Bug#1049960: ITP: half -- C++ library for half precision floating point arithmetics
Hi Mathieu, On 2023-08-22 13:29, Mathieu Malaterre wrote: > On Thu, Aug 17, 2023 at 1:27 PM Christian Kastner wrote: >> * Package name: half >> Version : 2.2.0 >> Upstream Author : Christian Rau >> * URL : https://sourceforge.net/projects/half/ >> * License : MIT >> Programming Lang: C++ >> Description : C++ library for half precision floating point arithmetics >> >> This is a C++ header-only library to provide an IEEE-754 conformant >> half-precision floating point type along with corresponding arithmetic > > What's the difference with https://packages.debian.org/sid/libimath-dev ? I don't know the particular differences. MIOpen depends on this particular implementation. >> operators, type conversions and common mathematical functions. It aims >> for both efficiency and ease of use, trying to accurately mimic the >> behaviour of the builtin floating point types at the best performance >> possible. It automatically uses and provides C++11 features when >> possible, but stays completely C++98-compatible when neccessary. > > I believe gcc+c++20 provides _Float16 already. Who needs c++98 compat ? I guess I could raise that at MIOpen upstream. In any case, the package will remain in experimental until MIOpen itself is ready. >> This is needed by MIOpen, which is also in the process of being >> packaged. Best, Christian
Bug#1049960: ITP: half -- C++ library for half precision floating point arithmetics
On Thu, Aug 17, 2023 at 1:27 PM Christian Kastner wrote: > > Package: wnpp > Severity: wishlist > Owner: Christian Kastner > X-Debbugs-Cc: debian-de...@lists.debian.org, debian...@lists.debian.org > > * Package name: half > Version : 2.2.0 > Upstream Author : Christian Rau > * URL : https://sourceforge.net/projects/half/ > * License : MIT > Programming Lang: C++ > Description : C++ library for half precision floating point arithmetics > > This is a C++ header-only library to provide an IEEE-754 conformant > half-precision floating point type along with corresponding arithmetic What's the difference with https://packages.debian.org/sid/libimath-dev ? > operators, type conversions and common mathematical functions. It aims > for both efficiency and ease of use, trying to accurately mimic the > behaviour of the builtin floating point types at the best performance > possible. It automatically uses and provides C++11 features when > possible, but stays completely C++98-compatible when neccessary. I believe gcc+c++20 provides _Float16 already. Who needs c++98 compat ? > This is needed by MIOpen, which is also in the process of being > packaged. > > This will be maintained by the Debian ROCm Team. >
Bug#1049960: ITP: half -- C++ library for half precision floating point arithmetics
Package: wnpp Severity: wishlist Owner: Christian Kastner X-Debbugs-Cc: debian-de...@lists.debian.org, debian...@lists.debian.org * Package name: half Version : 2.2.0 Upstream Author : Christian Rau * URL : https://sourceforge.net/projects/half/ * License : MIT Programming Lang: C++ Description : C++ library for half precision floating point arithmetics This is a C++ header-only library to provide an IEEE-754 conformant half-precision floating point type along with corresponding arithmetic operators, type conversions and common mathematical functions. It aims for both efficiency and ease of use, trying to accurately mimic the behaviour of the builtin floating point types at the best performance possible. It automatically uses and provides C++11 features when possible, but stays completely C++98-compatible when neccessary. This is needed by MIOpen, which is also in the process of being packaged. This will be maintained by the Debian ROCm Team.