Slashdot Mirror
It's All About the Pentium (4)
Submissions about the P4 flow in like the tides, so here's a batch
of them.
Rooster sent us
the Hot Hardware take.
TBM sent us Ace's extensive comparison of the P4 and K7.
Piete submitted a
fairly negative review of the chip (between the RDRAM thing, the motherboard thing, and the fact that the chip just isn't much faster for normal use, that's not surprising).
Slashdot Minion sent in
Hard OCP and
Sharky Extreme's respective reviews (including 200fps Quake).
Kudos to the AMD team.
--
A feeling of having made the same mistake before: Deja Foobar
Blue Man Group will have to get another member?
spawn_of_yog_sothoth
> It's going to be fast. Very fast. At the moment, it isn't significantly faster than the P3 or the Athlon, but remember this is the first release of an entirely new core.
So, over the long haul, how did their past core compete with AMD?
AMD is beating them on the ground, today. AMD will continue to evolve as well. This release gives us no reason at all to believe that Intel has made a technicological comeback.
In order to do that, they need to quit releasing overpriced, overclocked crap and actually, well, actually make a technological comeback. Then they can brag all they want about how the big company with a huge bank account and unbounded name recognition beat the little guy after a mere year's effort. Or two years' effort. Or three...
Sheesh, evil *and* a jerk. -- Jade
If anyone sees a site with a review and the only benchmark I really care about, please post it.
:)
I want to know how fast it compiles a kernel (with everything enabled/modulized.
-----
If Bill Gates had a nickel for every time Windows crashed...
The Pentium Pro was a dog of a chip for running the 16-bit code that was still prevalent back when it came out, and people jumped all over the chip and bashed the hell out of it. But a heck of a lot of servers were sold using 32-bit code and Pentium Pro processors, and we were very happy with the way they performed in a 32-bit world. I still have several dual Pentium Pro servers around, and they run very nicely still.
I think it's similar to the situation with the Pentium 3 and Pentium 4. The Pentium 3 is designed to take advantage of today's memory systems and bus technology, and the Pentium 4 is designed to work best with technology that really isn't in popular use yet (and may well never get there). So right now, pund for pund, the Pentium 4 looks like a bowser. Given code that's designed and optimized for the Pentium 4/Rambus combo, I'm sure it'll look much nicer than it looks running current apps. Nobody's bothered optimizing for that sort of environment yet.
What'll be interesting is what happens in the competition while Intel strives for Pentium 4 market acceptance. When the Pentium Pro came out, there was no competition in the high-end chip category, so Intel could afford to bide their time and wait for the marketplace to catch up. With the pressure AMD is applying in the high-end with Athlon, Intel can't afford to just sit and wait. They're going to have to be a lot more aggressive with Pentium 4 pricing, and push to get Rambus RDRAM pricing down in order to build any sort of demand.
Remember how 2000 was supposed to be the year of 64-bit computing? Looks like the priorities have shifted in the market.
- -Josh Turiel
-- Josh Turiel
"2. Do not eat iPod Shuffle."
Those who comparison shop may see the lack of benefits and not purchase it, but I'm sure plenty of others will.
Developers: We can use your help.
Yeah, the difference is about $1200....
Think outside the... Hey, where'd the friggin' box go?
No. Anything above movie-fps (25, if I'm not mistaken) is generally wasted, and anything above 60 is very wasted.
AMD is srill looking better. The initial benchmarks show only a slight performance increase by using a P4 over the AMD Thunderbird and now that AMD will be releasing SMP support for the Thunderbird and Intel will not support SMP for atleast a year and a half it just makes it easier to spend the $500 less on the AMD chip and spend the savings on extra RAM or larger hard drives.
"If ignorance is bliss, why aren't there more happy people in the world?"
I'm becoming a tad irritated with people who keep bringing up that moot point.
First off, in complex scenes filled with gibs, smoke, and the like, framerates drop drastically.
Second of all, when things such as FSAA are enabled, visual quality increases and framerates drop accordingly.
Third of all, 60 fps now will mean about 15fps in new games in 2 years. I remember these exact same comments when the Voodoo 2 debuted. Are people inherently this nearsighted?
Please, people, think ahead.
I sure would like to get a P4....so I can heat up my house this winter....
Burn Hollywood Burn
I'm surprised to not see Tom's review of the P4 listed in this roundup. See: Intel's New Pentium 4 Processor at Tom's Hardware Guide.
