Hi kateiacy

A definite no. It's the first thing I check after setting "While processor usage is less than" from 25 to 0

I am being short changed on hours on the Atom with HT enabled runiing xubuntu but I can hardly complain as they are just "virtual" ones

I popped by to share my favourite sentence of the week. It's from PC Pro's April 2011 Edition (Time travelling print press ).

It reads


"ARM rather forced Microsoft's hand with the way it's adding power to it's chips" says Wes Miller, research vice president at analyst Directions on Microsoft.


I am sure I don't need to explain why I delight in it's construction

It's bliss


Dave

Thanks
Although I guess most of the credit goes to people large and small who have been pushing for better efficiency over the years, and to the engineers who make it happen.

Thanks for reminding me about the Ubuntu efficiency stats, I'll be posting them within a couple of days.

I agree that these comparisons are not perfect, but as long they are in the ballpark of reality they can still be useful.
In the same sense, I suppose, as points or runtime stats do not translate directly to "scientific usefulness". But the stats have the advantage of being simpler to get, and an approximate indicator is presumably better than nothing

http://www.xbitlabs.com/news/cpu/display/2011...s_for_Next_Gen_Chips.html

50% more number crunching per core

Same power envelope????


Will need a new motherboard - but the old Phenom II's will be forwards compatible

Each AMD Bulldog module has 2 complete integer cores. This is much faster than hyperthreading. However, each integer core is slower than the AMD Phenom II core. Adding 2 integer cores to each FPU adds less than 5% to the silicon area, just as Intel adding hyperthreading to a core adds less than 5% to silicon area.

My personal opinion is that AMD is making things much easier for the CPU architects. The original Core 2 CPU did not have hyperthreading because it took the architects a long time to add it to a new core design. This is a good design for the future.

Lawrence

Yep, the shared FPU might be key to how it performs when it comes to crunching. Applications supporting Advanced Vector Extensions (AVX) should do very nicely, but if not/until they do performance will take a hit.
There is a good chance we will start to see more performance segregation between floating point and integer intensive apps for different CPU architectures. So one CPU would be better at crunching for one project than another. CPU choice will come down to which project(s) you want to crunch for, if we are informed about which projects are fp or int intensive.

The SB i7-2600K is almost 20% faster than a mid-range i7-800 series CPU (both 95W TDP), so for now a well designed SB system should be the most efficient CPU for crunching, despite the stupid on-die GPU. When BD is released it might depend on what you are crunching, and by the end of the year the 6core/12thread SB systems will probably be more efficient again.