Sign in with
Sign up | Sign in
Your question

Intel unveils new chip design to challenge AMD

Last response: in CPUs
Share
August 19, 2008 10:41:37 PM

Once again, another article which states Intel is pickin' up the rear.

AMDs losses are attributed to Intel's dirty pool tactics, of course.
When the courts rule on that, AMD won't have any difficulty dealing
with Intel.

I support AMD because I think 30 years is too long a time to be regurgitating the x86 architecture.

If Intel isn't going to give us something new, I believe AMD will.

The title of the below article is in contradiction with what the article reveals.
Intel isn't bringing anything new to the table, AMD beat them to it!

=

Intel unveils new chip design to challenge AMD

By JORDAN ROBERTSON – 34 minutes ago

SAN FRANCISCO (AP) — Intel Corp. cracked the lid Tuesday on a new chip design that is at once a big challenge to smaller rival Advanced Micro Devices Inc. and an admission that AMD nailed a key design feature before it slipped into a severe financial slump.

Intel, the world's largest computer chip maker, showed off the new blueprint, known as a microarchitecture, for its chips at a developers conference in San Francisco.

Though some of the details were already known, the design's formal unveiling represented another demonstration of Intel's advantage over AMD in cranking out new chip designs once every two years, a factor that helped send AMD's stock price down 5 percent in an overall down day for technology shares.

AMD has racked up nearly $5 billion in losses during the past 18 months and last month replaced Hector Ruiz, who had been running AMD for six years, with a new chief executive, Dirk Meyer.

The details of Intel's microprocessor architecture are always highly technical. But they're also closely watched because of the ubiquity of Intel's chips in personal computers and corporate servers.

One of the most significant changes was already known. Intel now plans to build a part called an integrated memory controller — which moves information between the microprocessor and the computer's memory — directly into the processor itself.

That's a key change because processors are asked to do more and more, and any lag in communication can seriously hurt performance. AMD has already been incorporating integrated memory controllers into its processors.


Because of that and other tweaks, Intel said its new design, which is code-named Nehalem, will triple the speed at which data can be written to memory or read back, compared to previous generations. Intel says Nehalem also will have nearly double the 3-D animation capabilities as past chips, and better utilize the multiple "cores," or processing engines, on each chip.

Chip makers are adding multiple cores to their chips, essentially jamming many separate processors onto the same slice of silicon, to make sure they're able to continue ramping up performance without running into overheating problems.

Intel said four-core Nehalem chips, which are due to be in production by the end of 2008 and will first target servers and desktop computers and later laptops, have the ability to turn individual cores on and off and can be programmed to boost the speed of active cores when the workload ramps up.

Intel shares fell 39 cents, or 1.6 percent, to $23.62. AMD shares fell 31 cents, or 5.3 percent, to $5.60.

http://ap.google.com/article/ALeqM5ha5ufZX6hs1eRF_jSR3r...
August 19, 2008 11:05:43 PM

Intel had it first! (they really did.... like how many years before AMD even thought of it....?)
August 19, 2008 11:23:01 PM

Are we talking about Timna? Did they ever manage to get that thing on the market? I can't remember.
Related resources
August 19, 2008 11:49:06 PM

piesquared said:
Are we talking about Timna? Did they ever manage to get that thing on the market? I can't remember.


No 386 generation.

Word, Playa.
August 20, 2008 12:54:58 AM

enigma067 said:
I support AMD because I think 30 years is too long a time to be regurgitating the x86 architecture.

If Intel isn't going to give us something new, I believe AMD will.

Something new? You mean like a completely new microarchitecture, wholly removed from the x86 32 bit extension of a 16 bit extension of an 8 bit architecture? Oh, that's right - Intel did. It's called IA-64. And while Intel was working on that, AMD chose to add 64-bit extensions onto the x86 architecture, guaranteeing you'd be using it for at least another 10 to 20 years.

