Worst CPUs ever, now with poll!

[inf64]

The numbers are correct. Since FX8170 won't launch we have to consider PD based FX.
If you want to know how fast Vishera will score in both of those charts,it's rather simple. Take 10% IPC improvement (even tho L3-less Trinity is 10-15% faster than FX8150 at same clock), assume 4Ghz/4.6GhzTurbo clock like this source claims and you end up with: 4Ghz/3.6Ghz x 1.1=1.22 or 22% faster in MT workload and 4.6/4.2x1.1= 1.2 or 20% faster in ST workload.

To get time based value in above charts multiply FX8150 results in seconds with ~0.80 and ~0.78 factors:
Single-thread runtime
FX8350 @ 4/4.6Ghz - 673x0.8=~538s or 8:58s
Multi-thread runtime
FX8350 @ 4/4.6Ghz - 991x0.78=~773s or 12:53s

i7 3960x gets 481s and 724s respectively. Therefore it should be ~11% faster in ST and 6.4% faster in MT workload than FX8350@ 4/4.6Ghz.

 

[TekDemon]

Prescott indeed extended pipelines further but that's also part of the story. ALL Pentium 4 still employed a write-through cache policy which is known to be a poor algorithm. As I recall my Athlon XP @ 1.8 Ghz was equivalent or sometimes faster than a Pentium 4 NORTHWOOD @ 2.4 Ghz. So from pure efficiency point of view - even at the time of north wood we had better alternatives. And don't forget the 8 and 16 K L1 cache on Northwoods vs 64 K L1 on Athlons (that's 4 and 8 times as much respectively!!!). And for the early Pentium 4, don't forget they performed worse than their Pentium III counterparts which is a shame considering they were clocked much higher than Pentium III (and in contrast Bulldozer didn't boost frequency much compared to Phenom IIs).

I owned both an Athlon XP 1600+ (the legendary overclocker, I can't quite remember what I had it clocked at but it was heavily overclocked on a pricey Alpha heatsink) and a Northwood system (clocked at 2.4Ghz actually) and the Northwood was the faster system by far. By modern standards they're both uselessly slow of course but...memories.

Actually still have both chips in a closet somewhere lol.

 

[stevech]

need to add VIA to the CPU list

 

[kool kitty89]

I don't understand Bulldozer. A refined K10 design could've potentially been great. A Phenom III X8? Phenom II with SSE 4.1, 4.2, AVX, and all of the minor core tweaks from Llano could've potentially been quite competitive. If the 32nm node shrink could increase clocks nearly as much as the 45nm node shrink to Phenom II did, a 32nm K10 could realistically hit 4.5 GHz stock.

Then again, maybe there was some bizarre issue AMD saw coming when Thuban was developed. I thought Thuban was a step in the right direction. It could've likely benefitted from some extra L3 cache, but otherwise it was great.

Agreed. They had a competitive platform and Llano proved it, with better IPC. Not sure, what exactly happened.

This is a very good point . . . why not continue with 32 nm K10 chips in parallel with Bulldozer?

Has AMD made any comments on this in press releases?

I owned both an Athlon XP 1600+ (the legendary overclocker, I can't quite remember what I had it clocked at but it was heavily overclocked on a pricey Alpha heatsink) and a Northwood system (clocked at 2.4Ghz actually) and the Northwood was the faster system by far. By modern standards they're both uselessly slow of course but...memories.

Actually still have both chips in a closet somewhere lol.

Faster for what applications? And what RAM/FSB set-ups were used?

IIRC, RDRAM used about half the traces per channel (data), and it was equivalent per channel. Power, yeilds, and latency got better over time. 1066MHz 32ns RDRAM was no slouch...but who would buy it?

Perhaps on the motherboard itself, but looking at the modules:
RIMM single channel (16-bit), 184 pins
RIMM dual channel (32-bit), 232 pins

DIMM SDR 64-bit, 168 pins
DIMM DDR1 64-bit, 184 pins
DIMM DDR2 and DDR3 64-bit, 240 pins

worst cpu, relative to it's time, would be bulldozer...period. Pentium 4 was trumped by the original athlon 64s back in the days, but I doubt the gap between intel and amd has ever been as great as it is today. But I hope AMD will get their @$$ in gear and get things right with pile driver. :\

I'm not sure how you could be thinking this since AMD was an underdog in the x86 market until the late 90s, considerably more so than today. (albeit, their strongest point in the pre-Athlon k6 vs Celeron/PII days would probably be relatively close)

 

[Homeles]

need to add VIA to the CPU list

What have they made that sucked?

 

[kool kitty89]

What have they made that sucked?

VIA (or rather the Centaur/Winchip team they bought) made the "Cyrix III" or C3, and later C7 among other embedded chips.

Whether they sucked is a matter of perspective, but in terms of mainstream performance (and especially anything beyond light-duty office/browser use), the Cyrix-III/C3 and C7 sucked, though the (socket 370 based) Cyrix/C3 parts are the only ones that could really be considered desktop parts.
Clock for clock they were very poor performers and much poorer still for games/multimedia. (well beyond the gaming performance issues of the old Winchip and Cyrix chips -in their respective eras)
And on a note of cost effectiveness, the original 180 nm CIII/C3 parts actually took more silicon than AMD's 180 nm K6-2+ (and possibly the 3+), which were vastly better performers across the board.

There were also compatibility issues with desktop S370 motherboards, including those using VIA's own chipsets.

