Slashdot Mirror
Intel's New Architecture Too Late?
rts008 writes to tell us that TG Daily has an interesting interview with Randy Allen, AMD's vice president of the server products division, about (among other things) AMD's recent stellar fourth quarter numbers. From the article: "Responsible for that shrinking lead is especially AMD's server products group. Intel's CEO Paul Otellini recently acknowledged that Intel had to give up market shares to AMD and will likely be forced to hand over more shares until the next generation of server chips arrives. [...] AMD's Randy Allen explains in this conversation with TG Daily why he believes that Intel will need much more than a new processor to be able to slow AMD's growth."
AMD exec says AMD is better than its competitor. Earth shattering news!
Well it's a good thing that Apple is using Intel now otherwise they would really be in a hole.
I heard from someone that the Pentium-M is better than any of AMD's offerings for mobile CPUs; is there any truth to this?
I know that the new MacBook is running on the Core Duo line, and I understand that's a whopper of a mobile CPU, but I thought that AMD had a strong competitor to the Pentium-M?
Another perspective here:
1 3.html
http://www.thestreet.com/_yahoo/tech/semis/102631
Allen: Absolutely
Don't you just love it when a corporation thinks our legal system is just another subsidiary of their marketing department.
Both Intel and Advanced Micro Devices Inc. released dual-core desktop processors in 2005, but consumers are just now beginning to upgrade into the realm of increased hypertasking efficiency. The transition hasn't been entirely smooth--particularly for Intel, whose Pentium D series of dual-core processors was the target of frequent snipes from technocritics disappointed by the limitations inherent in processors' architecture.
From: http://www.informationweek.com/industries/showArt
Basically it boils down to this: servers like 64-bit processors, and AMD simply beat Intel when it came to affordable 64-bit computing. At the same time, AMD enjoys more support within the Linux/Unix community. Windows64 took ages to arrive and anyone who wanted a 64-bit server in the meantime simply bought AMD and installed linux. In the dual core deathmatch, once again AMD fragged Intel multiple times over.
And next gen is just a politically correct way of saying, "we are simply making all this faster than ever". The CPU equivalent of next-gen should be more like Nintendo's Revolution than like XBox 360. Alpha64/AMD64 were next-gen, dual cores were next-gen. But now, there seems very little "real" next gen CPUs in line. Intel's house is built of cards.
That depends upon what your goal is. The Pentium-M is superior in energy conservation. If this isn't your concern might I suggest: http://www.hypersonic-pc.com/FX7/
--Neth
Two of my imaginary friends reproduced once
Comment removed based on user account deletion
fact is, once the PHB's realise the "Intel inside" sticker is not an absolute requirement, Intel will have a hard time getting the cat in the bag again...
10 ?"Hello World" life was simple then
-William Brendel
You have to give Intel credit for ruling the mobile CPU market. AMD doesn't even come close in this area. And with everything becoming lighter, smaller, more portable, and dependant on lower power consumption... you can't count Intel out at all. Perhaps we're starting to see two companies that used to compete directly with the same kinds of chips begin to specialize at what they do best: performance for AMD and mobility for Intel.
In all the flamewars that will ensue, i think the one thing to note is that the CPU industry is very much a cyclical one, especially for the big players. While a design mistake by a small company might potentially be fatal (or get taken over by a bigger competitor), big companies can afford some minor slip-ups and still come back strong. The only problem is, the cycle is usually big (at least in internet terms, which is more like dog-years), and it easily takes a large corporation a few years to do this. All said and done, both these companies have their traditional strengths and selling points that are fairly distinct (process technology vs architecture, features and stability vs performance and value for money) etc. Given these unique USPs and perceptions that these companies have developed and nurtured over the years, it will always be a see-saw battle, IMHO. Of course, these perceptions also change over time, but again, it takes time and even this is cyclical as well.
One more thing: Can we please have a genuinely unbiased discussion/argument, for a change, please??
I believe the mobile market is about to explode, if it hasn't done it yet, and Pentium M's laptops are an overhelming majority over the AMD's powered (Turion?). And now here are the Yonah... AMD has a long road to be the king of the hill.
--
Superb hosting 20GB Storage, 1_TB_ bandwidth, ssh, $7.95
It seems there are always plateaus.
Dual core chips are still pretty new, right now it seems the emphasis is becoming getting more done per clock cycle and reducing power consumption as well as dual cores and other features.
I wish I could say I remember but I am probably imagining reading people talkinga bout quad core processors eventually. Could that be the next step in the march forward?
Would that not possibly be what the next-gen chips will look like?
"We're so tough we're made of nerf!" --D&D Character Tagline
Intel's CEO Paul Otellini recently acknowledged that Intel had to give up market shares to AMD and will likely be forced to hand over more shares
'Market share' is an idea that should be expressed in singular terms.
The Opteron strategy was a masterstroke and they have the edge in the technology battle for the fastest processor. But the company has a long way to go before it beats Intel.
I particular getting dual core onto the desktop first isn't going to be as big a coup as AMD thinks. Business generally doesn't seem to want dual core desktops yet and is certainly unwilling to pay a price premium for it.
So while Craig Barratt sidelined everyone with the Itanium processor AMD got the edge, but they may not keep it...
I don't mean the current Core-branded Yonah processors, but rather the forthcoming Conroe and the cores to follow. When it ships it will beat the X2 clock-for-clock and consume less power, and suddenly the things that have been novel about the K8 won't be. Instead of AMD having a clear leg up on Intel (the Pentium D is rather embarrassing at this point), AMD will have to compete on price with Intel having a clear advantage in process and volume.
All of which could have been said not THAT many years ago about AMD.
Jeeeeezus, can we all grow up? Please?
How about some INSIGHTFUL post and less hypercritical bs to let everyone know how S M A R T you are.
Intel has beat on AMD by: 1 Dumping memory prices to hurt AMD's bottom line. AMD's working on corporate restructuring to fix this. 2 Having a proprietary lead in tech. AMD has a better 64bit option, so Intel lost this battle. 3 Advertising/Corporate mindshare. AMD's making significant inroads with the blue suits AND gamers. 4 Better Manufacturing/Better product delivery. Intel still beats out AMD. Even now, AMD is NOT producing enough Athlon64's in the 3000-3500+ speeds to keep up with demand. Intel retains a significant operational lead, but not design and tech. Having an operational lead does NOT justify higher prices and margins on the free market.
Uh, Sun have a 8-core cpu for sale right now. See their t1000/t2000 servers.
Obligatory Futuram quote: If we hit that bullseye, the rest of the dominos will fall like a house of cards. Checkmate.
And this is one that looks to be messy for AMD. Intel's VT is more mature than AMD's Pacifica - not by as wide a margin as AMD's dual core beat Intel's, but Intel does have the virtualization edge. And once virtualization takes off - once VMware, Xen, and anyone else can run interesting stuff on generic chips - THEN we will need a whole new batch of processors.
Not immediately - I don't think the current generation of VT designs are good enough yet, I think it will take another iteration. But there is a wide market that Intel is going to hit just a hair before AMD. Which is a shame, because I really like AMD's chips.
Generational thought: Athlons / P3s and P4s were 7th generation, Opterons and 64-bit Pentiums are 8th generation. I don't know what 9th generation will look like yet - and I hope Intel / AMD don't try to claim it prematurely.
A witty [sig] proves nothing. --Voltaire
Well I stand corrected, though I imagine they're a bit out of the range of the average desktop machine buyer.
However, looking at it from a server pricing standpoint they are actually quite reasonable.
Interesting reading, thanks again.
"We're so tough we're made of nerf!" --D&D Character Tagline
And next gen is just a politically correct way of saying, "we are simply making all this faster than ever". The CPU equivalent of next-gen should be more like Nintendo's Revolution than like XBox 360. Alpha64/AMD64 were next-gen, dual cores were next-gen. But now, there seems very little "real" next gen CPUs in line. Intel's house is built of cards.
Don't count them out yet.
The new "Core Duo" laptop CPU shows really nice performance, similar to a dual-core Athlon at same clock speed but with less power consumption. If they bring the desktop version "Conroe" this summer as promised, AMD may need the step to 65nm just to keep competitive in that part of the market.
C - the footgun of programming languages
For one, Intel just won a major contract: Apple. Though Macs are a small minority of total systems sold when compared to all PCs, they are a reasonable player compared to other OEMs, and they've now decided they are all Intel, all the time. I suppose this could change, but not any time soon. Apple just cast in with a new provider, it'd be foolishness to switch again before they've even completed their transition (plus there's probably a contract involved).
Also Intel is allegedly releasing a new chip lineup in about 6 months. Supposedly not P7 (Pentium 4) core based. Given that the Core Duo/Solo line are targeted for low power uses, I'd bet that the new lineup will be performance chips. Will they beat AMD's offerings then? Who knows, but it's silly to think that Intel is just sitting on their thumbs.
In the dual core deathmatch, once again AMD fragged Intel multiple times over.
Not when it came to cruching multiple application simultaniously. Then the Intel chips came out slightly on top. That said AMD was 30% faster on running a single app, and used less power. But the Intel chip was cheaper!
May the Maths Be with you!
Better interview with Henri Richard from AMD here instead:
http://www.dailytech.com/article.aspx?newsid=295
The TGP interview is OK, but that guy is just a mouthpiece.
HJ
While I am a strong AMD (and generally "underdog") fan and I have predicted 2005 as a good year for AMD , I am afraid that 2006 can still bring surprises ... some good but many potentially bad ...
My analysis:
1. AMD will probably remain the leader on desktop machines at performance/$ and maybe (but not so certain) on performance/watt and overall performance;
2. however AMD is still far behind Intel in the notebook market, and totally out of the picture in the "thin and light" segment - that should become an important target for AMD!!!
3. more important AMD seems far behind Intel in the 65 nm transition - and without that 2006 can be a bad year for AMD;
4. the problem is not so much the speed gain on 65 nm but more on the L2 cache (which remains far behind Intel) and MOST IMPORTANT OF ALL on the PRICE - AMD is slowly giving up the most important weapon they had against Intel and without some cuts on price for the X2 line AMD might seriously loose market share to Intel in 2006 !!!
Not so sure on your final piece of cynicism, in that it appears that power (and therefore battery life) are the new target, rather than speed. Of course that's trying to change the rules once you've lost the battle (interesting to note that all the console makers have gone Power rather than either flavor of x86).
I also thought the line was that while Yonah is based on Pentium-M, Merom/Conroe is supposed to be the genuine 'next gen' CPU - i.e. a full re-design, from a different team (including some of the old DEC Alpha team) - and supposedly the most significant redesign since they launched NetBurst (2000). That's the story anyway, I'll reserve judgement until they deliver.
What that does say to me is that Yonah is a rush-job - a me-too product they've had to bring to market.
The Turion 64 ML-40 is 1/2 the price of the Pentium 780, around 350-400 dollars more (via froogle and pricewatch) The core-due also has 2 megs cache per core of cache.
Also where is benchmarks and battery life when each cpu is clocked at its lowest power saving? If you are going to do a benchmark for batterylife, how about actually doing the most important tests..
And, AMD laptops are opendesign to OEM vendors, so they can use any hardware and save money. Intel wants to control this to more of a degree for the centrino name, thus higher costs.
Now, I'm not bashing Intel, I cant wait for the 900 chips with dual core and virtualization. But that article could use some more info and less "Intels new chip is awesome compared to Old hardware on different platforms, blah chipset etc.."
No, it's a case of "The Nuts and Volts of News for Nerds"
So you are implying yourself that you never heard of mobile computing? You should try it, it's pretty fancy, you can do the same thing as on this big box you have in your room, but then without any cords! Amazing!
molmod.com - computing tips from a molecular modeling
I do remember, Intel lists only 75% of the TDP while AMD tells the truth and lists the peak value.
Quote taken from: http://www.businessweek.com/the_thread/techbeat/a
and MOST IMPORTANT OF ALL on the PRICE - AMD is slowly giving up the most important weapon they had against Intel and without some cuts on price for the X2 line AMD might seriously loose market share to Intel in 2006 !!!
Real world experience with marketing demonstrates that there are much more important things than 'price' in selecting a product. Wally World puts their cheapest "microwave" in the main trafficways. Mesmerized Customer says, "hey, good idea, I could use a new 'microwave', and damn, this thing's dirt cheap. But it's probably a POS, so I wonder what else they have..." Then they go into the isle and pick out a more expensive microwave, which has a significantly higher profit margin for WW, which is probably cheaper elsewhere (source: Frontline documentary on WalMart).
If price is all you push, your company will end up like General Motors ("We just lowered the sticker price on EVERYTHING!"), mismanaged into the ground, and have to give away your product at a loss...
Learn the rules so you know how to break them properly.
www.teslabox.com
I'm not convinced that virtualization is going to be such a big deal for either company.
:-), I don't think anyone is really going to care.
On the server side, most organizations have no need for it. Computers are cheap, and it's lower risk to run separate hardware. Those that are savvy enough to make good use of it tend to use every last bit of their hardware, so they have nothing left to spend on virtualized instances of other stuff anyway.
In the home market (and, FWIW, I don't consider the Slashdot crowd the "home market"
Virtualization has been around for a long time on much larger machines. Outside of the IBM mainframe world, you don't really see a whole lot of it in the grand scheme of things -- and that's on machines with six- and seven-figure price tags, where it would really make sense.
--S
-- sigs cause cancer.
I never knew that having all the facts could be this simple! Thanks, Slashdot!
--S
-- sigs cause cancer.
I switched from Intel years ago.
And WHY did I switch from Intel? WHY did I choose to go with a processor that, at that time, long ago, was not known for being as rock solid as Intel?
Because it was about $100 less than the same chip from you that ran at the same speed.
It's quite simple Intel. If you want to gain back marketshare from AMD, all you have to do is sell your processors for less than they sell theirs for. I don't care if you make the fastest processors or the pocessors with the most features. All I care about is getting the most bang for my buck. I never buy the fastest processor available, because it is a waste of money. I weigh speed vs. price, and choose the processor which makes the most sense in that context. And gueess what? It's ALWAYS AMD, because pound for pound, their processors cost less for the same computational power.
The same goes for the motherboards Intel. The motherboards for your chips are always more expensive as well. I take BOTH chip price and motherboard price into account when upgrading, and you're alays more expensive.
I really don't care if you licensed music from the Blue Man Group. The commercials were cute, but they're not going to make me change my mind about what I want to buy. I already know your name. Who are you trying to convince anyway? Joe Consumer who doesn't know anything about PC's buys them pre built from Best Buy. He doesn't know or care what processor is in it. So who is it you are trying to lure in with a cool image? A few hardcore people with a lot of disposable income who are already loyal to you?
was cheaper
a lot of the time, that's how you tell who's losing
Don't forget the Corporate Enviroment, the Financial Institution I work for, had 0 (Zero) AMD CPUs in its 500,000+ desktops or it's 250,000+ Laptops
I lookd at the Catalog and see the offerings that HP, Dell, IBM/Lenovo and Toshiba sell us and see that they are not offering us anything that has an AMD in it, well there are 3 or 4 server models from HP with opterons, but as for Corp. Desktops/Mobiles, 0 (Zero) is the offering
isn't the reason that Dell doesn't carry AMD is because they can't keep up with Dell's demand. it's great that AMD is growing but have they expand their capacity to keep up with the growing demand?
Intel=Nintendo.
Sony=AMD
Intel rules the portable markets, just as Nintendo does.
AMD rules the desktop markets, just as Sony rules the consoles.
Who said quantum computing was years away?
Might be a bit off topic but the name 'AMD' is not very catchy or impressive. I would change the name if I were running AMD.
How about something like "AdvanceChip" or "PowerChip" or something like that instead?
He who knows best knows how little he knows. - Thomas Jefferson
Inside Intel - Paul Otellini's plan will send the chipmaker into uncharted territory. And founder Andy Grove applauds the shift
In short, Intel has found themselves unable to keep up with AMD on the technology side and is now rebranding its products as "platforms" and elevating the role of marketing to be as important as engineering within the company. I've always felt AMD has been terrible at marketing, remaining largely an unknown to the general home user populace.
rooooar
AMD must be doing something right, there growing faster than intel, and they don't advertise, intel advertises as much as mcdonalds, AND intel got the mac contract. i personally think AMD is a much better brand.
I know that every time I read any article that mentions AMD v Intel that there will be people from the AMD side that come on and say there is not a single compeling Intel product and no reason to have a non-AMD processor in anything. I saw similar zealotry in the comments to anandtech.com's review of the processor - and they are a pretty unbiased and trustworthy source. Here is what they had to say...
... We continue to see that the Core Duo can offer, clock for clock, overall performance identical to that of AMD's Athlon 64 X2 - without the use of an on-die memory controller." And it accomlishes this with power consumption that is along the same lines at the previous generation high-end Pentium M chips.
"Our initial analysis still holds true, that for a notebook processor, the Core Duo will be nothing short of amazing for professionals. Looking at the performance improvements offered everywhere from media encoding to 3D rendering, you're going to be able to do a lot more on your notebook than you originally thought possible (without resorting to a 12-pound desktop replacement). In the past, power users on the go had to sacrifice mobility for CPU power, but with the Core Duo, that is no longer the case
I would think that as technology enthusiasts that we would be able to give credit where credit is due and recognize that, at the moment, Intel has a better processor lineup for laptops and AMD has a better line for desktops and servers - that it is possible for each to have strenghts and weaknesses as their produts evolve and change in different ways. Keeping an open mind and an up-to-date understanding of the strengths and weaknesses of each helps us to choose the right tool for the job and the budget.
I just can't believe how many are unwilling to concede even one success for Intel in their belief AMD is always better... The competition is helping all of us in spurring on better products and prices and the variety of options allows us to choose the right tool for the job.
So.... It's like listening to Cheese - Outside?
Quibble: P3 and P-M are 6-th generation (PPro-based chips). Athlon/Opteron are 7-th generation (K7-based chips), as is the P4 (32-bit or 64-bit). Conroe/Merom will be Intel's first true 8th generation chip. AMD doesn't have an 8th generation chip yet, though a new core is planned for '07.
A deep unwavering belief is a sure sign you're missing something...
So what? Every product could be said to be a rush job they had to bring to market. It only matters if the product is desirable, and Yonah appears to be.
Every next generation promises to be better than the last. Now we hear that Intel's processor 6 months from now will be better than today's. You're surprised by that? It suddenly makes today's part a me-too rush job?
Firstly, not such an insightful article as it's obviously biased. Sure it helps us learn more about AMD who rightfully deserve more of a market share than they have, but at the same time it also plays to the ears of the parent poster to make unsubstantiated comments like this and be modded +5 for it! Sure the AMD desktop and server CPU's are better than Intel's offerings in this space, but if you take a look at the Pentium M and the new Core benchmarks, you'll see that they outperform anything else in the mobile market at the moment!
They only held of 64-bit extensions for Core because of power concerns. The Core processors already have virtualization technology and 64-bit will be coming with the new Merom technology later in the year.
This slides neatly offtopic.
Does anyone else figure those PSP commercials are Sony's way of targeting the retarded market? I've seen some terrible ads before, but these manage to make me want to not buy the product.
Slashdot - where whining about luck is the new way to make the world you want.
