| Age | Commit message (Collapse) | Author | Lines |
|---|
|
TypeTree support in autodiff
# TypeTrees for Autodiff
## What are TypeTrees?
Memory layout descriptors for Enzyme. Tell Enzyme exactly how types are structured in memory so it can compute derivatives efficiently.
## Structure
```rust
TypeTree(Vec<Type>)
Type {
offset: isize, // byte offset (-1 = everywhere)
size: usize, // size in bytes
kind: Kind, // Float, Integer, Pointer, etc.
child: TypeTree // nested structure
}
```
## Example: `fn compute(x: &f32, data: &[f32]) -> f32`
**Input 0: `x: &f32`**
```rust
TypeTree(vec![Type {
offset: -1, size: 8, kind: Pointer,
child: TypeTree(vec![Type {
offset: -1, size: 4, kind: Float,
child: TypeTree::new()
}])
}])
```
**Input 1: `data: &[f32]`**
```rust
TypeTree(vec![Type {
offset: -1, size: 8, kind: Pointer,
child: TypeTree(vec![Type {
offset: -1, size: 4, kind: Float, // -1 = all elements
child: TypeTree::new()
}])
}])
```
**Output: `f32`**
```rust
TypeTree(vec![Type {
offset: -1, size: 4, kind: Float,
child: TypeTree::new()
}])
```
## Why Needed?
- Enzyme can't deduce complex type layouts from LLVM IR
- Prevents slow memory pattern analysis
- Enables correct derivative computation for nested structures
- Tells Enzyme which bytes are differentiable vs metadata
## What Enzyme Does With This Information:
Without TypeTrees (current state):
```llvm
; Enzyme sees generic LLVM IR:
define float ``@distance(ptr*`` %p1, ptr* %p2) {
; Has to guess what these pointers point to
; Slow analysis of all memory operations
; May miss optimization opportunities
}
```
With TypeTrees (our implementation):
```llvm
define "enzyme_type"="{[]:Float@float}" float ``@distance(``
ptr "enzyme_type"="{[]:Pointer}" %p1,
ptr "enzyme_type"="{[]:Pointer}" %p2
) {
; Enzyme knows exact type layout
; Can generate efficient derivative code directly
}
```
# TypeTrees - Offset and -1 Explained
## Type Structure
```rust
Type {
offset: isize, // WHERE this type starts
size: usize, // HOW BIG this type is
kind: Kind, // WHAT KIND of data (Float, Int, Pointer)
child: TypeTree // WHAT'S INSIDE (for pointers/containers)
}
```
## Offset Values
### Regular Offset (0, 4, 8, etc.)
**Specific byte position within a structure**
```rust
struct Point {
x: f32, // offset 0, size 4
y: f32, // offset 4, size 4
id: i32, // offset 8, size 4
}
```
TypeTree for `&Point` (internal representation):
```rust
TypeTree(vec![
Type { offset: 0, size: 4, kind: Float }, // x at byte 0
Type { offset: 4, size: 4, kind: Float }, // y at byte 4
Type { offset: 8, size: 4, kind: Integer } // id at byte 8
])
```
Generates LLVM:
```llvm
"enzyme_type"="{[]:Float@float}"
```
### Offset -1 (Special: "Everywhere")
**Means "this pattern repeats for ALL elements"**
#### Example 1: Array `[f32; 100]`
```rust
TypeTree(vec![Type {
offset: -1, // ALL positions
size: 4, // each f32 is 4 bytes
kind: Float, // every element is float
}])
```
Instead of listing 100 separate Types with offsets `0,4,8,12...396`
#### Example 2: Slice `&[i32]`
```rust
// Pointer to slice data
TypeTree(vec![Type {
offset: -1, size: 8, kind: Pointer,
child: TypeTree(vec![Type {
offset: -1, // ALL slice elements
size: 4, // each i32 is 4 bytes
kind: Integer
}])
}])
```
#### Example 3: Mixed Structure
```rust
struct Container {
header: i64, // offset 0
data: [f32; 1000], // offset 8, but elements use -1
}
```
```rust
TypeTree(vec![
Type { offset: 0, size: 8, kind: Integer }, // header
Type { offset: 8, size: 4000, kind: Pointer,
child: TypeTree(vec![Type {
offset: -1, size: 4, kind: Float // ALL array elements
}])
}
])
```
|
|
|
|
|
|
|
|
subtree-update_cg_gcc_2025-08-26
|
|
subtree-update_cg_gcc_2025-08-04
|
|
default data address space
|
|
subtree-update_cg_gcc_2025-06-30
|
|
Add SIMD funnel shift and round-to-even intrinsics
This PR adds 3 new SIMD intrinsics
- `simd_funnel_shl` - funnel shift left
- `simd_funnel_shr` - funnel shift right
- `simd_round_ties_even` (vector version of `round_ties_even_fN`)
TODO (future PR): implement `simd_fsh{l,r}` in miri, cg_gcc and cg_clif (it is surprisingly hard to implement without branches, the common tricks that rotate uses doesn't work because we have 2 elements now. e.g, the `-n&31` trick used by cg_gcc to implement rotate doesn't work with this because then `fshl(a, b, 0)` will be `a | b`)
[#t-compiler > More SIMD intrinsics](https://rust-lang.zulipchat.com/#narrow/channel/131828-t-compiler/topic/More.20SIMD.20intrinsics/with/522130286)
`@rustbot` label T-compiler T-libs A-intrinsics F-core_intrinsics
r? `@workingjubilee`
|
|
subtree-update_cg_gcc_2025-06-28
|
|
|
|
subtree-update_cg_gcc_2025-06-18
|
|
|
|
|
|
It is only used within cg_llvm.
