Slashdot Mirror
Intel Pushes Pentium 4 Past 3 GHz
denisbergeron writes "Yahoo has the news about the new P4 who will run at nothing less than 3.06 GHz. But the great avance will be the hyperthreading technology (already present in Xeon) that allows multiple software threads to run more efficiently on a single processor."
It's interesting to see what the cutting edge is capable of, but you pay such a stupidly massive premium for the latest processor that only fools would use their own money to buy it.
In the UK you usually have the ultimate latest Intel at about 700 UKP- the sweet spot in the price/performance trade-off tends to be around the 200 UKP mark, which will probably be the 2.5Ghz by the time this 3Ghz one is out.
graspee
And yet while running enterprise class systems I can't find a system with too little power.
:)
Well I have a citrix farm full of quad Xeons and 4 gigabytes of RAM, and we'd still love some more power, thanks.
Maybe you don't want 3.06 GHz for what you're working on, but our "Enterprise Class Systems" (Win2k application servers) can use all the CPU we can throw at them. Everyone has different needs, and for a lot of folks, faster processors are a good thing.
(I've seen this troll a few times over the last four or five AMD/Intel product announcements. And it's still getting modded up.)
Disregarding all of the comments on the 3.06 typo. Geez, I remember the day when we use to comment on processors, peripherals, parts. Now the community is stuck on whining about typos. Read it, chuckle to self,move on.
Anyway they have ramped up the speed, and added something that could have always been, hyperthreading. Xeon has always had it. This is not progress, this is almost not worthy reporting.
Puto
The Revolution Will Not Be Televised
> 3.06 MHz is over 3 times faster than a C64... ...but the C64 still got better sound.
>
I agree: despite al the claims about Moore's Law and technological advances, this proves that tripling the speed of a good ol' 6510 CPU has some disadvantages as well: Give a little to gain a little. 8-)
More seriously: graphics have improved since the early eighties, but what about gameplay? Isn't Mame the only thing that really justifies buy PC hardware now and then?
--
Money is the root of all evil (Send $30 for more info)
Correct - Word, Excel, and general Windows GUI operations - or, if you prefer, OpenOffice, KDE etc - won't show any improvement, and indeed the benefit for these apps (in terms of what the end-user perceives) began to diminish once processors passed the 6-700MHz mark.
.PST archive file whilst WinZIP-ing a large folder. All quite credible usage scenarios (who here can say, "that's preposterous! no-one would EVER want to do such a thing!"..?) and the difference between an HT- and non-HT-enabled system is dramatic - of the order of 20-30% time saved.
Where this kind of power shows, though, is in more intensive processing - as the article suggests, Photoshop, Cinema4D, and so on.
As for Hyperthreading / SMT technology - it absolutely does make a difference. I've been running an HT-enabled system (pre-release silicon) for some time, and there are specific usage models where it shines. Forget about single-threaded or single-tasking type environments - the user who loads Word, does some typing, then loads Photoshop, does some filtering, etc, isn't going to see any benefit at all from HTT. However, once you get into multi-tasking scenarios the story is very different. For example, run a series of Photoshop filters/macros whilst simultaneously virus-scanning the system; or, export a large Outlook folder to a
the huge number of story errors that keep popping up. You'd think that the story editors would try to mantain some kind of quality control.
However, it's also possibly a ploy to keep people posting indigant comments about errors. 50% of posts on these kinds of stories seem to be pointing out these glaring errors. Like the recent story about PS2 games on an Xbox which was nothing to do with the Xbox at all.
Come on guys, wise up!
Just dealing with the Microsoft OSen:
This is not a problem for anyone running XP - both Home and Pro variants will only count PHYSICAL processors for licensing purposes. This means, interestingly, that with HT, you can have "dual processors" running on XP Home. And, of course, 4 processors (2 physical, but 4 "logical") on XP Pro.
Unfortunately, the Windows 2000 variants don't make that distinction. So it's easy to exceed the licensing CPU limit if running 2000.
