backtop


Print 24 comment(s) - last by maroon1.. on Jun 2 at 7:25 AM

AMD beat its time to market estimates for "Istanbul"

The sluggish global economy has meant that not only were consumer computer sales down, but sales in the enterprise computing space for servers and data center computers were down significantly as well. The only thing for chipmakers like AMD and Intel to move ahead with new processors that offer more performance with power savings that help offset the cost of upgrading to new hardware.

Last week, Intel previewed its 8-core Nehalem EX server CPUs set to launch in the second half of 2009. AMD is now hitting the market with its new high-end 6-core x86 server processors codenamed Istanbul. The Istanbul Opteron processor is aimed at server markets with four or more sockets. AMD points out that the Istanbul processors are ready to go now, months before Intel is expected to begin shipping its Nehalem EX CPUs.

AMD's Istanbul CPUs are expected to start shipping this week and many of the top OEMs will be rolling out server systems powered by Istanbul processors. AMD was able to beat its original time to market estimation for Istanbul, something that will go a long way towards erasing memories of the troubled launch of the AMD Barcelona line of processors.

AMD reports that the Istanbul processors will give users a 30% increase in performance per watt with an overall performance improvement of 40 to 50% all within the same price and thermal envelope of its predecessor.

The Istanbul processors also come with other new technologies like HT Assist. AMD's John Fruehe wrote in a blog post, "[HT Assist] can give you much better throughput over the HyperTransport technology connections by reducing the amount of traffic generated by the processors in seeking the shortest path to data that they need."

HT Assist is a feature eWeek reports will make servers using the new technology more appealing to companies in the high performance computing space.

Analyst John Spooner from Technology Business research said, "The trend we're seeing is that there is a certain percentage of customers who are looking to scale down from a RISC/Itanium/mainframe-type machines to the high end of the x86 market. These customers are moving into top-of-the-line four-way and higher x86 servers."



Comments     Threshold


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

RE: Good Start
By amanojaku on 6/1/2009 12:04:32 PM , Rating: 5
Raw MHz performance isn't the only reason to get a CPU these days, even in the datacenter. Considering the average modern CPU is only 5%-10% utilized (Gartner group estimates, confirmed by my own evaluations as a virtualization consultant) most systems don't need high performance. This is one of the reasons multicore CPUs have lower MHz speeds than their counterparts with less or one core. Of course, newer cores with less MHz can outperform the older cores with more MHz due to architecture changes, but I digress.

Anyway, more datacenters need to utilize less power, and that is a major strength of modern CPUs. Second, virtualization is here to stay, and multiple cores and generally more useful than faster cores in this area. This is why the AMD vs. Intel MHz war has taken a backseat to the cores and power consumption wars.


RE: Good Start
By Amiga500 on 6/1/2009 12:11:34 PM , Rating: 2
I wonder what the breakdown of CPU sales is for datacentres vs. HPC (across the 2/4/8 skt areas).

A HPC would be much more biased towards speed as CPU utilisation would be well up.


RE: Good Start
By Motoman on 6/1/2009 12:23:25 PM , Rating: 2
The purchasing decision is going to be made based on performance, efficiency, and price as a combined function.

Or, more likely, people just buy whatever Dell tells them to.

But in the interest of winning the performace/efficiency/price function, I'm still not holding my breath.


RE: Good Start
By icanhascpu on 6/1/2009 3:42:30 PM , Rating: 1
quote:
Considering the average modern CPU is only 5%-10% utilized


Thats a bunch of horse poop. Considering that people only care about when they need the processing power is the only time it really matters, and when they are actively using it its usually at 50%-100% utilization, I would say your numbers are highly impractical.

Sure, if a 386 could sit at 100% for a week, then when I want to use it it would magically burst out a saved up chunk of processing time it saved up, that would be GREAT! But things dont work like that here in the 4th+ dimension.


"There is a single light of science, and to brighten it anywhere is to brighten it everywhere." -- Isaac Asimov

Related Articles













botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki