Tukwila delayed until 2010
IT LOOKS LIKE Intel is delaying Tukwila yet again, this time to Q1 of 2010. The ostensible reason is to allow greater application scaling.
Given that it was up and running over a year ago, and scheduled for release late last year, well, someone must have found something pretty bad. Just when you thought Itanic was going to finally shed the title for industry running joke, someone felt the need to press the big red 'whoops' button, pictured in the link above. Don't blame us, we told them not to.
In any case, onward to victory, next time. And a bit later. Really. The next one will kick butt, trust us. Please ignore the Nehalem EX that will beat it not only to market, but like the proverbial drum as well. Has the industry found something to grab the crown from Duke Nukem Forever?
The entire press release is below. µ
------------------------------------------------------------------
Here is an update to the Tukwila Itanium® schedule. As you know, end users choose Itanium-based servers for their most mission-critical environments, where application scalability is paramount. During final system-level testing, we identified an opportunity to further enhance application scalability. As a result, the Tukwila processor will now ship to OEMs in Q1 2010.
In addition to better meeting the needs of our current Itanium customers, this change will allow Tukwila systems a greater opportunity to gain share versus proprietary RISC solutions including SPARC and IBM POWER. Tukwila is tracking to 2X performance versus its predecessor chip. This change is about delivering even further application scalability for mission-critical workloads. IDC recently reported that Itanium continues to be the fastest-growing processor in the RISC/Mainframe market segment.
Share this:
ShareBusiness Intelligence Software
Customer Relationship Management (CRM)
Enterprise Accounting Software
Enterprise Resource Planning (ERP)
Enterprise Systems Management
He couldn't have, there's no spelling mistakes or puerile sniping at Nvidia...
Are they implying that Itanium is the opposite of proprietary, that is, open? Or am I supposed to understand "proprietary" to mean bad and non-Intel, while Intel means "industry standard" because they say so?
Although, I haven't worked in HP since 2000 (I was with the original McKinley design team), I have quite a few contacts still within the company.
It turns out that in the period between 2006-2008, HP's high-end computer development group in Richardson Texas was decimated in order to consolidate power in Fort Collins Colorado. As a result, many if not most of their best firmware developers decided that maybe there are better places to work than HP (join the crowd I guess). Now it turns out that it is pretty difficult to train up new hires on the nuances of high-end server firmware design - after all, there's only so many places to gain such experience. Guess what happened to the schedule for the firmware for their high-end platform. Oops, that was "unexpected consequence"
Now the management team that killed HP's server team has been sacked, but that doesn't really solve the schedule problems they're having. I guess this is the end-result of putting major **ck-ups in positions of power in big companies.
So, there's been two major schedule slips in the program, and each time, within a few weeks, Intel has come and announced that, well, maybe Tukwila isn't ready after all. Go figure!
"So, there's been two major schedule slips in the program, and each time, within a few weeks, Intel has come and announced that, well, maybe Tukwila isn't ready after all. Go figure!"
Sound like HP and Intel are playing schedule chicken - just wait for the other guy to blink and confess first.
You can't find something to bitch about, so you bitch about the fact that there is nothing to bitch about.
Did you ever say anything nice to anyone or are you always this negative?
During final system-level testing, we identified an opportunity to further enhance application scalability
Once the translator activated with a premium bullshit filter applied as well, I got this translation:
"When we finally got to test it on live system it &^ucked up so badly we had to give in another 9 or 10 months to figure out how to get out of the mess."
So 2x the performance vs predecessor is great when moving from 2c to 4c? For a CPU that doesn't show up after a few months with twice the core count, but a few YEARS? I would have expected (a lot) better than twice the performance. After that much time, I would have expected each core to have twice the speed (and the connection to the RAM), resulting in nearly four times the performance for a 4c-CPU...
By 2010 Montvale times two will be as good as one core in a 8c-Xeon :-)
So much for the 16 core Itanic...
Hehe yes its quite funny, its more or less what you said I think, because its Intel, and also prior to the launch of Itanium it was going to be the industry standard enterprise CPU that everyone would use instead of those silly companies with SPARC, PowerPC etc. In the end I was trying to think which OS's are now really used on Itanium and came to the conclusion HP-UX and VMS, VMS is being wound down anyway and HP-UX is being beat by AIX, Solaris and Linux. Im sure HP-UX would be in much better shape (in user base) if it had stuck with PA-RISC :S. There is also Tru64, not actaully sure if that ever got ported, Windows (like anyone wants to run that on Itanium) and Linux, some BSDs etc (dito).
Anyway, they cant say anything good about it thats true so doesnt really matter what bollocks they put in their propeganda ;)
The Itanium servers are absolutely brilliant.
I have six of these running OpenVMS and this combination is spectacular, great performance and great reliability. I disagree with many other comments made here, my thinking is that Intel usually makes good technical decisions and I am very confident that Tukwila will benefit from this decision.
see this:
http://insidehpc.com/2009/05/29/qa-with-hpc-virtualization-software-maker-scalemp/
HPC users can have access to large-scale SMP based on Nehalem, which is a much faster processor to begin with.
of course, for all those buying Superdome machines for large DB deployments... the waiting period still stands (extended).