Intel's March 30 release of the Xeon 7500-series is significant in more ways than one. Publicly, the company is boasting of taking its 30-year-old x86 architecture into new ground--the mission-critical space previously dominated by mainframes and RISC-based systems like Sun's SPARC and IBM's POWER processors.
The Xeon 7500 is not intended as a file and print or Web server. You can use a low-power Xeon 3400 or even a Core i7 for that. The 7500 is aimed at the high-performance market of servers that must stay up and running with no downtime and have the capacity to handle thousands of simultaneous users. It's new territory for x86, but not for Intel.
Intel is already in that space with its other architecture, the EPIC architecture in the Itanium. EPIC is not technically RISC in design, but is frequently lumped into that category. While Intel has not said publicly that the Xeon 7500 will replace the Itanium, and no one is expecting it to in the short run, the Xeon 7500 does show that Intel has the ability to continue to evolve the x86 architecture and move it up the performance food chain.
The Xeon 7500 family, developed under the codename "Nehalem-EX," is a monster of a chip, with 2.3 billion transistors used in eight cores connected with high-speed interconnects, four very fast memory channels, and Hyper-Threading, allowing each core to run two threads at a time. It adds more than 20 features that have been in the Itanium for some time, giving the Xeon 7500-series levels of performance and reliability other Xeons can't match.
The Itanium has never been a big seller for Intel, but the few machines that were sold were absolutely valuable. Itanium was used in servers that defined "mission-critical." They had to run 24x7 and not be brought down by a crash. Itanium servers were usually multi-million dollar beasts that ran multi-petabyte Oracle databases or line-of-business applications that had to always run.
This meant something known as RAS: reliability, availability and serviceability. The Xeon 7500 has more than 20 new RAS features normally found in Itanium processors, marking the first time they have been used in a Xeon.
The most significant among them is Machine Check Architecture (MCA) Recovery, a feature that allows the CPU to work with the operating system to isolate errors that would otherwise crash the machine and keep the machine operating. Other features include memory corruption protection like SMI Lane Failover to handle memory errors and QPI self-healing for errors during interprocessor communication.

nice article btw thank you.
4 memory channels, 16 DIMMs per CPU, damn. I imagine you'd spend more on the 16GB DDR3 DIMMs than you would the processors though.
Also nice to hear that these scale well in 4P/8P boxes.
But I must ask, why are the 7500 chips in 45nm? Is the 32nm process still too immature to make a 2 billion transistor chip with any decent level of success?
Assuming a 8P box, all CPU's clocked to 3.5GHz (~120GFlop per CPU, ~1TFlops total), you could run a few games purely in software mode and still get good performance. Damn.
When did Itanium change from being a VLIW architecture to being a RISC architecture?... It was designed to overcome some RISC architecture limitations of the day. XScale was Intel's big RISC mistake...
"In the course of one week, two separate events signaled what may be the end of Intel's grand experiment with RISC architecture. Intel released the Xeon 7500-series processor family, a RISC-based design developed in partnership with Hewlett-Packard containing many features found in the Itanium, and Microsoft ended its support of Itanium."
There, all better!
Compare the AMD 6128 (8 cores, 2.0 GHz, $266 list price) against the Xeon X7550 (8 cores, 2.0 GHz, $2729 list price) and you'll see what I mean. The XEON cost more than 10 times as much! Sure it's faster, but not 10 times faster. Not even 2 times faster.
You clearly didn't understand a word of this article.
You sir, are an idiot. RAM is MUCH more expensive than these CPUs. Even 16gb of desktop DDR3 memory costs about $800. Now these mobos generally have more than 4 dimms per cpu- more like 8, so $1600 for RAM makes a $266 CPU seem really really cheap. Now server memory is always more expensive, so I think it would make perfect sense to spend $2000 more in order to have a system with fewer bottlenecks.
It still has reliability features far exceeding the Nehalem-EX, and they are still greatly supported by the largest computer maker in the world, which, by the way, also was the original designer.
It's not going anywhere.
Two-socket 16 core Opteron system with 32 GByte RAM:
Tyan S8230 ... $460 (source: Froogle)
two Opteron 6128 ... $266 x 2 = $532 (source: AMD)
sixteen 2GByte DDR3 ECC DIMMs ... $60 x 16 = $1000 (source: Froogle)
case, P/S, HD ... $400 (estim.)
Grand total ... $2400 for a 16 core, 32 GByte server ($150 per core)
Two-socket 16 core Intel "Beckton" system with 32 GByte RAM:
motherboard ... $478 (assumed price, not available yet)
two Xeon E6550 ... $2461 x 2 = $4922 (source: Intel)
sixteen 2GByte DDR3 ECC DIMMs ... $60 x 16 = $1000 (source: Froogle)
case, P/S, HD ... $400 (estim.)
Grand total ... $6800 for a 16 core server, more than $400 per core
Again, nobody doubts that the "Beckton" system will be faster.
But, hey, $2400 vs. $6800 is a heck of a price difference.
This does not take a genius to figure out !
There are countless mission critical scenarios where second best in speed/reliability will cost companies money + get people fired(IT guy who chose the 'better deal' . Financial transactions with the NYSE, Amazon.coms inventory, Credit Card companies customer's accounts......
That's good news for small businesses ... as costs for the low and mid range systems will drop due to competition.
N-EX based systems will nevertheless cost you an arm, leg, and possibly a kidney ... no competition there to dissuade Intel to drop the price.
And why should they? Must have cost a squillion on the R&D to make this puppy pass inspection... many engineers toiling into the wee hours.
Good article ... a tremendous chip too.
petabyte - how many zeroes after the 1?
Imagine if we have these X7500 CPU they would be just sleeping. LOL
Imagine if I had the X7500 CPU in these
While 8P systems amount to a tiny fraction of the server/workstation market - mostly 2P and some 4P - evolution is always welcomed.
Even accustomed with the usual THGossip wintel bias and hype, micro$uxx' relevance in the mission-critical market is more than overblown... they're plain irrelevant.
No more Itanic support? Good riddance - only gamer loader diehards would've used the crap from redmond anyway.
All critical servers/workstations implement ECC. If not, it would be just a russian roulette game, until data corruption or crashes occur.
It's not about "if" - errors are guaranteed to occur - but about "when" - and how to reliably detect, and correct them, and if uncorrectable ones appear, to take appropriate measures.
Even if the drm infested games loader doesn't support RAS features, it doesn't mean that serious work shouldn't be done, where they are mandatory.
petabyte - how many zeroes after the 1?
1 PB = 1,000,000,000,000,000 or 1000 TB's