Slashdot Mirror
Intel Scraps Plan For 4 Ghz P4 Chip
bizpile writes "It was reported earlier that Intel would be delaying the release of their 4Ghz Pentium 4 chips, but it now appears that they will be cancelling them altogether. The announcement came Thursday and Intel says they are going to rely on approaches besides faster clock speed to improve the performance of chips. Engineers are working to add additional cores to a single chip and improving the efficiency in how the chips interact with the rest of the system. Intel spokesman Chuck Mulloy said, "Those are the sort of things where you get more capability out of a processor by designing specific silicon solutions as opposed to just keep turning the clock faster." In the meantime, Intel is planning on releasing a 3.8 Ghz chip with 2mb of cache."
Mhz do not always = performance!
Good job. Now I might be able to get a decent bus speed.
Wasn't that the entire reason behind AMD's use of the P-ratings? That performance was measured in more than just MHz.
Hell, Intel has spend DECADES convincing the public that MHZ is king and now they are (once again) following AMD's lead.
HA!
-Charles
Learning HOW to think is more important than learning WHAT to think.
But what about Moore's law? Is nothing sacred?
Seriously though, this seems like just what geeks have been saying for about a decade now -- clock speed isn't the be-all-and-end-all of CPU wars. Looks like Intel is agreeing with us!
--
Free gMail invites! (with references from the folks who're already there)
"..Intel says they are going to rely on approaches besides faster clock speed to improve the performance of chips."
:-)
Like getting rid of all the bloat in Windows and Office?
32bit is sooooo 1998
I knew this would catch up with them. I'm glad Intel is off the MHz thing. This doesn't mean the general populace will be more informed when buying a processor, but at least they might be looking at other features that may matter more (i.e. shared video memory, backside cache, etc.). Maybe.
Per Square Mile, a blog about density
Actually, Moore says that chip complexity will double along with relative performance, not clock speed. If Intel goes ahead with dual cores, and maybe quad cores later, then Moore's law is safe...for now
Does anyone really care about clock speed anymore? Yes, I know some applications need all the muscle they can get, such as video manipulation and scientific computing. However, it seems the interest in clock speed has waned considerably since the 1 GHz mark was hit. Basically, unless you are doing high end gaming or one of the aforementioned activities, increasing clock speed does very little for you. Consequently, it seems to me that the inevitable increases don't garner the same excitement they once did--going from 133 to 166 MHz was a big deal. Going from 3.0 to 3.8 GHz isn't nearly as useful, though the percentages are the same.
Sounds like they want to balance a pc like a mac. Bout time.
there is no such thing as common sense. If sense was common, everyone would have it. -unk
What this means is that Intel will probably be releasing a multi-core HT product in the same market window that the 4MHz part occupied.
Isn't this a full quarter in advance of what we expected? Won't this put their release in the same window as AMDs multi-core release?
AMD decides to follow suit. After the next round or two of evolutions, it'll be right back to square one of the Mhz battles. Somebody help me set my VCR back to 1992, I want to see how this one turns out.
You can't increase clock speed indefinitely. There's a fundamental limit we're brushing up against here, and it's called 0.8c.
Electrons on copper travel 3cm per nanosecond. At four Gigahertz, each clock cycle, the electrons can only travel a theoretical maximum of 0.75cm. I don't even think that covers the diameter of a single core these days.
You can't turn up the clock much faster than it's already going without getting into nanotechnology. The only viable solution is to optimize chip efficiency through other means, and add more cores to the chip working in parallel.
occultae nullus est respectus musicae - originally a Greek proverb
It just kills ppl when they see my Pentium Pro box keeping up with XP on a P4, for desktop stuff.
C|N>K
They've been doing this for a long time; basically all this says is that they're attempting to change the focus of their marketting from clock speed to other measures. I predict that consumers won't like it, and they'll go back to cranking up the marketting-clock-speeds ASAP.
Have you read my blog lately?
Or... if you RTFA, you would be able to do better than just "guess" that Intel would be releasing a 4GHz part. ;)
Meanwhile management will be on patrol with whips, screaming such motivational phrases as "Don't you be letting my stock price drop", "We kick AMD's a** or I'll come back and kick your a**" and always a favorite "Yamhill! I mean, Yah Mule, get along there, keep working, that's the stuff *crack*"
What the article doesn't tell you is about Intel's pile of unsold processors. Hey, do what automakers did, sell out your future today, slash prices, provide %0 financing and move them little doggies out.
A feeling of having made the same mistake before: Deja Foobar
looks like we are approaching a moore's law limit. i looked at http://www.spec.org benchmark submissions and found that 2002 4th quarter had the fastest CPU clocked at 3.06 GHz and in 2004 3rd quarter it is 3.6 GHz. This is just 18% increase over 21 months.
ofcourse, the CPU performance is still increasing at a rapid pace; but as far as MHz (or rather GHz) is concerned, we are reaching the limit faster.
WOW, my pc is 2 yrs old and I am not planning an upgrade!
Running 2.8GHz overclocked (aircooled) to 3.4GHz here...
There will always be the 1337 among us who want the most recent and fastest chip, no matter who makes it or what makes it fast, as long as they can get 4 more fps in Doom 3 and run a Seti@home task in 3 minutes less. If Intel sells some chips and steals some business from AMD et al., which they almost certainly would have with this thing, why not do it? They've already put time and money into R&D, so why not just go ahead with the thing?
I mean come on. We all know their engineers knew that MHz != better cpu. It just took them this long to finally convince their PR department to give up on the multi-billion dollar investment they have made in making "consumers" know that MHz == better cpu.
We were all warned a long time ago that MS products sucked, remember the Magic 8 Ball said, "Outlook not so good"
No, Intel switched to processor numbers when they realized that we realized that MHz don't tell the full story.
About stinkin' time. How much of the typical users' workload is CPU-bound? Let's work on some of the other parts, like through-put to RAM and to disks.
See what I've been reading.
Considering the fact that my 3 year old PC died, I replaced the 1.4GHz Athlon (T-bird) with the Socket 754 Sempron 3100+.
Same RAM, same disk, same video, but a new motherboard.
I *feel* like I'm getting more than a 28% speed boost from it, so it's clearly not just the clock speed that's doing it. Making a chip run faster never was the right idea, and I'm glad to see that they're walking away from that.
Now, if we can just get a core like the Pentium M, but for desktops, then maybe we'll see some real competition.
Raptor
"Procrastination is great. It gives me a lot more time to do things that I'm never going to do."
This is a pro-competition, anti-monopoly post btw... Micro$oft Monkey Boy sit down and take notes - it's spelled C, O, M, P, E... oh dammit I forgot my spell checker. Better load up Word, or maybe OpenOffice. God competition is confusing....
I've noticed that everyone who is for abortion has already been born - Ronald Reagan
...thats' to improve the P4 through mid 2005 and get the most bang for the buck. And move engineers onto the cpu Intel intend to roll-out Midish 2005 to rival the AMD64 product. Why mid 2005? Because that's around the time when Microsoft's 64bit Operating System will likely be launched and Intel doesn't see the need to launch prior to this. Sorta makes a lot of sense. All Intel needs to do is get their new 64 cpu RIGHT first time and moving engineers over is one way to ensure this happens ON TIME.
Multi-cores (i.e. parallel processing) is clearly the correct approach. The only fly in the ointment is a few software packages that charge on a per cpu basis, and count each core separately.
Ouch!
I think we've pushed this "anyone can grow up to be president" thing too far.
This sounds like a score for the good guys, with Intel finally realizing what others (like AMD) have realized alot earlier.
But...
Lets see what is actually going to happen. There are plenty of previous examples of Intel changing direction, and it is not always for obvious reasons. Remember slot1 and slot2, that Intel praised as a superior way to interface cpu's to motherboards as opposed to sockets, and when all came down to it, it was nothing but a stunt to try and make life harder for competitors.
Could this be a forced move by Intel, because they aren't capable of increasing the clockspeed and keeping cpus stable?
...of Microsoft realizing it had missed the boat with the Internet back in the '90s. Let's hope the paraniod play fair.
What if Digg added local news and a Slashdot inspired comment karma system? ---
http://houndwire.com
Intel says they are going to rely on approaches besides faster clock speed to improve the performance of chips
Strange, I thought the point of the big numbers was to sell more chips, not to make them faster. Wasn't part of the reason that Intel made the P4 pipeline as long as it is so that they could keep cranking the MHz up for a long, long time so they'd have lots of generations of P4 processors to sell? Because I don't think you really need that long a pipeline for purely performance reasons.
I wonder if AMDs inroads into the 64 bit market have Intel getting a bit scared about the future?
Intel continued to use the MHz race because the public was on board, and simply because they were able to maintain a demonstrable lead in the race due to their process technology lead. They preserved their enormous market share and high margins by spending decades convincing the public that MHz was the key.
It will be difficult for them to apply as much inertia into another simple metric that the public will understand and by whose measure they will be able to remain the clear leader. They need to come up with another marketing story that pushes yet another metric that is again closely tied to their process superiority. I don't know what this is, but I'm sure they have a new story that we will see when they do their multi-core HT rollout.
AMD did not exactly "win" simply because they gave up the MHz war so soon. Yes, they were the first, but they didn't have much of a choice since they knew they could not scale to 65nm process geometry like Intel could. They had to alter their architecture earlier. Intel did not, and it worked in their favor for more years.
It is obvious from the past that Intel's marketing story will never resemble AMD's. They are not "following AMDs lead" unless by that you mean they were able to scale clock speed for a longer time than AMD was.
Intel spokesman Chuck Mulloy said, "Those are the sort of things where you get more capability out of a processor by designing specific silicon solutions as opposed to just keep turning the clock faster." In the meantime, Intel is planning on releasing a 3.8 Ghz chip with 2mb of cache."
So to sum up:
1) We've realized it's dumb to just keep increasing the clock speed.