Comment removed based on user account deletion
You forgot some!
7. Intel sucks! Intel has kittens for breakfast! AMD owns!
8. AMD sucks! AMD has a secret contract with Satan! Go Intel!
"73% of quotes on the Internet are made up" -Ben Franklin
AMD is fully planning on transitioning to 65nm in 06. Also, many fanboys are speculating (hoping?) that prices will decrease with the new socket (M2) and the opening of Fab 36 in the next few months.
Of course they're not quite the same as an octuple core AMD64; they're fairly simple in-order-execution UltraSparc cores depending on chip-multithreading to keep throughput up on highly concurrent workloads, which is somewhat different from the rather complex deeply pipelined out-of-order cores with good single-process throughput we're used to.
Itanium, the next gen 64 bit chip from intel, for all its promise has ended up a huge commercial failure.
As you probably can imagine, intel put alot of R&D resources into this project at the expense of x86 development.
Thats probably the real reason they've fallen behind on the x86 x86-64 chips is because they thought everyone would be using itanium PCs by now. AMD just kept the focus on developing x86 chips.
It was a mistake by intel, but at least they tried to get us off x86. If intel can't do it, I don't think anyone can. which is sad.
AMD should match Core Duo (reminds me of a superhero called Duo Damsel) by converting all their future names to common Esperanto words and terms. This way they can be Core Du, and look even trendier than Intel.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
Not that I would be opposed to it, but I really don't. Their last decent architecture was the Pentium 3. After that, they fell into the "Longer pipeline, higher MHz" trap that they're still in. What did they try to do to pull themselves out of that? Hyperthreading. Well, that didn't turn out to be much use. So what did they try after that? Another P4 variant that did even *less* per clock cycle. That didn't help much, so what did they do? They added more cache.
All in all, once they fell into the MHz pit, they haven't been able to come up with anything really useful to pull themselves out with - all they're doing is trying the same clubs that they've tried over and over. But wait - the Itanium! Another idea that didn't go very well.
In fact, the only thing they've got that still makes their chips worthwhile is the fact that SSE1/2/3 instructions are pretty speedy on them - so if you're encoding video, chances are that Intel will be a better offering than AMD. For most other types of work, however, AMD gives you more performance for less money.
Intel sure hasn't come up with anything good for quite some time. There's still a chance that they could come up with something decent, but they're really going to have to change their mindset if they want to turn themselves around.
Oh, you're not stuck, you're just unable to let go of the onion rings.
Wincing and engaging in extreme histrionics, Kirk, on a time-travel mission and in the AMD executive conference room, emotes, "TELL me EXacTLY hoW iTEL crossed the RENdering BArieR AND redistRIBUted that inCUbus MYOOvie of MINE. THIS time in ESpeRANT-o..."
(Oops, he might pass another gallstone..."
word image: astute
Previously: "Linux... Toward the Sunrise..." Now: "Linux... Toward the-- No, now, part of Every Sunrise"
Um, why would Intel think we'd be using Itanium by now? If they'd release a good performing Itanium chip that was actually priced competitively with x86, I'm sure there'd be a lot of takers. The basic problem with Itanium (and most RISCS, btw) is that you cannot make one at a workstation pricepoint that competes with x86 in terms of performance. Without being able to offer that, how do they expect people to switch?
A deep unwavering belief is a sure sign you're missing something...
I'm of the camp that says that the recent AMD Turion's are relatively close to a Pentium M in battery life, the #1 feature in a laptop for many people, including me. Six months ago, before Turions hit retail, Intel Pentium M's were clearly better than anything from AMD in the mobile space that was readily available for purchase -- which, IMHO, is why Apple choose Intel over AMD.
It's important to remember, though, that the Celeron M's are very hamstrung in power consumption compared to their Pentium brothers. They don't scale their CPU frequency -- so my cheap Celeron M laptop only gets 1.5-2.5 hours of battery life, whereas, a Pentium M chip might get me an extra hour or two (for light office / web work) of battery time.
