v0.8.1

GPUCompiler v0.8.1

Diff since v0.8.0

Merged pull requests:

• Don't print multiple identical compiler invocations during reflection. (#102) (@maleadt)

v0.8.0

GPUCompiler v0.8.0

Diff since v0.7.3

Closed issues:

• Revisit kernel wrapping / byval lowering (#38)
• Remove TBAA helpers (#56)
• Manual compile cache (#88)

Merged pull requests:

• Fix application of SPIRV calling convention attrs. (#84) (@maleadt)
• Update manifest (#87) (@github-actions[bot])
• Ability to freeze compiled kernels (#89) (@maleadt)
• [GCN] Fake target early, rewrite allocas later (#90) (@jpsamaroo)
• Update manifest (#91) (@github-actions[bot])
• Deduplicate byval argument handling. (#94) (@maleadt)
• Work around Intel driver issue with byval. (#95) (@maleadt)
• Fix SPIRV trap handling (#96) (@maleadt)
• More byval work (#98) (@maleadt)
• Create structs and enums in the correct context. (#99) (@maleadt)
• CompatHelper: add new compat entry for "Scratch" at version "1.0" (#100) (@github-actions[bot])

v0.7.3

GPUCompiler v0.7.3

Diff since v0.7.2

Merged pull requests:

• Update manifest (#80) (@github-actions[bot])
• Catch some additional early errors. (#81) (@maleadt)
• Write the runtime to disk atomically. (#82) (@maleadt)
• Retry runtime library compilation if it failed to load. (#83) (@maleadt)

v0.7.2

GPUCompiler v0.7.2

Diff since v0.7.1

Merged pull requests:

• Update manifest (#77) (@github-actions[bot])
• Latency improvements (#78) (@maleadt)
• Update manifest (#79) (@github-actions[bot])

v0.6.1

GPUCompiler v0.6.1

Diff since v0.6.0

Closed issues:

• Make Julia's optimization passes configurable (#23)

Merged pull requests:

• Fix AMDGPU reverse CI (#34) (@jpsamaroo)
• Adapt to LLVM.jl changes for stateless codegen. (#70) (@maleadt)
• GCN: Rewrite allocas from AS 0 to 5 (#71) (@jpsamaroo)
• UB LLVM to 3.0, bump version to 0.7.0 (#72) (@jpsamaroo)
• Fix GCN alloca test (#73) (@jpsamaroo)
• bump DataStructures version (#75) (@DhairyaLGandhi)

v0.7.1

GPUCompiler v0.7.1

Diff since v0.7.0

Closed issues:

• Make Julia's optimization passes configurable (#23)

Merged pull requests:

• Fix AMDGPU reverse CI (#34) (@jpsamaroo)
• GCN: Rewrite allocas from AS 0 to 5 (#71) (@jpsamaroo)
• Fix GCN alloca test (#73) (@jpsamaroo)
• bump DataStructures version (#75) (@DhairyaLGandhi)

v0.7.0

GPUCompiler v0.7.0

Diff since v0.6.0

Merged pull requests:

• Adapt to LLVM.jl changes for stateless codegen. (#70) (@maleadt)
• UB LLVM to 3.0, bump version to 0.7.0 (#72) (@jpsamaroo)

v0.6.0

GPUCompiler v0.6.0

Diff since v0.5.5

Merged pull requests:

• Reinstate Julia 1.3 support. (#67) (@maleadt)
• Don't rename unofficial LLVM intrinsics. (#68) (@maleadt)

v0.5.5

GPUCompiler v0.5.5

Diff since v0.5.4

Merged pull requests:

• Sanitize all function names again. (#64) (@maleadt)

v0.5.4

GPUCompiler v0.5.4

Diff since v0.5.3

Merged pull requests:

• Fix pygmentize detection. (#63) (@maleadt)