However, all HT platforms - if properly designed, according to Intel specifications - have a BIOS switch or similar mechanism to disable the HT technology, in which case the processor reports as just a normal uniprocessor core. Users of "older" operating systems are encouraged to use this feature.
Hyperthreading is a complex proof of the limitations of todays CPU architectures. I belive in a CPU architecture containing many small CPU cores on one chip, instead of just multiplying the issue and commit parts and sharing the execution units.
It would be more scaleable and easier to implement to use several complete CPUs. The biggest drawback (compared to hyperthreading) would of course be that in special situations some CPU cores would be idle, but this simply corresponds to pipe-line bubbles in the hyperthreaded case. This is easily compensated by two facts: 1) multiple CPUs can be made very scalable and 2) most computer systems today always runs multiple threads (i.e. utilization will be good).
Of course, for Intel to maintain their market lead, everything has to be compatible, so they'll have to pay, time after time, for the errors they made in the eighties (the 286 paging + the CISC ISA). By breaking Amdahl's law time after time (SSE, MMX, etc.) they have made an even more complex beast. The only area where they really excel is in the production processing. They can squeeze out high frequencies and pack the transistors tight. For that, I'll give 'em cred. For their CPU ISAs, I'll just laugh...
Great, so now we'll see nerds nitrogen-cooling these things to get an extra performance boost as well? What a waste of time.
This is all pointless. The entire pentium "architecture" (more like a shanty-town) needs to be dumped entirely. We NEED a clean start.
Even moreso, why is no one addressing the fundamental problem--that the PC is just horribly designed? There are better ways of doing things than just ramming everything through a single CPU. This is 2002--why are we not pursuing better computer design? The "PC" is the bottleneck for crying out loud. 10 years from now will we be reading about the new 10 Ghz PVII chip, still running in 30-year-old hardware? Wonder if I can still get a "Missing Basic ROM" error on my desktop machine...
Be, Inc. tried to redesign the "PC"...they had a very nice design, but they killed it before it's time. And how about Amiga...yeah everyone is sick of hearing about the Amiga but it WAS intelligently designed. Instead of shoving everything through the CPU the Amiga used coprocessors to deal with much of the stuff that bottlenecks PCs, leaving the CPU free for more important stuff. It was a great idea, and it actually WORKED.
I don't care who does it--I want to see a better machine being built. If done right, the Ghz of the CPU won't matter nearly as much.
Moderator tip: When the user clicks "Read more," and the article says "0 comments," he takes a reasonable amount of risk in making obvious humor posts like this.
:)
Just because 4 other people clicked SUBMIT with the same content before above user did is really no reason to penalize him.
Besides, he's a really nice guy in person...
"You won't see a heck of a difference in Word..."
:)
So why are you asking "How can Word appear any faster at 3GHz?" It won't, but Adobe Photoshop blah blah will use as much as you can give.....
Stu.
Don't mind me I drank too much at tonight uni function. tee hihee.
As long as there are enough processes both solutions will work well. If you run only one thread the multiple CPUs solution will have idle CPUs, while the hyperthreading solution will have bubbles.
Still, the multiple CPU solution will be vastly more scaleable and far less complex.
By changing the ISA of the CPUs, one can avoid lots of the bubbles (all if one is mean to the compiler). Just introduce branch delay slots and you lose a whole lot of bubbles and complexity. Just imagine how simple a CPU without branch prediction would be...
Maybe you don't want 3.06 GHz for what you're working on, but our "Enterprise Class Systems" (Win2k application servers) can use all the CPU we can throw at them. Everyone has different needs, and for a lot of folks, faster processors are a good thing.
Are they actually CPU bound, or are they slowed by memory access and bus bandwidth? Apart from certain numerical computations, I have rarely seen cases in which the CPU is really fully occupied, altho' the tools often report that it is. For example, tools will report if the CPU is idle waiting for a page fault to the swapfile, but not if it's waiting for data to get to or from main memory, it just looks like the CPU is occupied.