I get the feeling that it may be time to short Intel's stock. It looks like they're going to have another miserable failure on their hands. Why would anyone want to buy something that is more expensive for about the same performance, but with a guaranteed retirement of all the components in 6 months?
Software sucks. Open Source sucks less.
"Watch in awe as MP3s download more quickly and graphics flow more smoothly."
What a load of crap! it's no wonder Joe Consumer keeps buying Intel's overpriced junk. As long as Intel's marketing Juggernaut keeps tossing around flashy words like "NetBurst Architecture" and "HyperPipeline", Intel will continue to sell chips. I fully believe that they could package shit on a stick, give it a nice marketing spin, and show some weird abstract commercials during prime time, and people would continue to buy it.
-atrowe: Card-carrying Mensa member. I have no toleranse for stupidity.
Okay, this is in regards to that negative review... These guys... *sigh* i dont know who gave them the permission to write that column but... geeez... They comment on the pentium 4's lackluster performance in business software. Since when has it become industry standard to run desktops on more than a 400 mhz machine??? Did i miss something? They say its only really good for graphics intensive programs and 3d games etc. THOSE ARE THE TARGET AUDIENCE!!!!! If i wanted to open M$ word faster, i would go use win2k on a duron 700 mhz and pump a gb of ram into it. not buy a 1.5 ghz processor!! And lord only knows how excited companies are when they hear "Intel recalled their 1.13 ghz chips, but today they release 1.5 ghz!!". Yep. i am sure all sorts of companies are jumping for joy.
Actually, the threshhold for human flicker resolution is about 72 fps. See previous slashdot coverage or the original article.
WARNING: there is a trojan on your
The point isn't whether the Pentium IV is going to scale up to ungodly speeds.
The point is that the Pentium IV 1.5Ghz is hardly competetive with what's on the market. Let me rephrase: you'd be an idiot to actually purchase a PIV 1.5GHz.
Maybe the Piv will scale to faster clock speeds, but what we have right now sucks. Buying a Piv right now is like buying an early 80s BMW 320 because it's the predecessor to a really nice car. It's irrational. You're better off with the Athlon. Intel better have set aside a heafty chunk of change for advertising, because it's going to take a lot to get people to buy these.
This is the first *really* new core from Intel since the P2. The P2 first shipped at 120Mhz. This puppy is going to clock and clock. Expect to see 2Ghz by Q2 2001, 3Ghz before Q2 2002.
I agree that this release is strictly for the lunatic fringe, but this is the core that Intel are relying on to regain them bragging rights. Don't underestimate it.
Meanwhile, I'm still hoping for my 1024-way UltraSparc 3 box...
Am I the only person who finds all it annoying that most anti-Pentium retorts always include something positive about AMD? It's not like AMD is breaking amazing new ground. They're still keeping the x86 family alive, with all of its standard troubles (too many instructions, overly complex addressing modes, too much legacy baggage, too few registers, stack-oriented design of the floating point processor). The processors from both Intel and AMD are too expensive for what you get, and use too much power, especially when compared against other chip designs outside the x86 world. So both these companies are having a high-end pissing contest that only seems to be advancing the so-called state of the art in minor, expected ways, trading more power consumption and die space for speed increases of a few percent. Yee-haw!
I would love for another company to walk in and set things straight. Too bad Motorola seems to have trouble figuring out where to go with the PowerPC.
This is a bad kind of message to post, I think, considering the preponderance of crazed AMD supporters. But let's not let fanaticism for a corporation get in the way of real progress, okay?
The Pentium Pro was the last new core from Intel. And may I remind you - the first issue of the PPro - It beat the Alpha! For like a month, until DEC moved to a new process, Intel beat the king of the RISC chips. So I don't think there is the precedence you seem to see - when Intel brings out a new core, I expect bang - not "it'll scale".
.25um to a .18um process and kick some butt.)
ROFL!
No one noticed it, but you got bang today as well.
The original PPro was, for about a month, and only barely, the fastest chip in the world in SPECint95. The P4/1500 is...the fastest chip in the world in SPECint2000. Its SPECint2k scores are 522/535 base/peak; the fastest previously available processor in the world is an Alpha EV67/833 which scores 511/533. Considering Intel will almost certainly release a P4/1600 before Compaq finally releases a faster clocked Alpha, this gap will almost certainly become even larger for the P4. (And then Alpha will *finally* move from a
Even more spectacularly, the PPro shocked the MPU world by being somewhat competitive with the fastest RISC chips in SPECfp95--about 75% the top Alpha scores. Meanwhile, the P4/1500 put up SPECfp2000 scores of...549/558 base/peak, or roughly 90% those of the fastest Alpha.
