triton.language.associative_scan
1. OP 概述
简介:triton.language.associative_scan 对输入tensor沿指定轴应用关联扫描操作,使用combine_fn函数组合元素并更新进位值。
triton.language.associative_scan(input, axis, combine_fn, reverse=False, _semantic=None, _generator=None)
2. OP 规格
2.1 参数说明
参数 |
类型 |
含义说明 |
|---|---|---|
|
|
输入tensor,可以是单个tensor或tensor元组 |
|
|
沿着哪个维度进行关联扫描操作 |
|
|
用于组合两个标量tensor组的函数(必须用@triton.jit标记) |
|
|
是否沿轴的反方向应用关联扫描 |
|
|
保留参数,暂不支持外部调用 |
|
|
保留参数,暂不支持外部调用 |
返回值:
tensor:对输入tensor沿指定轴应用关联扫描操作,使用combine_fn函数组合元素并更新进位值之后的tensor。
2.2 支持规格
2.2.1 DataType 支持
uint8 |
int8 |
uint16 |
int16 |
uint32 |
int32 |
uint64 |
int64 |
fp16 |
fp32 |
bf16 |
bool/int1 |
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
GPU支持 |
✓ |
✓ |
✓ |
✓ |
✓ |
✓ |
✓ |
✓ |
✓ |
✓ |
✓ |
✓ |
Ascend A2/A3 |
✓ |
✓ |
× |
✓ |
× |
✓ |
× |
✓ |
✓ |
✓ |
✓ |
✓ |
2.2.2 Shape 支持
结论:在 Shape 方面,GPU 与 Ascend 平台无差异。
2.3 特殊限制说明
相对社区能力缺失且无法实现 reverse=True是否沿轴的反方向应用关联扫描,该功能需要tl.load加载数据时对齐,即不使用mask过滤掉多余数据索引,即如下面示例代码:
tl.static_assert(
numel_x == XBLOCK, "numel_x must be equal to XBLOCK in this kernel"
)
tl.static_assert(
numel_r == RBLOCK, "numel_r must be equal to RBLOCK in this kernel"
)
idx_x = tl.arange(0, XBLOCK)
idx_r = tl.arange(0, RBLOCK)
idx = idx_x[:, None] * numel_r + idx_r[None, :]
x = tl.load(in_ptr0 + idx)
2.4 使用方法
以下示例实现了对2Dshape的tensor进行associative_scan运算:
@triton.jit
def bitwise_and_fn(a, b):
return a & b
@triton.jit
def bitwise_or_fn(a, b):
return a | b
@triton.jit
def bitwise_xor_fn(a, b):
return a ^ b
@triton.jit
def minimum_fn(a, b):
return tl.minimum(a, b)
@triton.jit
def maximum_fn(a, b):
return tl.maximum(a, b)
@triton.jit
def triton_kernel_2d_scan(
out_ptr0,
in_ptr0,
dim: tl.constexpr,
reverse: tl.constexpr,
numel_x: tl.constexpr,
numel_r: tl.constexpr,
XBLOCK: tl.constexpr,
RBLOCK: tl.constexpr,
combine_fn_name: tl.constexpr,
):
tl.static_assert(
numel_x == XBLOCK, "numel_x must be equal to XBLOCK in this kernel"
)
tl.static_assert(
numel_r == RBLOCK, "numel_r must be equal to RBLOCK in this kernel"
)
idx_x = tl.arange(0, XBLOCK)
idx_r = tl.arange(0, RBLOCK)
idx = idx_x[:, None] * numel_r + idx_r[None, :]
x = tl.load(in_ptr0 + idx)
if combine_fn_name == "maximum_fn":
ret = tl.associative_scan(x, axis=dim, reverse=reverse, combine_fn=maximum_fn)
elif combine_fn_name == "minimum_fn":
ret = tl.associative_scan(x, axis=dim, reverse=reverse, combine_fn=minimum_fn)
elif combine_fn_name == "bitwise_or_fn":
ret = tl.associative_scan(x, axis=dim, reverse=reverse, combine_fn=bitwise_or_fn)
elif combine_fn_name == "bitwise_xor_fn":
ret = tl.associative_scan(x, axis=dim, reverse=reverse, combine_fn=bitwise_xor_fn)
elif combine_fn_name == "bitwise_and_fn":
ret = tl.associative_scan(x, axis=dim, reverse=reverse, combine_fn=bitwise_and_fn)
tl.store(out_ptr0 + idx, ret)