https://github.com/crystal-lang/crystal/pull/7214 looks promising, but due to lack of deep understanding of how it works, I’d like to ask how it’s going to affect the language (if it’s merged)? From my point of view, the only good thing the PR brings is faster switching between fibers, but fibers th…

Is it possible? It’d be a 128bit atomic read on 64bit and a 64bit atomic read on 32bit. Not all architectures support that.

Actually, it’d be an unbounded size atomic read, since pointers are already safe: we can save the pointer to a register then read the type ID from the class header. The problem comes from unions of structs. For an Array(Struct1 | Struct2), where Struct1 and Struct2 are large structs, then how can th…

That’s why I think that segfaults and other undefined behavior can and will happen in Crystal if you don’t synchronize access to shared resources.

I’d rather disallow shared memory than have random segfaults and undefined behaviour. A crystal program which doesn’t use pointers or other explicitly unsafe constructs should never segfault.

So, I just tried this code: require "llvm" ctx = LLVM::Context.new mod = ctx.new_module("main") i256 = ctx.int(256) mod.functions.add("foo", ([] of LLVM::Type), i256) do |func| func.basic_blocks.append do |builder| x = builder.alloca i256 ld = builder.load x ld.ordering = :sequential…

Reading up on __atomic_load, I think it will probably be quite slow. But I think that’s fine, this is probably the best solution.

llvm-mirror/compiler-rt/blob/master/lib/builtins/atomic.c /*===-- atomic.c - Implement support functions for atomic operations.------=== * * The LLVM Compiler Infrastructure * * This file is dual licensed under the MIT and the University of Illinois Open * Source Lic…

I guess we can use that and slowly optimize things. For example the optimization is only needed for instance variables holding unions that aren’t read-only. Right now the compiler doesn’t detect read-only instance variables but it could, or we could add some keyword for that (for example readonly). …

Actually, it’s only needed for instance variables holding unions which contain structs, which is rare. Unions of classes are stored as simple pointers iirc.

I have a split brain on this topic. On one side, I hate “we should have safe types in multithreaded environment”. Multithreading IS THE PAIN and it is unavoidable! Even if we remove segfaults our programs will not magically became correct. Multithreaded program could be correct only if you thought …

MT schedulers impact

Crystal Contrib

rogerdpack October 14, 2021, 5:46am 36

Appears there is a (non documented) same_thread option to spawn Multithreaded Crystal initial thoughts - #3 by Blacksmoke16 I remember there was some discussion of work stealing at some point FWIW.

Topic		Replies	Views
Fiber is danger or not on Crystal?	31	2714	September 27, 2019
Wait-groups fiber pooling in 1.19	10	391	January 21, 2026
Crystal multithreading support Crystal Contrib	40	2983	March 20, 2026
Timeline for multithreading support Crystal Contrib	31	2794	September 24, 2024
Multi-thread program: Unhandled exception: Cannot allocate new fiber stack: Cannot allocate memory (Errno) Help & Support	21	1992	March 8, 2021

MT schedulers impact

Related topics