v0.0.162

What's Changed

• mlir: Modify BatchOpInterface to allow generating more than one operation by @Pangoraw in #2142
• Fix memtransfer of anything by @wsmoses in #2152
• Improve canonicalization error message by @wsmoses in #2154

Full Changelog: v0.0.161...v0.0.162

v0.0.161

What's Changed

• Update BCLoader.cpp by @wsmoses in #2148
• BCLoad: use enzyme_math by @wsmoses in #2149
• Fix named bcloader by @wsmoses in #2150

Full Changelog: v0.0.160...v0.0.161

v0.0.160

What's Changed

• Activity analysis: induction by @wsmoses in #2143
• Remove bad print by @wsmoses in #2144
• Mark force recompute by @wsmoses in #2145
• Fix enzyme_nocache by @wsmoses in #2146
• Cast of addr13 by @wsmoses in #2147

Full Changelog: v0.0.159...v0.0.160

v0.0.159

What's Changed

• Update benchmarks by @ZuseZ4 in #2035
• Fix iv of constant by @wsmoses in #2141

Full Changelog: v0.0.158...v0.0.159

v0.0.158

What's Changed

• mlir: Func call reverse diff by @Pangoraw in #2127
• mlir: implement forward mode for func.call by @Pangoraw in #2134
• mlir: cache and reuse reverse funcs by @Pangoraw in #2133
• Fixes warnings and adds missing header guards by @MattBolitho in #2124
• Improve cache index error message by @wsmoses in #2138
• tgamma error improvement by @wsmoses in #2139
• Implement tgamma derivative by @wsmoses in #2140

New Contributors

• @Pangoraw made their first contribution in #2127

Full Changelog: v0.0.157...v0.0.158

v0.0.157

What's Changed

• Adapt to const tblgen by @wsmoses in #2128
• Fix undef value storage by @wsmoses in #2129
• Improve runtime activity err message by @wsmoses in #2132
• Fix combined temp cache for reverse by @wsmoses in #2131

Full Changelog: v0.0.156...v0.0.157

v0.0.156

What's Changed

• Add gcloaded TT by @wsmoses in #2125

Full Changelog: v0.0.155...v0.0.156

v0.0.155

What's Changed

• Add header files to ClangEnzyme target by @jandrej in #2062
• Fix blas decl updater indexing by @wsmoses in #2123

New Contributors

• @jandrej made their first contribution in #2062

Full Changelog: v0.0.154...v0.0.155

v0.0.154

What's Changed

• Main tmp by @wsmoses in #2095
• Adapt to const tblgen main function by @wsmoses in #2099
• Update MustExitScalarEvolution.cpp for upstream LLVM changes by @dklimkin in #2098
• Fix llvm main build by @wsmoses in #2100
• add sinh/cosh/tanh support for llvm19+ by @ZuseZ4 in #2096
• Size returning new handler by @wsmoses in #2101
• MLIR: permit locals in mlir tblgen by @wsmoses in #2102
• MLIR: improved tensor type error messages by @wsmoses in #2103
• MLIR: orig to new map cf by @wsmoses in #2104
• MLIR: fix invert pointer by @wsmoses in #2105
• MLIR: AD Dataflow by @wsmoses in #2106
• MLIR: Recursive readonly by @wsmoses in #2107
• MLIR: Consider readonly in activity analysis by @wsmoses in #2108
• MLIR: improve dataflow interface by @wsmoses in #2110
• MLIR: prioritize dataflow interface by @wsmoses in #2111
• MLIR: fix result positions to shadow by @wsmoses in #2112
• Preserve primal shadow info by @wsmoses in #2113
• [mlir] Fix build for LLVM 20 by @pengmai in #2115
• MLIR: Use terminator location for arg-less reverse function by @wsmoses in #2116
• LLVM integrate for mlir changes by @wsmoses in #2118
• Support more Julia 1.11 functions by @wsmoses in #2121
• Fix handle sync by @wsmoses in #2122
• Improve unknown function error messages by @wsmoses in #2120

New Contributors

• @dklimkin made their first contribution in #2098

Full Changelog: v0.0.153...v0.0.154

v0.0.153

What's Changed

• Disable sparse tests on 18+ by @wsmoses in #2093
• Subnull even with runtime activity by @wsmoses in #2094

Full Changelog: v0.0.152...v0.0.153