• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

intel working on hammer clone

Adul

Adul

Elite Member

http://www.geek.com/news/geeknews/2001dec/bch20011213009285.htm


NEWS
Mad Mike over at The Inquirer has posted a blurb in which he now believes Intel is working on a Hammer clone. If true, this will not mark the first time Intel has taken cue's from AMD. Back in the 1990s AMD developed their powerful 3DNow! extensions for the x86 instruction set, and did so long before Intel ever dreamt of it. However, Intel (with its marketing clout) took the idea, dubbed it MMX, put a few dancing bunny suits on TV and ... well, we all know the rest of that story. Still, if true this *WILL* be the first time Intel has copied an entire CPU architecture from AMD. Honestly, I think it only makes sense. C't reported that a mere 500 Itanium systems have been sold (with around 3,000 being made as demo or introduction systems). By any sense of the word Itanium is a failure. The project cost Intel billions to produce, it's significantly behind schedule (years), running around 1/2 the clock speed it was originally promised, still on 0.18 micron technology, using a pitiful 266 MHz FSB ... and now we find out they've only sold a few hundred of them? C'mon! Intel is no dummy. They are fiscally responsible and would go to great lengths to ensure they don't lose market share when they don't have to. Craig Barrett (Intel's infamous CEO--who owns a horse named Pentium, by the way) is on record stating that "only the paranoid survive." So, in light of the above evidence I must concur fully with Mike's assessment of the situation. Intel *IS* working on a compatible chip to x86-64. They *ARE* hedging their bets against some kind of (almost inevitable) cancelation of the Itanium project. And we *WILL* see a compatible Hammer chip from Intel sometime in 2003. On a personal note, it's nice to see the tail wagging the dog for a change. 🙂
 
oh man. only 500 units? yikes, that's really terrible. if i was intel, i'd be thinking about doing something drastic too. AMD will kill them unless they do something fast... sounds like a resonable choice on their part.
 
Sounds like BS to me.

The Itanic isn't meant to be a big seller, it's intended as an introduction to IA-64 more than anything else.

And AMD has a LOOONG way to go before being in any kind of position to "kill" Intel, Intel still covers nearly 100% of the high margin markets(that is servers and workstations).

I'll believe it when I hear it from some credible source.
 
< it's intended as an introduction to IA-64 more than anything else. >

LOL, you keep believing that...hahahaha

Introductions usually don't go this long without the product actually working like it was intended...


I agree amd wont kill Intel ever, but the closer they make it to a even market share the more that will feul competition and keep prices low for the consumer. intel has lived high on the hog for too many years, and now it just got interesting...
 
Duvie, McKinely is what will be the first "serious" product in the IA-64 line.

I doubt anyone will deny that IA-64 is horribly late though.
 
Sometimes in the computer industry when it takes this long and so much money to develop something that is not working near half of what was intended you need to "cut your losse" maybe revisit later as technological advancements warrant...
 
They only way I can see Intel producing a x86-64 compatible chip is if AMD can get most of the major software developers to write x86-64 software only and you and I both know AMD has neither the market share, clout or money to make that happen.
 
They only way I can see Intel producing a x86-64 compatible chip is if AMD can get most of the major software developers to write x86-64 software only and you and I both know AMD has neither the market share, clout or money to make that happen.

The issue here is that AMD doesn't need as much software support as Intel does with EPIC. What do you think Intel is going to do, watch as AMD introduces a fast 64-bit processor with even faster 32-bit performance into the desktop, workstation and mobile markets? Of course not. Unless Intel can find a way to get IA-64 into the desktop/workstation/mobile market (that seems far off) it's not all that weird to hear that Intel may be copying a form of x86-64. The question remains, how is Intel going to compete with Hammer with all the extra features it brings? Surely a P4 doesn't stand a chance against Hammer (feature wise at the very least, and likely performance wise).
 
Sales of the Itanium is a near non-issue. The real beginning of the IA64 line is the Mckinely, that's when Intel needs to see purchasers lining up. Itanium is little mopre then a pilot of the IA64 technology, Mckinely is the true beginning. And by all indications Mckinely is looking to be vastly superior to the quite unimpressive Itanium.... whether it sees enough software support and performs decently enough is yet to be seen.

It wouldnt surprise me if they were planning a Hammer clone on the off-chance that IA64 fails and X86-64 takes off. At the moment though, outside of the Linux environment software support for X86-64 is virtually nil.... though that's not quite as big of an issue as faces IA64 as X86-64 always has the fallback to decent performing 32bit performance with the obvious software support.

It wouldnt surpise me if their planning it in some small corner of Intel, but I sincerely doubt it's a high priority nor do I think Intel need be overly much concerned about X86-64 unless MS and other such developers start paying a lot more attention to the 64bit component of it.

IA64 is clearly an incredibly expensive total failure thus far.... but it's still very early for IA64, and the Itanium is but a place shadow of future IA64 processors.
 
I don't know much about this debate, but from what I can tell from the little bit i've read on the subject. I do think that x86-64 still supporting 32-bit apps has a HUGE leg up on the Intel platform which only supports 64-bit apps. I guess is depends most on what software is developed.
 


<< in the 1990s AMD developed their powerful 3DNow! extensions for the x86 instruction set, and did so long before Intel ever dreamt of it. However, Intel (with its marketing clout) took the idea, dubbed it MMX, put a few dancing bunny suits on TV and ... well, we all know the rest of that story >>

*sigh*....where do people get off writing this stuff? First of all, neither Intel nor AMD developed the idea of SIMD computing, and especially not with 3DNow, MMX, or SSE. SIMD computing dates back to the early 1970s, most notibly with Univ. of Illinois' Illiac IV in 1972. Cray did the groundbreaking work in vector computing in 1976. Secondly, MMX was introduced with the Pentium MMX in January of 1997, 3DNow was introduced with the K6-2 in May of 1998. Thirdly, the P3 was released in February of 1999...this timescale is *FAR* too short for Intel to merely "copy" 3DNow and dub it SSE.



<< So, in light of the above evidence I must concur fully with Mike's assessment of the situation. Intel *IS* working on a compatible chip to x86-64. They *ARE* hedging their bets against some kind of (almost inevitable) cancelation of the Itanium project. And we *WILL* see a compatible Hammer chip from Intel sometime in 2003 >>

All we have now is complete speculation from the Inquirer, hardly a respectable publication. The Inquirer's only line of "proof":

<< Intel has a secret skunk works in Beaverton, US, developing X86-64 compatible processors, the INQUIRER believes. >>


I hardly see how this proves that Intel *IS* working on an x86-64 CPU.
 
Maybe Intel is, maybe it's not, time will tell. However, it should be noted that in the late 80's Intel developed a 32bit processor using much of the same technology that the Itanium uses. That processor also needed it's own OS/software and was released with great expectation. It was a dismal failure. This doesn't mean that Itanium/McKinley will fail, but it does show that Intel has an uphill battle and won't just dominate because of superior(?)design or the fact that they happen to be Intel.
 


<< oh man. only 500 units? yikes, that's really terrible. if i was intel, i'd be thinking about doing something drastic too. AMD will kill them unless they do something fast... sounds like a resonable choice on their part. >>



I don't want AMD to kill intel. If they do, AMD would have a monopoly and they'd be able to make slow chips and charge $500 for them (like what intel is doing now.)
 
Thanks guys. 🙂

What a lot of people don't realize is that many of the revolutionary concepts in computer architecture; pipelining, superscaling, RISC computing, SIMD, out-of-order processing, register renaming, etc, etc, etc; were first researched at universities, and first implemented in high-end mainframes in the 1960s and 1970s (the exception being RISC, which was developed by Patterson and Hennessey of Berkeley in the early 80s, and was quickly implemented for workstation and server CPUs). It takes a long time for these ideas to be implemented in microprocessors....the 486 was the first pipelined Intel CPU, yet pipelining was first developed for the IBM Stretch project in (IIRC) the late 1950s.

I'm not trying to bemoan what Intel, AMD, Motorola, etc, do....implementing these ideas in microprocessors is no small feat. But the reason the MPU manufacturers come out with CPUs using similar concepts at the same time is not because they're copying off of each other, but because they read the same publications from university researchers 🙂. Many CPUs will be released soon using chip-level multiprocessing and simultaneous multithreading, simply because much of the research in these areas was published in 1997 and 1998.

BTW, I didn't mean to go a little off-topic, Adul....but what that guy at Geek.com wrote was too stupid to pass up. 🙂
 
sandorski - i960?


and i don't see why intel couldn't be looking into x86-64. in fact, they probably should just to keep up to date with the competition. assigning some engineers to look into the relative merits of such a design does not mean that they will market a product based on it.
 


