Print 24 comment(s) - last by RandomUsername.. on Feb 6 at 2:42 PM

Intel says chip is fine, it only wants to add engineering enhancements

Intel is the biggest CPU maker in the world with AMD taking a distant second place. Intel is most known for its x86 CPUs, but the chipmaker also produces CPUs that are RISC-based. The company has announced delays in its next RISC-based processors.

Intel says that the chip, code-named Tukwila, is fine and that the delays aren’t caused by any issues with the processor itself. Intel says that the delays are to allow the addition of certain engineering enhancements.

The two most notable of these enhancements are compatibility with DDR3 memory; the original Itanium Tukwila design was compatible with DDR2 only, and an upgrade to allow the Tukwila CPU to use the same socket as future Itanium processors.

The future processors that Intel is talking about are code-named Poulson and Kittson. Not much is known about these two future processors other than the Poulson part will be built on a 32nm process and will have more than four cores per processor.

Tukwila is built on the 65nm process and has 2 billion transistors. The CPU will also have 30MB of on-die cache along with dual integrated memory controllers for balanced performance.

An Intel spokesman told InformationWeek, "The processor itself is fine, but Intel has made the decision to add some engineering enhancements."

The move to the same socket for the Tukwila CPU as the future parts is a good one for customers. That means that as upgrades to the faster parts are needed adding new Itanium processors will be as simple as removing the old CPU and plugging the new one in.

Experts say the changes are being added because the Tukwila CPU is slower in its original form than the IBM Power6 parts already on the market.

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

itanium refresh delayed?
By RamarC on 2/5/2009 2:37:33 PM , Rating: 3
so, how many 'titanic'ium shops will this affect? 6, 7, 10? :)

RE: itanium refresh delayed?
By gstrickler on 2/5/2009 2:50:59 PM , Rating: 2
Don't exaggerate, there are at least 25 Itanic users.

RE: itanium refresh delayed?
By mrubermonkey on 2/5/2009 3:50:22 PM , Rating: 5
I will buy an Itanium based system just to spite the both of you and make it 26 users.

RE: itanium refresh delayed?
By mrubermonkey on 2/5/2009 3:50:46 PM , Rating: 2

RE: itanium refresh delayed?
By overzealot on 2/5/2009 11:01:03 PM , Rating: 1
A what?

By RandomUsername3463 on 2/6/2009 2:42:48 PM , Rating: 2
Oh really, did you win the lottery or something?

By IntelUser2000 on 2/6/2009 8:15:26 AM , Rating: 2
so, how many 'titanic'ium shops will this affect? 6, 7, 10? :)

Putting the joke aside, according to IDC data a little over 160K Itanium CPUs have been sold Q1-Q3 2008. Xeon MP CPUs are roughly quadruple that so the delta isn't far away as many think.

I thought
By afkrotch on 2/5/2009 2:26:08 PM , Rating: 2
Wasn't some other company the number 1 chip manufacturer in the world? Like TSMC or something.

RE: I thought
By acase on 2/5/2009 2:30:42 PM , Rating: 5
Yah, pretty sure it's Frito-Lay.

RE: I thought
By UltraWide on 2/5/2009 2:53:09 PM , Rating: 3

RE: I thought
By nafhan on 2/5/2009 2:55:53 PM , Rating: 4
Nope... Intel is #1 by a large margin over #2 Samsung. TSMC is at 6.

Here's an article with a list that includes TSMC (had a hard time finding a 2008 chart, this is 2007):

How Itanium killed the computer industry
By chartguy on 2/5/2009 3:12:53 PM , Rating: 2,2817,2339629,

Dvorak may be a cranky old geek, but I think he's right on this one.

RE: How Itanium killed the computer industry
By RamarC on 2/5/2009 3:33:24 PM , Rating: 2
sorry, but this is classic dvorak nitwitism.

most of his argument is based on intel's stock price and idc's predictions. to both, i say "so what."

intel's SALES and PROFIT have soared since 2001 (current recession not withstanding). they recovered from their mistakes and regained the dominant-male leadership role in the industry. so how can dvorak indict intel on any level?

if anyone is to blame for this perceived ruination of the computer industry, it's the stock analysts and the trade rag media who placed much too lofty expectations on itanic. dvorak's probably ticked because he bought intel at $90/share and has since also lost his once lustrous position as a media pundit.