And yet, just as when the PPro was launched, all we hear about is how the P4 is a failure because it performs poorly on legacy apps. The P6 launched to universal derision from the mainstream computer press because it wasn't any faster than an ordinary P5 at 16-bit apps (yeah, maybe eventually there might be *some* 32-bit apps, but who's going to rewrite their code just to optimize for some new-fangled processor?); it was perhaps the most successful MPU design in history, as predicted by its astonishingly good SPEC95 scores.
The P4 is launching to universal derision from the mainstream computer press because it isn't any faster than an ordinary PIII or Athlon at x87 apps and apps which use instructions which the P4 explicitly deemphasizes in favor of faster replacements (yeah maybe eventually there might be *some* SSE2 and P4 optimized apps, but who's going to recompile or even rewrite some of their code just to optimize for some new-fangled processor?); its SPEC2000 scores are just as astonishing as the PPro's were, if not more so.
We'll see how it plays out this time.
finally ventured away from the PentiumPro core! Yay Intel, you go. The concept of SSE2 is interesting, I'm still waiting for software using SSE other than a small handful of Photoshop filters and a 3D model of the Soalr System. I guess to be a wet blanket I'll complain about lack of innovative architecture. The P4 and Athlon haven't inspired me at all. Wow more powerful scalar operations yay Intel, you go. Why hasn't anyone building x86 chips stuck a vector unit in one yet? Is there some architectural design limit I'm missing? Whatever happened to the insta-recompile code morphing the P4 was supposed to have (I haven't read all the articles posted yet, only Sharkey's)? Since they lowered the IPC and lock you into a go-nowhere processor line with extraordinarily priced memory and hardware who exactly is the tarket sucke..market for the P4?
I'm a loner Dottie, a Rebel.
Back in the dark ages (say, 4 years ago) you needed a high-end PC if you just wanted to surf the Web and print a document at the same time. But these 1 gHz+ machines have are overkill for general-purpose users. Combine that in with mediocre reviews and recent evidence that PC market growth is finally leveling off, and it can only translate to sluggish sales.
The best part of all of this is that people who have been chugging along with older P2s will be able find moderately-clocked P3 chips cheap.
Is there anyone out there that is planning on getting a P4 for non-corporate use? Or even corporate use, for that matter?
-
Next time I buy hardware, I want to make sure that not a dime of my money goes to Rambus. That means I want no memory from any of the memory manufacturers who caved to their demands, no chipset from manufacturers who caved to their demands and of course no RAMBUS RAM. Is it even possible to set up such a system anymore?
I'm trying to teach myself to set people on fire with my mind... Is it hot in here?
You and me know that the PIV is slower than a PIII, but the average Joe Doe doesnt know this.
/me can't wait till we finally drop all the old legacy stuff on the x86.
Before you email me, remember: "There is no god!"
From the ZDNet UK article:
Intel's new Pentium 4 "Willamette" processor (Willy for short) will become public news: it's really not worth buying. At a clock speed of 1.5 GHz -- Guy Kewney says it's barely faster than a Pentium 3 at 1 GHz Intel, in short, has a little Willy. Go to AnchorDesk UK for the news comment.
--K
---
Looks like everyone seems to have come to the same conclusions:
1) A 1.5 GHz AMD product probably would do it better.
2) The Pentium 4 is too damn expensive
3) It isn't going to be in YOUR hands anytime soon
I find it funny that the one article was comparing an overclocked 750 MHz AMD (clocked up to 1.1 GHz) to a 1.5 GHz P4 and not seeing that big of difference.
Anywho, good to see AMD is going to stay competitive and push Intel to work a little harder...can't but help you and I.
I love this: Grapes on fire
Check this link. Gives a good simple explanation of what plasma is, as well as containing links to where you can find it, and what its current practical uses are.
Specifically, plasma is a "collection of free electrons and ionized atoms." They remain in a homogeneous mixture, so the electrical charges are balanced, but the particles have *way* too much energy for the electrons to "reattach" themselves to the atoms. Check it out, this stuff is pretty interesting.
It may look like I'm doing nothing, but I'm actively waiting for my problems to go away.
--Scott Adams
It's like I'm not even listening...
