Was the Pentium 4 the worst processor design in history?

[dennilfloss]

It was better than the crippled cacheless original Celeron 300 (Not A).

 

[DrMrLordX]

Guys, are we forgetting some of Cyrix's amazing competitors for the original Pentium?

Did any of you remember the 5x86 processors that would sometimes burn out at stock settings with factory cooling solutions? I do. Cyrix produced some truly awful chips.

Then there was IDT and their amazing Winchips.

 

[Gikaseixas]

Worst cpu line ever gotta be the Celeron's. I don't how good they're now, i haven't messed with one lately but in Intel's defense i loved my Pentium III 450 mhz with 128 mb of ram, 10 gig hard drive...

 

[eelw]

Only the original cacheless Celerons are bad. Current Celeron Ds are great value processors and fit the bill for the large majority of computer users.

 

[LouPoir]

P4 Northwood rocks

 

[IEC]

Originally posted by: eelw
Only the original cacheless Celerons are bad. Current Celeron Ds are great value processors and fit the bill for the large majority of computer users.

I got a 3.2GHz Celeron D recently just to overclock... *my* first computer was a Celeron 300A OC'd to 450MHz. I'm waiting on a s775 HSF so I can see how far it'll go...

 

[InlineFive]

It wasn't terrible, only the Prescott sucked.

 

[myocardia]

The original P4's were one of the worst cpu's ever released. By the time they got to the P4C's, that no longer applied. Then they released the first horrible Preshott's, which had a higher clockspeed, but were about as bad as the first P4's, clock for clock. But, by the time the C2D was released, the last P4's weren't so bad again.

I vote that netburst, overall, was neither the best nor the worst cpu ever; it was, however, the most up-then-down architecture ever. Every time it seemed like they had it right, they'd release an "improvement", which was always a letdown, until it had a year or more to mature.

 

[Zap]

Originally posted by: DrMrLordX
Guys, are we forgetting some of Cyrix's amazing competitors for the original Pentium?

Did any of you remember the 5x86 processors that would sometimes burn out at stock settings with factory cooling solutions? I do. Cyrix produced some truly awful chips.

Then there was IDT and their amazing Winchips.

Going back even farther, how about the 486SLC chips? These were basically (IIRC) Cyrix/IBM clones of 386SX chips with 16 bit interface, with a 486 compatible instruction set, were pipelined and had a tiny cache (1k or 2k?). So, what makes a chip a 486? Is it a minimum performance level? If so, then these 486SLC chips were not "486" level. Is it instruction compatible? If so, then these 486SL chips really were "486" chips.

pics of 486SLC

So basically these were really fast... for a 386SX.

 

[myocardia]

Originally posted by: Zap
Going back even farther, how about the 486SLC chips? These were basically (IIRC) Cyrix/IBM clones of 386SX chips with 16 bit interface, with a 486 compatible instruction set, were pipelined and had a tiny cache (1k or 2k?). So, what makes a chip a 486? Is it a minimum performance level? If so, then these 486SLC chips were not "486" level. Is it instruction compatible? If so, then these 486SL chips really were "486" chips.

pics of 486SLC

So basically these were really fast... for a 386SX.

Yeah, but don't we all agree that Cyrix was absolutely the worst cpu manufacturer ever? They actually made "laughing stock" not sound like such a bad thing, IMO.:laugh:

 

[ingenue007]

worse was pentium with math coprocessor error back in 93?

 

[secretanchitman]

the pentium 4 2.4C - 3.4C (northwood socket 478) were the best pentium 4 chips IMO. hyperthreading, moderately cool, and high overclocks.

p4 2.4C overclocking FTW!!!

 

[myocardia]

Originally posted by: secretanchitman
the pentium 4 2.4C - 3.4C (northwood socket 478) were the best pentium 4 chips IMO. hyperthreading, moderately cool, and high overclocks.

p4 2.4C overclocking FTW!!!

