[docs] fastmem draft

Signed-off-by: lizzie <lizzie@eden-emu.dev>
This commit is contained in:
lizzie 2025-09-03 09:10:18 +00:00 committed by crueter
parent 21ca974f8a
commit 1106494f6b

View file

@ -4,7 +4,7 @@ The main way of accessing memory in JITed programs is via an invoked function, s
The solution? Delegate invalid accesses to a dedicated arena, similar to a swap. The main idea behind such mechanism is to allow the OS to transmit page faults from invalid accesses into the JIT translator directly, bypassing address space calls, while this sacrifices i-cache coherency, it allows for smaller code-size and "faster" throguhput.
Many kernels however, do not support fast signal dispatching (Solaris, OpenBSD, FreeBSD). Only Linux and Windows support relatively "fast" signal dispatching. Hence this feature is better suited for them only.
Many kernels however, do not support fast signal dispatching (Solaris, *BSD). Only Linux and Windows support relatively "fast" signal dispatching. Hence this feature is better suited for them only.
![Host to guest translation](./HostToGuest.svg)