Knowing what I know of Citrix, it alone is far bigger than the L2, and that's before even considering the user applications. It requires the CPU to switch context heavily, and constantly flush and reload its L1/2/3 caches. After all, if you need 4G of RAM to run the applications you are using, and you have say an 8M cache, the CPU is going to be spending a lot of time managing its cache rather than doing useful work. Given that, it is bound by memory access, not raw CPU.
Manufacturers, driving by consumer marketing which believes that higher Mhz == better product, are optimizing in the wrong areas. If they want to talk numbers, they should be pushing fast memory and buses which are actually a useful measure of a machine's performance, not CPU Mhz which isn't.
The fact is that for work a 700MHz PIII is usually fast enough given the rest of the system, as well as being reasonably cool and quiet.
So what is the point of this advert? Is it the result of a kind of desperation on the part of Intel? Marketing departments insisting on announcing ever smaller "feature creeps" in an effort to create a buying climate run the risk of the very buyer turnoff they want to avoid. It's like the old Indian auto industry, where the big new feature for each year was something like a differently shaped tail-light molding.
Panurge has posted for the last time. Thanks for the positive moderations.
Why not spend more R&D money in increasing the speed of the bus? It would give us way better performance.
Users do stupid things.
I've seen vector graphics with millions of lines inserted into word. Fine for a drawing package or desktop publishing app, but god awful slow in Word. Not really a fault of MS, it's these people should be using a desktop publishing application, Word is for wordprocessing.
The true fact in the matter is that intel are going to rely almost entirely on the marketability of a big number with the P4, as it's handling is rather unimpressive when compared to such ordinary designs as those from AMD, which clock poorly, yet crunch happily.
I need not mention about G4's and other well designed chips as some GHz bunny is certain to point out that they are only at 1.25GHz at the moment.
I know that there are some of you on here that will flame me saying that you DO use that power. And that's fine, you are the 1% of the population I mentioned earlier. But to do it (like most of you would... admit it) just to get another 4fps in UT2003 or whatever, it's just sick. Yes, eventually I will buy a new computer, but only when my needs exceed the resources in my computer, which hasn't happend just yet (it's getting close though...). If any of you can actually tell the difference between this 3.06GHz P4 and the 2.5GHz P4 (without using a stopwatch that measures in the milliseconds) I have a bridge to sell you. Don't let Intel make you think that you need to buy a new computer right now. It may help the economy in the short term, but you will just be wasting precious electricity (in this case gobs of it) just to say you have the latest and greatest. It's becoming a disease!
today is spelling optional day.
I recently purchased a laptop with a super duper 1.7Mhz Pentium 4 mobile processor. It's a nice box but it won't play GTA III worth a damn. Why? Well because the video hardware is crap. So I slapped together a cheap Athlon box, running at a lower clock speed but with a GeForce4 video card and all of a sudden I have high-res frame rates that leave my laptop in the dust. But both boxes have about the same performance for running Oracle.
If you think about it these chips are somewhat ridiculous. How many of them are going to be installed in motherboards that still have an IEEE 1284 parallel interface on them? How many of them will be installed in motherboards that still have PS/2 keyboard and mouse ports? How many of them will be installed on motherboards that have floppy controllers? You've got all of this insanely great hardware being strapped onto motherboards that are loaded down and dragged down with a bunch of legacy technology from the 1970s. I for one would be a little more excited if Intel were to dedicate itself to eliminating the cruft in the PC architecture. This would do a lot to improve performance and ease of use and would probably improve performance more than just slamming in a faster CPU every six months.
cheap labor conservatives - they want to keep you hungry enough to be thankful for minimum wage.
Is your receptionist sitting out where people can see her/him? That's often why these things happen. The company is investing in its image.
Say what you will about .NET and Java, but there are hoardes of programmers waiting for the average PC to be fast enough to run managed code without breaking a sweat. When entry level computers are in the 2-3GHz range, managed applications will be very popular.
I also agree with the other poster about word processing being a much more complicated task today than you think.
Yes, they are slowly improving, but modern PCs are still behind where workstations were years ago, and a modern Intel based server is well behind a SPARC based machine.
The bus and memory bandwidth has improved pretty much in lockstep with the CPU computational ability. While it might be nice on paper to have 16GB of memory bandwidth, and it might look good on a ridiculously synthetic memory bandwidth benchmark, in practicality such a imbalance would be just a monstrous waste of money: Generally processors actually do something with the data that they're processing, so the two factors have to balance: You need a system design that can keep the processor satiated. In the Athlon world such a situation was demonstrated superbly recently with the ramping up of the memory subsystem speed, DDR ramping up from 266Mhz to 400Mhz...what improvement did it demonstrate? Virtually none. The processor simply had no real need for the additional memory bandwidth, though I'm sure it will as they come out with the next generation.
Intel and AMD will spend their money on whatever generates the most ROI. They have collectively spent literally billions of dollars convincing Joe Public that CPU Mhz is the best way to measure the speed of a system - they aren't going to throw that away. A competent manager with R&D dollars to spend will therefore spend them on increasing Mhz.
While I have spent considerable effort in the past disputing the Mhz-is-king myth (especially in regards to the P4 versus the Athlon), I think you're promoting just as false of an claim. CPU speed DOES matter. By your claims, shouldn't these benchmarks show no improvement as the CPU power ramps up, given your claims that it's starved for throughput?
Yes, they are slowly improving, but modern PCs are still behind where workstations were years ago, and a modern Intel based server is well behind a SPARC based machine.
Are you kiddding me? You must work for Sun's PR machine.
Any modern PC versus a modern workstatation [including SUN] is a joke. Nearly every application runs significantly faster on a PC than a workstation [large CAD packages, image processing, databases, etc]. Yes, the workstations have larger L1/L2 caches and such, but the sheer speed of the core processors in the PCs overcome this limitation quickly as they are significantly faster than their workstation counterparts. Also, the memory architecture on the SUN's isn't some form of wizardary; rather, the latest SUN workstations use interleaved PC133 RAM [the expensive SUN OEM version of course though] - not exactly revolutionary.
Anyone with their choice of a PC 'workstation' running Win32/linux versus a IRIX/SUN workstation will choose the PC hands down if they have used both.
The true fact in the matter is that intel are going to rely almost entirely on the marketability of a big number with the P4, as it's handling is rather unimpressive when compared to such ordinary designs as those from AMD, which clock poorly, yet crunch happily.
r s/hartsteina_optimum_pipeline_color.pdf r s/Deep%20Pipes.pdf
/.
I disagree. Intel's strategy of designing for higher clock speeds has given them a much more scalable chip, and that is evidenced by Intel's ability to increase the clock speeds frequently while AMD is struggling. And if you look at the last Toms hardware review (its a couple of weeks old), the P4 2.8 GHz pretty much tied with the Athlon 2800 (they both won about 14 benchmark tests). But that is much less meaningful when you realize that Tom was testing an Intel chip that has been available for 2 months with an Athlon that won't be available until December. If you compare the 2.8 GHz P4 with the fastest available Athlon today, the P4 beats it in over 90% of the benchmarks (I'd imagine that a comparison between the 3.06 GHz HT chip and the Athlon 2800+ would be similar). So Intel's strategy is working for performance, and it is more marketable to boot.
And there is a lot of research right now about the optimal pipeline depth, and the conclusion was that the current pipelines are not deep enough. The optimal pipeline depth for the x86 architecture is around 40-50 stages.
http://systems.cs.colorado.edu/ISCA2002/FinalPape
http://systems.cs.colorado.edu/ISCA2002/FinalPape
BTW- thanks to fobef for these links- I read them yesterday on
"The defense of freedom requires the advance of freedom" - George W Bush