2) Buy our new Pentium 4! It's going to have a higher clock speed!
~Philly
Consumers are looking for bigger numbers. The number they have been trained to is Clock Speed. That's really all that counts.
What about drive space or RAM? They don't know the difference. It's the CPU.
I may be generalizing, but CPU speed has always been the main selling point, aside from price.
Intel's finally playing catch up with AMD. Bravo!
Non impediti ratione cogitationus.
That a 4ghz clocked Pentium 4 melted through the motherboard every time they tested it. Ok, I'm just kidding, but its great to see a company so headstrong about the Megahertz myth finally admit that it takes more than just core clock speed to make a computer run faster. With this said, I wonder what surprises we'll see coming out of Intel.
"Some fight for law. Some fight for justice. What will you fight for? One day, you will see."
Okay, maybe it's not the hard disk if you're running Linux, but for Windows users, it's disk i/o that seems to bog things down the most.
Well, I used to do molecular dynamics simulations that were, but that's just the exception that proves the rule.
I tried to run Seti@home on a 2.8 GHz P4. Wonderful speed, but that damn fan noise (quiet - Ramp Up - REALLY LOUD - Ramp Down - quiet) bugged me, so I got rid of it.
The clearance system sounds logical. It is not. It is completely arbitrary. -- John Bolton
I recall some earlier discussions about how Intel was finally starting to wise up and design processors that are efficient, rather than just raise the clock speed.
The first incarnation of this is the Banias, also known as the Pentium M. It's basically a P3 pipeline, but with P4 branch prediction (and some other technologies). The P4 has to have very advanced branch prediction in order to even HOPE to get reasonably efficient use of its pipeline. Applying this to the P3's shorter pipeline results in a much higher IPC.
In other words, something philosophically like the Athlon.
Since then, I haven't heard anything about it. And then there's this article. Is there any relationship?
I just want a desktop Pentium M system, without having to browse some Japanese-only Hitachi site.
I don't want more power, I want a fast enough machine that runs silently.
I guess it's my fault for waiting for Intel to provide this instead of just buying a Mac.
I don't know about you guys, but I always thought that it was Moore's _observation_ about the field and not a _law_. Nobody in their right mind would think it will last forever. It's just a matter of _when_ it will be irrelevant.
.....
Oh well
Nah, they switched when AMD's speed ratings were higher/faster than any current or planned P4s. Really, if Intel could go from 3Ghz to 4 or 5 or 6 Ghz as quickly as it went from 2Ghz to 3Ghz do you think they'd ditch Mhz altogether? For all the campaigning Apple and AMD have done, the average person STILL thinks Mhz is the only think to look for in a PC. There's no way Intel would ditch that marketing advantage if they could help it. With Athlons quickly scaling to 3600 and 3800 speed numbers and little to no possibility of a P4 with a higher number in the near future, Intel's higher Mhz weren't much of an advantage anymore, so they switched to a naming scheme that couldn't be easily compared with Athlons (or previous Pentiums, for that matter).
Almost nothing in the parent makes any sense.
Both AMD and Intel are currently producing 90nm chips, and both are heading down the nm-road pretty much in parallell. AMD and IBM are working together on both the 65nm and... the 45nm process.
"[AMD] knew they could not scale to 65nm process geometry"? Really? What the fuck are you guys smoking?
Sheesh.
anyone know if the 3.8 ghz chip will be for socket 478? or just for 775? i couldn't find any info on this.
Marge, get me your address book, 4 beers, and my conversation hat.
Intel is just admitting what the rest of the processor industry has known for years. AMD stopped playing the Mhz game with their 64 bit chips. IBM, and Sun have had 64 bit chips for years and are already shipping multicore CPUs. Sun has plans for dozens of cores per die. Intel will have to work overtime to catch up with these other companies.
Not only was that a cool joke but I doubt theres somebody in this world who have an infinit quantity of liquid nitrogen to every few hours (or whatever time it takes) to put some more on the cpu and whichever chip he decided to also cool using LN2.
Then again, doing that everytime would create more muscle in the arms than clicking that mice.
Getting four more frames per second in Doom 3 is about as useful as getting a turd that smells half as long as they usually do.
Aopen sells cheaper Pentium M mobo
Belief is the currency of delusion.
For 3D Rendering all you need to do *is* just turn up the clock speed. It doesn't matter how fast the memory bus is... or even how much cache is on a chip beyond a certain mimimal level.
You can build super cheap (except for processors) computers to use in a renderfarm.. (I use Lightwave 3D, Modo, and SoftImage XSI)... and hard drive speed / graphics card speed / Memory speed / Cache on die, Do nothing to speed up a render once you hit that "Render" button. Sure... SSE extensions and the like do speed it up if the code is optimized... but there isn't really a way to optimize the code with this new direction Intel is going.
I am Jack's HTTP Server
It's 24 cm/ns and 6 cm at 4Ghz.
Besides, it's energy, not electrons, that travels fast in a circuit.
No, they only realize that their processors are not scaling as well as they used to. As a result, they run very hot, hotter than AMD chips. This makes them very hard to cool. OEM's don't like that. Therefore, it's time to change course, and make it sound like it was planned all along....
If they (and AMD) weren't having a heat problem, it WOULD be business as usual....
Remember intel has done other things to increase speed other than just MHz increase. Such as: 1) Increase Front Side Bus (in the p4's case 400 -> 533 and now 800MHz) 2) Increase Cache (256 -> 512 -> 1024 -> 2048kb) 3) SSE 1, 2 and 3 4) HyperThreading
But the droids blinkered by intel FUD put their fingers in their ears sang "lalalalala" and barked "NO - faster clock speed is a FASTER CHIP!!!"
Now, suddenly: oOooooo - cycles per second isn't as important!
Oh well. It will certainly be very interesting to see what Intel does over the next few years.
Here's an interesting question, related to this topic:
Assuming they go multicore (like IBM and Power[x] chips) what are the limits involved there? What would logically stop the development of multicore chips from increasing their number of cores?
And: What next?
RS
Shoes for Industry. Shoes for the Dead.
Am I the only one who's dissapointed with the announcement of multi-core chips? Certainly, they'll net performance gains in the short-term, but the process has to cieling at some point. And then what? The companies are only doing this to keep down the amount of money they have to spend on researching new process technologies (such as carbon) while keeping their share prices high. Why are we praising them for it?
There was a long (as in still hasn't ended yet) discussion with mostly HPC types who are of the opinion that desktop apps aren't sufficiently parallelizable to benefit from multi-core cpus.
You should hear the fans on our new Prescott Dells. If you don't run the processor it's pretty quiet, but once utilization goes over 50% on one HT "core" the fan kicks up and continues to speed up until you could hear them from 30 feet away with two doors in between. We demanded and got new fans from Dell (and some crazy heatsinks (5-6" tall heat pipe jobbies) that must have cost Dell more than $20. I can see why they might be having a few problems with heat at 4 GHz.
Degaussing scares the bad magnetism out of the monitor and fills it with good karma.
If Intel's primary motivation behind going from the Nortwood core to the hotter & less efficient Prescott core (longer pipelines result in a Presocott chip with double the cache of an equally clocked Nortwood actually being slower) was that the Pressy would allow them to scale to higher clockspeeds than the Northwood would allow does this make the Prescott a failure?
my sig's at the bottom of the page.
I don't think Intel can just decide to move up the dual-core Pentium 4 by a quarter. Presumably they were already working on it as fast as they could.
Intel released their Q3 results late Tuesday. In their conference call they were evasive about a suprising drop in their tax rate and also about the amount of their inventory writeoff. Intel claimed their inventory was down $43 million to $3.2 billion with an unspecified writeoff amount. Investors were happy to see inventory did not go up again and the stock went up Wednesday. In several different articles people are working out the mystery of the writeoff amount. Normally Intel's "cost of sales" is a steady number. Any writeoff will add to this number. So you can estimate the writeoff just by seeing how much this increased. With this calculation, it seems Intel had a writeoff of $472 million.
"Intel Scraps Plan For 4 Ghz P4 Chip"
Duh...they're still going to 4ghz, but it'll be a pentium FIVE! Because they havnt upped the number in a year or two...
And what speed processor, pray tell, do I need to run OSX today?
I run Windows XP on a 384 meg/450mhz K6-III, a 192 meg/400mhz pentium III, and a 80meg/133mhz pentium. All three machines run it with good responsiveness.
Intel is in TROUBLE. The focus of their entire company, their raison d'etre, no longer exists.
Fans wear out. Does the Athlon have automatic underclocking on overheat like the P4 has? Intel has won businesses over by focusing on reliability.
Maybe they realized they weren't going to be able to reliably cool the netburst architecture at those speeds so they're going to have to switch to the lower-clocked, possibly multicore Pentium-M arch.
They'd be FORCED to use a numbering scheme because any conspicuous lowering of the MHz would cause Joe Shmoe to say "What the hell?"
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
They liscence Hypertransport...
Maybe I am the only one who chuckles about this unlikely possibility, but sharing is just plain nice.
Do they offer classes based on Tanenbaum Structured Computer Organization in High School yet? They should.
Oh and maybe require a computer drivers liscense!
Sorry, the NK Hackers have me worried, ignore all my base.
Just like any complex system, there are a number of factors that influence the speed of a processor. Clock speed is a very large player.
It's silly for people to think that clock speed doesn't matter, why else would people go through the trouble of overclocking their systems?
Intel should just bite the bullet and spend some more R&D on alternative active cooling solutions like liquid. They have the leadership to produce a standard for connections, case layout, etc. and make it easier for us.
Yes, there are other areas in which performance can be enhanced.
Larger L2 caches only go so far. Intel has already made large cache P4s available to us. Perhaps doing some work increasing the L1 cache sizes would be beneficial.
Multiprocessor cores would help considerably for some. For 3D rendering, this would be a godsend, two processors are around 98% faster than a single one. But we are talking large die sizes and therefore expensive. Intel should just let us run two P4s instead of having to buy the expensive Xeon solution.
Go clock speed! I'm the only one still rooting for you!
What the hell are you talking about?
OpenGL vs. Direct3D has nothing to do with the choice between Intel vs AMD. Either you're pulling this shit out of your ass, or you're very confused.
And the AMD for games and Intel for business/video logic is equally retarded... (I will allow that the best video encoding WILL be found on P4s most times)
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
The static power (quiescent) is certainly higher than for, say, 130nm processes, but the dynamic (operating) power isn't really that much more.
But don't laptops and other battery-powered devices really need a low static power drain?
The Pentium was not copying anything done by AMD (AMD was pushing a P6-like design from a company they purchased in the K5/K6 cores)
Now the P6 could be considered a copy of the K6. That I'll believe.
And the EMT64 was not really a "copy" but a tweak in the microcode of the P4 to appease Microsoft and consumers who wanted to take advantage of the larger register set and memory space that x86_64 long mode affords. Internally, the P4 has the "features" required to make this work, but Intel did not try especially hard to make these extended features run fast in their CPUs, as if to downplay it.
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
But not until later, when they were tweaking the system bus to run at 60MHz. This yielded the 60 MHz, 90 MHz (I had this one), 120 and 180 MHz pentiums.
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
And the EMT64 was not really a "copy" but a tweak in the microcode of the P4 to appease Microsoft and consumers who wanted to take advantage of the larger register set and memory space that x86_64 long mode affords.
Actually, that's dead wrong. It came out a few months ago that not only is EMT64 a copy of AMD's 64 bit extensions, it's a *direct* copy, even to the point of copying and pasting AMD's literature on the instructions. Those instructions were copied verbatim.
It's not at all compatible with OpenFirmware, and that's a damn shame.
Because what makes OpenFirmware so great is that you can take any hardware (graphics card, scsi card, etc.), plop an FCODE ROM on it, and it can be used AT BOOT by any type of machine that is OpenFirmware compliant. Now that doesn't remove the need to write native OS drivers, but at least you wouldn't have that needless Mac AND PC versions of video cards or whatever... just OF-compatible cards.
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
We have had an informal pool as to when Intel would release a 4.77 GHz chip. My date was June of next year. This delay will probably knock me out of the running.
I was going to load up my copy of the Mix C compiler to see it run 1000 times faster.
Why do they keep increasing clockspeeds on 32 bit processors?
A while ago I heard they were going to work on making them smaller, now I hear of 4 ghtz.
They need to get in gear and start working on 64 bit, or at this rate, AMD might starting taking a bigger chunk in the industry.
http://www.macinhack.com
Any app that isn't parallelizable will run just fine on a single ~1.5GHz Pentium M, for example. If you have a four-core system with those processors it's going to be fast as hell. And, you can run four of those unparallelizable programs at once without choking your ability to retire instructions.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Actually, you're wrong. "Moore's law" merely states that the transistor count will be experience exponential growth. Which actually an observation rather than anything concrete.
Nothing about chip complexity or performance.
Sadly, this beloved phrase is rapidly losing value with inflation. Back in its day, "dollars to donuts" meant REALLY steep odds. But today you're only betting about 2:1, which isn't much to get excited about.
I am aware that the EMT64 extensions are exactly the same as x86_64. IF THEY WEREN'T THEY'D BE USELESS FOR RUNNING THE 64 BIT CODE... which is why Intel included them.
But Intel didn't change any of their cores, IIRC. Just the microcode inside so it could decode and emulate those instructions.
They weren't "copying" them, because they were just giving into OEM/custmer pressure and re-provisioned the internal chip resources into decoding the instruction set. Of course, doing this all quietly with little fanfare because of the NIH syndrome.
As an aside, one thing that could save the netburst core would be if they truly changed the microcode to run a vastly different IA with a much larger register file, nay a windowed register file... one that puts integer, FPU and SIMD instructions all on equal footing too. This could provide chances for greater internal parallelism and less traffic to the caches.
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
It'll be interesting to see what sort of an effect this trend has on emulation, which is all about the MHz and basically non-parallelizable. MAME already supports some arcade systems that cannot be emulated at full speed on any CPU yet manufactured. It may be that systems that exist now or in the near future will never be possible to emulate at a reasonable speed.
Omnes arx vestrum sunt adiuncta nobis.
To me this seems like the same problem HD's have, trying to cram more density in less space.
Personally I wouldn't mind a computer that was the size of a small refrigerator if I knew it could pump out 1THZ and did a good job at power consumption.
BTW I do a lot of computation hungry coding, so speed is important.
... Intel has announced that they will be ditching their 8mm training tapes, switching instead to VHS.
When asked about this move, an Intel spokesman said "As you can tell from our other directional changes, we have decided to step boldly into the 1980s!"
Yes the MHz == performance thing promulgated by ...)
...).