Compare a Turion to a Celeron M and I'm pretty sure that the AMD chip will be miles ahead on power use, and at a similar price.
And if you claim that a Conroe/Merom is more advanced than a modern Opteron, you REALLY need to look at your processor architectures. It ain't - it's pulling even. Barely.
Generational effects I'm aware of: /P4): uOps-based architecture
3rd (386): virtual memory
4th (486): pipeline
5th (Pentium): superscalar (multiple execution units)
6th (PPro, K6): out-of-order execution
7th (Athlon, P3 (PM)
8th (Opteron, Pentium D): 64-bit
A witty [sig] proves nothing. --Voltaire
The Opteron shows up as a K8, but its just a very straightforward 64-bit translation of the K7 core, plus an integrated memory controller. The P4 64-bit and 32-bit are almost identical internally as well, which is why I categorize both as 7th generation cores. There are no 64-bit P-Ms.
Conroe/Merom are in a way more advanced than a modern Opteron. As I said, the Opteron isn't a new core, its a 64-bit K7 with an IMC. It's a very advanced chip, but its not "modern", in the sense that the basic design is quite old. Merom is a completely new design, different from either the P4 or the P6, although it draws elements from both.
My organization of the generational effects is below. The differences in the definition of the 7th and 8th generation make the categorization easier to correlate with the historical progression of RISC architectures.
3rd-6th: Same.
7th: uOps-based architecture doesn't really fit, since both the PPro and the K6 did translation to uOps (RISC86 in K6 lingo). A better distinguishing factor would be that both are massively out-of-order chips, having many more instructions in flight than the P6 core. This classification is useful because it allows these x86 chips to be grouped with other massively OOO RISC chips, like the 21264.
8th: Merom falls here, because its the first 4-issue x86 chip, and the first one to heavily use instruction grouping as a means of increasing the real issue rate. The K8 uses groups to a limited extent, in that it combines LOAD + OP uOps into a single "double" instruction, but Merom will use the technique much more extensively.
A deep unwavering belief is a sure sign you're missing something...
Itanium isn't risk. It's closer to VLIW.
Intel and AMD actually did this with the Pentium Pro and K5, respectively (AMD actually beat Intel to market with with the first uOp-cored-x86-decoder CPU by a few months, using the technology they bought from NexGen - pity the K5 was a bit of a dog in actual performance).
I didn't say it was. My point was that Itanium, like the majority of other RISC architectures, have had limited success because they don't scale well down to smaller machines. Itanium is quite dependent on its huge L1 caches to offer competitive performance. Without those, its a very mediocre performer relative to consumer x86 chips. The same is true for Power5 or SPARC64 --- these chips offer very good performance partially because they can afford to have 400mm^2 dies with very expensive system interconnects. Strip those things away, by making the caches smaller and the interconnects simpler (eg: as IBM did with the PowerPC 970), and the chip that's left isn't that great relative to x86.
A deep unwavering belief is a sure sign you're missing something...
7th: accepted, that's what I get for trying to go from memory. Having pulled out my references... PPro through PIII/Xenon are P6 cores, 6th gen. P4 is NetBurst core, 7th gen; NetBurst is the deeper pipeline (higher inflight) and register renaming for major improvements, and some tweaks (deeper btb, adaptive prefetch, increasing ALUs from 5 to 7). Athlon matches the renaming, and beefed up the ALUs to go as fast without the deeper pipeline - yes, all 7th gen. (And now I've found my old notes that say 7th gen is massively OOO, d'oh!)
Re: 8th gen... going from 3-issue to 4-issue isn't going to do much. P6 can't even keep a 3-issue fully saturated; a 4-issue will show some improvement, but far below the 33% theoretical gain. (Yes, Intel claims they'll make 4 IPC under load. I don't believe them. I suspect 4 IPC under ideal optimized load, <2 IPC under real load. How can you make 4 IPC when real code averages a branch every 5 instrs?). And I don't see convoying as a good thing. Itaniums were VLIW, but the compilers still aren't caught up; Merom can't expect compiler support (most compilers now target 5th gen Pentiums, no one will compile for Merom) except for their publicity benchmarks, and HPC users. Convoying looks good in benchmarks, but terrible outside HPC environments. It won't help on the desktop OR the server.
Adding 64-bit is not a mere port. 100+ registers to support renaming, major ISA changes (doubling GP registers and adding 4-level page tables) - the gain from implementing those is astounding. Easily more than a 4-issue will show. Opterons are 8th gen, no doubt about it. P4 seems to smear across 7th and 8th (they had a lot of cores, though no major arch. changes). Merom is a good 8th gen core designed from the ground up; Opteron is a so-so 8th gen core designed from the ground up (first system effect); P4 is a 7th gen core with 8th gen features bolted on.
I would actually claim the real 8th-gen features are x86-64 (the ISA extensions), MATURE register renaming (source of Opteron's performance gain in 32-bit), and good power management (selectively disabling unused parts of the chip). Merom is better than Opteron at these, but evolutionarily so.
A witty [sig] proves nothing. --Voltaire
Unfortuneately lower speed can only last so long. Amd stuck with outside processor "HYPER TRANSPORT" that can't take much more than DDR 400, basic work in processor (before cache is even involved) can now reach much higher speeds than AMD works with.Higher speed pci express with 40 or 96 lanes leaves AMD confused & in dust bin. AMD has zram & other exciting pumps, besides going DDR-2, yet hyper transport bus won't do in its present speed, gAME cARDS ARE ABOUT TO GO ABOVE HYPER TRANSPORT SPEEDS.IT JUST WON'T WORK WITH THAT "CLOG". http://www.geocities.com/tsvondrashekmd/WASHINGTON .html
Signed:PHYSICIAN THOMAS STEWART VON DRASHEK M.D.
WINDOWS XP Service Pack -X- 396 mb. http://www.geocities.com/tsvondrashekmd/WASHINGTO
Adding 64-bit is not a mere port. 100+ registers to support renaming, major ISA changes (doubling GP registers and adding 4-level page tables) - the gain from implementing those is astounding. Easily more than a 4-issue will show.
The reason I consider 4-issue (+ micro-ops fusion) a bigger change than 64-bit is because the former is a much more fundamental change in the internal CPU design. The issue width is a fundamental property of a particular core design. Increasing the issue with exponentially increases complexity, and causes rippling changes throughout the entire CPU. The same thing for micro-ops fusion. Once the basic unit of instruction scheduling inside the CPU goes from being an instruction (as in the P4), to a batch of instructions (as in Merom), the nature of the core is changed.
Opteron is mostly a wider Athlon with an IMC. The Opteron has the same execution units as the Athlon, in the same layout, they are just wider. It handles register renaming using the same mechanism as the Athlon, the register file is just larger. It handles branch prediction the same way as the Athlon, the history buffer is just larger. It handles instruction scheduling and queing the same way as the Athlon, the queue is just deeper. Merom is a completely different core from the P-M. Changing the issue width changes how your handle dispatching instructions, which changes your your scheduling mechanism, which changes the layout of your execution units, etc.
A deep unwavering belief is a sure sign you're missing something...
The last time 4-issue was used was in the AMD K5. It offered no significant performance benefit, and it had lots of problems. Design and production problems held it back, and AMD released the K5-133 a full year later than the Pentium 133, even though it ran at 100 MHz. The K5 tends to get very hot, also because it requires 3.52 V core voltage. It was designed for the 3.3 V Socket 5 pentium socket, but stability issues forced AMD to up the core voltage. If your motherboard didn't offer 3.5 V next to the standard 3.3 V, you were out of luck. The K5 was never popular due to stability issues and poor overclock potential. A 100 MHz K5 rarely runs reliably at 110 MHz, if it does it produces a massive amount of heat.
After this fiasco both Intel and AMD abandoned the idea of 4-issue and continued with 3-issue ever since.