Agreed, what AMD did was the clever thing to do (they couldn't afford to let the market move to IA-64, because they couldn't follow that route). But stop with the AMD is all things new and wonderful and "Intel isn't bringing anything new to the table" tripe.
August 20, 2008 1:10:50 AM

Troll value = 9.9
a c 123 à CPUs
a b À AMD
August 20, 2008 3:15:33 AM

sonoran said:
Something new? You mean like a completely new microarchitecture, wholly removed from the x86 32 bit extension of a 16 bit extension of an 8 bit architecture? Oh, that's right - Intel did. It's called IA-64. And while Intel was working on that, AMD chose to add 64-bit extensions onto the x86 architecture, guaranteeing you'd be using it for at least another 10 to 20 years.

Agreed, what AMD did was the clever thing to do (they couldn't afford to let the market move to IA-64, because they couldn't follow that route). But stop with the AMD is all things new and wonderful and "Intel isn't bringing anything new to the table" tripe.


this OP is known to be a troll or at least post useless BS. You are right on many counts here and Intel and HP teamed together to create IA-64 but thats the main problem. People didn't want to do a massive change, mainly the software makers.

Yes technically we can blame AMD for getting us stuck with x86 for a bit longer and I doubt AMD has any brilliant ideas on something to change it all.

Timna did make it to market but only to a few select vendors. After that it was cancelled and never seen again. And spud is right. Intel has had and been working on a IMC since the 386 which was in the 80s.

And to the OP, AMDs losses right now are of their own doing. They spent WAY too much on ATI (even though I like ATI better than nVidia and still buy them today) and did it at the worst point. They were dealing with Core 2 for over a year and instead of invest the money into R&D and FABs they decided to purchase ATI. Thats where the losses come from.

I will agree its a great idea but their timing was horrible. Plus add on the Barcy delays then it is released but with a bug that can affect servers so it was reaclled and all that just caused AMD to loose money in the most important sector, servers.
August 20, 2008 3:26:03 AM

enigma067, stop drinking the Kool-Aid.
a c 123 à CPUs
a b À AMD
August 20, 2008 3:32:33 AM

^I think he moved from the kool-aide to the "green" brownies.....
August 20, 2008 3:39:34 AM

Zorg said:
enigma067, stop drinking the Kool-Aid.


I often wonder what the dose of LSD they have mixed into that Kool-Aid.... Same for any person who is blindly loyal to a brand name (any brand name)

August 20, 2008 6:12:32 AM

Some people need to put the crack pipe down....not pointing any fingers here.....

Neh, there have been alternatives around to x86 for quite a while. Anyone remember 3DO? That gaming/Computer system was based off of a 32bit risc processor, and that was back in the early 90s. IBM's Power chips are all Risc 32 and 64 bit based. The Cell BE processor, I'm not quite sure what it's based off, haven't researched it much. In fact for a long time during the late 90s early 2000's many Computer analysts considered 32bit and 64 bit native RISC processors to be the thing of the future. But they never caught on due to the added difficulty and extra coding required for the reduced instruction set computer processors.

Just look at the things the PS2 could do with that 242Mhz 128bit RISC Emotion engine processor, that allowed it to keep up graphically and game play wise with the Xbox. Also consider the fact that there are really no computers in existence on the consumer level that could competently emulate that system through software, short of the ps3, and even that requires a bit of extra hardware.

Yes, Intel was the first to start researching the IMC, back in the day. But they never managed to produce a working one that wasn't hampered by one thing or another. You gotta remember that at the time AMD had access to all said IP and research due to the Cross license. The Tinma IMC wouldn't work properly due to issues with the translater hub used to communicate with the DDR SDRam, or was it to communicate with the RDRam(can't remember off the top of my head). But then again Tinma also never reached retail production either, maybe a few ES's but that was about it.

In order to move away from X86 it's going to take more than just Intel or AMD. It's going to need MS to move away from CISC based programming to something else. Seeing as how they're one of the main OS makers. Not to mention it would require every program, and game maker out there to move away from X86 CISC programming. That would be something that couldn't be forced. But, in the end, some kind of transition does need to happen.

For those who don't know CISC= Complex Instruction Set Computer= X86.
RISC = Reduced Instruction Set Computer= native 32bit and up.

RISC has been capable of Higher clock speeds and extremely high IPC compared clock for clock to x86. But, X86 processes more complex and longer instructions per clock cycle.
August 20, 2008 6:32:26 AM

Mathos, I'm not even sure you can call the current generation of X86 processors CISC. They don't processor stuff like the older CISC processors. They break everything down into uOPS. What are uOPS? simple instructions that can be done very fast. Just like RISC instructions. I would call the X86 processors today to be very very RISC like.

