It’s all about performance: Using Visual C++ 2012 to make the best use of your hardware
- Date: October 31, 2012 from 11:15AM to 12:15PM
- Day 2
- B92 Odyssey
- 3-013
- Speakers: Don McCrady, Jim Radigan
- 47,281 Views
- 4 Comments
Loading User Information from Channel 9
Something went wrong getting user information from Channel 9
Loading User Information from MSDN
Something went wrong getting user information from MSDN
Loading Visual Studio Achievements
Something went wrong getting the Visual Studio Achievements
Right click “Save as…”
Slides (view online)If you attend this talk, you will learn about the latest and greatest C++ performance features. All hardware being produced by Intel, ARM and AMD is now mulit-core with each core supporting a rich instruction set supporting vectorization of ordinary C++ that uses 128-bit registers. Additionally, on chip GPU’s are going through their 3rd iteration in many hardware product lines thus making general purpose computing on the GPU important. This talk provides an accessible overview of all the new hardware and how C++ allows the developer to take advantage of it. We start with the engineering of auto-vectorization and auto-parallelization for existing unaltered C/C++ programs, progress to PPL and then tie in the new C++ AMP language extensions. This is a “go-fast” talk that’s also a great intro into computer architecture and C++ compilers.
Already have a Channel 9 account? Please sign in
Follow the Discussion
great talk! the effort that went into vs2012 seems tremendous...
Nice talk!
Looking forward to see such optimizations also available to the JIT and NGEN compilers in the .NET platform.
Really liked the talk,
It's quite amazing how the hardware evolved and how the VC team compiler catch up with auto vectorization and parallel constructions with PPL and AMP. Visual Studio 2012 looks amazing. Can wait for this to be mainstream.
One thing that really bothered me was the use of "movps" and "pmax". That should probably be "movups" or maybe "movaps" (the other slide used _mm_loadu_ps though) and "pmaxsd".
And I'm left wondering, is that vectorizor going to emit unaligned loads all over the place? Because that kind of sucks. Ok unaligned loads got a little faster recently, but still.
Remove this comment
Remove this thread
close