Omar SobhandClaude Sonnet 4.6
ef786c0ab1
CI / Clippy Check (push) Failing after 8s
Documentation / Build User Guide (push) Successful in 7s
Documentation / Build API Documentation (push) Failing after 9s
Performance Benchmarks / Run Benchmarks (push) Successful in 1m29s
CI / Format Check (push) Failing after 15s
CI / Build (ubuntu-latest) (push) Failing after 42s
CI / Build CPU-Only (Explicit) (push) Failing after 3m17s
CI / Build (macos-latest) (push) Failing after 30s
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 1s
Continuous batching (rtx-serving-api): - ContinuousBatchingConfig: enable_mid_batch_injection (default true), injection_check_interval (default 1), max_injections_per_step (default 4) - ContinuousBatchingController: inject_into_active_batch() + try_inject_pending() allow new sequences to join a running decode batch after each step - BatchingError::BatchFull variant; 3 new tests PagedAttention v2 defrag (rtx-memory): - PageTable::fragmentation_ratio() — hole-counting (sandwiched free pages / total) - PageTable::defragment() — in-place left-compaction of physical page metadata, consistent lock order (free_pages -> physical_pages -> sequences); GPU KV copy stub comment; DefragStats return value; re-exported from lib.rs - 4 defrag tests; fixed 2 pre-existing compile errors in gpu_oom.rs + gpu_transfer.rs - 192 tests pass Fused RoPE kernel (rtx-transformers): - build_cos_sin_table() + rope_forward_cpu() CPU reference (norm-preserving) - RopeFusedKernel wrapper; rope_forward.cu CUDA kernel (1 block per (B,H,T), 1 thread per dim pair, NVRTC compiled) - Replaced apply_rope_rotation() mul_scalar(0.99) stub with real pairwise rotation - build.rs for NVRTC kernel tracking; layers/mod.rs wired; 8 tests pass Co-Authored-By: Claude Sonnet 4.6 <[email protected]>
27 lines
927 B
Rust
27 lines
927 B
Rust
// build.rs for rtx-transformers
|
|
//
|
|
// When the "cuda" feature is active this build script compiles the
|
|
// `rope_forward.cu` kernel using NVRTC at *runtime* (via `cudarc`), so there
|
|
// is no hard build-time dependency on nvcc. The build script only instructs
|
|
// Cargo to re-run when the kernel source changes.
|
|
//
|
|
// If CUDA is not enabled (the default) this build script is a no-op.
|
|
|
|
fn main() {
|
|
// Only relevant when the cuda feature is active.
|
|
#[cfg(feature = "cuda")]
|
|
cuda_setup();
|
|
|
|
// Always tell Cargo to re-run if *this* script changes.
|
|
println!("cargo:rerun-if-changed=build.rs");
|
|
}
|
|
|
|
#[cfg(feature = "cuda")]
|
|
fn cuda_setup() {
|
|
// Track the kernel source so Cargo rebuilds when it changes.
|
|
println!("cargo:rerun-if-changed=src/cuda_kernels/rope_forward.cu");
|
|
|
|
// Emit the cuda feature flag for conditional compilation inside the crate.
|
|
println!("cargo:rustc-cfg=feature=\"cuda\"");
|
|
}
|