kernels

I can't sleep, so let's talk about kernels and threads

I'm working on a [i]next generation ENTERPRICE QUALITY microkernel based on Mach^tm.

The difference with Mach is that I have removed complitely the mach threads/cthreads shit so the user is forced to create new process and communicate via message passing. But I as far as I know modern CPUs can run in parrallel many threads, but what happens if they are different processes? I was thinking to just create user-space threads when needed but I don't think that will help.

Also I'm thinking to emulate pthreads/C11 threads via userspace threads, what do you think?

If you really want a next-gen kernel, try to implement Viengoos to work on Mach. I'd love to see a system where programs can tell the kernel about how important is some bits of memory so that the kernel can dynamically allocate and reclaim memory to cater to the demands of the programs in the system.


http://walfield.org/papers/2009-walfield-viengoos-a-framework-for-stakeholder-directed-resource-allocation.pdf

>>2
Thank you for the link sir. That sounds nice, it may be my next project

>>1
Multiple cores are supposed to being able to run multiple processes seperately.

Read Kunle Olukotun papers on multi-core processing.

He was the first to demonstrate that multi-core processors could outperform big superscalar processors in most applications, due to course-grained parallelism.

Part of his results on the multi-core chip were calculated by running multiple copies of a benchmark on the processor.

>>4
Found the paper: Kunle Olukotun, Basem A. Nayfeh, Lance Hammond, Kenneth G. Wilson, Kunyung Chang: The Case for a Single-Chip Multiprocessor. ASPLOS 1996: 2-11

>>4-5
Thank you for the paper but I was wondering about things like hyper-threading https://en.wikipedia.org/wiki/Simultaneous_multithreading

█

>>7
Troll.

>>8
█████████/prog/███████