Intel was BS (and BS that seemed to flow so
deep that they would engineer in superlong
pipelines etc
And yes, the computer architectures that can now
flourish with the Mhz race slowing down will
be exciting (In years past many such research
projects died because by the time they were done
the march of Moores law had resulted in a faster
conventional architecture).
BUT: There is all kinds of exciting ideas that
were being made possible by faster and faster
processors (including immersive VR words,
new kinds of programming, new UI models, image
recognition
If this is genuinely a community of computing
enthusiasts should we also not be sad if this
is slowing down.
- Mudchicken
Okay, this means it's time for the CPU performance increases to take a back seat. Maybe now the rest of the computer can have some time to catch up better with the CPU. I am talking about bus and memory bandwidth. This is one hurdle that needs to be overcome.
Low latency and high bandwidth up the wazoo is one aspect that supercomputers for example have over standard pc components, besides massive parallelism of course.
It would be cool to see intel start making inroads from R&D on the memory front. I'm not talking about on-die cache, that is a given. The questions to be answered are how to get the main memory up to snuff with the rest of the system.
If the current state of the art in CPU power stagnated from here until 5 or more years from now, it really wouldn't be an issue if the same efforts during that time were put into lower latencies across the whole sytem architecture itself.
So what am I saying? The CPU has had enough innovation in it's current form. It's time to focus on other lagging components. Pci-x is a step in the right direction, but it is nothing without main memory advances and other mainboard bus architectural improvements.
>>>>>> Chewie, take the professor in the back and plug him into the hyperdrive.
I don't know if I'd agree exactly with this comment. While a 3.8 GHz P4 does not perform as highly as a 3.8GHz Athlon chip would, an AMD chip can not physically run at these speeds. The pipeline would not support it.
The slashdot crowd is quick to attack Intel because they're the big guys, but the NetBurst architecture is an extremely powerful and (gasp!) good architecture. While the engineers designing it designed a processor for maximum pipelinability (over 30 stages now) this is not really a bad thing. Pipelining a processor is a good thing in general. Its main claim to usage is that it allows a processor to run at a higher clock speed. That is what pipelining was created for; to break down the time into smaller slices so more can occur in parallell. This process works great when each stage is of approximately equal length, and I have enough faith in the Intel engineers that no single stage was much longer then the next longest stage.
Back to the point though the pipeline does have downsides. A processor with 20 stages will lose ~ twice as many cycles on a branch missprediction (and more on a cache miss, but that number varies further) when compared to a 10 stage processor. However assuming that by using 20 stages we cut the cycle length by even 50% the additional stages were worthwhile. Cache misses are not a "common" event and branch prediction is in the 95+% range now, so the stalls added there are not as large as you'd think.
What the pentium 4 has done was manifest these to a larger scale. Unfortunately the engineers desiging the processor did not realize the massive leakage currents that are seen with processors at the speeds Intel is using. From a computer architect's standpoint they build upon past assumptions, and more stages in a pipe generally help out, so thats what they did. While the end result is not as impressive as they were hoping the end result is not a poor product.
Now what has the NetBurst architecture offered to the consumers? Well one of the main offerings its had is building an SMT processor (hyperthreading in marketing speak). SMT is more then mere marketing hype. It was not an afterthought thrown onto the P4 due to less then stellar performance as people have hinted at. SMT was originally designed for the Alpha ev8 chip that was scrapped. Intel however bought the alpha design team and used the SMT technology (albeit to a lesser extent then some would hope for) in the NetBurst architecture.
What else has NetBurst added? The trace cache is a wonderful feature as well. This removes the x86 decode logic from the runtime pipeline for most instructions.
So where can Intel go from here? My hope isn't so much in the multicore logic that some talk about. While multicore is interesting, I personally would rather see a wider P4 core (more execution units) and have them extend their implementation of SMT to allow for more concurrent threads of execution. a 4 or 8 way SMT processor could show some real results.
And for those of you who are going to question what I'm saying... No I don't work for Intel. And no my desktop processor is not an Intel processor either (I run an athlon 1600 for my workstation). However in my lab I am working on algorithms designed specifically around SMT processors (as well as cache aware/prefetching enabled applications). Intel's processors happen to enable quite a bit of optimization if done properly.
While I never agreed with Intel playing the MHz game, or their ridiculous prices, I would not say that the engineers were completely against the super-pipelining of the NetBurst architecture. While they may have questioned the reasons behind it, the real world performance gain does exist do to it.
Philip Garcia
Rendering operations can almost always be done via asynchronous parallel operations.
They're known for this.
So much so that the term "renderfarm" has been coined. So doubling the width of the CPU bus will allow for double the calculations (assuming you have ops to handle it that way - to make one 64 bit CPU act like 2 32 bit CPUs, or 4 16 bit CPUs).
Similarly, doubling the number of CPUs on one die will give the same advantages until you reach the harddrive bottleneck.
Also, a lot of operations (blurring, sharpening, scaling, rotating) can be done using convolution, and convolution can be done using dynamic programming techniques, reducing multiplications while also increasing memory usage. Since most frames are larger than the average cache while uncompressed, a larger cache will speed this up - especially if you can get one that's the size of the buffer. I think that most rendering engines don't consider this because to get the real significant improvements, you have to know low-level details of how memory is fetched.
Mod me down and I will become more powerful than you can possibly imagine!
It seems to me that there has got to be maximum rate at which we can push the clock.
I have a 3.2 GHz Pentium 4. How far can light travel in one clock cycle at that speed?
186000 miles / 3.2 billion is about 3.7 inches isn't it?
The race isn't always to the swift... but that's the way to bet!
what it is, is a 'Central Processing Multi-Core Unit' or CPMCU, *technically* still a (one) 'Central Processing' Unit :-) ... that has the capability of processing more 'threads' at a time due to its 'multi-core' architecture
I don't claim I know more than I know, and if you know you know more than I know, then by all means, let me know.
...you shoulda thoughta that before you deepened your pipeline. Idiots.
I think the main reason why the Sempron 3100+ runs quite a bit faster than your Athlon 1.4 GHz CPU is that the CPU architecture is extensively based on the highly-efficient Opteron/Athlon 64 design, which means very efficient system memory access and very efficient access to the on-CPU die L1/L2 memory cache. That evens even though the Sempron 3100+ runs only 400 MHz faster CPU clock speed compared to the Athlon 1.4 GHz, other CPU architectural improvements improve overall performance quite a bit.
By the way, you might want to know that future Pentum 4 variants will base their core on the Pentium M CPU architecture; you'll see these new Pentium 4 variants arrive on the market in 2005.
Is this the response to what the industry has been saying for many years about the x86 / performance limitations. Couple intels egotistical approach to the market.
This has been seen with the Opterons onchip memory management etc etc. There is always room to grow sideways which results in an upwards growth.
Only 2 millibytes of it. People writing articles here at /. should know how Mega is spelled.
Good post. You should drop by comp.arch on usenet. Maybe you already do.
In the Windows operating system, there's a 32-bit routine that returns the processor speed in Hz.
However, the largest number a 32-bit value can represent is 4.29 billion; in other words, 4.29 GHz. This imposes an absolute speed limit for Intel processors, because clearly Intel is unwilling to break backwards compatibility with their obsolescent x86 architecture. So they are putting the brakes on now, so they can coast slowly toward this upper limit.
Pity. I would have liked to see the creative ways people could find ways to make software run slow on a 50GHz processor. (Then again, perhaps that software already exists, under the moniker of Microsoft Word?)
Weeks of coding saves hours of planning.
This is excellent news.
The MHz war is over we hope. If they are truly going to focus on smarter chips which "guzzle less gas" then that is wonderful.
The new war is efficiency and power. As a start, I think all CPUs and computer components should have an energy rating, based on how much energy is used up. I think I heard about a C7 chip that ran 1 GHz for 3.5W, which is amazing. Intel need to make smaller, more efficient chips. Doesn't a P4 consume 80W for 3GHz ?
If the new war was how much 'power' (instructions per clock tick) per Watt, wouldn't we all benefit ?
L
Intel realized their 4Ghz chip needed more sheilding and cooling then a nuclear reactor.
Pipelining may be generally good but adding shifting stages but do note that out of the seven added stages in Prescott, two are dedicated to buffering signals, presumably because they are too slow to reliably get cleanly across the chip between two ticks.
One would have thought that process shrink and these extra stages would have given Prescott significantly more headroom than what Northwood already can achieve but such is not the case in reality, a first in Intel's history AFAIK.
"Customers have become numb to mHz numbers, so we're refocusing our efforts on a core-count arms race instead."
While there are good reasons to go some way in that direction, I couldn't help remembering the battle for the hearts and minds of transistor-radio purchasers away back when. The marketing people got a lot of consumers to believe that more transistors == better radio, and then it was found that some "10-transistor" radios had only two or three functional transistors; the rest were just soldered to isolated pads, or used as diodes, or something. Let's not go *there* again.
Sure, in high-availability situations [automatic underclocking on CPU cooler failure is] essential. But for the average user, who never bothers to read their logs anyway
If anything can save them, it's the Pentium-M, which is an extraordinary piece of technology out of their Israeli branch.
The king is dead... long live the king.
Bullshit (sorry for that)
For less than 5% gain in performance they put a major "complication" into the core.
They will need to haul it around long after they have any use of it (compatibility).
There was a talk by one of the Intel chief designers where he told that complexity of Intel's top of the line processors has reached such levels that they have only 2-3 guys who have clear understanding of the beast and who are able to debug it. And that the large number of these small gain hacks leads to side effects in form of functional interference so that the system behave as it has it's own agenda (like those hard to explain and eliminate performance drops for few hundred cycles, etc.)
He said that in terms of the overall complexity of the core it looks like it is the end of the road (i.e. major "rethinking" is in order)
The main bottleneck is CPU - RAM interface. Latent delays of several hundred cycles just tower over these small hacks (the problem HT supposed to help with but it doesn't help much).
