I had already collected pointers given by Paul A. Lalonde or Bakul Shah about GPUs, because they do introduce, de facto, multi heterogeneous cores with NUMa even on commodity hardware, and GPUs are inherently manycore, NUMa, multithreaded and parallel (SIMD) changing programmation paradigms for this kind of task.
So questions (questions, not necessarily answers: what applies to GPUs can not apply to tightly coupled cores handled by a kernel) arise and are linked to the Nix approach. I have found a very good in-depth but with reasonable length exposition, written by John Nickolls and David Kirk (NVIDIA) in Patterson and Hennessy "Computer Organization and Design, the hardware/software interface", 4th edition, about x386 (this is why I bought this as a used book). But it happens that this appendix, unchanged, is still available as on online resource for new editions (I have the RISC-V and even the MIPS ones) here: https://www.elsevier.com/books-and-journals/book-companion/9780128203316 (look for appendix B). I have added the link in: http://notes.kergis.com/nix-os.html#bibliography For people like me with a very vague knowledge in this area, it is worth reading. -- Thierry Laronde <tlaronde +AT+ kergis +dot+ com> http://www.kergis.com/ http://kertex.kergis.com/ Key fingerprint = 0FF7 E906 FBAF FE95 FD89 250D 52B1 AE95 6006 F40C ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T051d4ea26fa16841-M94ee974364c863533972c87c Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
