v0.4.0

UnsafeAtomics v0.4.0

Diff since v0.3.4

UnsafeAtomics 0.4 is the front-end for atomics on CPUs and GPUs. Each operation emits one LLVM atomic instruction with the requested ordering and syncscope. Ptr and LLVMPtr now share a single implementation, and GPUCompiler (2.12 and later) legalizes that instruction for each GPU back-end.

Breaking changes

• UnsafeAtomics now depends on LLVM.jl. The UnsafeAtomicsLLVM extension is gone, and LLVMPtr support is always available.
• Floating-point max! and min! follow Julia's semantics, so a NaN propagates. Version 0.3 emitted atomicrmw fmax/fmin on LLVMPtr. The new fmax! and fmin! keep the IEEE maxNum/minNum behavior.
• The default scope is renamed from none to system. none remains as an alias, but code that dispatches on LLVMSyncScope{:none} has to switch to LLVMSyncScope{:system}. Scopes no longer print as LLVM IR. The default scope is still system on every pointer type. GPU front-ends that want device-scope atomics should pass device.
• On x86, fence(seq_cst) is a plain fence instruction, as in Threads.atomic_fence(). 0.3 used inline assembly before LLVM 20.
• Code compiled by Julia for the host CPU uses the system scope for every scope except singlethread. The CPU back-ends don't support other scopes: on x86, for example, a seq_cst fence in any other scope compiled to no instruction.
• Orderings, scopes and keyword arguments have to be compile-time constants. Anything else results in a dynamic call. Back-ends that pass them through their own functions can call the Val-based primitives (UnsafeAtomics.Internal.llvm_load and friends), which are now public.

New features

• Public scopes singlethread, subgroup, workgroup, device and system, plus UnsafeAtomics.SyncScope(name) for target-specific scopes. Every operation and fence takes one.
• Orderings and scopes can be passed as Symbols.
• The volatile, align and weak keywords, and a failure_order for compare-and-swap.
• New operations: inc_wrap!, dec_wrap!, sub_cond!, sub_sat!, fmax! and fmin!.
• Support for 8-bit integers, Bool, Float16, BFloat16, Int128 and pointers. Operations without a native instruction fall back to a compare-and-swap loop, so every operation works for every supported type.
• With GPUCompiler 2.12, device code gets every instruction the IR can express, with the requested scope. GPUCompiler then renames the scopes for each target.

Fixes

• Operations on Core.BFloat16 that needed a compare-and-swap loop threw a MethodError.

v0.3.4

UnsafeAtomics v0.3.4

Diff since v0.3.3

v0.3.3

UnsafeAtomics v0.3.3

Diff since v0.3.2

Merged pull requests:

• CI: run on main and test released Julia 1.13 (#26) (@maleadt)
• Support LLVM.jl 10 in the LLVM extension (0.3.3) (#31) (@maleadt)

v0.3.2

UnsafeAtomics v0.3.2

Diff since v0.3.1

Merged pull requests:

• Fix fence elision on Julia 1.10 and 1.11 (#23) (@maleadt)
• Preserve fence behavior across Julia backports (#24) (@maleadt)

v0.3.1

UnsafeAtomics v0.3.1

Diff since v0.3.0

Merged pull requests:

• update to opaque ptr (#22) (@vchuravy)

v0.3.0

UnsafeAtomics v0.3.0

Diff since v0.2.1

Merged pull requests:

• Use atomic_pointer* methods when available (#14) (@jpsamaroo)
• Use a package extension for UnsafeAtomicsLLVM (#15) (@vchuravy)
• Fix load and store for sizeof(T) == 0 (#18) (@vchuravy)
• Implement UnsafeAtomics.fence (#19) (@vchuravy)
• implement syncscopes everywhere (#20) (@vchuravy)
• Dispatch to LLVM atomic OP (#21) (@pxl-th)

Closed issues:

• Properly support fence (#16)
• Support for passing synchronization scopes (#17)

v0.2.1

UnsafeAtomics v0.2.1

Diff since v0.2.0

Closed issues:

• Missing support for floating point operations (#8)

Merged pull requests:

• Add floating point operations (#9) (@jpsamaroo)
• Rename: .github/workflows/{test.yml -> ci.yml} (#10) (@tkf)
• Setup Aqua and code-style CI jobs (#11) (@tkf)
• Support and test primitive types (#12) (@tkf)
• Simplify fallback cas! (#13) (@tkf)

v0.2.0

UnsafeAtomics v0.2.0

Diff since v0.1.0

Merged pull requests:

• Fix modify!; return a Pair (#6) (@tkf)

v0.1.0

UnsafeAtomics v0.1.0

Merged pull requests:

• Setup CI (#1) (@tkf)
• Improve dispatch pipeline for RMW (#2) (@tkf)
• Vendor some Julia internals (#3) (@tkf)
• Test print (#4) (@tkf)