While that is indeed very nice, the reality is that Real Life IPCs are dependent on low cache miss rates, extremely good branch prediction and the "cost" of flushing a pipeline in the case of a miss (which scales with the length of that pipeline).

It would be interesting to see SPEC numbers from AMD's next gen and maybe an architecture brief. During the pre Core-uarch days AMD generally ruled the roost with a stubbier pipeline and a ruthlessly efficient cache, but Intel caught up after the shitfest that was Prescott. It'll be nice to see what AMD has planned for next gen.

Is there going to be a tech and process brief soon? If so it would be nice to see it.

I'm quite interested with the die stacking and thermal performance of the new memory subsystem.

It has been years since I've written a full review and technical dissertation of a graphics cards but the slew of new chips from both AMD and NVidia are making me interested.

I've had a look at the Carrizo launch slides and it is very interesting.

I really like the idea of the IGPU in the APU itself taking up the graphical load in tandem with an existing graphics card for an extra boost in performance. Very elegant in principle but I suspect that there's a lot of work done in the background and by the drivers to ensure both coherency and scheduling within a graphical load.

AMDmy Do you have a programming/optimization guide for developers available in regards to this new feature?

Also the 10 to 15 percent reduction in leakage current is very good. As a rule of thumb, nearly all power a processor dissipates is via leakage current as FETs are voltage operated devices. Thus not only is the increase in efficiency due to the better branch prediction and prefetch very welcome, but the new proc will run cooler.

I like this new Carrizo. Slap it on a Global Foundries 14nm FINFET process and might be the new go to processor for mobile devices.

Review sample for keepsies!

But then again, AMD/ATI owez kedekut review sample, which never endeared them to me when I used to review their graphics cards.

Depends on what you're using and what you're using it for.

Power usage in my house can be chalked down to a few things. Namely my fridges, storage water heater and aircon.

Mostly computers don't take much up power even while gaming. The heaviest load I've had was actually running ANYSYS 15 FLUENT and MECH using my badly programmed personal pipe mixer benchmark (running all models, on 1024 steps, full parallel), computer topped out at 550-ish watts for an SLI-ed GTX-980 (haven't tested on the TIs) measured via Fluke 43B Power analyzer.

At the highest tariff of 33 sen per kWh. An absurdly specced computer running at 500 watts running for 12 hours a day at full blast will cost 2 ringgit a day. But most gaming computers top out at about 250 watts, so you're looking at 1 ringgit a day realistically.

Nice, what are the temps like on it?

Hooked up the SLI and just realized that I don't have a 3-way SLI bridge lol!

Definitely will be watercooling the rig. The tight space between the first and second GC means it's about 10 celsius hotter than the other 2 cards.