2396806165
Add LLVM bitcode writer
2024-02-21 16:24:59 +01:00
4653fc4bb4
LLVM Builder: Add dbg.declare and dbg.value intrinsics
2024-02-21 16:24:59 +01:00
15029590ee
LLVM Builder: Add integer values to more enums fields
2024-02-21 16:24:59 +01:00
ec5a433fdf
Fix FastMath packed struct
...
* Added missing legacy field (unused_algebra)
* Made struct correct size (u32 -> u8)
2024-02-21 16:24:59 +01:00
7cb8813743
Added value_indices and valueIndex to Function
...
value_indices keeps track of the value index of
each instruction in the function (i.e skips
instruction which do not have a result)
2024-02-21 16:24:59 +01:00
423c2c3a27
Made .block = false in WipFunction.hasResultWip
2024-02-21 16:24:59 +01:00
c57b4e70b0
Builder: Add function_attributes_set
2024-02-21 16:24:59 +01:00
52e8434022
Added opcode functions to Instruction/Constant.Tag
2024-02-21 16:24:59 +01:00
ff76ba64d6
Added values to AtomicRmw.Operation enum fields
2024-02-21 16:24:59 +01:00
6df8302be6
Fixed values in AtomicOrdering enum
2024-02-21 16:24:59 +01:00
fd3b81f93b
LLVM Builder: Add toLlvm helper to Alignment
2024-02-21 16:24:59 +01:00
2801bf6400
LLVM Builder: Add strtab helper to String
2024-02-21 16:24:59 +01:00
9ccd7158b9
LLVM Builder: Make Type.Simple reflect LLVM codes
2024-02-21 16:24:59 +01:00
1d94e9ef83
LLVM: Make sure child types get added first
...
The LLVM bitcode requires all type references in
structs to be to earlier defined types.
We make sure types are ordered in the builder
itself in order to avoid having to iterate the
types multiple times and changing the values
of type indicies.
2024-02-21 16:24:59 +01:00
8feae5d2d5
LLVM: Assign correct values to enum/union tags
2024-02-21 16:24:59 +01:00
ec7293422d
llvm: fix c abi for structs not passed in registers
...
Closes #18916
2024-02-20 23:36:47 +01:00
97290e0bfc
cbe: make more use of InternPool.Index.Slice
...
Attempts to fix the current CI failures introduced by #18983 .
2024-02-19 19:32:06 -08:00
dd4d320eb9
Merge pull request #18948 from alichraghi/vector
...
spirv: use extended instructions whenever possible
2024-02-17 14:01:07 +01:00
6f08e17229
InternPool: make more use of NullTerminatedString.Slice
...
This should avoid the random pointer invalidation crashes.
Closes #18954
2024-02-16 00:27:25 -08:00
23b7075656
spirv: add capability to compile with x86_64 backend
2024-02-15 17:28:15 +03:30
44c31194e3
spirv: use extended instructions whenever possible
2024-02-15 17:25:44 +03:30
6fe90a913a
llvm: fix lowering of recursive debug info
...
This change allows recursing over types that are currently being
resolved fully with a second pass of forward resolution.
Closes #16414
2024-02-15 02:46:18 -08:00
a9f738e56b
x86_64: implement c abi for bool vectors
2024-02-12 05:25:07 +01:00
72bd1cd378
spirv: remove now-redundant isUnused calls from AIR handler functions
2024-02-10 09:50:02 +03:30
5ee96688a7
spirv: emit vectorized operations
2024-02-09 16:40:34 +03:30
ddcea2cad4
Merge pull request #18857 from alichraghi/shader
...
spirv: make rusticl the primary testing implementation
2024-02-09 14:11:31 +02:00
37b0aa600a
spirv: make rusticl the primary testing implementation
2024-02-09 09:27:08 +03:30
eb2d61d02e
spirv: merge construct(Struct/Vector/Array) into constructComposite
2024-02-09 09:27:05 +03:30
3122fd0ba0
Merge pull request #17634 from ianprime0509/type-erased-writer
...
Add type-erased writer and GenericWriter
2024-02-07 23:52:53 -08:00
38c2a25735
Builder: fix memory leaks
2024-02-06 22:31:13 -05:00
78f15bc714
compiler: rename value.zig to Value.zig
...
This commit only does the file rename to be friendlier to version
control conflicts.
2024-02-05 18:13:07 -07:00
739108c9f0
spirv: support enum integer values in Assembler
2024-02-05 11:55:14 +03:30
b41aad0193
spirv: emit vectors whenever we can
2024-02-05 11:55:14 +03:30
afa7793351
spirv: basic shader support
2024-02-05 11:55:14 +03:30
7634a115c5
Merge pull request #18580 from Snektron/spirv-more-vectors
...
spirv: more vector operations
2024-02-05 09:24:49 +01:00
eaa6218f09
x86_64: fix errors compiling the compiler
...
This fixes issues targetting both `x86_64-linux` and `x86_64-macos` with
the self-hosted backend.
2024-02-04 22:58:38 -05:00
2511106150
spirv: air vector_store_element
2024-02-04 19:09:34 +01:00
1d548aa2aa
spirv: air splat
2024-02-04 19:09:33 +01:00
76d5696434
spirv: air abs
2024-02-04 19:09:32 +01:00
631d1b63a8
spirv: fix shuffle properly
2024-02-04 19:09:32 +01:00
9641d2ebdb
spirv: vectorize max, min
2024-02-04 19:09:31 +01:00
9f0227a326
spirv: vectorize int_cast, trunc
2024-02-04 19:09:31 +01:00
408c117246
spirv: air is_(non_)null_ptr, optional_payload_ptr
2024-02-04 19:09:30 +01:00
7dfd403da1
spirv: air mul_add
2024-02-04 19:09:30 +01:00
345d6e280d
spirv: air int_from_bool
2024-02-04 19:09:29 +01:00
77ef78a0ef
spirv: clean up arithmeticTypeInfo a bit
...
- No longer returns an error
- Returns more useful vector info
2024-02-04 19:09:29 +01:00
54ec936549
spirv: wrap strange its before instead of after operation
...
Wrapping strange integers before an operation was initially
done as an attempt to minimize the amount of normalizations
required: This way, there would not be a normalization
necessary between two modular operations. This was a
premature optimization, since the resulting logic is more
complicated than naive way of wrapping the result after
the operation.
This commit updates handling of strange integers to do
wrapping after each operation. It also seems slightly
more efficient in terms of size of generated code, as
it reduces the size of the behavior tests binary by
about 1%.
2024-02-04 19:09:28 +01:00
b67d983abd
spirv: vectorize add/sub overflow
2024-02-04 19:09:27 +01:00
761594e226
spirv: reduce, reduce_optimized
2024-02-04 19:09:27 +01:00
2f815853dc
spirv: shlWithOverflow
2024-02-04 19:09:26 +01:00
antlilja