As for Timna. The IMC worked just fine. A friend of mine at Intel even has a Timna processor and motherboard. The whole issue is that the cost of RDRAM never came down. The Timna processors was designed for entry level systems but if the memory cost more than the processor and board combined it could not be used as an entry level system. Also, the issue with the MTH is correct. Intel could never get it to work correctly with the DDR memory of that time.
August 20, 2008 7:03:03 AM

pausert20 said:
Mathos, I'm not even sure you can call the current generation of X86 processors CISC. They don't processor stuff like the older CISC processors. They break everything down into uOPS. What are uOPS? simple instructions that can be done very fast. Just like RISC instructions. I would call the X86 processors today to be very very RISC like.

As for Timna. The IMC worked just fine. A friend of mine at Intel even has a Timna processor and motherboard. The whole issue is that the cost of RDRAM never came down. The Timna processors was designed for entry level systems but if the memory cost more than the processor and board combined it could not be used as an entry level system. Also, the issue with the MTH is correct. Intel could never get it to work correctly with the DDR memory of that time.


Yeah, very true. That is part of what x87 out of order execution is based off of. I was reading an article a while back that was posted back in......02 I think it was. (couldn't remember what RISC stood for so thats how I found it). That same article also said something about how x86 CISC and RISC was starting to intertwine in some ways, and become very similar. If you look at it you could almost say the k8 was more RISC based while the P4 was more CISC, as the k8 had to break up 128kbit ops into 2 small er 64kb ops. Not necisarally a true anology but you get the point. ANd I can't spell right now, I'm tired, sorry bout all the errors.

And to add a bit to the my previous post, actually there aren't too many differences between IA64 and x86-64, other than x86-64 supports a couple more op commands.
a c 123 à CPUs
a b À AMD
August 20, 2008 1:44:43 PM

Mathos said:
Yes, Intel was the first to start researching the IMC, back in the day. But they never managed to produce a working one that wasn't hampered by one thing or another. You gotta remember that at the time AMD had access to all said IP and research due to the Cross license. The Tinma IMC wouldn't work properly due to issues with the translater hub used to communicate with the DDR SDRam, or was it to communicate with the RDRam(can't remember off the top of my head). But then again Tinma also never reached retail production either, maybe a few ES's but that was about it.


Timna started with a IMC on it and RDRAM, and worked great except the cost of it. And from what I remember RDRAM had the ability to be directly linked to the IMC in Timna.

Due to costs Intel redesigned it for the cheaper SDRAM (not sure if it was DDR or not) But since the SDRAM could not be directly linked to the IMC thus why the MHT was needed and THATs what ended up causing the problems with it.

Timna was a great idea and design but due to high costs of memory (damn RAMBUS) and so forth it got cancelled a bit before production.
August 20, 2008 1:59:06 PM

sonoran said:
Something new? You mean like a completely new microarchitecture, wholly removed from the x86 32 bit extension of a 16 bit extension of an 8 bit architecture? Oh, that's right - Intel did. It's called IA-64. And while Intel was working on that, AMD chose to add 64-bit extensions onto the x86 architecture, guaranteeing you'd be using it for at least another 10 to 20 years.

Agreed, what AMD did was the clever thing to do (they couldn't afford to let the market move to IA-64, because they couldn't follow that route). But stop with the AMD is all things new and wonderful and "Intel isn't bringing anything new to the table" tripe.


+1

QFT
August 20, 2008 4:47:03 PM

Ah yes, need to change what I said about something. IA-64, developed by HP and Intel did indeed come first, it was the architecture used for the Itanium 64 bit processors. Only problem being it couldn't emulate 32 bit x86 code to save it's life. Which means there was no real backwards compatibility for the architecture to be able to run things like Windows or any x86 based programs. AMD developed x86-64 changed the name to AMD64 as an alternative to IA64, which would allow perfect compatibility to current 32 bit x86 programs and future 64 bit code. Intel responded to AMD64 with EM64T, which was basically x86-64 adapted to the P4, basically x86-64 without the full instruction set, was missing a couple command ops that didn't work well with the netburst uarch.

!