572 lines
21 KiB
Python
572 lines
21 KiB
Python
# Licensed to the Apache Software Foundation (ASF) under one
|
|
# or more contributor license agreements. See the NOTICE file
|
|
# distributed with this work for additional information
|
|
# regarding copyright ownership. The ASF licenses this file
|
|
# to you under the Apache License, Version 2.0 (the
|
|
# "License"); you may not use this file except in compliance
|
|
# with the License. You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing,
|
|
# software distributed under the License is distributed on an
|
|
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
|
|
# KIND, either express or implied. See the License for the
|
|
# specific language governing permissions and limitations
|
|
# under the License.
|
|
|
|
import tvm
|
|
import tvm.testing
|
|
from tvm.relax.transform import LegalizeOps
|
|
from tvm.script import relax as R
|
|
from tvm.script import tirx as T
|
|
|
|
|
|
def test_quantize_fp32_to_int8():
|
|
@tvm.script.ir_module
|
|
class Quantize:
|
|
@R.function
|
|
def main(
|
|
data: R.Tensor((2, 4), "float32"),
|
|
scale: R.Tensor((2,), "float32"),
|
|
zp: R.Tensor((2,), "int8"),
|
|
) -> R.Tensor((2, 4), "int8"):
|
|
out = R.quantize(data, scale, zp, axis=0, out_dtype="int8")
|
|
return out
|
|
|
|
@tvm.script.ir_module
|
|
class Expected:
|
|
@T.prim_func(private=True, s_tir=True)
|
|
def quantize(
|
|
A: T.Buffer((T.int64(2), T.int64(4)), "float32"),
|
|
B: T.Buffer((T.int64(2),), "float32"),
|
|
C: T.Buffer((T.int64(2),), "int8"),
|
|
quantized: T.Buffer((T.int64(2), T.int64(4)), "int8"),
|
|
):
|
|
T.func_attr({"tirx.noalias": True})
|
|
# with T.sblock("root"):
|
|
for i0, i1 in T.grid(T.int64(2), T.int64(4)):
|
|
with T.sblock("quantized"):
|
|
v_i0, v_i1 = T.axis.remap("SS", [i0, i1])
|
|
T.reads(A[v_i0, v_i1], B[v_i0], C[v_i0])
|
|
T.writes(quantized[v_i0, v_i1])
|
|
quantized[v_i0, v_i1] = T.Cast(
|
|
"int8",
|
|
T.max(
|
|
T.min(
|
|
T.round(A[v_i0, v_i1] / B[v_i0]) + T.Cast("float32", C[v_i0]),
|
|
T.float32(127),
|
|
),
|
|
T.float32(-128),
|
|
),
|
|
)
|
|
|
|
@R.function
|
|
def main(
|
|
data: R.Tensor((2, 4), dtype="float32"),
|
|
scale: R.Tensor((2,), dtype="float32"),
|
|
zp: R.Tensor((2,), dtype="int8"),
|
|
) -> R.Tensor((2, 4), dtype="int8"):
|
|
out = R.call_tir(
|
|
Expected.quantize, (data, scale, zp), out_ty=R.Tensor((2, 4), dtype="int8")
|
|
)
|
|
return out
|
|
|
|
mod = LegalizeOps()(Quantize)
|
|
tvm.ir.assert_structural_equal(mod, Expected)
|
|
|
|
|
|
def test_quantize_fp16_to_uint8():
|
|
@tvm.script.ir_module
|
|
class Quantize:
|
|
@R.function
|
|
def main(
|
|
data: R.Tensor((2, 4), "float16"),
|
|
scale: R.Tensor((2,), "float16"),
|
|
zp: R.Tensor((2,), "int8"),
|
|
) -> R.Tensor((2, 4), "uint8"):
|
|
out = R.quantize(data, scale, zp, axis=0, out_dtype="uint8")
|
|
return out
|
|
|
|
@tvm.script.ir_module
|
|
class Expected:
|
|
@T.prim_func(private=True, s_tir=True)
|
|
def quantize(
|
|
A: T.Buffer((T.int64(2), T.int64(4)), "float16"),
|
|
B: T.Buffer((T.int64(2),), "float16"),
|
|
C: T.Buffer((T.int64(2),), "int8"),
|
|
quantized: T.Buffer((T.int64(2), T.int64(4)), "uint8"),
|
|
):
|
|
T.func_attr({"tirx.noalias": True})
|
|
# with T.sblock("root"):
|
|
for i0, i1 in T.grid(T.int64(2), T.int64(4)):
|
|
with T.sblock("quantized"):
|
|
v_i0, v_i1 = T.axis.remap("SS", [i0, i1])
|
|
T.reads(A[v_i0, v_i1], B[v_i0], C[v_i0])
|
|
T.writes(quantized[v_i0, v_i1])
|
|
quantized[v_i0, v_i1] = T.Cast(
|
|
"uint8",
|
|
T.max(
|
|
T.min(
|
|
T.round(A[v_i0, v_i1] / B[v_i0]) + T.Cast("float16", C[v_i0]),
|
|
T.float16(255),
|
|
),
|
|
T.float16(0),
|
|
),
|
|
)
|
|
|
|
@R.function
|
|
def main(
|
|
data: R.Tensor((2, 4), dtype="float16"),
|
|
scale: R.Tensor((2,), dtype="float16"),
|
|
zp: R.Tensor((2,), dtype="int8"),
|
|
) -> R.Tensor((2, 4), dtype="uint8"):
|
|
out = R.call_tir(
|
|
Expected.quantize, (data, scale, zp), out_ty=R.Tensor((2, 4), dtype="uint8")
|
|
)
|
|
return out
|
|
|
|
mod = LegalizeOps()(Quantize)
|
|
tvm.ir.assert_structural_equal(mod, Expected)
|
|
|
|
|
|
def test_quantize_fp32_to_int8_symbolic():
|
|
@tvm.script.ir_module
|
|
class Quantize:
|
|
@R.function
|
|
def main(
|
|
data: R.Tensor((4, "n"), "float32"),
|
|
scale: R.Tensor(("n",), "float32"),
|
|
zp: R.Tensor(("n",), "int8"),
|
|
) -> R.Tensor((4, "n"), "int8"):
|
|
out = R.quantize(data, scale, zp, axis=-1, out_dtype="int8")
|
|
return out
|
|
|
|
@tvm.script.ir_module
|
|
class Expected:
|
|
@T.prim_func(private=True, s_tir=True)
|
|
def quantize(var_A: T.handle, var_B: T.handle, var_C: T.handle, var_quantized: T.handle):
|
|
T.func_attr({"tirx.noalias": True})
|
|
n = T.int64()
|
|
A = T.match_buffer(var_A, (T.int64(4), n))
|
|
B = T.match_buffer(var_B, (n,))
|
|
C = T.match_buffer(var_C, (n,), "int8")
|
|
quantized = T.match_buffer(var_quantized, (T.int64(4), n), "int8")
|
|
# with T.sblock("root"):
|
|
for i0, i1 in T.grid(T.int64(4), n):
|
|
with T.sblock("quantized"):
|
|
v_i0, v_i1 = T.axis.remap("SS", [i0, i1])
|
|
T.reads(A[v_i0, v_i1], B[v_i1], C[v_i1])
|
|
T.writes(quantized[v_i0, v_i1])
|
|
quantized[v_i0, v_i1] = T.Cast(
|
|
"int8",
|
|
T.max(
|
|
T.min(
|
|
T.round(A[v_i0, v_i1] / B[v_i1]) + T.Cast("float32", C[v_i1]),
|
|
T.float32(127),
|
|
),
|
|
T.float32(-128),
|
|
),
|
|
)
|
|
|
|
@R.function
|
|
def main(
|
|
data: R.Tensor((4, "n"), dtype="float32"),
|
|
scale: R.Tensor(("n",), dtype="float32"),
|
|
zp: R.Tensor(("n",), dtype="int8"),
|
|
) -> R.Tensor((4, "n"), dtype="int8"):
|
|
n = T.int64()
|
|
out = R.call_tir(Expected.quantize, (data, scale, zp), out_ty=R.Tensor((4, n), "int8"))
|
|
return out
|
|
|
|
mod = LegalizeOps()(Quantize)
|
|
tvm.ir.assert_structural_equal(mod, Expected)
|
|
|
|
|
|
def test_quantize_fp32_to_int8_scalar_param():
|
|
@tvm.script.ir_module
|
|
class Quantize:
|
|
@R.function
|
|
def main(data: R.Tensor((2, 4), "float32")) -> R.Tensor((2, 4), "int8"):
|
|
out = R.quantize(
|
|
data, R.const(2.0, "float32"), R.const(1, "int8"), axis=-1, out_dtype="int8"
|
|
)
|
|
return out
|
|
|
|
@tvm.script.ir_module
|
|
class Expected:
|
|
@T.prim_func(private=True, s_tir=True)
|
|
def quantize(
|
|
A: T.Buffer((T.int64(2), T.int64(4)), "float32"),
|
|
quantized: T.Buffer((T.int64(2), T.int64(4)), "int8"),
|
|
):
|
|
T.func_attr({"tirx.noalias": True})
|
|
# with T.sblock("root"):
|
|
for i0, i1 in T.grid(T.int64(2), T.int64(4)):
|
|
with T.sblock("quantized"):
|
|
v_i0, v_i1 = T.axis.remap("SS", [i0, i1])
|
|
T.reads(A[v_i0, v_i1])
|
|
T.writes(quantized[v_i0, v_i1])
|
|
quantized[v_i0, v_i1] = T.Cast(
|
|
"int8",
|
|
T.max(
|
|
T.min(
|
|
T.round(A[v_i0, v_i1] / T.float32(2)) + T.float32(1),
|
|
T.float32(127),
|
|
),
|
|
T.float32(-128),
|
|
),
|
|
)
|
|
|
|
@R.function
|
|
def main(data: R.Tensor((2, 4), dtype="float32")) -> R.Tensor((2, 4), dtype="int8"):
|
|
out = R.call_tir(Expected.quantize, (data,), out_ty=R.Tensor((2, 4), dtype="int8"))
|
|
return out
|
|
|
|
mod = LegalizeOps()(Quantize)
|
|
tvm.ir.assert_structural_equal(mod, Expected)
|
|
|
|
|
|
def test_quantize_fp32_to_int8_scalar_1d_param():
|
|
@tvm.script.ir_module
|
|
class Quantize:
|
|
@R.function
|
|
def main(data: R.Tensor((2, 4), "float32")) -> R.Tensor((2, 4), "int8"):
|
|
out = R.quantize(
|
|
data,
|
|
R.const([2.0, 1.0], "float32"),
|
|
R.const([4, 5], "int8"),
|
|
axis=0,
|
|
out_dtype="int8",
|
|
)
|
|
return out
|
|
|
|
@tvm.script.ir_module
|
|
class Expected:
|
|
@T.prim_func(private=True, s_tir=True)
|
|
def quantize(
|
|
A: T.Buffer((T.int64(2), T.int64(4)), "float32"),
|
|
B: T.Buffer((T.int64(2),), "float32"),
|
|
C: T.Buffer((T.int64(2),), "int8"),
|
|
quantized: T.Buffer((T.int64(2), T.int64(4)), "int8"),
|
|
):
|
|
T.func_attr({"tirx.noalias": True})
|
|
# with T.sblock("root"):
|
|
for i0, i1 in T.grid(T.int64(2), T.int64(4)):
|
|
with T.sblock("quantized"):
|
|
v_i0, v_i1 = T.axis.remap("SS", [i0, i1])
|
|
T.reads(A[v_i0, v_i1], B[v_i0], C[v_i0])
|
|
T.writes(quantized[v_i0, v_i1])
|
|
quantized[v_i0, v_i1] = T.Cast(
|
|
"int8",
|
|
T.max(
|
|
T.min(
|
|
T.round(A[v_i0, v_i1] / B[v_i0]) + T.Cast("float32", C[v_i0]),
|
|
T.float32(127),
|
|
),
|
|
T.float32(-128),
|
|
),
|
|
)
|
|
|
|
@R.function
|
|
def main(data: R.Tensor((2, 4), dtype="float32")) -> R.Tensor((2, 4), dtype="int8"):
|
|
cls = Expected
|
|
out = R.call_tir(
|
|
cls.quantize,
|
|
(data, R.const([2.0, 1.0], "float32"), R.const([4, 5], "int8")),
|
|
out_ty=R.Tensor((2, 4), dtype="int8"),
|
|
)
|
|
return out
|
|
|
|
mod = LegalizeOps()(Quantize)
|
|
tvm.ir.assert_structural_equal(mod, Expected)
|
|
|
|
|
|
def test_quantize_fp16_to_int8_scalar_param():
|
|
@tvm.script.ir_module
|
|
class Quantize:
|
|
@R.function
|
|
def main(data: R.Tensor((2, 4), "float16")) -> R.Tensor((2, 4), "int8"):
|
|
out = R.quantize(
|
|
data, R.const(2.0, "float16"), R.const(1, "int8"), axis=-1, out_dtype="int8"
|
|
)
|
|
return out
|
|
|
|
@tvm.script.ir_module
|
|
class Expected:
|
|
@T.prim_func(private=True, s_tir=True)
|
|
def quantize(
|
|
A: T.Buffer((T.int64(2), T.int64(4)), "float16"),
|
|
quantized: T.Buffer((T.int64(2), T.int64(4)), "int8"),
|
|
):
|
|
T.func_attr({"tirx.noalias": True})
|
|
# with T.sblock("root"):
|
|
for i0, i1 in T.grid(T.int64(2), T.int64(4)):
|
|
with T.sblock("quantized"):
|
|
v_i0, v_i1 = T.axis.remap("SS", [i0, i1])
|
|
T.reads(A[v_i0, v_i1])
|
|
T.writes(quantized[v_i0, v_i1])
|
|
quantized[v_i0, v_i1] = T.Cast(
|
|
"int8",
|
|
T.max(
|
|
T.min(
|
|
T.round(A[v_i0, v_i1] / T.float16(2)) + T.float16(1),
|
|
T.float16(127),
|
|
),
|
|
T.float16(-128),
|
|
),
|
|
)
|
|
|
|
@R.function
|
|
def main(data: R.Tensor((2, 4), dtype="float16")) -> R.Tensor((2, 4), dtype="int8"):
|
|
out = R.call_tir(Expected.quantize, (data,), out_ty=R.Tensor((2, 4), dtype="int8"))
|
|
return out
|
|
|
|
mod = LegalizeOps()(Quantize)
|
|
tvm.ir.assert_structural_equal(mod, Expected)
|
|
|
|
|
|
def test_dequantize_int8_to_fp32():
|
|
@tvm.script.ir_module
|
|
class Dequantize:
|
|
@R.function
|
|
def main(
|
|
data: R.Tensor((2, 4), "int8"),
|
|
scale: R.Tensor((2,), "float32"),
|
|
zp: R.Tensor((2,), "int8"),
|
|
) -> R.Tensor((2, 4), "float32"):
|
|
out = R.dequantize(data, scale, zp, axis=0, out_dtype="float32")
|
|
return out
|
|
|
|
@tvm.script.ir_module
|
|
class Expected:
|
|
@T.prim_func(private=True, s_tir=True)
|
|
def dequantize(
|
|
A: T.Buffer((T.int64(2), T.int64(4)), "int8"),
|
|
B: T.Buffer((T.int64(2),), "float32"),
|
|
C: T.Buffer((T.int64(2),), "int8"),
|
|
dequantized: T.Buffer((T.int64(2), T.int64(4)), "float32"),
|
|
):
|
|
T.func_attr({"tirx.noalias": True})
|
|
# with T.sblock("root"):
|
|
for i0, i1 in T.grid(T.int64(2), T.int64(4)):
|
|
with T.sblock("dequantized"):
|
|
v_i0, v_i1 = T.axis.remap("SS", [i0, i1])
|
|
T.reads(A[v_i0, v_i1], C[v_i0], B[v_i0])
|
|
T.writes(dequantized[v_i0, v_i1])
|
|
dequantized[v_i0, v_i1] = (
|
|
T.Cast("float32", T.Cast("int32", A[v_i0, v_i1]) - T.Cast("int32", C[v_i0]))
|
|
* B[v_i0]
|
|
)
|
|
|
|
@R.function
|
|
def main(
|
|
data: R.Tensor((2, 4), dtype="int8"),
|
|
scale: R.Tensor((2,), dtype="float32"),
|
|
zp: R.Tensor((2,), dtype="int8"),
|
|
) -> R.Tensor((2, 4), dtype="float32"):
|
|
out = R.call_tir(
|
|
Expected.dequantize, (data, scale, zp), out_ty=R.Tensor((2, 4), dtype="float32")
|
|
)
|
|
return out
|
|
|
|
mod = LegalizeOps()(Dequantize)
|
|
tvm.ir.assert_structural_equal(mod, Expected)
|
|
|
|
|
|
def test_dequantize_int8_to_fp32_scalar_param():
|
|
@tvm.script.ir_module
|
|
class Dequantize:
|
|
@R.function
|
|
def main(data: R.Tensor((2, 4), "int8")) -> R.Tensor((2, 4), "float32"):
|
|
