And yet another version from AMD

[intelamduser]

http://www.tech.co.uk/computing/upgrades-and-peripherals/motherboards-and-processors/news/amd-quad-core-debuts-august-but-only-at-2ghz?articleid=2118719296

They would rather focus on performance per watt than speed, what a load of carp.

 

[NMDante]

http://www.tech.co.uk/computing/upgrades-and-peripherals/motherboards-and-processors/news/amd-quad-core-debuts-august-but-only-at-2ghz?articleid=2118719296

They would rather focus on performance per watt than speed, what a load of carp.

Better than performance per mm squared.

 

[Hatman]

According to that it sais the quad core desktop prossessors which im guessing it means the Agena x4 will be around early next year, that saddens me as I was hoping it would be out before X-mas.

 

[chunkymonster]

I might have agreed with you 3 years ago when gighertz was a the only valid indication of a processor's potential performance. The days of rating a processors solely on it's core speed are long gone. For example, the X2 3800 is "only" 2GHz and it out-performs PentiumD's rated at higher clock speeds; and nowadays, the E6600 @ 2.4GHz bests an X2 4800+ @ 2.4 GHz, so...read this article, it puts the gigahertz question into perspective.

Performance per watt is a very valid measure of a processors performance, especially in the context of a datacenter and doing everything you can to cut costs, use less energy, and still deliver the performance that the business needs.

 

[lghtmup98]

I agree. It seems there are 2 major schools of thought.

Ultimate performance: Performance at any cost.(power or dollar)

Efficient performance: Best performance within energy consumption constraints.

Depends on what you're goal is.

 

[morg]

intel first anounce about conroe was it's performance/watt ratio...

this technology leads to performance for sure.

68W per core is an awesome step for multi-processing.
it makes 136W tdp for 8 cores running at 2.0ghz while outperforming actual opterons

barcelona is designed to be specifically efficient in floating point operations, SSE applications.... it will perform at least normally on other application types.

i imagine quad socket mobo would be nice with 16 cores at 2ghz for 390$ each... 1560$ for 4 cpu and it can use a good ammount of ram increasing max server power.

 

[intelamduser]

My remarks were only to point out the spin.

 

[liquidx]

The biggest consumers of processors are really the companies running thousands of them. We as regular everyday joes only make up a much smaller portion of the overall scheme of things. In that right, if you replace 1000 computers with ones that use half the electricity of previous ones that greatly reduces your TCO, thus making it a wise business decision.

Now on the everyday consumer level, that benefits us, because lower electricity saves us money, and in general means the proc produces less heat, which also especially for people like us, usually gives great headroom for overclocking. Ex: C2D's

 

[bixplus]

I might have agreed with you 3 years ago when gighertz was a the only valid indication of a processor's potential performance. The days of rating a processors solely on it's core speed are long gone. For example, the X2 3800 is "only" 2GHz and it out-performs PentiumD's rated at higher clock speeds; and nowadays, the E6600 @ 2.4GHz bests an X2 4800+ @ 2.4 GHz, so...read this article, it puts the gigahertz question into perspective.

Performance per watt is a very valid measure of a processors performance, especially in the context of a datacenter and doing everything you can to cut costs, use less energy, and still deliver the performance that the business needs.

This is only relevant where the better performing part (per watt) can produce an acceptable level of performance to begin with. In other words, if a lesser performing part (per watt) finishes a task quicker than its competitor, it actually is the better part because it uses less or equal 'total' power. This, of course, would require lot's of measuring, but I think you get the concept without having to go into much further detail.

 

[crow_smiling]

i imagine quad socket mobo would be nice with 16 cores at 2ghz for 390$ each... 1560$ for 4 cpu and it can use a good ammount of ram increasing max server power.

The Barcelona pricing that has been suggested is for the 2P versions, if you want the versions for 4P and above expect to pay much more.