Omar SobhandClaude Sonnet 4.6
082a50e3a0
CI / Build (ubuntu-latest) (push) Failing after 8s
Performance Benchmarks / Run Benchmarks (push) Successful in 8s
CI / Clippy Check (push) Failing after 8s
CI / Build CPU-Only (Explicit) (push) Failing after 8s
Documentation / Build API Documentation (push) Failing after 7s
CI / Format Check (push) Failing after 9s
GPU Tests / Check GPU Availability (push) Successful in 0s
Documentation / Build User Guide (push) Successful in 7s
GPU Tests / CUDA Tests (11.8) (push) Has been skipped
GPU Tests / CUDA Tests (12.1) (push) Has been skipped
CI / Build (macos-latest) (push) Failing after 14s
CI / Test (macos-latest) (push) Has been skipped
CI / Test (ubuntu-latest) (push) Has been skipped
CI / Python Bindings (maturin) (macos-latest) (push) Has been skipped
CI / Python Bindings (maturin) (ubuntu-latest) (push) Has been skipped
CI / WASM Build + Size Check (push) Has been skipped
CI / Distributed Training Tests (push) Has been skipped
CI / CI Success (push) Failing after 0s
GPU Tests / Metal Tests (push) Has been skipped
FP8 GPU FFI (rtx-tensor) - `fp8_cast.rs`: replaced `not_implemented` stubs with real cudarc 0.18.2 PTX launches; `cast_bf16_to_fp8_e4m3` and `cast_fp8_e4m3_to_bf16` now dispatch to NVCC-compiled `fp8_cast.ptx` via `LazyLock` module cache, matching the `inplace_ops` pattern - `build.rs`: `create_dummy_ptx` now emits `fp8_cast.ptx` alongside `element_wise.ptx` so `include_str!` resolves cleanly when NVCC is absent FlashAttention-3 typed kernel launch (rtx-flash-attention) - `flash_v3_forward.rs`: `forward()` now takes typed `CudaSlice<bf16>` Q/K/V/O + `CudaSlice<f32>` LSE buffer; dispatches via `stream.launch_builder` with block_dim=(128,1,1), grid_dim=(ceil(seq_len/64), batch*heads, 1), shared_mem_bytes=0 (PTX metadata-resolved) - `simple.rs`: added `has_flash_v3()` + `flash_attention_v3_forward_raw()` dispatch - `Cargo.toml`: `half` added as optional cuda-gated dependency CUDA Graphs stream threading (rtx-transformers) - `training_loop.rs`: added `cuda_stream: Option<CudaStreamHandle>` field; `set_cuda_backend()` now creates a non-default capture stream; capture step calls real `begin_capture(stream)` + `end_capture(stream)`; added `set_cuda_stream()` override; replay unchanged (no stream needed) SnapKV + prefix cache BatchScheduler wiring (rtx-inference) - `scheduler.rs`: added `prefix_hit_pages: Option<Vec<PageId>>` + `evicted_positions: Vec<usize>` to `SchedulerRequest`; `BatchScheduler` gains `kv_cache` + `snapkv_eviction` fields; `submit_request` does non-blocking `try_lock` prefix lookup; added `notify_prefill_complete` (registers prefix + runs `select_evict_positions`), `set_kv_cache`, `set_snapkv_eviction`, `get_evicted_positions`, `get_prefix_hit_pages` — +5 new integration tests Test results: 22 + 42 + 75 + 102 = 241 tests, 0 failures Co-Authored-By: Claude Sonnet 4.6 <[email protected]>
74 lines
2.0 KiB
TOML
74 lines
2.0 KiB
TOML
[package]
|
|
name = "rtx-flash-attention"
|
|
version = "1.0.0"
|
|
edition.workspace = true
|
|
rust-version = "1.92"
|
|
authors.workspace = true
|
|
license.workspace = true
|
|
repository.workspace = true
|
|
description = "Revolutionary Flash Attention implementation with Quantum/Neuromorphic/Edge variants"
|
|
|
|
[dependencies]
|
|
# Core RustyTorch dependencies
|
|
rtx-tensor = { workspace = true }
|
|
rtx-runtime = { workspace = true }
|
|
rtx-kernel = { workspace = true }
|
|
|
|
# GPU acceleration
|
|
cudarc = { workspace = true, optional = true }
|
|
half = { workspace = true, optional = true }
|
|
|
|
# Core utilities
|
|
anyhow = { workspace = true }
|
|
thiserror = { workspace = true }
|
|
tracing = { workspace = true }
|
|
serde = { workspace = true }
|
|
|
|
# Performance and memory management
|
|
parking_lot = { workspace = true }
|
|
rayon = { workspace = true }
|
|
dashmap = { workspace = true }
|
|
|
|
# Math and numerics
|
|
nalgebra = { workspace = true }
|
|
|
|
# Async support
|
|
async-trait = { workspace = true }
|
|
tokio = { workspace = true }
|
|
|
|
[target.'cfg(target_os = "macos")'.dependencies]
|
|
objc2 = { version = "0.6", optional = true, features = ["std"] }
|
|
objc2-metal = { version = "0.3", optional = true }
|
|
objc2-foundation = { version = "0.3", optional = true }
|
|
block2 = { version = "0.6", optional = true }
|
|
|
|
[dev-dependencies]
|
|
criterion = { workspace = true }
|
|
proptest = { workspace = true }
|
|
tokio-test = { workspace = true }
|
|
tempfile = { workspace = true }
|
|
|
|
[build-dependencies]
|
|
bindgen = "0.69"
|
|
cc = "1.0"
|
|
|
|
[[bench]]
|
|
name = "flash_attention_bench"
|
|
harness = false
|
|
|
|
[[example]]
|
|
name = "flash_attention_demo"
|
|
path = "examples/flash_attention_demo.rs"
|
|
required-features = ["disabled_tests"]
|
|
|
|
[features]
|
|
# Note: No default GPU features - select cuda or metal explicitly based on your platform
|
|
default = []
|
|
cuda = ["rtx-kernel/cuda", "rtx-tensor/cuda", "rtx-runtime/cuda", "dep:cudarc", "dep:half"]
|
|
metal = ["rtx-kernel/metal", "rtx-tensor/metal", "dep:objc2", "dep:objc2-metal", "dep:objc2-foundation", "dep:block2"]
|
|
variants = []
|
|
experimental = []
|
|
disabled_tests = []
|
|
[lints]
|
|
workspace = true
|