RE: How Itanium killed the computer industry
By Parhel on 2/5/2009 3:46:56 PM , Rating: 2
Itanium was a good idea that the market just didn't accept.

Reading that article reminded me of a rant my 70 year old COBOL professor would go on about how IBM killed the mainframe market. They probably don't teach COBOL anymore, so you can tell this was while back.

His former employer made a competing mainframe, and IBM put them out of business. Over 20 years had passed, and he was still bitter.

By rmlarsen on 2/5/2009 5:50:40 PM , Rating: 2
Regardless of whether it was a good idea or not, the Itanium simply didn't deliver, performance wise. Never did, never will. Sure, it gets very high instruction level parallelism on certain (mostly HPC) workloads, but the 128 bit SSE engine in Conroe basically eliminated that advantage.

These days, Penryn-based quad Xeons are running circles around the Itanium because the Xeons are clocked much higher. Add to that the fact that Itanium is HUGE (well, the cpu core is pretty small but requires oodles of cache to run efficiently) and hot, I'm surprised they still bother sinking money into developing it.

Nehalem-based Xeons will just completely trash Itanium.

NOT RISC based - EPIC is a VLIW architecture
By eilersr on 2/5/2009 3:25:19 PM , Rating: 5
Contrary to the article, the Itanium family is not a traditional RISC based architecture. It uses EPIC (Explicitly Parallel Instruction Computing), which is a form of VLIW (Very Long Instruction Word) architecture. HP and Intel developed EPIC to get around the limitations of RISC based architectures.

RISC uses lots of (relatively) short, fixed-length instructions.

VLIW produces very long instructions that have many components (you can think of them as individual RISC instructions packaged together}.

EPIC is a form of VLIW that gets around branching issues by looking all possible branches (hence the "Explicit") and throwing out the paths not needed. This actually places a high load on the compiler. Some argue that its this reliance on compilers plus the expense of the hardware to execute the instructions, that have led to the less-than-stellar acceptance of the Itanium family.

By icrf on 2/5/2009 5:07:43 PM , Rating: 2
The big idea was that, at compile time, you had many orders of magnitude more time and processing power to choose which branch should be taken than you would with the standard real time branch predictors in x86 chips. It makes compiler/assembler writing much more difficult and important, but had the potential to have incredible gains (and probably did under certain work loads).

My guess was they just never got good enough at it to make a difference in general cases, and it takes too long to port the multitude of code from x86 to IA64. The emulator just plain didn't work anywhere near fast enough. And so, no one used it, and it's largely been a money sink.

By segerstein on 2/5/2009 4:28:47 PM , Rating: 2
Isn't this delay because Intel wants to make Xeon and Itanium sockets & motherboards compatible?

RE: Socket
By trisct on 2/6/2009 2:30:31 PM , Rating: 2
Can't. Too many differences between the chips for motherboards to deal with economically. While it is probably feasible to make sockets compatible, it wouldn't be useful because the motherboards couldn't be. The thermal and power envelopes are too different, meaning differences in how you must deliver power and cooling. Therefore differences in how you design the motherboard.

By Spectator on 2/5/09, Rating: -1
RE: Recession
By tdawg on 2/5/2009 3:26:49 PM , Rating: 2
What? This post makes absolutely no sense. Not in relation to this article, or anything else. I actually feel stupider for reading it.

RE: Recession
By acase on 2/5/2009 3:33:39 PM , Rating: 3
Yah, he definately should have remained a Spectator.

RE: Recession
By Spectator on 2/5/09, Rating: -1
RE: Recession
By eetnoyer on 2/5/2009 7:18:44 PM , Rating: 4
Mr. Phelps?

Excuse me, could I get your autograph? And would you mind passing the bong?

"If you mod me down, I will become more insightful than you can possibly imagine." -- Slashdot
Related Articles

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki