Intel’s Next Generation Processor: 'The Biggest Leap Since the Pentium Pro'

At a press gathering this morning, Pat Gelsinger, senior vice president and general manager of Intel's (INTC) Digital Enterprise Group, revealed details about a next generation processor, Nehalem, which brings an entirely new microarchitecture (and motherboards), and will include onboard memory and graphics controllers.

Nehalem-based desktop, server and mobile processors and systems are slated to be available in 2008. The Nehalem effort is part of Intel's strategy to step up its chip cadence.

"The new systems architecture is a major shift is system architecture, and a stunning value proposition," Gelsinger beamed. "It's the biggest leap since the Pentium Pro [which reached the market in 1996]." It's also a big shift for those who create memory and graphic controllers, such as AMD, which just acquired ATI.

"We will integrate the memory controller on the die, with both a buffered and native version," Gelsinger explained. For many years, AMD's claim to fame has been its integrated memory controller, which helps to reduce latency. "We have the best memory hierarchy on planet today," Gelsinger said in response to a question about AMD's HyperTransport processor design approach. "The best cache is more important than an integrated memory controller, which is why Intel wins on benchmarks."

On a technical front, each core utilizes two threads and the chips are designed specifically for the 45-nanometer process, unlike the Penryn family due later this year that was a bridge between 65 and 45 nanometer. Nehalem designs are underway with eight cores, and two thread each, Gelsinger said, and the processors will exist in the same thermal envelope as previous generations.

For software developers, Gelsinger said, "Nehalem system server applications are very throughput-oriented already and will take full advantage of the simultaneous multithreading. On client systems, we have been on a multithreaded focus since 2000, when we launched hyperthreading. Now we are reaping the benefits and seeing good parallelism in games, media and even in things like Microsoft Office 2007, but it is still heavy lifting to move the software community together to take advantage of threading."

Gelsinger also outlined the Penryn family of processors, due in the second half of this year, that are based on the 45-nanometer Hi-k process technology and high-k + metal gate transistor design. In plain English, the Penryn family, which will include six processors (dual and quad core Core and Xeon systems for server, desktop and mobile) means chips that are faster and more energy efficient.

The 45-nanometer processors will have 820 million transistors, and the dual core die size will be 25 percent smaller than the 65-nanometer equivalent.

Comments

[Tech:]

should have been titled :
Intel biggest AMD follow-up! ;o)
I. Integrated memory controller (an AMD first for this family of processor)
II. Hypertransport like communication channel (AMd introduced and supported by third parties years ago!)
III. Integrated graphic controller (already announced by AMD during ATI acquisition)
IV. 8 cores (AMD Native four cores (Barcelona) will arrive months in advance of Intel!)
45nm , 32nm, 22nm all those technologies will be available to AMD through technology partnership with IBM, Sony,... (top world R&D labs!)
Intel need absolutely those finer technologies (and spend alone the cost associated with the development and new production tools ), because they must use 65nm techno to fight 90nm AMD better architectures! and so will need 45nm to reach performance achieved by actual AMD 65nm production!

2007 Mar 29 09:04 AM

[Wannabe Maven:]

Sure, AMD invented the microprocessor and created the entire electronics industry. All Intel did was make some money, but I thought that's why we visit this site.

2007 Mar 29 10:05 AM

[Tech:]

you may have noticed the smiley in my message ;o) be cool!
and I never said what you mention!
for sure, Intel made an awfull lot of money! (who paid ?)
it is partly why we should not encourage too monopolistic positions
and the press should recognize that Intel endorsing some (at a time criticised!) AMD technical solutions finally recognize that even with a lot smaller budget , innovative companies can bring something to the party.
that's all.

2007 Mar 29 01:28 PM

[eduardo gon...:]

I agree with you AMD gave us consumers great value for our money releasing superior processors at lower prices. They introduced much needed competition in the market. But they slept in the laurels and now Intel reclaimed the lead.

Lets not let our simpathy for the "underdog" cloud our judgement: Intel kept on top offering inferior products with worse design and about the same litho tech, thanks to its financial/sales/market... muscle. Now they have the best design and a 12 month lead in microlithography process. Even if Barcelona was 40% faster clock by clock than Clovertown as they say (I suspect this is the case in a couple handpicked scenarios) ... it will run at 2.8 Ghz. This Clovertown thingy can already be clocked high enough to deny any speed advantage.

The best we can hope is AMD to survive this round to deliver a nice surprise. Unfortunately, even that possibility is not completely clear.

2007 Mar 29 05:59 PM

[Tech:]

Hi Eduardo,
I agree partly with you , but it is so easy to hide behind the x00 pounds gorilla, that I prefer taking care and bet on the small ;o
As I said microlithography drawing size alone is not all. and it was my first remark, Why would intel finally choose some AMD innovations if they could win only with lithography size?
I was criticizing the title, and "Market Watch" (TM) titled yesterday:
"Intel takes aim at AMD chip architecture"
which was exactly my point!
I am not persuaded that Intel has a 12 month lead in process and again, AMD partnering with IBM, SONY and I believe Toshiba alltogether may have similar R&D budget on process
concerning 2.8Ghz, with different architectures, Max Ghz is not the only metric.
I agree that with actual AMD communication we have not enough visibility.
but Intel has so much money in the bank to fund R&D that it may be impossible for AMD to show their future too early.
We can only hope that AMD if necessary may now take advantage of Intel innovations to continue to improve the "value for money" factor for the end customer or propose "new of its own!" innovations ;o))

2007 Mar 30 09:11 AM