/. crew for some permanent space to soapbox?
/., I hadn't really thought about it, especially since I feel that there are many people out there who have a whole whole lot more knowledge on MPU design than me. Of course, as I seem to be one of the few who actually tries to enlighten the more software-minded crowd at /., I suppose it might be worth a thought...
Don't worry; the SPEC scores have been very poorly reported, while the P4's rather poor performance on non-optimized code has gotten all the press. You are by far not the only one to have missed the SPEC scores and assume that the P4 is a dud. Of course, in some ways this is valid, since SPEC scores are more indicative of the potential of the P4 core than of how well a P4 will perform on today's code. Still, as it turns out, the Alpha scores I was comparing the P4 against in my original post are for chips that won't be released until January; so technically, the P4 has not just the SPECint2000 crown (base and peak) but the SPECfp2000 crown as well (base only)!
Okay, I'm not heavy into hardware, I just wanted to point out the numerous problems with the P4 - it seems that the processor itself is not one of them!
I certainly wouldn't go out and buy a P4 today--a DDR Athlon is a much better deal for today's software. But the SPEC scores show that once we get some P4-optimized software, it's gonna kick butt. So, mediocre as a current product, great as a debut for a new core.
Tolu, you always have something insightful to say about chip design, but you have to repeat yourself fairly often across articles - have you thought about bugging the
Thanks!
Eh, I do get carried away too easily I suppose. It's always a problem when you feel like you have all this relevant information that many people reading may not know, and you don't know how much of it to repeat. (I generally tend to go for "all of it".) As for submitting an essay to
That is not the point.
Intel is not trying to beat the competition immediately, despite appearances to the contrary. They are, instead, looking on the Pentium IV as a long-term solution.
Take a look at the chip. The whole thing is designed to run at faster and faster clock speeds. Now, I am not taking a stand on whether AMD will be able to out-clock Intel (though personally, I hope so) but their CPUs do not sacrifice as much to clock speed at the moment. That is, AMD prefers to produce more complex, slightly less highly-clockable CPUs.
Of course, these chips could be clocked higher than Intel's Pentium III chips, and they were more stable as well. But now Intel has redesigned.
Really, the question comes down to how well AMD can scale to faster clock speeds. If AMD can hit even only 3.5 gigahertz by the time Intel hits 5 gigahertz, AMD will have won. But it is quite possible that AMD will not be able to do this, at least not without a redesign. Of course, if AMD can match Intel Ghz-for-Ghz, Intel is in serious trouble.
And that, my friends, is the point.
Oceania has always been at war with Eastasia.
And to reiterate - killing it more than direct comparison to the Athlon is that the associated parts are just out-and-out defunct.
The RDRAM system is hideously expensive; when implemented (correctly) by Intel it didn't meet Rambus's projected specs; anyone who gets in bed with Rambus gets a nasty case of lawsuits.
Entirely new boards, no compatibility, either backwards or forwards.
Chip itself is massively more expensive, perhaps to produce as well.
No SMP - goodbye server market. Remember how long it took to get 4-way Xeons? SMP will not be kludged in easily.
VIA, Intel, Rambus, and others are in a really screwed up relationship ATM.
Intel has some large problems here, more than can be overcome by one chip, even the most important chip, scaling to infinity. But hey, I hope they do - what would AMD do without competition?
Of course, in many cases the Pentium4 still doesn't look all that good compared to even the released Athlon and PentiumIII solutions, but there is another explanation. Remember that is is an entirely new architecture -- they always look bad when they first come out. Remember the PentiumPro? On one hand, it was faster than all the RISC CPUs of the time. On the other hand, it was getting beat by the vanilla Pentium in Windows 16-bit benchmarks.
The same thing applies, here. Take the RC5 benchmark where the Pentium4 is a lot slower than the PentiumIII. RC5 has been hand optimized for every single popular CPU architecture. Of course it looks bad on the Pentium4, because hand optimizations aren't available for it. Give it a few months for optimizations appear then see how well it runs. The same story goes for the SSE2 optimizations.
Also noticed what happened with RAMBUS. The i850 chipset isn't all that different from the i840 (the PentiumIII's dual-RAMBUS chipset), yet the i850 has dramatically higher memory scores. Intel decided to go with RAMBUS for the Pentium4, and designed the the new CPU around RAMBUS. In particular, the CPU is designed to have multiple outstanding requests to the memory subsystem, a feature supported by RAMBUS but not by SDRAM. This means that the PentiumIII can never take advantage of RAMBUS. It is even possible that these multiple outstanding transactions will allow the Pentium4 to show LOWER latency than a PentiumIII/SDRAM solution. (Oh, and the 400MHz bus improve absolute bandwidth as well :-).
In short: if I were buying a computer today, I'd go for an Athlon (or dual-PentiumIII). However, I bet 6 months from now, I'd probably be looking at the Pentium4.
(PS: ...and of course, I think Intel DID make tradeoffs for meaningless MHz increases for marketing reasons.)
Well, anyways, just thought I'd throw out my opinion, based on the real world use I would have put on the comp anyways. This just reinforces my thought that the P4 is gonna flop, and take a large chunk of Intel's marketshare with it. The only thing that's gonna save it is if Intel gets the speed up to 2gHz+ very soon, and even then I still have my doubts. My advice is that noone buy this version of the chip, since once the chip moves to the .13 micron process, the existing chips and MB's are gonna be completely worthless, since they are not gonna be forwards compatible.
Well, anyways, feel free to disagree with me. I don't claim to be some computer genious, I'm just someone with enough knowledge to use a computer and do some basic testing.
-C
"Well kids, you tried your best, and you failed. The lesson is, never try." -Homer Simpson
Make sure to check out recent articles here and here. There's also a very imformative one right here as well. I think it will be interesting to see how the race pans out against AMD. There have been quite a few differing reports on the performance benchmarks versus the Athlon thus far. Stay tuned!!
Interested in open source engine management for your Subaru?
That said, the machines which will be sold by Dell, et al. will already be evolutionary dead ends.
--
A feeling of having made the same mistake before: Deja Foobar
I think its a quote from the "It's All About The Pentiums" song from Weird Al. Watch the video here.
Let's see what Dr. Tom, a noted Rambus hater, has to say...
................What do I think of the components around Pentium 4? I have got to admit it, but with Pentium 4 Rambus is finally able to deliver for the first time. If you look at Pentium 4's design closely enough, you can see that it's engineered to live with RDRAM in perfect harmony. The memory benchmarks from above show that Pentium 4 really requires the 3,200 MB/s of data bandwidth supplied by the two Rambus channels. I doubt that it will perform as well with DDR-SDRAM, unless two channels will be used. One DDR-SDRAM channel offers 'only' 2,122 MB/s of data bandwidth, which might make quite a difference with Pentium 4."
"You can see that the memory speed does indeed have a major impact on all the benchmark results except of the 3D Studio Max scores. In some cases the difference between the slowest and the fastest score is more than 10%! This proves clearly that Pentium 4 lives from the high memory bandwidth that RDRAM is finally able to deliver. Keep that in mind in case someone wants to sell you PC600 RDRAM!
The technology works as advertised, maybe we can stop all the mindless bashing now.
THE YEAR WAS 2081, and everybody was finally equal...
Also note, performance-wise to compare to the Athlon - in 1Q2001 AMD releases the Palomino version of the Athlon, which should perform even better than the current Thunderbird Athlon, plus be at higher clock rates. At the same time, SMP systems will start coming online for the Athlon, which you can use Durons, Thunderbird Athlons, or Palomino Athlons with, so you can grow your system slowly if money is tight. Take two 900mHz Durons and start SMP slowly. Most people probably won't need to go to the Athlon at all! Sweet.
Also note: DDR will get better - current comparisons are being done with systems using, I think CAS 2.5-3-3 memory. Faster DDR is coming soon (though, of course, that'll obviously cost more). I don't mind paying for performance, but I _do_ mind paying for uneven performance (better in some ways than old, worse in others, like Rambus DRDRAM and the Pentium 4).
This all adds up to some pain for Intel in 2001.
As the P4 and the K7 recognized each other, each would vie for more and more system resources. This would cause a radical increase in power consumption, which is of course eventually released as heat. The intense heat from the resource battle would strip all electrons from your video card, essentially reducing it to a glob of molten plasma. Due to the strange interactions between the Intel and AMD products, both would begin feeding off of this plasma as a power source, and it would grow, eventually consuming everything it comes in contact with, including, but not limited to, your house, your pet cat, and your neighbour's swimming pool. Essentially, you will be accelerating the eventual heat death of the universe by approximately 82%.
It may look like I'm doing nothing, but I'm actively waiting for my problems to go away.
--Scott Adams