I highly doubt that you'll get any argument whatsoever about that. I think that the 2.4C @ 3.5 Ghz was the first OMG overclock, assuming we discount Celeron the 300A@450/366@550.

 

[harpoon84]

Originally posted by: myocardia

Originally posted by: secretanchitman
the pentium 4 2.4C - 3.4C (northwood socket 478) were the best pentium 4 chips IMO. hyperthreading, moderately cool, and high overclocks.

p4 2.4C overclocking FTW!!!

I highly doubt that you'll get any argument whatsoever about that. I think that the 2.4C @ 3.5 Ghz was the first OMG overclock, assuming we discount Celeron the 300A@450/366@550.

It can be said that the entire Northwood line was pretty decent as far as overclocking goes. Remember the 1.6A almost overclocked to 2.4GHz every time.

 

[Avalon]

Originally posted by: myocardia

Originally posted by: Zap
Going back even farther, how about the 486SLC chips? These were basically (IIRC) Cyrix/IBM clones of 386SX chips with 16 bit interface, with a 486 compatible instruction set, were pipelined and had a tiny cache (1k or 2k?). So, what makes a chip a 486? Is it a minimum performance level? If so, then these 486SLC chips were not "486" level. Is it instruction compatible? If so, then these 486SL chips really were "486" chips.

pics of 486SLC

So basically these were really fast... for a 386SX.

Yeah, but don't we all agree that Cyrix was absolutely the worst cpu manufacturer ever? They actually made "laughing stock" not sound like such a bad thing, IMO.:laugh:

Cyrix's 6x86 was a good chip.

 

[myocardia]

Originally posted by: Avalon
Yeah, but don't we all agree that Cyrix was absolutely the worst cpu manufacturer ever? They actually made "laughing stock" not sound like such a bad thing, IMO.:laugh:

Cyrix's 6x86 was a good chip.[/quote]
True, but making one good chip hardly qualifies you as a good cpu manufacturer, at least IMO.

 

[Viditor]

Originally posted by: myocardia

Originally posted by: Avalon
Yeah, but don't we all agree that Cyrix was absolutely the worst cpu manufacturer ever? They actually made "laughing stock" not sound like such a bad thing, IMO.:laugh:

Cyrix's 6x86 was a good chip.

True, but making one good chip hardly qualifies you as a good cpu manufacturer, at least IMO.
[/quote]

The one thing Cyrix did do (an I agree that it was a very good chip for it's day) was to embarass Intel a bit...
Andy Grove had given a speech wherein he stated that Intel had determined that a FSB faster than 66MHz was just not physically possible...2 months later, Cyrix released a 6x86 at 75MHz FSB.
That said, it was the first and last time I ever saw Andy Grove with egg on his face...

 

[dguy6789]

Originally posted by: myocardia
The original P4's were one of the worst cpu's ever released. By the time they got to the P4C's, that no longer applied. Then they released the first horrible Preshott's, which had a higher clockspeed, but were about as bad as the first P4's, clock for clock. But, by the time the C2D was released, the last P4's weren't so bad again.

I am going to have to say that this is completely untrue. To compare Prescotts(no matter how much people hate their heat) to the performance of the Willamette is absolutely unjustified. The Prescott, due to it's longer pipeline, did perform worse than the Northwood in some applications. Many people have only read reviews of the Prescotts for very brief instances. The fact is, if you look at many reviews and look at all of the benchmarks, the Prescott tends to win more than the Northwood does overall, although not by any significant degree in any scenario.

The Prescott improved upon the Northwood in many areas(Better branch predictor and more advanced implementation of hyper threading just to name a few). They increased the pipeline stage count from 20 to 31, and still managed to maintain similar performance. There were so many improvements in the design that many people don't even know exist. The only problem with Prescott is that it had an unexpectedly high level of heat leakage. From a price vs. performance standpoint however, it always did remain in the shadow of the Athlon 64. Intel did have answers for the Athlon 64s, but just not at the proper price. The 3.8Ghz Pentium 4 outperforms the Athlon 64 4000+ by a slight margin in most scenarios, but the price premium was astronomical, far too great to make it worth someone's money.