Basically, two much simpler cores will do much better than one which tries to pretend it's two.
Plus there is major problem with such things - software which supposed to use it.
Here's the news - software is orders of magnitude more expensive than hardware.
Very few projects will go into optimizations of such level. Automatic (compiler based) will not give you much. And people will not rewrite existing code. Companies would rather spend few thousand dollars on additional hardware than hundreds of thousands on software optimizations (which is also very risky).
Cache misses are not a "common" event
Actually it's _SO_ common that by designers own description it's a shame.
That's why HT was invented in the first place - to do something when they wait for "missed" data.
> For less than 5% gain in performance they put a major "complication" into the core.
5%?
This depends on instruction mix and data set sizes/locality. In cases where the instruction mix has generally poor instruction parallelism, SMT can do wonders well beyond 5%... but on highly tuned non-threaded code, it can come with a slight penalty due to shared resources.
The big point hehind SMT is that whenever one thread stalls by waiting for RAM/IO, the CPU can continue working on the other threads instead of waiting.
SMT does not produce the same typical gains as SMP. It does cause most single tasks to be slightly slower but overall system throughput while running multiple apps usually increases by far more than 5% - from what I remember, Intel's projections were in the 30-50% range typical and from my personal experience and limited benchmarking, this feels just about right.
A processor with 20 stages will lose ~ twice as many cycles on a branch missprediction
I assume branch prediction means that it tries to guess which way an "if" will go?
If so, what are the usual techniques?
branch prediction is in the 95+% range now
That sounds pretty damn good.
Is there any sort of common wisdom about certain languages, algorithms, coding styles etc. that can make this better or worse?
That's what a law is, in the context of science. An observation that holds universally true.
Wrong. The Prescott is actually a better performer than the Northwood above about 3.8 GHz. What Intel has admitted - a little late, considering Tejas should have told them this some time ago - is that the problem is Precotts are so damn hot they cannot scale past 4.0 without extreme cooling measures.
If somehow Intel could have gotten Pressys to scale to 5+ GHz without having to include a phase change unit in the retail box, Prescotts would be in the roadmap for another 2 years. But they just can't, nor get enough these little hot plates to work at the 4+ range within vcore and temp specs.
Yes, MHz doesn't always equal performance, but the same chip at a higher speed does - unless it is on fire.
Slashdot "libertarians": Small government for me, big government for those I disagree with. -1, I disagree with you
hum.. wondering... what number would be assigned to a CPU beyond the 10 Ghz barrier? ;)
Long ago, they would have two tracks, so that the 486 team upon release would move to the pentium pro/2, while the pentium team would be in the middle.
I think the Itanium tripped them up. It was supposed to be another generation, and they took too long to abandon it.
That might explain the P4 being slow, and the P-M deriving from the P3.
What were they smoking? I don't know but I think the problem is that it was being smoked in large, long pipes.
Apple is still around partially because they took a RISCy approach, so do more per GHz, and can step up the GHz since they don't have that heiroglyphic like x86 instruction set to decode.
From what I have been reading on Ars and other sources, the x86 emulation layer is actually very small now. All the current-gen "X86" chips aren't actually X86 anymore... internally, they are RISC. They have a conversion engine that converts the x86 instruction set into 'real' instructions, called micro-ops. They definitely have some overhead and difficulty from the register renaming and speculative execution stuff, which wouldn't be as hard with a cleaner architecture, but overall, to my understanding, it's not that much of the transistor budget.
Apple's chips haven't been going up very fast in Ghz anymore either, if you notice. They just now hit 2.5ghz. AMD is around 2.4ghz on their FX-53. Intel is quite a lot faster, up at 3.8ghz, but the less efficient architecture soaks up nearly all of the clock difference.
All the chip manufacturers seem to have hit a wall here, and I think it's entirely likely that we're nearly at the end of Moore's Law... from here on out, it's likely to be diminishing (rapidly diminishing?) returns. We can go multicore, sure, but desktops the way we're used to using them aren't going to benefit much past about four cores.
The moniker "intel" is becoming a misnomer. Per haps they should do a name change. Let's see "stupid" would seem more appropriate.
How many screw-ups, delays and paper launches can a company do before they lose cedibility?
Seems to me that there is no evidence here that any fundamental limit has been reached. AMD, with a fraction of the resources, is doing amazing stuff. AMD just did not go hell-for-leather clockspeed, instead taking a balanced approach to the issues. We have a winner, its called Athlon64. Intel (Stupid), on the other hand amassed great teams of technicians (I hestitate to call them engineers) and did the electronic equivalent producing one baby from 9 women in one month.
MHz is not only a simple measure of performance that a consumer can grasp, its a simple (and misleading) index of what the R&D people are producing for marketing to plug into their spreadsheets and powerpoint shows. Its pepsi time at Apple all over again!
Now, if you please, time to upgrade all of my systems to Athlon64. I can't tie my company to intel anymore.