gahow/xserv
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

style: format Rust workspace

Browse Source
This commit is contained in:
Gahow Wang
2026-06-18 18:11:58 +08:00
parent 013465fc06
commit 531cd3fe08
57 changed files with 4045 additions and 1204 deletions
Download Patch File Download Diff File Expand all files Collapse all files

56
crates/xserv-kernels/src/embedding.rs
View File

@@ -2,10 +2,24 @@ use std::ffi::c_void;
use xserv_tensor::{DType, Device, Tensor};
unsafe extern "C" {
fn launch_embedding_f32(table: *const c_void, token_ids: *const c_void, out: *mut c_void,
num_tokens: i32, hidden_size: i32, vocab_size: i32, stream: *mut c_void);
fn launch_embedding_bf16(table: *const c_void, token_ids: *const c_void, out: *mut c_void,
num_tokens: i32, hidden_size: i32, vocab_size: i32, stream: *mut c_void);
fn launch_embedding_f32(
table: *const c_void,
token_ids: *const c_void,
out: *mut c_void,
num_tokens: i32,
hidden_size: i32,
vocab_size: i32,
stream: *mut c_void,
);
fn launch_embedding_bf16(
table: *const c_void,
token_ids: *const c_void,
out: *mut c_void,
num_tokens: i32,
hidden_size: i32,
vocab_size: i32,
stream: *mut c_void,
);
}
/// Embedding lookup: table[token_ids[i]] for each i.
@@ -18,8 +32,14 @@ pub fn embedding(table: &Tensor, token_ids: &[u32]) -> Tensor {
let hidden_size = table.shape()[1];
let num_tokens = token_ids.len();
let vocab_size = table.shape()[0];
assert!(num_tokens <= i32::MAX as usize, "too many tokens for i32 kernel param");
assert!(hidden_size <= i32::MAX as usize, "hidden_size too large for i32 kernel param");
assert!(
num_tokens <= i32::MAX as usize,
"too many tokens for i32 kernel param"
);
assert!(
hidden_size <= i32::MAX as usize,
"hidden_size too large for i32 kernel param"
);
// Upload token_ids to GPU
let ids_bytes = unsafe {
@@ -28,11 +48,15 @@ pub fn embedding(table: &Tensor, token_ids: &[u32]) -> Tensor {
num_tokens * std::mem::size_of::<u32>(),
)
};
let mut ids_gpu = xserv_cuda::allocator::cached_alloc(ids_bytes.len()).expect("alloc token_ids");
let mut ids_gpu =
xserv_cuda::allocator::cached_alloc(ids_bytes.len()).expect("alloc token_ids");
ids_gpu.copy_from_host(ids_bytes).unwrap();
for &tid in token_ids {
assert!((tid as usize) < vocab_size, "token_id {tid} out of bounds (vocab_size={vocab_size})");
assert!(
(tid as usize) < vocab_size,
"token_id {tid} out of bounds (vocab_size={vocab_size})"
);
}
embedding_device_ids(table, ids_gpu.as_ptr() as *const c_void, num_tokens)
@@ -53,14 +77,22 @@ pub fn embedding_device_ids(table: &Tensor, ids_gpu: *const c_void, num_tokens:
unsafe {
match table.dtype() {
DType::F32 => launch_embedding_f32(
table.data_ptr() as _, ids_gpu,
table.data_ptr() as _,
ids_gpu,
out.data_ptr() as *mut c_void,
num_tokens as i32, hidden_size as i32, vocab_size as i32, xserv_cuda::current_stream_raw(),
num_tokens as i32,
hidden_size as i32,
vocab_size as i32,
xserv_cuda::current_stream_raw(),
),
DType::BF16 => launch_embedding_bf16(
table.data_ptr() as _, ids_gpu,
table.data_ptr() as _,
ids_gpu,
out.data_ptr() as *mut c_void,
num_tokens as i32, hidden_size as i32, vocab_size as i32, xserv_cuda::current_stream_raw(),
num_tokens as i32,
hidden_size as i32,
vocab_size as i32,
xserv_cuda::current_stream_raw(),
),
_ => panic!("unsupported dtype for embedding"),
}