Skip to content

GitLab

Commit 1d5d035c authored by Shucai Xiao's avatar Shucai Xiao
Browse files

clang format

parent ea9776f5
Hide whitespace changes
Inline Side-by-side
Showing with 60 additions and 40 deletions
src/quantization.cpp
View file @ 1d5d035c
......@@ -153,15 +153,16 @@ void quantize_int8(program& prog, const std::vector<std::string>& ins_names)
// operation, if it has 3 inputs, then the last one should
// be converted to int32_type
shape::type_t quant_type = shape::int8_type;
auto param = int8_param[param_index++];
auto param = int8_param[param_index++];
if(ins->name() == "dot" and inputs.size() == 3 and input == inputs.back())
{
quant_type = shape::int32_type;
}
auto s = input->get_shape();
auto s = input->get_shape();
if((s.type() == shape::float_type || s.type() == shape::double_type ||
s.type() == shape::int32_type) && s.type() != quant_type)
s.type() == shape::int32_type) &&
s.type() != quant_type)
{
// if the input is a convert operator, uses its input
// as its current input
......@@ -204,66 +205,74 @@ void quantize_int8(program& prog, const std::vector<std::string>& ins_names)
// equal)", we need additional calculation for the adjustment
if(ins->name() == "dot")
{
auto dot_op = any_cast<op::dot>(ins->get_operator());
auto dot_op = any_cast<op::dot>(ins->get_operator());
float new_alpha = dot_op.alpha / (int8_param[0].first * int8_param[1].first);
float new_beta = dot_op.beta;
// We need additional checking about the quant_alpha value. If
float new_beta = dot_op.beta;
// We need additional checking about the quant_alpha value. If
// abs(quant_alpha) > 50 (some tmp value set here), we can convert
// it to an integer as the new_alpha in the quant_dot
float threshold = 50.0f;
if (fabs(new_alpha) >= threshold && fabs(new_beta) >= threshold)
if(fabs(new_alpha) >= threshold && fabs(new_beta) >= threshold)
{
int32_t quant_alpha = static_cast<int32_t>(new_alpha);
int32_t quant_beta = static_cast<int32_t>(new_beta);
shape quant_shape = compute_shape(op::quant_dot{1, 0}, converted_inputs);
if (quant_shape.type() == orig_type)
int32_t quant_beta = static_cast<int32_t>(new_beta);
shape quant_shape = compute_shape(op::quant_dot{1, 0}, converted_inputs);
if(quant_shape.type() == orig_type)
{
prog.replace_instruction(ins, op::quant_dot{quant_alpha, quant_beta}, converted_inputs);
prog.replace_instruction(
ins, op::quant_dot{quant_alpha, quant_beta}, converted_inputs);
}
else
{
auto quant_dot = prog.insert_instruction(ins, op::quant_dot{quant_alpha, quant_beta}, converted_inputs);
auto quant_dot = prog.insert_instruction(
ins, op::quant_dot{quant_alpha, quant_beta}, converted_inputs);
prog.replace_instruction(ins, op::convert{orig_type}, quant_dot);
}
}
// only alpha can be quantized, quantization of beta will cause
// big error, so we have to manually do the multiplication and
// only alpha can be quantized, quantization of beta will cause
// big error, so we have to manually do the multiplication and
// addition
else if (fabs(new_alpha) >= threshold)
else if(fabs(new_alpha) >= threshold)
{
int32_t quant_alpha = static_cast<int32_t>(new_alpha);
int32_t quant_beta = 0;
if (orig_type == shape::int32_type)
int32_t quant_beta = 0;
if(orig_type == shape::int32_type)
{
if (inputs.size() == 2 or dot_op.beta == 0.0f)
if(inputs.size() == 2 or dot_op.beta == 0.0f)
{
prog.replace_instruction(ins, op::quant_dot{quant_alpha, quant_beta}, converted_inputs);
prog.replace_instruction(
ins, op::quant_dot{quant_alpha, quant_beta}, converted_inputs);
}
// if there are 3 inputs, we need to consider the third argument
else
{
auto q_dot = prog.insert_instruction(ins, op::quant_dot{quant_alpha, quant_beta}, converted_inputs);
auto q_dot = prog.insert_instruction(
ins, op::quant_dot{quant_alpha, quant_beta}, converted_inputs);
std::vector<float> vec_beta(q_dot->get_shape().elements(), dot_op.beta);
auto l_beta = prog.add_literal(literal{orig_type, vec_beta});
auto beta_c = prog.insert_instruction(ins, op::mul{}, l_beta, inputs.back());
auto beta_c =
prog.insert_instruction(ins, op::mul{}, l_beta, inputs.back());
prog.replace_instruction(ins, op::add{}, q_dot, beta_c);
}
}
else
{
if (inputs.size() == 2 or dot_op.beta == 0.0f)
if(inputs.size() == 2 or dot_op.beta == 0.0f)
{
auto q_dot = prog.insert_instruction(ins, op::quant_dot{quant_alpha, quant_beta}, converted_inputs);
auto q_dot = prog.insert_instruction(
ins, op::quant_dot{quant_alpha, quant_beta}, converted_inputs);
prog.replace_instruction(ins, op::convert{orig_type}, q_dot);
}