out = R.dequantize(
|
|
data, R.const(2.0, "float32"), R.const(1, "int8"), axis=0, out_dtype="float32"
|
|
)
|
|
return out
|
|
|
|
@tvm.script.ir_module
|
|
class Expected:
|
|
@T.prim_func(private=True, s_tir=True)
|
|
def dequantize(
|
|
A: T.Buffer((T.int64(2), T.int64(4)), "int8"),
|
|
dequantized: T.Buffer((T.int64(2), T.int64(4)), "float32"),
|
|
):
|
|
T.func_attr({"tirx.noalias": True})
|
|
# with T.sblock("root"):
|
|
for i0, i1 in T.grid(T.int64(2), T.int64(4)):
|
|
with T.sblock("dequantized"):
|
|
v_i0, v_i1 = T.axis.remap("SS", [i0, i1])
|
|
T.reads(A[v_i0, v_i1])
|
|
T.writes(dequantized[v_i0, v_i1])
|
|
dequantized[v_i0, v_i1] = T.Cast(
|
|
"float32", T.Cast("int32", A[v_i0, v_i1]) - 1
|
|
) * T.float32(2)
|
|
|
|
@R.function
|
|
def main(data: R.Tensor((2, 4), dtype="int8")) -> R.Tensor((2, 4), dtype="float32"):
|
|
cls = Expected
|
|
out = R.call_tir(cls.dequantize, (data,), out_ty=R.Tensor((2, 4), dtype="float32"))
|
|
return out
|
|
|
|
mod = LegalizeOps()(Dequantize)
|
|
tvm.ir.assert_structural_equal(mod, Expected)
|
|
|
|
|
|
def test_dequantize_int8_to_fp32_symbolic():
|
|
@tvm.script.ir_module
|
|
class Dequantize:
|
|
@R.function
|
|
def main(
|
|
data: R.Tensor((2, "n"), "int8"),
|
|
scale: R.Tensor(("n",), "float32"),
|
|
zp: R.Tensor(("n",), "int8"),
|
|
) -> R.Tensor((2, "n"), "float32"):
|
|
out = R.dequantize(data, scale, zp, axis=-1, out_dtype="float32")
|
|
return out
|
|
|
|
@tvm.script.ir_module
|
|
class Expected:
|
|
@T.prim_func(private=True, s_tir=True)
|
|
def dequantize(
|
|
var_A: T.handle, var_B: T.handle, var_C: T.handle, var_dequantized: T.handle
|
|
):
|
|
T.func_attr({"tirx.noalias": True})
|
|
n = T.int64()
|
|
A = T.match_buffer(var_A, (T.int64(2), n), "int8")
|
|
B = T.match_buffer(var_B, (n,))
|
|
C = T.match_buffer(var_C, (n,), "int8")
|
|
dequantized = T.match_buffer(var_dequantized, (T.int64(2), n))
|
|
# with T.sblock("root"):
|
|
for i0, i1 in T.grid(T.int64(2), n):
|
|
with T.sblock("dequantized"):
|
|
v_i0, v_i1 = T.axis.remap("SS", [i0, i1])
|
|
T.reads(A[v_i0, v_i1], C[v_i1], B[v_i1])
|
|
T.writes(dequantized[v_i0, v_i1])
|
|
dequantized[v_i0, v_i1] = (
|
|
T.Cast("float32", T.Cast("int32", A[v_i0, v_i1]) - T.Cast("int32", C[v_i1]))
|
|
* B[v_i1]
|
|
)
|
|
|
|
@R.function
|
|
def main(
|
|
data: R.Tensor((2, "n"), dtype="int8"),
|
|
scale: R.Tensor(("n",), dtype="float32"),
|
|
zp: R.Tensor(("n",), dtype="int8"),
|
|
) -> R.Tensor((2, "n"), dtype="float32"):
|
|
n = T.int64()
|
|
out = R.call_tir(
|
|
Expected.dequantize, (data, scale, zp), out_ty=R.Tensor((2, n), dtype="float32")
|
|
)
|
|
return out
|
|
|
|
mod = LegalizeOps()(Dequantize)
|
|
tvm.ir.assert_structural_equal(mod, Expected)
|
|
|
|
|
|
def test_dequantize_int8_to_fp16():
|
|
@tvm.script.ir_module
|
|
class Dequantize:
|
|
@R.function
|
|
def main(
|
|
data: R.Tensor((2, 4), "int8"),
|
|
scale: R.Tensor((2,), "float16"),