If anyone is interested in reading up on the changes that took place going from Northwood to Prescott, here is a very good read:

http://www.anandtech.com/showdoc.aspx?i=1956

 

[serox]

The Pentium 4 was a good arch when it first hit the scene. The only problem with it was that Intel tried to hold onto it for so long that they essentially ran it into the ground.

 

[FelixDeCat]

This is a loaded question based on a stupid premise. Kind of like asking if Bill Clinton was the worst president in history or if Pepsi Cola is the worst soft drink in history.

How about this one: "Does Random Variable ask stupid loaded questions?"

 

[Arkaign]

The most disappointed I've ever been with a Cpu was with an AMD K5-PR90. I couldn't overclock that thing 10Mhz with a huge heatsink and a really loud 5000rpm fan! Performance was pretty sad too. Sticking with AMD, I 'upgraded' to a 100Mhz K5-PR133, which also wouldn't overclock even to 120Mhz (60x2), but was marginally faster. It wasn't until K6 that AMD wasn't embarrassing to have in my PC

P4 has had bad times (Willamette and Prescott), and good times (Northwood, and .. well .. Northwood!)

Disgusting seeing P4 1.5Ghz getting smashed to pieces by 1.3Ghz P3/Athlon.
Fun seeing 1.6A@2.13Ghz

Also had a 2.4B @ 3.3ghz for a bit

 

[dev0lution]

Nope. One word. Itanium.

haha..

 

[myocardia]

Originally posted by: dguy6789

Originally posted by: myocardia
The original P4's were one of the worst cpu's ever released. By the time they got to the P4C's, that no longer applied. Then they released the first horrible Preshott's, which had a higher clockspeed, but were about as bad as the first P4's, clock for clock. But, by the time the C2D was released, the last P4's weren't so bad again.

I am going to have to say that this is completely untrue. To compare Prescotts(no matter how much people hate their heat) to the performance of the Willamette is absolutely unjustified. The Prescott, due to it's longer pipeline, did perform worse than the Northwood in some applications. Many people have only read reviews of the Prescotts for very brief instances. The fact is, if you look at many reviews and look at all of the benchmarks, the Prescott tends to win more than the Northwood does overall, although not by any significant degree in any scenario.

The Prescott improved upon the Northwood in many areas(Better branch predictor and more advanced implementation of hyper threading just to name a few). They increased the pipeline stage count from 20 to 31, and still managed to maintain similar performance. There were so many improvements in the design that many people don't even know exist. The only problem with Prescott is that it had an unexpectedly high level of heat leakage. From a price vs. performance standpoint however, it always did remain in the shadow of the Athlon 64. Intel did have answers for the Athlon 64s, but just not at the proper price. The 3.8Ghz Pentium 4 outperforms the Athlon 64 4000+ by a slight margin in most scenarios, but the price premium was astronomical, far too great to make it worth someone's money.

If anyone is interested in reading up on the changes that took place going from Northwood to Prescott, here is a very good read:

http://www.anandtech.com/showdoc.aspx?i=1956

Where do you come up with this crap? Any speed Prescott vs. the same speed Northwood, and the Northwood is faster at everything, except video encoding. And the only reason the Preshott was faster at video encoding was a)larger L2 cache and b)SSE optimizations. BTW, Intel never made any 3.8 Ghz Prescott's. As a matter of fact, Intel never made any 3.8 Ghz P4's of any core.

 

[dancingwllamas]

Actually, they did release a 3.8 Ghz chip.

http://www.intel.com/cd/channel/reselle...processors/pentium-4/feature/index.htm

 

[soydios]

Clarification: Prescott (the second-generation Pentium 4 core) was terrible. The Northwood cores, especially the 2.4GHz ones, were easily overclocked. Prescott was supposed to take the P4 up to the 10GHz that had been previously promised by Intel, but they hit a power wall just short of 4GHz.