// if there are 3 inputs, we need to consider the third argument
else
{
auto q_dot = prog.insert_instruction(ins, op::quant_dot{quant_alpha, quant_beta}, converted_inputs);
auto q_dot = prog.insert_instruction(
ins, op::quant_dot{quant_alpha, quant_beta}, converted_inputs);
auto oq_dot = prog.insert_instruction(ins, op::convert{orig_type}, q_dot);
std::vector<float> vec_beta(q_dot->get_shape().elements(), dot_op.beta);
auto l_beta = prog.add_literal(literal{oq_dot->get_shape(), vec_beta});
auto beta_c = prog.insert_instruction(ins, op::mul{}, l_beta, inputs.back());
auto beta_c =
prog.insert_instruction(ins, op::mul{}, l_beta, inputs.back());
prog.replace_instruction(ins, op::add{}, q_dot, beta_c);
}
}
......@@ -272,10 +281,10 @@ void quantize_int8(program& prog, const std::vector<std::string>& ins_names)
{
auto q_dot = prog.insert_instruction(ins, op::quant_dot{1, 0}, converted_inputs);
std::vector<float> vec_alpha(q_dot->get_shape().elements(), new_alpha);
if (orig_type == shape::int32_type)
if(orig_type == shape::int32_type)
{
auto l_alpha = prog.add_literal(literal(ins->get_shape(), vec_alpha));
if (converted_inputs.size() == 2 or dot_op.beta == 0.0f)
if(converted_inputs.size() == 2 or dot_op.beta == 0.0f)
{
prog.replace_instruction(ins, op::mul{}, l_alpha, q_dot);
}
......@@ -283,17 +292,18 @@ void quantize_int8(program& prog, const std::vector<std::string>& ins_names)
else
{
std::vector<float> vec_beta(ins->get_shape().elements(), new_beta);
auto l_beta = prog.add_literal(literal(ins->get_shape(), vec_beta));
auto l_beta = prog.add_literal(literal(ins->get_shape(), vec_beta));
auto alpha_ab = prog.insert_instruction(ins, op::mul{}, l_alpha, q_dot);
auto beta_c = prog.insert_instruction(ins, op::mul{}, l_beta, inputs.back());
auto beta_c =
prog.insert_instruction(ins, op::mul{}, l_beta, inputs.back());
prog.replace_instruction(ins, op::add{}, alpha_ab, beta_c);
}
}
else
{
auto oq_dot = prog.insert_instruction(ins, op::convert{orig_type}, q_dot);
auto oq_dot = prog.insert_instruction(ins, op::convert{orig_type}, q_dot);
auto l_alpha = prog.add_literal(literal(ins->get_shape(), vec_alpha));
if (converted_inputs.size() == 2 or dot_op.beta == 0.0f)
if(converted_inputs.size() == 2 or dot_op.beta == 0.0f)
{
prog.replace_instruction(ins, op::mul{}, l_alpha, oq_dot);
}
......@@ -301,9 +311,10 @@ void quantize_int8(program& prog, const std::vector<std::string>& ins_names)
else
{
std::vector<float> vec_beta(ins->get_shape().elements(), new_beta);
auto l_beta = prog.add_literal(literal(ins->get_shape(), vec_beta));
auto l_beta = prog.add_literal(literal(ins->get_shape(), vec_beta));
auto alpha_ab = prog.insert_instruction(ins, op::mul{}, l_alpha, oq_dot);
auto beta_c = prog.insert_instruction(ins, op::mul{}, l_beta, inputs.back());
auto beta_c =
prog.insert_instruction(ins, op::mul{}, l_beta, inputs.back());
prog.replace_instruction(ins, op::add{}, alpha_ab, beta_c);
}
}
......@@ -323,23 +334,32 @@ void quantize_int8(program& prog, const std::vector<std::string>& ins_names)
shape quant_shape = compute_shape(op::quant_convolution{}, converted_inputs);
std::vector<float> vec_factor(quant_shape.elements(), adjust_factor);
auto fl = prog.add_literal(literal{{orig_type, quant_shape.lens()}, vec_factor});
if (quant_shape.type() == orig_type)
auto fl = prog.add_literal(literal{{orig_type, quant_shape.lens()}, vec_factor});
if(quant_shape.type() == orig_type)
{
if (adjust_factor == 1.0f)
if(adjust_factor == 1.0f)
{
prog.replace_instruction(ins, op::quant_convolution{padding, stride, dilation, padding_mode, group}, converted_inputs);
prog.replace_instruction(
ins,
op::quant_convolution{padding, stride, dilation, padding_mode, group},
converted_inputs);
}
else
{
auto quant_conv = prog.replace_instruction(ins, op::quant_convolution{padding, stride, dilation, padding_mode, group}, converted_inputs);
auto quant_conv = prog.replace_instruction(
ins,
op::quant_convolution{padding, stride, dilation, padding_mode, group},
converted_inputs);
prog.replace_instruction(ins, op::mul{}, quant_conv, fl);
}
}
else
{
auto quant_conv = prog.insert_instruction(ins, op::quant_convolution{padding, stride, dilation, padding_mode, group}, converted_inputs);
if (adjust_factor == 1.0f)
auto quant_conv = prog.insert_instruction(
ins,
op::quant_convolution{padding, stride, dilation, padding_mode, group},
converted_inputs);
if(adjust_factor == 1.0f)
{
prog.replace_instruction(ins, op::convert{orig_type}, quant_conv);
}
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment