Intel Now Shipping Dual-Core Atom 330 Processors
Intel has official begun shipping its new Atom Processor 330, a 45 nm dual-core processor for nettops.
On Friday, Intel officially began shipping its 45 nm Intel Atom Processor 330, Intel’s first Atom processor to feature dual-cores. With a 1.6 GHz clock-speed, 1 MB of L2 cache and support for DDR2 667, the Atom 330 looks like Intel just combined two Atom 230 processors. The added core does double the processor’s TDP unfortunately, an increase from 4 W to 8 W.
The Atom 330 will be available as an integrated part of the new Intel D945GCLF2 desktop motherboard, which features the Intel 945GC Express Chipset, Intel GMA 950 integrated graphics and a 533 MHz system bus. The Intel D945GCLF2 motherboard also offers a S-Video connector, gigabit Ethernet, 6-channel high definition audio and a single DIMM socket supporting up to 2 GB of DDR2 667/533 memory.
Intel Atom processors have been specifically designed by Intel to be extremely power-efficient, however many sacrifices were made to achieve this balance between performance and power. It turns out that multi-threading greatly helps Atom processors overcome the effects of some of the sacrifices that were made. As a result, Atom processors that use Hyper-Threading technology see large performance gains and the use of two actual cores along with Hyper-Threading, as used in the Atom 330, should result in even greater gains.
Intel has designed the Atom 330 for use in nettops, which are entry-level, Internet-centric desktop PCs. It is likely that a dual-core Atom processor will eventually make its way into the mobile market, but Microsoft may not allow Windows XP to be sold on such a system. Intel Atom processors are facing strong competition, such as from Via’s Nano processors and offerings by AMD.
2gb and vista basic with this dual core might JUST run (it will run, but performance wise for every day use...)
this could be an interesting piece of hardware
I realize that the architecture is probably different and these aren't C2Ds or Athlons...and I suppose the thermal loss may be exponential for faster chips...the small die of the atom may just be due to the reduced cache size...all that and whatever.
But still, they have a dual core 1.6 with 1mb of L2 in a die package this small that runs on 8W!! The greatest and fastest processors of the day are huge compared to these and suck tons of power, and yet the performance gains scale in much smaller proportions (even if you consider architecture and not just numbers).
I think it just blatantly points out how much more work can and should be done on todays top end chips to shrink them and reduce their power needs, which in a way is a really good thing.
When I am able to go to Newegg and buy a Nano L2100 @ 1.8GHz and pair it with a suitable motherboard, then it's competing. Currently it seems they just have integrated 1-1.5GHz versions that are fixed to the motherboard itself. The motherboards are also much more expensive than their Intel/AMD/NVidia counterparts while offering less capabilities. Until you can buy what they advertise on their site, and buy them separate from each other, I don't think it's going to be a big threat to either Intel or AMD. Conceptually I'm all for the Nano, but it seems like mostly vaporware.
Maybe it's the digital-signage market eating them all up faster than VIA can make them.
lol 8w cpu 20+ w chipset yey, what do you expect from a netburst era chipset
You do realize that the Atom is also BGA (soldered instead of using pins) and not socketed? In netbooks, handhelds, and MIDs, a socketed processor doesn't make sense yet. Even in super low-cost desktops or "nettops," the processors are likely to be soldered directly in place for cost reasons.
That's what both the Atom and the Nano are aimed at right now. Unless somebody has information on a socketed version of the Atom that I've missed.
Major points below (there are lots of minor points like aesthetics, what need for 4-pin CPU power header on an 8 watt TDP = 0.66A) ...
CPU
---
* It's dual core AND hyperthreading (switchable in the BIOS). That's 4 virtual cores. Lovely!
* Runs at 36C vs 21C ambient. Warm for a low-power rig. Culprit is the passive heatsink - the design is hopeless. A passive heatsink needs lots of airways (think, pins or fingers). This one is out of the ark and depends heavily on the N/B fan for a blast of air over the basic, very basic, fins. The position of the CPU cooler doen't even make good use of this dependancey - a good 40% of it misses the N/B airflow.
* If the N/B cooler fails ...
Chipset
-------
* Southbridge runs warm - 1st time ever I've seen that on a working Intel board. I had to use MOSFET heatsinks to cool it.
