Visual C++ Team Blog - All Comments

re: Deep Dive into NuGet Native (Part One)

Garrett Serack, MSFT — Wed, 19 Jun 2013 15:02:00 GMT

Sorry about that broken link -- fixed.

re: Deep Dive into NuGet Native (Part One)

Henrik — Wed, 19 Jun 2013 13:53:50 GMT

Your channel9 link links to a login page. This link works for me:

"channel9.msdn.com/Shows/C9-GoingNative/GoingNative-16-Garrett-Serak-Inside-NuGet-for-C"

channel9.msdn.com/.../GoingNative-16-Garrett-Serak-Inside-NuGet-for-C

re: Deep Dive into NuGet Native (Part One)

Morons — Wed, 19 Jun 2013 07:47:51 GMT

The forums or whateverthehell thatcrap is is just stupid. It is THE MOST stupid thing MS has ever done, even nore than Win8. Well.

re: Optimizing C++ Code : Overview

knowinfonow — Wed, 19 Jun 2013 07:38:17 GMT

Nice and very informative information on code cleaning, but needed one online tool to clean up codes of blogger to increase speed

www.knowinfonow.com

re: Deep Dive into NuGet Native (Part One)

Neil — Wed, 19 Jun 2013 00:34:15 GMT

Channel 9 link is a little off and references post editor:

blogs.msdn.com/.../GoingNative-16-Garrett-Serak-Inside-NuGet-for-C

re: Optimizing C++ Code : Overview

Jim Hogg — Tue, 18 Jun 2013 21:53:49 GMT

@Simon

We don't document, or support, feeding parsed source code into UTC, unfortunately. And yes, that was (one of) the many aims of Phoenix (where I spent 4 interesting years). Although Phoenix disappeared as a shipping product, that technology, and part of the team, live on within another group inside Microsoft.

re: Optimizing C++ Code : Overview

Mike — Tue, 18 Jun 2013 14:57:04 GMT

@Simon Buchan - well, that was the idea with Phoenix but unfortunately Phoenix disappeared...

re: Optimizing C++ Code : Overview

Simon Buchan — Tue, 18 Jun 2013 06:52:28 GMT

I realize this probably isn't something that will ever be supported, but being able to target C2.exe from my own front-end would be cool - get a bunch of optimizations for free, possibly also with whole-program optimization support and binary interop. Doing something more sane like targeting LLVM is probably a better idea though :)

I might try pulling apart those /Bk outputs for fun some time...

re: Profile Guided Optimization (PGO) – Under the Hood

Alexander Tretyak — Sat, 15 Jun 2013 15:54:37 GMT

Hi. Can you ask someone from your compiler dev-team to add this simple feature - analogue of __builtin_expect from GCC/ICC, or at least likely/unlikely macro functionality equivalent (likely(x) = __builtin_expect(!!(x), 1)), may be with other syntax, e.g. __pragma(unlikely) if (...) {}

As PGO is not always can be easily integrated for all type of applications, while there is some common heavy-used code, which can be optimized much better (without PGO training phase) if it were possible to somehow instruct compiler at most frequent cases (i.e. I suggest to just provide a way to manually guide a compiler when trained data is unavailable/absent for a specific branch in the source code, so this pragma need to work only when compiled with /LCCG:PGO option).

Thanks in advance!

re: Optimizing C++ Code : Overview

Jim Hogg — Sat, 15 Jun 2013 04:04:08 GMT

@Tom Kirby-Green

Gathering posts together? Yes - I am now realizing that some of the interesting discussion arises from questions, so a PDF summary would usefully require an editor pass to merge text and Q&A together.