The World's First 65 W Desktop Quad Core

[Guest]

As far as performance is concerned, AMD is lagging behind its competitor Intel. But, other than gamers, few users care only about performance.

The World's First 65 W Desktop Quad Core : Read more

 

[xx12amanxx]

VERY INTERESTING...

 

[custommadename]

People can still play games with this quad core. Sure, it's not as fast as a quad core from Intel, but it's evidently more than suitable to handle today's games. However, for about $200 it's a horrible value.

 

[Guest]

Just a quick question... at 1.8GHz is this chip even competitive with a higher speed dual core (even when considering 4 core vs 2 and considering multi-core optimized applications)?

It'd be interesting to add in a Intel mid to high range dual core and look at the power consumption and performance vs the relatively low clocked quad. Dropping the clockspeed so low would seem to give back most of the games from having 2 extra cores (and would be worse for SW that cannot use 4 cores).

 

[JSP78]

Im an Q9450 owner and to just drop the multipler on the QX9770 and use the same vid isent fair.
My Q9450 VID = 1.1v, EIST drops VID to 1.0375v
C1 stepping, slawr l806a762
I bet thats gonna make it hard for the phenom to beat

 

[thuan]

Page 12 title is wrong. It's 9.8% not 8.8%. Just mention it, as it seems no one's noticed it yet.

 

[royalcrown]

This just goes to show how crappy "Barceloney" really is.

Consider the fact that one can buy an e8400 or 8500 which uses the same power, can keep up with AMD's quad cores (even when they are over clocked), and also DESTROY this thing at gaming. WHY exactly would I buy this piece of crap again Frank ?!?!

I guess if I ran Cinebench benchmarks all day and did NOTHING else, then this cpu would look good.

People that buy quad cores are not going to care about 35 watts, they want to brag about speed OR get their work done ASAP, and AMD is crap for that right now !

 

[Reynod]

An excellent review Frank ... well done.

Very impartial ... showing the strength's and weaknesses of the low power unit.

The overclocking limit of 2.4 would be interesting to explore further.

Can you elaborate on what settings you went to ... and the HTT frequency please?

Cheers,

 

[Reynod]

An excellent review Frank ... well done.

Very impartial ... showing the strength's and weaknesses of the low power unit.

The overclocking limit of 2.4 would be interesting to explore further.

Can you elaborate on what settings you went to ... and the HTT frequency please?

Cheers,

 

[OMA]

Why did you underclock the l3 cache and memory controller? The ht link is ment to be overclocked, not underclocked! L3 runs the same speed as ht link not 2,4 ghz. Slow L3 is a big bottleneck in games.

 

[skittle]

comparing power consumption of a 1.8ghz phenom with a stock q6600 and touting the phenom as the power consumption king is a very inaccurate conclusion. The articles here just keep getting worse and worse... seriously who hires these clowns?

 

[sandmanwn]

Its performance is perfectly adequate for office work and entertainment, though not for games.

I hope that statement was intended for this particular CPU only, or the fanboyism of this site just went up 10 notches to ludicrous stupidity.

 

[royalcrown]

skittles, even if it is accurate, so what because INTEL could just under clock the q6600 and draw 65 watts to, but the performance would suck also.They're just pushing efficiency because AMD is getting destroyed on the performance front and there is no real news right now on any performance oriented front.

 

[Guest]

The real point of this (chip and article) is twofold:

1, If you're really going to run multithreaded apps 24/7 (or not much less) this processor is really perfect. This points to the fact, that K10 was really made for server segment: it is scaling very well in multisocked configurations, and at lower clocks it is very energy-efficient.

2, The biggest deficiency of the current dekstop Phenoms is that they are way too power-hungry. Maybe the SOI is to blame, as I've read the SOI process has a frequency tipping point after which the dissipation grows tremendously. Anyway, the situation seems curable, as this 9100e can be clocked to perform on par with 9750 yet consuming substantially less. That brings some hope for AMD.

 

[royalcrown]

@Rachotilko

Uh what about the fact that they SUCK (x4's) ?! This architecture is AMD's version of Netburst. This design sucks no matter how you try to spin it; here is hoping the next cores from AMD fare better. Where is the advantage in buying this piece of junk ? Maybe if I had a bank of servers, maybe...as far as any real desktop apps go, I'd be just as well of going core 2 duo as I would on this thing, winning in one benchmark doesn't impress me because I do lots of things with my desktop.

 

[Mathos]

Eh, 9100e still has at least one ace up it's sleeve. If you use bios 1.2 with that chip, you can lower the VID on the northbridge/IMC and it will cut the Load and Idle power a lot. Depending on what it's stock set at, there should be plenty of room to lower it and still keep it stable. My 9600 BE was doing 2.4ghz on the NB at the stock VID, and could easily do 2ghz on the nb at 1.1v VID, stable. Only problem is, changing the VID I dont know if C&Q will still work. Still the biggest issue with Phenom Power consumption is the IMC, until they make it throttle or lower voltages like the rest of the chip.

 

[apaige]

Any chance on finally getting your Sandra memory bandwidth benchmarks straight?
http://www.legitreviews.com/article/617/1/

 

[qasdfdsaq]

I don't see the significance of a 65W Phenom running at 1.8Ghz. Intel has already announced their QX9300, a 35W TDP Core 2 Quad running at 2.53Ghz:

http://www.theinquirer.net/gb/inquirer/news/2008/05/14/intel-updates-gaming-notebook

Even if AMD have a slight lead now, they will be totally raped come Q3/08.

 

[etstech]

According to Intel, they have a Quad-core Xeon with a rated TDP of 50W.
http://processorfinder.intel.com/details.aspx?sspec=slap4

 

[wild9]

I think AMD is even more efficient than most tests claim; the memory controller is intergradted into the processor and not the Northbridge part of a motherboard's chipset. Extra traffic and external components running on 65nm (or higher) chipsets is surely to add to the power consumption of all Intel processors especially when dealing data storage facilities or super-computers.

 

[jckrieger]

I'm not sure about this article. From the benchmarks, it looks like this cpu is slower and consumes more power than a Q6600 which is Intel's old 65nm cpu. Performance per dollar and performance per watt still seems to lean in Intel's favor. The only thing I see worth mentioning in this article is AMD released the world's slowest quad core that still costs as much as a respectable Q6600. As the others pointed out, if you don't need performance, get a reasonably clocked dual core and keep general application performance at a higher level.

 

[Guest]

Is it me or have Tom's reviews seem to have a bias against Intel lately? Ever since the site design change, they haven't really had anything nice to say about them. It's usually something like "Intel MAY lead in performance, BUT...."

 

[DXRick]

The type of users that use office apps, do not play the latest games, and care about power are mostly buying laptops these days.

 

[Guest]

JSP78 is right. Q9450 doesn't have 1.2875V, but max 1.2V. This would translate 92W from article to less than 80W

 

[frozenlead]

AMD might just be getting in back together. In no way does it match intel, but compared to previous overclocking margins, 33% is huge!