Adel Kara Slimane comments

Results 68 comments of


                                            Adel Kara Slimane

Static type hashing of std::complex

> I am talking about the overload of static_type_hash Alright, yeah I do not have any overload of `static_type_hash` > As you mentioned, the data layout of std::complex depends only...

Static type hashing of std::complex

My bad, my workaround with `reinterpret_cast` didn't work because of a mistake on my end. It does work and there's no need to overload `static_type_hash` for it to work. I...

media-libs/libvmaf: add slotted 3.0.0, update ffmpeg ebuilds

@juippis - [NVTX is something else](https://github.com/NVIDIA/NVTX) that doesn't look like is provided by `dev-util/nvidia-cuda-toolkit` so it cannot be merged in the same useflag. - I tried implementing the `cuda` useflag...

media-libs/libvmaf: add slotted 3.0.0, update ffmpeg ebuilds

@juippis thanks for your feedback. [Cuda already got requested](https://bugs.gentoo.org/927296), it didn't take long haha.

Lutris 0.5.17 Fails to Correctly set the Vulkan ICD Loader

I was having stutters for a few months and I realized just today that I was running on `amdgpu-pro` all this time, turns out the ICD loader choice in Lutris...

Safety concerns regarding compiler optimizations that introduce FPU/SIMD code that may bypass kernel_fpu_begin/end() guards designed to prevent the kernel from clobbering userspace FP/SIMD register context

Hey! If you check this line in the kernel's make file https://github.com/gregkh/linux/blob/master/arch/x86/Makefile#L70 ```Makefile # # Prevent GCC from generating any FP code by mistake. # # This must happen before...

Safety concerns regarding compiler optimizations that introduce FPU/SIMD code that may bypass kernel_fpu_begin/end() guards designed to prevent the kernel from clobbering userspace FP/SIMD register context

I just went and checked [the patch](https://github.com/jevinskie/linux-cobs-muxed-ttys/commit/7551ceede430a2650aeaec50779b8d4372604f98#diff-086e0dad4e257bfda1470cd74f56c17876dfc12270ec207efd31f94520580522) @jevinskie referenced and he is aware of those lines in the `Makefile` and adds some extra flags to disable newer AVX instructions (`-mno-avx2`...

Safety concerns regarding compiler optimizations that introduce FPU/SIMD code that may bypass kernel_fpu_begin/end() guards designed to prevent the kernel from clobbering userspace FP/SIMD register context

I think the Linux kernel is relying on the fact that `-mno-avx` disables all the other ones. Otherwise they would've updated those lines ? Or maybe that's what everyone is...