* Northbridge heatsink is wobbly. According to Intel tech docs this is deliberate (!) as it prevents shock damage to the N/B chip during shipping/transit.
* RAM bandwidth is awful. I have a 4-year old S478 875P PC3200 dual-channel m/b that stomps all over this board.
Graphics
--------
* Work. Will play benchmarks as well as some older graphics cards
* Better Vista score than my older cards (FX5500 etc)
Overall
-------
Solid, dependable but a great CPU is throttled by a poor chipset / RAM implementation. Really cheap tho'
Plan
----
Will fit a low-profile PCI graphics card (128mb ATI 9200) for retro gaming test.
Fit into an old Flex-ATX desktop. Will also make a custom-fit to see how warm it runs when fully cased.
1) all psu's have that 4 pin connector with dedicated rails, why not use it?
2) intel certifies most of its mobile chipsets to run at upto 100ºc, that 36ºc is nothing, and passive SUPPOSED TO BE HOT but stable! run a burn in test - still works? duh cause its designed like that to run and operate like that!
3) Southbridges on all the newer boards run hot - there also supposed to run like that etc - iv seen some ASUS P5KPL-CM boards run at 60ºc+ on the SB - no big deal.
4) Northbridge - look under it, its a (less then) 1cm square die containing the MCH etc and yes they want it to flex rather then break
5) DDR-400 actualy benchmarks better then DDR2-533, and that cpu actually runs that memory at or most likely 533mhz, vs your old rig (equal to a low latency 800mhz etc or 2x400) on an ultra efficent cpu, ofcourse its going to bench lower
what do you expect from it? ITS NOT A DESKTOP DESIGN
The Northbridge heatsink fan has failed ... after just two weeks of light use (on my HSPC testbench, pics here http://s65.photobucket.com/albums/h216/Ray_GTI-R/PC/HSPC/).
After some research this is a known issue with the board. Intel are reputedly not offering a replacement (tuppny) fan, rather requiring the whole board be returned to the dealer for replacement. Presumably, that's because of the deliberately wobbly N/B heatsink fixing that could damage the N/B chip if any attempt is made to replace the tuppny fan in situ. Hmmm ...
So I'll be fitting a Papst then!
Ray.
Yeah, right! So why have Intel and most others' passive & active N/B heatsinks in desktop* boards for the past 6 years been rigidly fixed?
FWIW 36ºc appears to ref the CPU not the chipset. (DUH!)
Do you mean 200mhz FSB quad bumped (4 x 200mhz = 800mhz) whereas DDR2 is not quad bumped, just marketing hype to fool people into thinking DDR2 is somehow inherently faster than old-fashioned DDR? Answer - it seems to have worked in at least one instance.
ROTFLMAO - maybe I should fit a hairdryer inside my rig! That kind of "news" would put all the watercooling companies out of business, uhh?
*
See
Specifically, have a peep at the text inside The Red Ellipse kindly added in for those visually challenged amongst us by someone who appears to know what they are talking about.
LUV & LOL
1jupiterway
-------------------------------------------
"It's better to be thought a fool than open your mouth and prove it." So, if I have done so, please excuse my mere mortal condition AKA "Welcome to the human race."
http://www.intel.com/Assets/Image/prodlarge/D945GCLF2.jpg and
http://s65.photobucket.com/albums/h216/Ray_GTI-R/?action=view¤t=box1.jpg
Southbridge (MOSFET) heatsinks are doing the job very well i.e., cool.
Working links are now here;-
http://i65.photobucket.com/albums/h216/Ray_GTI-R/PC/D945GCLF2/box1.jpg
and
http://i65.photobucket.com/albums/h216/Ray_GTI-R/PC/D945GCLF2/DSCN3746.jpg
(Links Checked OK)
I wonder where apache_lives gets that info?
LUV & LOL
1jupiterway
-------------------------------------------
"It's better to be thought a fool than open your mouth and prove it." So, if I have done so, please excuse my mere mortal condition AKA "Welcome to the human race."
It shows that this is an Intel Desktop board.
Just for fun I re-connected the original fan to the chassis fan header(not fixed to the N/B heatsink) and ... it's OK. ??? After 4 days it's still OK.
Can anyone explain how a fan can apparently fix itself sitting idle on a flat surface for a couple of weeks?
Thanks!