Hello everyone, with Delphi 10.3 release
Hello everyone, with Delphi 10.3 release
I'd like your opinion, please.
Finally we will have an optimized, modern compiler
with vectorization support, SIMD, AVX2 and etc?
I see in this list below a series of compilers
with its advantages and disadvantages.
https://en.wikipedia.org/wiki/List_of_compilers
In this list we have mixed/cross-compilers C ++ and CUDA
Will I find this with Delphi 10.3?
Embarcadero could create a Wrapper and compilation parameters to facilitate the I work with AVX, because I have to do it all manually ever.
What are the performance advantages of my clients
will receive by me to develop using this new pascal syntax
will it generate more optimized code or will it be more of the same?
Thank you.
https://en.wikipedia.org/wiki/List_of_compilers
I'd like your opinion, please.
Finally we will have an optimized, modern compiler
with vectorization support, SIMD, AVX2 and etc?
I see in this list below a series of compilers
with its advantages and disadvantages.
https://en.wikipedia.org/wiki/List_of_compilers
In this list we have mixed/cross-compilers C ++ and CUDA
Will I find this with Delphi 10.3?
Embarcadero could create a Wrapper and compilation parameters to facilitate the I work with AVX, because I have to do it all manually ever.
What are the performance advantages of my clients
will receive by me to develop using this new pascal syntax
will it generate more optimized code or will it be more of the same?
Thank you.
https://en.wikipedia.org/wiki/List_of_compilers
Jennifer Powell Thank you for your great words!
ReplyDeleteAs they say here in my country, You summed up the opera!
I believe I should "look" at PGI as an alternative! because the focus of the embarcadero is following other ways!
I will reflect on all this, thank you for your generosity!
If you dont want the bloat of the CUDA SDK install, you can get OpenCL and what you need from Khronos Group.
ReplyDeleteOne of the nice things about the OpenCL dll is that it contains a procedural C compiler and it will let you compile a execute a CUDA image on the fly, even for x86.
There is also a version for Android.