<t>QuoteExactly! What about the VS community? In that case PPL It is part of VS2010.AMD Core Math Library is free and includes LAPACK/BLAS and few other things.Does VS express license allow commercial distribution? MKL costs $399 and there are no royalties. It seems there is no extra charge for runn...

<t>QuoteWell, I programmed QR, SVD etc. before (in Fortran) and it is part of graduate numerical analysis studies. No problem, and we only need a small subset.Sorry, but that isn't very convincing. Can you deliver SVD implementation that is not 50 times slower than MKL? And who decides which subset ...

<t>QuotePortable, easy to install libraries, not necessarily optimised. BTW Math Toolkit's focus was accuracy, initially.Does anyone is planning to run this project on anything but x86? And what do you mean by "easy to install"?LAPACK/BLAS interface must be available to developers. All serious linea...

<t>Not sure about nice part, but LAPACK/BLAS and corresponding C-API (as it is found in MKL 10.3 and newer) are standardized interface for linear algebra.uBlas is no replacement for LAPACK/BLAS. As far as I know it doesn't work well with Intel/AMD/ATALS BLAS. uBlas is OK for small matrices, but too ...

<t>I should have said "...can't publish yet". Once this project gets started I will contribute my code.In the context of this project, I am interesting only in things that can't be done easily without C/C++. Right now there are several issues that need to be resolved:1. AVX capable CPU/compiler/OS.2...

Excellent reference. There are also several books by Stevens & friends.

I wrote some LMM code (both plain and Longstaff-Scwartz) that can run almost as fast as CUDA (or even faster, depending on hardware), but can't publish it due to lack of time and some other reasons.

<t>Linus is a kernel guru. Maybe he doesn't need C++ to get things done, but most people can't write kernels. It makes no sense to ask them to code everything in C.Now, since somebody already mentioned Java...I think Java is less OO than C++ by any measurable standard. And here is why.1) C++ support...

Let's be conservative and stick to C++03 for a while (next 2-3 years?).

<t>If 1-factor solver is SIMD friendly then 0.01s to call 1D solver should be possible and it is very conservative estimate. Single threaded scalar throughput is about 3 GFlops. With 0.01s per call it gives 30M floating point operations to solve 1-factor. With 4 SIMD capable cores we get around 20-5...

<t>CPU dispatching not used. Users can add CPU specific flags to their make file, but in my experience it doesn't make much difference.First, a big chunk of heavy duty number crunching is delegated to optimized libs. I think both MKL and ACML do CPU dispatching internally.Second, I observed that str...

<t>@Polter.No custom data types because STL containers and data types with alignment requirement don't mix well. See how Eigen deals with it. Not pretty.@outrun, Cuchulainn.What we need is the ability to gracefully fall back to SIMD=1. Those who don't want or can't write SIMD code should be able to ...

I more or less figured out parallel part, but SIMD part can get very user unfriendly. The main challenge is to shields users from low level coding. Some users can code SSE/AVX directly, but most users can't and don't want to do low level stuff. The challenge is how to make all users happy.

<t>Salvaging code from QL is an option, but I prefer a standalone library maintained by someone else. Reinventing the wheel is unproductive and hurts quality. I am all for licensing that permits commercial use, as long as it doesn't hurt the quality. At the end quality wins over quantity. And lets n...

I am guessing that GPL is no go. If so, we need quadratures similar to those found in GSL. I couldn't find anything non-GPL yet.