chipsets, "graphics" adapters, ...really anything they touch) - Do you hate stability? performance? compatibility? driver updates? future OS support? sanity?! VIA is there to help!!!
  • Cyrix MediaGX - To the guy that said the Geode was bad, the Geode was an EVOLUTION of this piece of garbage! Another fake "Pentium", this chip was actually a 486 in drag. To make things worse, everything was integrated into the core
Click to expand...

All good points, though 5x86 did support Socket 3 and therefore worked on a number of PCI equipped motherboards. I had one in a hand me down Compaq that had PCI slots.
DracoDan said:
Missing CPUs that *I* think belong on this list:
  • Intel P4 Willamette - Very hot, power hungry, slower than PIII, socket 423 was replaced after barely a year (no upgrade path!), Initially required crazy expensive and effectively slow RAMBUS, support for old PC133 added later while AMD was already on DDR.
  • AMD Am5x86 - You thought the K5 was bad? Cute... AMD tried to pass this off as a Pentium competitor, it was no coincidence that it used a 486 chipset and socket. We had one of these that ran at 133mhz, AMD was honest-ish by giving it a PR score BELOW its actual MHz speed! The 133mhz part was rated at "PR75". In reality it was about half the speed of a P75 for any floating point, which more and more programs had started to use. Also, since it was 486 based almost all compatible motherboards were limited to ISA (no PCI).
  • Transmeta Crusoe - most people have never even heard of these, they were ultra-low power CPUs used in a handful of laptops (like the tiny Sony VAIO PCG series. Unfortunately they were also ultra-low performance. It wasn't an actual x86 CPU, but instead used "code morphing software" to translate x86 instructions to it's own VLIW architecture. The designers intended to be able to "upgrade" the CPU by updating the firmware (aka code morphing software) but this never came to be. It was great in theory, but it just couldn't keep up with even the lowest end AMD and Intel CPUs of the time.
  • Anything VIA (applies to their CPUs, chipsets, "graphics" adapters, ...really anything they touch) - Do you hate stability? performance? compatibility? driver updates? future OS support? sanity?! VIA is there to help!!!
  • Cyrix MediaGX - To the guy that said the Geode was bad, the Geode was an EVOLUTION of this piece of garbage! Another fake "Pentium", this chip was actually a 486 in drag. To make things worse, everything was integrated into the core (graphics, sound, memory controller and PCI controller). Sure it was low power, but the performance was astonishingly bad, partially thanks to having no L2 cache. Since everything was integrated on the CPU, the only option was embedded.
CPU that I think didn't belong on the list:
  • AMD Bulldozer - Was it great? No. Was it good? ...well, no, not really, but was it decent for the price it was sold at? absolutely! This chip allowed for some very decent gaming systems for dirt cheap.
Click to expand...

All good points, though 5x86 did support Socket 3 and therefore worked on a number of PCI equipped motherboards. I had one in a hand me down Compaq that had PCI slots. It also had 4 30pin SIMMs amd 1 72 pin SIMM memory slots, a real blend of old and new (at the time) tech.
 
  • Like
Reactions: N4CR
like this
Nice editorial. It definitely gets you to think about the CPUs of past.
Any plans/thoughts on an editorial about best or worst overclocking CPUs?
 
Don't understand the hate for the Atom. I still have an HP Mini with an Atom from around 2011 and as a whole it's still more useful than a current iPad with almost the same battery life. <shrug>
 
Don't think Kyle ever reviewed an Atom?

Atoms had their place, often times you could get a solid working foundation with a weak CPU. That opened up a lot of builds for me personally but the first few iterations (early dual cores) left a lot to be desired.
 
The Willamette P4 could easily replace Bulldozer on the list. It had all the drawbacks listed for Bulldozer and had the added double whammy of:

1) RAMBUS - Nuff said.

2) A shockingly short product life. It was realistically just over 6 months, as most held out the last couple months because the platform was crap. Anyone how bought into it was royally screwed down the road. Forget replacing a motherboard, just a simple RAM upgrade had you shopping against an entire core build.

The Tualatin (or whatever it was, the PIII 1133, don't care enough to look it up) fiasco was pretty crappy, but it was always the last hurrah on that socket, whereas the Willamette was supposed to be the next big thing. Peeps got burned so bad.

