backtop


Print 36 comment(s) - last by Hare.. on Mar 14 at 2:53 AM

Intel announces energy-efficient quad-core processors that consume 50-watts of power

Intel today released two new energy-efficient quad-core Xeon processors for multi-processor servers. The new Intel Xeon L5320 and L5310 operate at 1.86 GHz and 1.60 GHz respectively. Energy-efficient Xeon models consume 50-watts of power, which translates to 12.5-watts of power per core. Intel’s regular quad-core Xeon 5300-series consume 120-watts of power.

Energy-efficient Intel Xeon L5320 and L5310 processors are nearly identical to their higher-clocked counter parts. The energy-efficient models have 8MB of total L2 cache, 4MB of shared L2 per pair of cores, as with other Xeon 5300-series models. Front-side bus of the Xeon L5320 and L5310 are clocked at 1066 MHz, similar to the normal Xeon E5320 and E5310.

Pricing for the energy-efficient Intel Xeon L5320 and L5310 is $519 and $455 in quantities of 1,000, respectively. Intel Xeon L5320 and L5310 processors are drop-in compatible with Intel’s Bensley server platform.


Comments     Threshold


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

Do low power CPUs over clock better?
By nurbsenvi on 3/12/2007 6:02:16 PM , Rating: 2
Do low power CPUs over clock better than the normal ones?

How do they make low-power CPUs? is it just good batch of CPUs down clocked or do they actually do something to reduce the leakage?

Sorry to ask so many questions...




RE: Do low power CPUs over clock better?
By Shintai on 3/12/2007 6:09:44 PM , Rating: 2
They dont OC better per default. They are just running at a lower voltage. Sometimes using slower switching transistors that takes less power.

There was a person at the XS forum running a C2D at 0.88V at stock speed of 2.4Ghz. I doubt it used over 15-20W. So low voltage or ultra low voltage is the key.


By yehuda on 3/12/2007 9:28:55 PM , Rating: 2
Here's the link:

http://www.xtremesystems.org/forums/showthread.php...

I wish we could hear more often about undervolting.


By D4rr3n on 3/13/2007 12:25:57 AM , Rating: 2
nurbsenvi: Obviously nothing is guaranteed in overclocking, but yes a low power/low voltage version of CPU should overclock better than the higher power version the vast majority of the time. This low power version creates less heat and uses a lower stock voltage (which are interrelated) to achieve the same speeds. Being stable at a lower voltage for the same speeds generally means a better/more efficient CPU also. And 2 of the major things that limit your maximum overclock are heat and the maximum safe voltage you can give your CPU. So if one chip can do stock speeds at lower voltage and giving off less heat you can generally assume that these differences will remain as you continue to overclock. Basically you have more "wiggle room" at the top end with the low power versions. Which is why so many overclockers have chosen in the past to run low power mobile versions of desktop CPU's in their main setup. But, like I said nothing is guaranteed.

Yehuda: While I agree and understand the benefits undervolting can have in various scenarios (especially passively cooled silent pc's, sff setups, and HTPC's for example). It can actually be quite dangerous (even more so than overvolting) depending on the type of setup you have. A64's have been known to die running low vcore and high vdimm. The combination of a low cpu voltage and high memory voltage did not play nice with the on die memory controller and a lot of people lost their A64's very fast unfortunately. This is no longer an issue with the much lower voltage used in DDR2, however if it is to be discussed more people need to be aware of possible dangers related to certain configurations.


By Hoser McMoose on 3/13/2007 9:43:59 AM , Rating: 2
There's as much reason to suspect that it might be WORSE for overclocking, not better.

There are two tricks to lowering power consumption. The first and most important is to reduce the voltage. By running at a lower clock speed and lower voltage the same chip will consume less power. Since power tends to go up with the square of voltage, this factor is more important.

However beyond that there are some tweaks made at the manufacturing level. Making processors is a tricky business and there are a number of adjustments that can be made in the system. Some of these adjustments can improve maximum clock speed, yields, or power consumption.

Intel (and AMD, IBM, etc.) will tweak these various parts of their manufacturing process to optimize production for various parts. One run of wafers might get tweaked for maximum clock speed and sold as high-end desktop chips. Another run might get tweaked for maximum yields to be sold as mainstream or value processors. A third run might get optimized for minimum power consumption to be sold as mobile chips and these low-power server chips.

Then, after all is said and done, it comes down to a measure of tolerances. Intel has followed AMD's lead and just specified a few "thermal design powers" for all their processors, with that number now simply being a maximum that will not be exceeded for any processors in that range. Of course, individual chips will vary. So, for example (numbers pulled purely out of my ass), out of 100 Core 2 Duo/Quad/Xeon dies at 2.6GHz, Intel might have 5 that consume between 60 and 65W of power, 20 consuming between 50 and 60W, 50 consuming between 40 and 50W, 20 that consume between 30 and 40W and the last 5 consuming less then 30W. Now Intel could sell ALL of these chips as Core 2 Duo chips with a 65W maximum, or they could split off the best 25% or so and sell those at a premium and rated with a 40W TDP, while selling the remainder as "standard" chips. Market condition play a big role in deciding whether this is practical/economical to do or not.


"I want people to see my movies in the best formats possible. For [Paramount] to deny people who have Blu-ray sucks!" -- Movie Director Michael Bay

Related Articles













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