<< However, it should be noted that in the late 80's Intel developed a 32bit processor using much of the same technology that the Itanium uses. That processor also needed it's own OS/software and was released with great expectation. It was a dismal failure >>

Yep, that was the i860. Though it should be noted it was a RISC processor, not VLIW like Itanium. It had a few quirks that set it apart from other CPUs; it had a dual-instruction mode turned on with a control bit, and pipelined load/stores. These features were difficult to use, and Intel made a fatal mistake of not coinciding the release of the i860 with software and compiler support. Likewise, Itanium depends a lot on compiler robustness, but this time around Intel and HP aren't making the same mistake (hopefully), and thus compiler maturity has been in development for a long time and will continue for a long time...though it is still up for debate whether the compiler can schedule integer operations effectively on a VLIW processor.
 
I thought we determined in this thread that geek.com hasn't a clue what's going on.

And I thought we determined in this thread that theinquirer article was full of crap.

Not only that but (and I don't want to get into this too much)... That number of Itanium systems sold that they quoted is incredibly inaccurate. And there is no plant (secret skunky or not) in Beaverton. There aren't even any Intel offices in Beaverton. I live in Beaverton, as do a few other AT members. Add that with the things that Sohcan has already pointed out, and you get a steaming pile of dogdoo.

C'mon people... Just because some schmuck posts it on the 'net, doesn't make it a fact. Especially if it comes from theinquirer or geek.com. And now we have one place quoting the other? 😕
 


<< They only way I can see Intel producing a x86-64 compatible chip is if AMD can get most of the major software developers to write x86-64 software only and you and I both know AMD has neither the market share, clout or money to make that happen.

The issue here is that AMD doesn't need as much software support as Intel does with EPIC. What do you think Intel is going to do, watch as AMD introduces a fast 64-bit processor with even faster 32-bit performance into the desktop, workstation and mobile markets? Of course not. Unless Intel can find a way to get IA-64 into the desktop/workstation/mobile market (that seems far off) it's not all that weird to hear that Intel may be copying a form of x86-64. The question remains, how is Intel going to compete with Hammer with all the extra features it brings? Surely a P4 doesn't stand a chance against Hammer (feature wise at the very least, and likely performance wise). >>



Intel doesn't have to compete with the Hammer they just have to nullify x86-64. And they can do that buy releasing a chip that has as good or better x86-32 performance then the Hammer and have it on the market before the Hammer. If Intel can do that then AMD will need to get software developers to write programs that run only on there chips and I don't think AMD will be able to that.
 
Not only that but (and I don't want to get into this too much)... That number of Itanium systems sold that they quoted is incredibly inaccurate.

Well Cnet would disagree with you Wingz:

In the third quarter--the first full quarter of Itanium sales--manufacturers sold just $13.7 million worth of servers containing the chip, which comes to less than 500 servers, according to market researcher IDC.

The Inquirer concurs with Cnet. I don't know if I should be asking you this, since you said "I don't want to get into this too much", but nevertheless, I'd like to know the reasoning behind your statement.
 
And they can do that buy releasing a chip that has as good or better x86-32 performance then the Hammer and have it on the market before the Hammer

Intel is not going to have a processor that supports something similar to x86-64 out on the market before Claw/SledgeHammer is out. That would be implying that Intel has been working on a type of x86-64 microprocessor for years now, and I don't think that's the case.

Therefore, I think Hammer has definite advantage in time-to-market of Hammer compared to if Intel decided to introduce something similar to Hammer software wise.
 
<<<< <I>in the 1990s AMD developed their powerful 3DNow! extensions for the x86 instruction set, and did so long before Intel ever dreamt of it. However, Intel (with its marketing clout) took the idea, dubbed it MMX, put a few dancing bunny suits on TV and ... well, we all know the rest of that story</I> >>

*sigh*....where do people get off writing this stuff? First of all, neither Intel nor AMD developed the idea of SIMD computing, and especially not with 3DNow, MMX, or SSE. SIMD computing dates back to the early 1970s, most notibly with Univ. of Illinois' Illiac IV in 1972. Cray did the groundbreaking work in vector computing in 1976. Secondly, MMX was introduced with the Pentium MMX in January of 1997, 3DNow was introduced with the K6-2 in May of 1998. Thirdly, the P3 was released in February of 1999...this timescale is *FAR* too short for Intel to merely "copy" 3DNow and dub it SSE.>>

aren't 3DNow and SSE build on top of MMX?
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top