Edit: Oh yea, and the Atom was actually pretty good for what it was. The early ones could be pretty chunky if you weren't good about closing things, but later models were fine for surf/stream systems. They only really fell down when peeps wanted to throw gaming or workstation loads at them.
 
gxp500 said:
No netburst p4?
Click to expand...

I knew I'd catch some flak on not having the Pentium IV in the list. The reality of it is that while Netburst wasn't great, there were some great chips in the lineup. The 2.4C, 3.0C, 3.06B, etc. were all great chips. Sure the early socket 423's were dogs and the Prescott cores were ovens, but some of those chips were pretty good even if they weren't necessarily the fastest out there. I ran P4's over Athlon XP's and would again if I had to. The platform was better and they just worked. The Pentium IV in a lot of ways was a low point for Intel, especially when saddled with RDRAM and lower speed DDR modules, but the Pentium IV gave us Hyperthreading and some of them overclocked well. Later iterations of the Pentium IV and Pentium D had their places too.

OldGator said:
Surprised nobody mentioned the Celerons that were based on the socket 423/socket 478 P4s- those were some miserable CPUs in terms of performance and features.
Click to expand...

True, but I felt like the ones in the list were worse than those were.
 
Unibomber13 said:
Nice editorial. It definitely gets you to think about the CPUs of past.
Any plans/thoughts on an editorial about best or worst overclocking CPUs?
Click to expand...

You'll be seeing one on some of the best overclockers of all time. I hadn't done one on the worst. I might though, if the mood strikes. :)

DracoDan said:
Missing CPUs that *I* think belong on this list:
  • Intel P4 Willamette - Very hot, power hungry, slower than PIII, socket 423 was replaced after barely a year (no upgrade path!), Initially required crazy expensive and effectively slow RAMBUS, support for old PC133 added later while AMD was already on DDR.
  • AMD Am5x86 - You thought the K5 was bad? Cute... AMD tried to pass this off as a Pentium competitor, it was no coincidence that it used a 486 chipset and socket. We had one of these that ran at 133mhz, AMD was honest-ish by giving it a PR score BELOW its actual MHz speed! The 133mhz part was rated at "PR75". In reality it was about half the speed of a P75 for any floating point, which more and more programs had started to use. Also, since it was 486 based almost all compatible motherboards were limited to ISA (no PCI).
  • Transmeta Crusoe - most people have never even heard of these, they were ultra-low power CPUs used in a handful of laptops (like the tiny Sony VAIO PCG series. Unfortunately they were also ultra-low performance. It wasn't an actual x86 CPU, but instead used "code morphing software" to translate x86 instructions to it's own VLIW architecture. The designers intended to be able to "upgrade" the CPU by updating the firmware (aka code morphing software) but this never came to be. It was great in theory, but it just couldn't keep up with even the lowest end AMD and Intel CPUs of the time.
  • Anything VIA (applies to their CPUs, chipsets, "graphics" adapters, ...really anything they touch) - Do you hate stability? performance? compatibility? driver updates? future OS support? sanity?! VIA is there to help!!!
  • Cyrix MediaGX - To the guy that said the Geode was bad, the Geode was an EVOLUTION of this piece of garbage! Another fake "Pentium", this chip was actually a 486 in drag. To make things worse, everything was integrated into the core (graphics, sound, memory controller and PCI controller). Sure it was low power, but the performance was astonishingly bad, partially thanks to having no L2 cache. Since everything was integrated on the CPU, the only option was embedded.
CPU that I think didn't belong on the list:
  • AMD Bulldozer - Was it great? No. Was it good? ...well, no, not really, but was it decent for the price it was sold at? absolutely! This chip allowed for some very decent gaming systems for dirt cheap.
Click to expand...

As for the MediaGX, I recall those. They were cheap and worked well enough for office type builds. It was dismal in a lot of ways, but it's not the worst thing ever. Willamette Pentium IV's almost made the list. As for Bulldozer, even for the price I can't agree. They'd get their asses kicked at gaming my Core i3's. That's fucking sad. Bulldozer, in retrospect is a bit better today now that some games are better at going over 4 threads now. Still, I hated those chips. They ran hot, didn't go much beyond stock clocks in some cases and the IPC was dismal. Lastly, the motherboard situation for those wasn't good either.
 
I love the list in the article and agree with it, but, for me as a system builder from 1998 to 2010 or so I often wondered about the bad CPUs "what would they have been like had they had good chipsets and motherboards to go with them?" It's too bad we'll never know that.

The bane of my existence for many of those years was dealing with the low end Compaqs and clones that used those processors. There were times when I opened up a system and saw one of those CPUs and thought "why did those companies build with these things?" I fought tooth and nail against my boss during those years to not sell or build with those CPUs. It was also tough explaining to the buyers why their computers would never likely be stable or reliable. Whenever most customers wanted a new build from me it was always Intel on an Intel board to avoid the headaches (unless they were a gamer or someone who demanded more then I'd build whatever they wanted). I was never anti-AMD but for me knowing that putting a great CPU on a sorry board would result in a crappy build and most average people would not understand the subtlety of that I tried to avoid selling them (except to people I knew were enthusiast and would pay for the few good boards). In terms of saving time when putting together new systems on a daily basis having to fuss with bad drivers and problems with Windows more than made up for the $40 or so in cost for the Intel tax.

I never sold a single Willamette based build to anyone thought I managed to get MaLabs and ASI to comp me enough parts to put a few together for my own knowledge and I can't recall ever seeing more than 1 or 2 among all the systems I worked on over the years. Hated every minute trying out those builds - not due to instability but the fact that they performed worse than the high end PIII/Tualatin (which the Tualatin was one of my favorite CPUs since it would even work without a heatsink installed on it - that was fun). RAMBUS sucked.
 
Northwood P4s etc all good but presshot? No mention? Come on..

They used to cause us a lot of work by needing dust cleaning much more often than everything, because they ran the fans so hard. They stuck those shitty extruded heatsinks on them and they'd just clog due to the sheer volume of air all the time at idle and any moderate load.

T-birds were pretty hot but performed well and first to 1GHz. Other than that they were pretty crappy for the heat at the time.

Dan_D said:
You'll be seeing one on some of the best overclockers of all time. I hadn't done one on the worst. I might though, if the mood strikes. :)
Click to expand...

I implore you to put the P4 mobiles and oppy 939s on that list, they were not so well known at the time outside of enthusiast circles. Got my oppy 165 duallie from 1.8 to 2.85GHz setting a WR on air. Those things all went hard, 30-40% OCs were normal. Binned sillicon FTW.
 
iamjanco said:
Intel Pentium 60/66MHz (1993)? I paid more than two grand at our local, newly opened CompUSA for a Packard Bell equipped with that chip back in the day. It was so good that it convinced me to move away from Apple.

'Course, that was before I really knew what I was doing (sometimes I wonder if that's still the case).
Click to expand...
I too had a Packard Bell but had the Pentium 75 I overclocked to 90mhz. I bought it at something that resembled costco. I assume it was samsclub or pace. It had a 850mb hard drive. Oh yea... 4X NEC Cd-rom and a Cirus Logic 5430 Onboard PCI that I added 1 extra meg of ram to. Lol
 
J3RK said:
VIA

:p
Click to expand...
VIA had their place. They practically invented ITX form factor.
As well they were ahead of the curve for power saving x86.

Never mattered as far as performance goes. However they had their niche under wraps for a time.

I'm not going to laugh at either of those things.
 
  • Like
Reactions: N4CR
like this
Fun fact: working on one of the official listings and one in the dishonorable-mention pile in the thread drove me from CPU design.

Really cool tech in both cases, ahead of the time. Too ahead, in both cases.

Having said that, both products had bleeding edge tech which did eventually work well and still in use today. Just not when it launched. I tell myself that as I read these lists and drink heavily.
 
Last edited:
PhaseNoise said:
Fun fact: working on one of Kyle's listings and one in the dishonorable-mention pile in the thread drove me from CPU design.

Really cool tech in both cases, ahead of the time. Too ahead, in both cases.

Having said that, both products had bleeding edge tech which did eventually work well and still in use today. Just not when it launched. I tell myself that as I read these lists and drink heavily.
Click to expand...
Details!?