|
|
|
|
|
|
|
|
This avoids having to get the function signature.
|
|
|
|
|
|
|
|
subtree-update_cg_gcc_2025-05-14
|
|
|
|
|
|
simd intrinsics with mask: accept unsigned integer masks, and fix some of the errors
It's not clear at all why the mask would have to be signed, it is anyway interpreted bitwise. The backend should just make sure that works no matter the surface-level type; our LLVM backend already does this correctly. The note of "the mask may be widened, which only has the correct behavior for signed integers" explains... nothing? Why can't the code do the widening correctly? If necessary, just cast to the signed type first...
Also while we are at it, fix the errors. For simd_masked_load/store, the errors talked about the "third argument" but they meant the first argument (the mask is the first argument there). They also used the wrong type for `expected_element`.
I have extremely low confidence in the GCC part of this PR.
See [discussion on Zulip](https://rust-lang.zulipchat.com/#narrow/channel/257879-project-portable-simd/topic/On.20the.20sign.20of.20masks)
|
|
|
|
subtree-update_cg_gcc_2025-04-18
|
|
|
|
remove `simd_fpow` and `simd_fpowi`
Discussed in https://github.com/rust-lang/rust/issues/137555
These functions are not exposed from `std::intrinsics::simd`, and not used anywhere outside of the compiler. They also don't lower to particularly good code at least on the major ISAs (I checked x86_64, aarch64, s390x, powerpc), where the vector is just spilled to the stack and scalar functions are used for the actual logic.
r? `@RalfJung`
|
|
|
|
intrinsics: unify rint, roundeven, nearbyint in a single round_ties_even intrinsic
LLVM has three intrinsics here that all do the same thing (when used in the default FP environment). There's no reason Rust needs to copy that historically-grown mess -- let's just have one intrinsic and leave it up to the LLVM backend to decide how to lower that.
Suggested by `@hanna-kruppe` in https://github.com/rust-lang/rust/issues/136459; Cc `@tgross35`
try-job: test-various
|
|
field in `LayoutData`.
Also update comments that refered to BackendRepr::Uninhabited.
|
|
things which are already immediates
That means it stops trying to truncate things that are already `i1`s.
|
|
|
|
|
|
intrinsic
|
|
Combined with [1], this will change the overflowing multiplication
operations to return an `extern "C"`-safe type.
Link: https://github.com/rust-lang/compiler-builtins/pull/735 [1]
|
|
|
|
subtree-update_cg_gcc_2025_01_12
|
|
Signed-off-by: acceptacross <csqcqs@gmail.com>
|
|
|
|
the behavior of the type system not only depends on the current
assumptions, but also the currentnphase of the compiler. This is
mostly necessary as we need to decide whether and how to reveal
opaque types. We track this via the `TypingMode`.
|
|
|
|
|
|
|
|
Add intrinsics `fmuladd{f16,f32,f64,f128}`. This computes `(a * b) +
c`, to be fused if the code generator determines that (i) the target
instruction set has support for a fused operation, and (ii) that the
fused operation is more efficient than the equivalent, separate pair
of `mul` and `add` instructions.
https://llvm.org/docs/LangRef.html#llvm-fmuladd-intrinsic
MIRI support is included for f32 and f64.
The codegen_cranelift uses the `fma` function from libc, which is a
correct implementation, but without the desired performance semantic. I
think this requires an update to cranelift to expose a suitable
instruction in its IR.
I have not tested with codegen_gcc, but it should behave the same
way (using `fma` from libc).
|
|
|
|
|
|
|
