New Itanium More Powerful, Power Efficient

Heir Of The Mess writes "Intel have a press release out about their new Itanium 2 Processor. The new processor doubles the performance, and improves performance per watt by 2.5 times compared to the existing single-core versions. The flagship model triples the cache and can execute 4 threads/instructions per processor enhanced by Hyper-Threading. Transistor count is a whopping 1.7 billion. Triples the previous SPEC_int_rate_base_2000 record. Retails for US$3692 for the top of the range.
So yes the Itanium crew are still pushing forward. I wonder if this could help save SGI?"

Comments

The ship has sailed

[powerlord]

Considering how little buzz this article seems to be generating (even on Slashdot), it seems like this might be a case of too little, too late.

The world has moved on from iTanic to x86-64.

Limited Scope

[tomstdenis]

The problem with the IA-64 platform is there are so few applications for which it is well suited. And even for problems it IS well suited it's a matter of figuring out how to extract the desired performance.

That said, a well tuned IA-64 application can smoke the best offerings from x86 world [on both sides of the fence]. But a $3700 USD price tag may push people away. Specially since processors like the Opteron 285 are nearly half the price and way more flexible. :-)

Tom

Working with the hardware directly

[drachenstern]

I can tell you that the processors are not even in production at the main server manufacturer backing the Itanium family (ie, not Intel, the other one*).

We have seen a few proto style units roll through, but they have all had serious problems and are not running at full speed. The engineering group either cannot or will not give us a reason why these units are running crippled, but we believe it to be a chipset issue. Hopefully we will see the servers rolling through our manufacturing process within the next 60-90 days, but no management timelines have been released.

Here's to all those of us who want that raw power and are looking to pay for it!!!

*Won't disclose the name since I don't know if this info violates my NDA, but screw it, the public should know this stuff. Information does deserve to be known.

-drach!

Re:Working with the hardware directly

[nkpatel]

I gotta call BS on this one. I work on the development team for many of these servers. (No, I'm not allowed to tell you when they'll ship, so don't ask). There are NO chipset issues I am aware of, and the units are not crippled. I'd be interested in finding out why you think this is the case. Case in point: ahref=http://www.tpc.org/tpcc/results/tpcc_result_ detail.asp?id=106071802rel=url2html-4549http://www .tpc.org/tpcc/results/tpcc_result_detail.asp?id=10 6071802>

Bumber for Intel

[sgt scrub]

I buy machines for our 100M product: 4 way opteron 870's with 8G Ram (2/proc) 2x73G Cheetas. Retail the box would cost about $7000. If one processor cost me $3500 life would suck. I can't see this processor doing the work of 4 duel core opterons.

Re:Sell off Itanium

[GnuPooh]

Intel didn't sell off an ARM division. ARM has always been a separate company. They BOUGHT the StrongARM part of Digital/Compaq years ago and renamed it XScale and still have it.

Secondly, the current Pentium-M/Core processors are decended from Pentium-III and Pentium Pro, so the thing they learned from P4 was that it was a dead-end architecture. I'm sure there are some elements here and there they "backported" to Pentium-III to make the current Dual Core 2, but you statement is not really accurate there either.

Price

[jefu]

It seems quite likely that if Intel were to push the Itanium more and get it used in more systems that the demand would rise and the price would therefore drop. Of course, this might require a lot of demand, but if the performance is good enough for the price, the demand will be there.

It might be worthwhile for Intel to find a way to drop the price enough to put these things into more places. Even give them away to visible web installations (like slashdot, fer'nstance). Get a bit more market penetration, convince some vocal people that its a good buy and it will start to take off. (I'd be glad to take an Itanium system for free for web service - even though my primary web presence is anything but big. Even better a couple of them to let my students use for compute bound projects.)

Re:Price

[tomstdenis]

Even with a sufficient market, right now that chip is going to be expensive. It's huge and the yield can't be that high.

But that's kinda the problem. The cost and the fact that's its a very niche processor will never make it as common place as Xeon or Opteron processors. Specially when both Intel and AMD are pushing SSE you can get a lot of the parallel SIMD like benefits of IA64 on x86.

Tom

Re:Price

[boner]

The Itanium has horrible price/performance for web-serving... the ad-hoc nature of web-serving does not do the architecture justice. The new dual-core x64 Conroe offering would be a much better choice.

Has anybody noticed that the Conroe effectively kills Itanium for most workloads?

Re:Price

[nkpatel]

While x64 may be better-suited for web serving, part of the reason x64 seems to be better than ia64 on most other workloads is that you may not have the OS and software optimized for the ia64 architecture - you're still writing and compiling in a x86-centric view and I don't think compiler technologies have evolved to the point where they can fully utilize ia64 ILP.

Re:Price

[LWATCDR]

"While x64 may be better-suited for web serving, part of the reason x64 seems to be better than ia64 on most other workloads is that you may not have the OS and software optimized for the ia64 architecture"
Or it could be the ia64 really isn't well suited to web serving.
The ia64 has very good double precision floating point. And seems to be very good at minimal branching / very predictable branching code. In other words not web serving. The ia64 seems to have the same problem that they had with the i860 in fact if you read the wikipedia entry for it it sounds like the exact same chip!
"One fairly unusual feature of the i860 was that the pipelines into the functional units were program-accessible, requiring the compilers to carefully order instructions in the object code to keep the pipelines filled."
and this seems to be even more on the mark "As a result of its architecture, the i860 could run certain graphics and floating point algorithms with exceptionally high speed, but its performance in general-purpose applications suffered and it was difficult to program efficiently (see below)."

My goodness it is hard to believe that Intel would go down this road again. I guess if at first you don't succeed try, try, again.

Re:Triple the cache

[tomstdenis]

24MiB of 6T SRAM is 1,207,959,552 transistors just for the cells. Not including tags, address decoders, etc...

By comparison, an Opteron uses 113,246,208 transistors for the 2048+256KB of cache [assuming they use a 6T structure which I don't know for a fact since I'm not privy to the details and technically I couldn't say even if I were, so don't assume what I said is verbatim fact, yada, I hate disclaimers] and the 4MB Duo (total of 4096+128KB of cache) needs 207,618,048 transistors for its cache.

Tom