|
|
zp: R.Tensor((2,), "int8"),
|
|
) -> R.Tensor((2, 4), "float16"):
|
|
out = R.dequantize(data, scale, zp, axis=0, out_dtype="float16")
|
|
return out
|
|
|
|
@tvm.script.ir_module
|
|
class Expected:
|
|
@T.prim_func(private=True, s_tir=True)
|
|
def dequantize(
|
|
A: T.Buffer((T.int64(2), T.int64(4)), "int8"),
|
|
B: T.Buffer((T.int64(2),), "float16"),
|
|
C: T.Buffer((T.int64(2),), "int8"),
|
|
dequantized: T.Buffer((T.int64(2), T.int64(4)), "float16"),
|
|
):
|
|
T.func_attr({"tirx.noalias": True})
|
|
# with T.sblock("root"):
|
|
for i0, i1 in T.grid(T.int64(2), T.int64(4)):
|
|
with T.sblock("dequantized"):
|
|
v_i0, v_i1 = T.axis.remap("SS", [i0, i1])
|
|
T.reads(A[v_i0, v_i1], C[v_i0], B[v_i0])
|
|
T.writes(dequantized[v_i0, v_i1])
|
|
dequantized[v_i0, v_i1] = T.Cast(
|
|
"float16",
|
|
T.max(
|
|
T.min(
|
|
T.Cast(
|
|
"float32",
|
|
T.Cast("int32", A[v_i0, v_i1]) - T.Cast("int32", C[v_i0]),
|
|
)
|
|
* T.Cast("float32", B[v_i0]),
|
|
T.float32(65504),
|
|
),
|
|
T.float32(-65504),
|
|
),
|
|
)
|
|
|
|
@R.function
|
|
def main(
|
|
data: R.Tensor((2, 4), dtype="int8"),
|
|
scale: R.Tensor((2,), dtype="float16"),
|
|
zp: R.Tensor((2,), dtype="int8"),
|
|
) -> R.Tensor((2, 4), dtype="float16"):
|
|
out = R.call_tir(
|
|
Expected.dequantize, (data, scale, zp), out_ty=R.Tensor((2, 4), dtype="float16")
|
|
)
|
|
return out
|
|
|
|
mod = LegalizeOps()(Dequantize)
|
|
tvm.ir.assert_structural_equal(mod, Expected)
|
|
|
|
|
|
def test_dequantize_int8_to_fp16_scalar_param():
|
|
@tvm.script.ir_module
|
|
class Dequantize:
|
|
@R.function
|
|
def main(data: R.Tensor((2, 4), "int8")) -> R.Tensor((2, 4), "float16"):
|
|
out = R.dequantize(
|
|
data, R.const(2.0, "float16"), R.const(1, "int8"), axis=0, out_dtype="float16"
|
|
)
|
|
return out
|
|
|
|
@tvm.script.ir_module
|
|
class Expected:
|
|
@T.prim_func(private=True, s_tir=True)
|
|
def dequantize(
|
|
A: T.Buffer((T.int64(2), T.int64(4)), "int8"),
|
|
dequantized: T.Buffer((T.int64(2), T.int64(4)), "float16"),
|
|
):
|
|
T.func_attr({"tirx.noalias": True})
|
|
# with T.sblock("root"):
|
|
for i0, i1 in T.grid(T.int64(2), T.int64(4)):
|
|
with T.sblock("dequantized"):
|
|
v_i0, v_i1 = T.axis.remap("SS", [i0, i1])
|
|
T.reads(A[v_i0, v_i1])
|
|
T.writes(dequantized[v_i0, v_i1])
|
|
dequantized[v_i0, v_i1] = T.Cast(
|
|
"float16",
|
|
T.max(
|
|
T.min(
|
|
T.Cast("float32", T.Cast("int32", A[v_i0, v_i1]) - 1)
|
|
* T.float32(2),
|
|
T.float32(65504),
|
|
),
|
|
T.float32(-65504),
|
|
),
|
|
)
|
|
|
|
@R.function
|
|
def main(data: R.Tensor((2, 4), dtype="int8")) -> R.Tensor((2, 4), dtype="float16"):
|
|
cls = Expected
|
|
out = R.call_tir(cls.dequantize, (data,), out_ty=R.Tensor((2, 4), dtype="float16"))
|
|
return out
|
|
|
|
mod = LegalizeOps()(Dequantize)
|
|
tvm.ir.assert_structural_equal(mod, Expected)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
tvm.testing.main()
|