The C7 and later parts are somewhat less legitimate to label as poor since they targeted more specialized niche embedded/industrial/mobile markets.

 

[Magic Carpet]

This is a very good point . . . why not continue with 32 nm K10 chips in parallel with Bulldozer?

Has AMD made any comments on this in press releases?

Nothing, that I am aware of. I am going to speculate and say, they did not have enough resources for that to happen. And perhaps, that decision had also been seriously influenced by the marketing people. After all, 8 cores > 6 cores. Actual engineers themselves had little to do with it.

In other news:
I've got a 65nm D0 Presler on the way, for a test. Supposedly, one of the "better" late Pentium 4 chips.

 

[kool kitty89]

Nothing, that I am aware of. I am going to speculate and say, they did not have enough resources for that to happen. And perhaps, that decision had also been seriously influenced by the marketing people. After all, 8 cores > 6 cores. Actual engineers themselves had little to do with it.

That's assuming they didn't introduce 8-core 32 nm K10 chips . . . which might still be as "slim" as bulldozer depending on the cache configuration.

Marketing influence does really screw things up sometimes . . . granted, weak marketing (or weak market reputation) can also ruin a technically good product too.

 

[essjayar]

My first...

A Cyrix 586 133

 

[kool kitty89]

My first...

A Cyrix 586 133

What's wrong with the 5x86 133? I can see an argument for Cyrix being mistaken to discontinue it (and the 5x86 in general) too soon when socket 3 boards were still common in the low-end, part of mid range, and upgrade market (and focusing on the 6x86 instead), but not much argument for it being "bad" in and of itself.

Plus, unlike the 6x86, its P-rating ended up being rather accurate for integer and floating point ratings since it has the same FPU as the 6x86, but was clocked much faster relative to the P-rating (due to the much weaker ALU and slower/narrow FSB).

I'd go as far as say it may have been better if Cyrix had pushed the 5x86 further and introduced a socket 5 version as well. Assuming the yields/clock speeds remained proportionally high to the 6x86 (as was the case at the beginning of 1996), it would have been a good low-cost pentium contender with much more balanced floating point performance. And (also unlike the 6x86), it would have been genuinely cheaper to manufacture than the Pentium, though profit margins would still almost certainly be lower. (more like K6 vs Pentium II/celeron)

 

[zaydq]

my pentium 3 demolished my pentium 4... so pentium 4 is the worst. My P3 also used Rambus Ram.... expensive friggin ram but it worked great!

 

[Gerilgfx]

ARM926EJ-S

 

[WT]

My nomination for worst chip = 386SX any

:thumbsup: I'm going with this as well. I owned one, but it never impressed me. The PC itself, yea that impressed me. But not the chip. Not a bit.

 

[VirtualLarry]

At least Pentium 4 wasn't a step back from Pentium III in practice when it was released.

Yes, it was. Willamette was slower than PIII, clock for clock. It wasn't until the Northwood revisions, did the P4 pull ahead decidedly.

 

[Abwx]

Worst was the infamous 1.3GHz P4...

 

[Blue_Max]

Gee, and no one wants to mention the cacheless celerons? shame on you nerds

though in the areas of "oops" which I think would make an honourable mention would be the Pentium D and it's hype of being a dual core when it was just two high powered cpus in the same package (sharing the FSB just like normal dual cpu units).

or the pentium bug that intel down played as being "pointless" to the masses.

Agreed... the Celeron 266 was awful... SO glad the guy allowed me to upgrade to the freakin' awesome 300A! Hard to imagine how far that 128k of cache went!

Why the 286 though?? I remember the WORLD of difference between my Tandy 1000SX (7.14MHz 8088) vs my friend's 1000TX (8MHz 80286 w/ 8-bit bus.) The difference was monumental!

 

[erunion]

though in the areas of "oops" which I think would make an honourable mention would be the Pentium D .

As a former owner of one, I was going to say the same thing.

 

[SOFTengCOMPelec]

The worst processor ever, is the 2012NECRO cpu, released about 3 years ago. But no one seems to have noticed.

 

[Stuka87]

You guys realize this is a rather OLD post right?

 

[SOFTengCOMPelec]

You guys realize this is a rather OLD post right?

No.

The worst processor ever, is the 2012NECRO cpu, released about 3 years ago. But no one seems to have noticed.

(Obviously I was joking, yes I noticed).

WARNING: 2.5+ Year old NECRO thread!

 

[Blue_Max]

ARM926EJ-S

3 posts... a rather odd offense... oh well, decent discussion all the same.

 

[jhu]

No.



(Obviously I was joking, yes I noticed).

WARNING: 2.5+ Year old NECRO thread!

I'm honored someone (not me) necroed this thread. 286 is on the list because it had a borked protected mode.

 

[Gerilgfx]

of course i noticed how old this thread is.

actually i arrived back here with google, i even had to reset my password to log-in, and post ARM926EJ-S, which is clearly the worst cpu in its context ever.

p4, bulldozer, and even iapx are just 1.5-3x slower than the competitions.

but this is 10-50x slower than other competiting CPU's with the same power consumltion/date/nanometer.

for example, it requires 2-4 minute to render this forum.

 

[master_shake_]

i guessed right!

where is my ipod!

i had a prescott p4 that was balls.

 

[ninaholic37]

I was going to ask why Broadwell wasn't on here, then realized it was from 2012.

Nice avatar Gerilgfx.