Btw, people, it's Daniel's list of worst CPUs.
 
I'm surprised Pentium D, aka Preshott doesn't even get a mention.
That processor was the literal definition of gluing together two CPUs.

Sure, the Pentium D 805 was a cheap dual core, even cheaper than some AMD chips at the time, but that thing ran extremely hot.
Add to that, Conroe, Core 2 Duo being right around the corner...
 
  • Like
Reactions: N4CR
like this
P4 Williamette, P4 Prescott, Pentium D, basically any Atom or Celeron. All chips not listed that belong in a dumpster.
 
SixFootDuo said:
Some of you old skool guys will remember the Matrox Mystique. At the time it was THE CARD to own. Then of course a few years later around 1996 you paired the Matrox with a Diamond Monster 3DFx Voodoo.

Ahhhh .... the sweet sweet dreams of going to bed at night dreaming of the morning when I can then wake up and play with my awesome Celeron 300A system.
Click to expand...

For a long time I would always use a Matrox Millenium then a Millenium 2. Then OpenGL came mainstream and many game supported it instead of Direct X. So, no more Matrox products. I got myself a Hercules GeForce 2 GTS Pro.

I believe the Celeron 300A, when used with the 440BX chipset, could be among the top 3 in a top 10 list
 
Based on reviews of Bulldozer I remember reading when compared to Intel's releases of the same years, Bulldozer was significantly lacking. However, based on personal experience the FX-8120, which I still use now, has been my greatest and here's why.
I had a Asus G50V-X1 laptop. (https://www.newegg.com/Product/Product.aspx?Item=N82E16834220379 won at ROGcon). I used it for just over 4 years. As new games were coming out that would not perform well on it, I decided to retire it. (still works to this day) I was on a budget. So I got the FX-8120 along with a Radeon HD 7750. The first game I played with my new rig was Bioshock Infinite and I was blown away by the performance.
 
dark_reign said:
I had the Matrox Millennium paired with a Diamond Monster, and at the time it was a huge game changer. Yup, I remember the graphics in QuakeGL being mind blowing. Also, that time period reminds of upgrading my gaming PC every year.
Click to expand...

remember borrowing a friends verite rendition card just for the amazing open gl performance I got in quake 2
 
Meeho said:
Details!?

Btw, people, it's Daniel's list of worst CPUs.
Click to expand...

I'll just say a lot of things are really good ideas in a general sense, and on the way to retail - shit happens. A fab can't do something as well as you wanted, compilers can't do things as well as promised, OSes are idiotic with how they handle things, etc.

I have no beef with this sort of shame list really, but it sucks just on a personal level knowing we had 9 out of 10 parts working great, and the world will forever remember the 10% which didn't go well.

Such is life. There's winners and there's everyone else.
 
MrDeaf said:
I'm surprised Pentium D, aka Preshott doesn't even get a mention.
That processor was the literal definition of gluing together two CPUs.

Sure, the Pentium D 805 was a cheap dual core, even cheaper than some AMD chips at the time, but that thing ran extremely hot.
Add to that, Conroe, Core 2 Duo being right around the corner...
Click to expand...

All Prescott, Cedar Mill, Presler and Smithfield cores were garbage compared to their AMD counterparts.

EDIT: They were garbage PERIOD.
 
How about the other part of the puzzle?
Worst and best chipsets...a buggy chipset could drag down a stellar CPU (looking at VIA).
 
Factum said:
How about the other part of the puzzle?
Worst and best chipsets...a buggy chipset could drag down a stellar CPU (looking at VIA).
Click to expand...

I never really got the VIA hate. My first PC I bought with my own money used an ECS board (pretty sure it had a via chipset AND it was bottom of the barrel for board manufacturers) paired with a Third 1.4 and I never saw an issue.

Same as now I use an ASRock board, but if you talk to people they say steer clear. I guess I have just had good luck with less than stellar choices lol
 
My friends Cyrix 6x86 chip was a joke even way back when....

Intel cornered the market for years because of Quake and floating-point.

History repeated itself with K6-2+ "3D Now" and Quake II verus P-II.

Remember when integer vs floating-point was a discussion point?
 
auntjemima said:
I never really got the VIA hate. My first PC I bought with my own money used an ECS board (pretty sure it had a via chipset AND it was bottom of the barrel for board manufacturers) paired with a Third 1.4 and I never saw an issue.

Same as now I use an ASRock board, but if you talk to people they say steer clear. I guess I have just had good luck with less than stellar choices lol
Click to expand...

I tried twice the horror of Via's 4in1 chipset + drivers...I never touched anything but an Intel CPU with an Intel chipset after that.
 
The Cyrix brings back memories of having to create a MASM executable running at startup to disable the onboard cache which stabilized the system. To this day I'm still unsure if it was the cache or the chipset implementation which was causing the stability issue.
 
Factum said:
I tried twice the horror of Via's 4in1 chipset + drivers...I never touched anything but an Intel CPU with an Intel chipset after that.
Click to expand...

2 times huh? Damn, quite the expert.

I used all AMD systems from a During 750 up to the Barton core. Until the Nforce2 it was all VIA KT chipsets. I don't remember a single bad board. Corrupting hard drives from to high of an FSB sure, but that's not surprising. Abit and Shuttle made good stuff back then.

KT133a, 266a, 333 were all good chipsets. I
 
Dan_D said:
I knew I'd catch some flak on not having the Pentium IV in the list. The reality of it is that while Netburst wasn't great, there were some great chips in the lineup. The 2.4C, 3.0C, 3.06B, etc. were all great chips. Sure the early socket 423's were dogs and the Prescott cores were ovens, but some of those chips were pretty good even if they weren't necessarily the fastest out there. I ran P4's over Athlon XP's and would again if I had to.
Click to expand...
Re: 2.4C/3.0C

From what I recall there was so much demand for the 2.4C that there were claims that Intel was taking chips that were earmarked to be 3.0Cs and setting the multiplier to 2.4C (in order to have enough chips to fill the demand). Enthusiasts would get ahold of those 2.4C chips (that were really 3.0C), and set the multiplier to 16x. Faithfully we'd read thread after thread stating 'ZMOG the 2.4C is a legendary overclocker!!!!!11' when in reality you squeezed 200 MHz out of it...

AMD will never own up to it, but I'm sure the same was true with the Barton 2500+ XPs vs 3200+ XPs. I never actually heard of anybody who owned a Barton 2500+ that could NOT run it at 3200+...which leads me to believe they were all 3200+ chips, it's just that some of them came in 2500+ boxes...
 
pfc_m_drake said:
Re: 2.4C/3.0C

From what I recall there was so much demand for the 2.4C that there were claims that Intel was taking chips that were earmarked to be 3.0Cs and setting the multiplier to 2.4C (in order to have enough chips to fill the demand). Enthusiasts would get ahold of those 2.4C chips (that were really 3.0C), and set the multiplier to 16x. Faithfully we'd read thread after thread stating 'ZMOG the 2.4C is a legendary overclocker!!!!!11' when in reality you squeezed 200 MHz out of it...

AMD will never own up to it, but I'm sure the same was true with the Barton 2500+ XPs vs 3200+ XPs. I never actually heard of anybody who owned a Barton 2500+ that could NOT run it at 3200+...which leads me to believe they were all 3200+ chips, it's just that some of them came in 2500+ boxes...
Click to expand...

That was the rumor about the 2.4C, but no one really knows for sure outside of Intel. I don't recall that ever being substantiated. That said, I have actually seen a 2.4C that wouldn't do more than 2.7GHz or so. I've also seen a 1.8GHz Celery hit 3.2GHz. It was one of the gimped cache socket 478 chips so it was still slow as hell, but the clocks were insane. You might be right on the Bartons as well. I don't know for certain. Using the 2.4C and 2500+ as examples, I wouldn't be surprised if what you are saying has some truth to it. It only makes sense for Intel and AMD to fill the orders they are getting. If more people are buying mid-range CPUs than high end ones and their yields are good enough to make high end chips out of all of them, it's the sensible thing since their costs on manufacturring for a 2.4C or XP 2500+ are the same as making a 3.0C or XP 3200+.
 
You must log in or register to reply here.
Back
Top