Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix int4 conversion #4983

Merged
merged 2 commits into from
Mar 29, 2021
+31 −44
Merged

Fix int4 conversion #4983

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
e081091
Fix int4 conversion
ilyachur Mar 26, 2021
de0184f
Simplify logic
ilyachur Mar 26, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
59 changes: 23 additions & 36 deletions inference-engine/src/transformations/src/transformations/convert_precision.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -406,7 +406,7 @@ struct EnumClassHash {
* int4 [1011] -> int8 [00001011]
* * For signed types we copy all bits (except sign bit) to destination type and after
* that for negative values we set to 1 all higher bits:
* int4 [1011] -> int8 [11110011]
* int4 [1011] -> int8 [11111011]
*
* @param src source value !!! the type must be unsigned !!!
* @param dst destination value !!! the type must be unsigned !!!
Expand All @@ -427,42 +427,29 @@ void convert_lp_value(const SRC& src, DST& dst, size_t src_offset, size_t src_si
// dst [10001111 00000100] offset 5 size 9
// new_val [00000000 00000000]
DST new_val = 0;
// If source type is signed
if (is_signed) {
// Get the sign of value
// sign [00000000]
// invert value in order to use XOR
SRC sign = (~(val >> (src_size - 1))) & 0b1;
// Calculate diff in order to clean bits which don't exist in the source value
// diff 5
size_t diff = sizeof(SRC)*8 - src_size + 1;
// Clean unnecessary bits
// val [10100000]
val = val << diff;
// val [00000101]
val = (val >> diff);

// Negative number
if (!sign) {
// val [11110101]
val |= (src_max << (diff - 1));
// new_val [00000001 11111111]
new_val = (sign << (dst_size - 1)) ^ (dst_max >> (sizeof(DST) * 8 - dst_size));
// new_val [00000001 11110101]
new_val &= (dst_max << sizeof(SRC)*8) | val;
} else {
// new_val [00000000 00000101]
new_val = val;
}

// Calculate diff in order to clean bits which don't exist in the source value
// diff 4
size_t diff = sizeof(SRC)*8 - src_size;
// Clean unnecessary bits
// val [11010000]
val = val << diff;
// val [00001101]
val = val >> diff;

// Get the sign of value
// sign [00000001]
SRC sign = (val >> (src_size - 1)) & 0b1;

// If source type is signed and negative
if (is_signed && sign) {
// val [11111101]
val |= src_max << diff;
// new_val [00000001 11111111]
new_val = dst_max >> (sizeof(DST) * 8 - dst_size);
// new_val [00000001 11111101]
new_val &= (dst_max << sizeof(SRC)*8) | val;
} else {
// Calculate diff in order to clean bits which don't exist in the source value
// diff 4
size_t diff = sizeof(SRC)*8 - src_size;
// Clean unnecessary bits
// val [11010000]
val = val << diff;
// val [00001101]
val = val >> diff;
// new_val [00000000 00001101]
new_val = val;
}
Expand Down
16 changes: 8 additions & 8 deletions inference-engine/tests/functional/inference_engine/transformations/convert_precision.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -656,11 +656,11 @@ TEST(TransformationTests, ConvertPrecision_ConstantConversion_I4ToI8) {
}

TEST(TransformationTests, ConvertPrecision_ConstantConversion_I4ToU8_neg) {
constant_convert_test(element::Type_t::i4, element::Type_t::u8, 171, 242);
constant_convert_test(element::Type_t::i4, element::Type_t::u8, 171, 250);
}

TEST(TransformationTests, ConvertPrecision_ConstantConversion_I4ToI8_neg) {
constant_convert_test(element::Type_t::i4, element::Type_t::i8, 171, -14);
constant_convert_test(element::Type_t::i4, element::Type_t::i8, 171, -6);
}

TEST(TransformationTests, ConvertPrecision_ConstantConversion_U4ToI32) {
Expand All @@ -680,11 +680,11 @@ TEST(TransformationTests, ConvertPrecision_ConstantConversion_I4ToI32) {
}

TEST(TransformationTests, ConvertPrecision_ConstantConversion_I4ToU32_neg) {
constant_convert_test(element::Type_t::i4, element::Type_t::u32, 171, -14);
constant_convert_test(element::Type_t::i4, element::Type_t::u32, 171, -6);
}

TEST(TransformationTests, ConvertPrecision_ConstantConversion_I4ToI32_neg) {
constant_convert_test(element::Type_t::i4, element::Type_t::i32, 171, -14);
constant_convert_test(element::Type_t::i4, element::Type_t::i32, 171, -6);
}

TEST(TransformationTests, ConvertPrecision_ConstantConversion_U4ToI16) {
Expand All @@ -704,11 +704,11 @@ TEST(TransformationTests, ConvertPrecision_ConstantConversion_I4ToI16) {
}

TEST(TransformationTests, ConvertPrecision_ConstantConversion_I4ToU16_neg) {
constant_convert_test(element::Type_t::i4, element::Type_t::u16, 171, 65522);
constant_convert_test(element::Type_t::i4, element::Type_t::u16, 171, 65530);
}

TEST(TransformationTests, ConvertPrecision_ConstantConversion_I4ToI16_neg) {
constant_convert_test(element::Type_t::i4, element::Type_t::i16, 171, -14);
constant_convert_test(element::Type_t::i4, element::Type_t::i16, 171, -6);
}

TEST(TransformationTests, ConvertPrecision_ConstantConversion_U4ToI64) {
Expand All @@ -728,11 +728,11 @@ TEST(TransformationTests, ConvertPrecision_ConstantConversion_I4ToI64) {
}

TEST(TransformationTests, ConvertPrecision_ConstantConversion_I4ToU64_neg) {
constant_convert_test(element::Type_t::i4, element::Type_t::u64, 171, -14);
constant_convert_test(element::Type_t::i4, element::Type_t::u64, 171, -6);
}

TEST(TransformationTests, ConvertPrecision_ConstantConversion_I4ToI64_neg) {
constant_convert_test(element::Type_t::i4, element::Type_t::i64, 171, -14);
constant_convert_test(element::Type_t::i4, element::Type_t::i64, 171, -6);
}

TEST(TransformationTests, ConvertPrecision_ConstantConversion_U1ToU8) {
Expand Down