[EISW-191805] Enable quantization for FP4 type #175
+106
−33
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -35,9 +35,8 @@ static Type parseStorageType(DialectAsmParser &parser, bool &isSigned) { | |
| if (auto intType = llvm::dyn_cast<IntegerType>(type)) { | ||
| isSigned = !intType.isUnsigned(); | ||
| storageTypeWidth = intType.getWidth(); | ||
| } else if (mlir::isa<Float8E5M2Type, Float8E4M3FNType, Float4E2M1FNType>(type)) { | ||
|
Contributor
There was a problem hiding this comment. ditto. just cast to FloatType? |
||
| storageTypeWidth = llvm::dyn_cast<FloatType>(type).getWidth(); | ||
| isSigned = true; | ||
| } else { | ||
| parser.emitError(typeLoc, "illegal quantized storage type alias"); | ||
|
|
@@ -132,12 +131,15 @@ static ParseResult parseStorageRange(DialectAsmParser &parser, Type storageType, | |
| const auto width = llvm::dyn_cast<IntegerType>(storageType).getWidth(); | ||
| defaultMin = QuantizedType::getDefaultMinimumForInteger(isSigned, width); | ||
| defaultMax = QuantizedType::getDefaultMaximumForInteger(isSigned, width); | ||
| } else if (mlir::isa<Float8E5M2Type>(storageType)) { | ||
| defaultMin = QuantizedType::getDefaultMinimumForF8E5M2(); | ||
| defaultMax = QuantizedType::getDefaultMaximumForF8E5M2(); | ||
| } else if (mlir::isa<Float8E4M3FNType>(storageType)) { | ||
| defaultMin = QuantizedType::getDefaultMinimumForF8E4M3FN(); | ||
| defaultMax = QuantizedType::getDefaultMaximumForF8E4M3FN(); | ||
| } else if (mlir::isa<Float4E2M1FNType>(storageType)) { | ||
| defaultMin = QuantizedType::getDefaultMinimumForF4E2M1FN(); | ||
| defaultMax = QuantizedType::getDefaultMaximumForF4E2M1FN(); | ||
| } else { | ||
| defaultMin = std::numeric_limits<int64_t>::max(); | ||
| defaultMax = std::numeric_limits<int64_t>::min(); | ||
|
|
@@ -150,7 +152,7 @@ static ParseResult parseStorageRange(DialectAsmParser &parser, Type storageType, | |
| } | ||
|
|
||
| // Explicit storage min and storage max. | ||
| // F8 and F4 min and max values are integers, so parseInteger() is used. | ||
| SMLoc minLoc = parser.getCurrentLocation(), maxLoc; | ||
| if (parser.parseInteger(storageTypeMin) || parser.parseColon() || | ||
| parser.getCurrentLocation(&maxLoc) || | ||
|
|
@@ -382,7 +384,7 @@ parseQuantParamListUntilRBrace(DialectAsmParser &parser, Type expressedType, | |
| /// block-size-info `,` scale-zero-tensor `>` | ||
| /// storage-spec ::= storage-type (`<` storage-range `>`)? | ||
| /// storage-range ::= integer-literal `:` integer-literal | ||
| /// storage-type ::= (`i` | `u`) integer-literal | `f8E5M2` | `f8E4M3FN` | `f4E2M1FN` | ||
| /// expressed-type-spec ::= `:` `f` integer-literal | ||
| /// axis-spec ::= `:` integer-literal | ||
| /// scale-zero ::= scale (`:` zero-point)? | ||
|
|
@@ -407,9 +409,9 @@ parseQuantParamListUntilRBrace(DialectAsmParser &parser, Type expressedType, | |
| /// scale-zero-list `>` | ||
| /// storage-spec ::= storage-type (`<` storage-range `>`)? | ||
| /// storage-range ::= integer-literal `:` integer-literal | ||
| /// storage-type ::= (`i` | `u`) integer-literal | `f8E5M2` | `f8E4M3FN` | `f4E2M1FN` | ||
| /// quantile-type-spec ::= `:` ((`i` | `u` | `f`) integer-literal | `f8E5M2` | | ||
| /// `f8E4M3FN` | `f4E2M1FN`) | ||
| /// expressed-type-spec ::= `:` `f` integer-literal axis-spec ::= | ||
| /// `:` integer-literal quantiles-list ::= `{` quantile (`,` quantile)* `}` | ||
| /// scale-zero ::= `:` float-literal `:` integer-literal | ||
|
|
@@ -641,6 +643,8 @@ static void printStorageType(QuantizedType type, DialectAsmPrinter &out) { | |
| out << "f8E5M2"; | ||
| } else if (type.getStorageType().isa<Float8E4M3FNType>()) { | ||
| out << "f8E4M3FN"; | ||
| } else if (type.getStorageType().isa<Float4E2M1FNType>()) { | ||
| out << "f4E2M1FN"; | ||
| } else if (isSigned) { | ||
| out << "i" << storageWidth; | ||
| } else { | ||
|
|
@@ -655,7 +659,9 @@ static void printStorageType(QuantizedType type, DialectAsmPrinter &out) { | |
| ? QuantizedType::getDefaultMinimumForF8E5M2() | ||
| : type.getStorageType().isa<Float8E4M3FNType>() | ||
| ? QuantizedType::getDefaultMinimumForF8E4M3FN() | ||
| : type.getStorageType().isa<Float4E2M1FNType>() | ||
| ? QuantizedType::getDefaultMinimumForF4E2M1FN() | ||
| : std::numeric_limits<int64_t>::max(); | ||
|
|
||
| int64_t defaultMax = | ||
| type.getStorageType().isa<IntegerType>() | ||
|
|
@@ -664,7 +670,9 @@ static void printStorageType(QuantizedType type, DialectAsmPrinter &out) { | |
| ? QuantizedType::getDefaultMaximumForF8E5M2() | ||
| : type.getStorageType().isa<Float8E4M3FNType>() | ||
| ? QuantizedType::getDefaultMaximumForF8E4M3FN() | ||
| : type.getStorageType().isa<Float4E2M1FNType>() | ||
| ? QuantizedType::getDefaultMaximumForF4E2M1FN() | ||
| : std::numeric_limits<int64_t>::min(); | ||
|
|
||
| if (defaultMin != type.getStorageTypeMin() || | ||
| defaultMax != type.getStorageTypeMax()) { | ||
|
|
@@ -685,6 +693,8 @@ static void printQuantileType(Type quantileType, DialectAsmPrinter &out) { | |
| out << ":f8E5M2"; | ||
| } else if (quantileType.isa<Float8E4M3FNType>()) { | ||
| out << ":f8E4M3FN"; | ||
| } else if (quantileType.isa<Float4E2M1FNType>()) { | ||
| out << ":f4E2M1FN"; | ||
| } else { | ||
| // Float types | ||
| out << ":" << quantileType; | ||
|
|
||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -126,6 +126,16 @@ func.func @parse() -> !qalias { | |
| // expected-error@+1 {{illegal storage type minimum: -500}} | ||
| !qalias = !quant.quantile<f8E4M3FN<-500:448>:f16:f32, {-1.0,1.0}:0.99872:127> | ||
|
|
||
| // ----- | ||
| // Illegal storage min/max: max > defaultMax | ||
| // expected-error@+1 {{illegal storage type maximum: 10}} | ||
| !qalias = !quant.quantile<f4E2M1FN<-6:10>:f16:f32, {-1.0,1.0}:0.99872:127> | ||
|
|
||
| // ----- | ||
| // Illegal storage min/max: min < defaultMin | ||
| // expected-error@+1 {{illegal storage type minimum: -10}} | ||
| !qalias = !quant.quantile<f4E2M1FN<-10:6>:f16:f32, {-1.0,1.0}:0.99872:127> | ||
|
|
||
| // ----- | ||
| // Illegal uniform params: invalid scale | ||
| // expected-error@+1 {{expected floating point literal}} | ||
|
|
||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -46,6 +46,15 @@ func.func @parse() -> !qalias { | |
| return %0 : !qalias | ||
| } | ||
|
|
||
| // ----- | ||
| // Default min/max value optimization for f4E2M1FN. | ||
| // CHECK: !quant.quantile<f4E2M1FN:f16:f32, {-1.000000e+00,-8.667000e-01,-7.333000e-01,-6.000000e-01,-4.667000e-01,-3.333000e-01,-2.000000e-01,-0.066699999999999995,0.066699999999999995,2.000000e-01,3.333000e-01,4.667000e-01,6.000000e-01,7.333000e-01,8.667000e-01,1.000000e+00}:9.987200e-01:127> | ||
| !qalias = !quant.quantile<f4E2M1FN<-6:6>:f16:f32, {-1.0000,-0.8667,-0.7333,-0.6000,-0.4667,-0.3333,-0.2000,-0.0667,0.0667,0.2000,0.3333,0.4667,0.6000,0.7333,0.8667,1.0000}:0.99872:127 > | ||
| func.func @parse() -> !qalias { | ||
| %0 = "foo"() : () -> !qalias | ||
| return %0 : !qalias | ||
| } | ||
|
|
||
| // ----- | ||
| // Required per-layer params specified: | ||
| // [unsigned] storageType, expressedType, scale | ||
|
|
@@ -92,6 +101,15 @@ func.func @parse() -> !qalias { | |
| return %0 : !qalias | ||
| } | ||
|
|
||
| // ----- | ||
| // Storage type: f4E2M1FN | ||
| // CHECK: !quant.quantile<f4E2M1FN:f16:f32, {-1.000000e+00,-8.667000e-01,-7.333000e-01,-6.000000e-01,-4.667000e-01,-3.333000e-01,-2.000000e-01,-0.066699999999999995,0.066699999999999995,2.000000e-01,3.333000e-01,4.667000e-01,6.000000e-01,7.333000e-01,8.667000e-01,1.000000e+00}:2.000000e+02> | ||
| !qalias = !quant.quantile<f4E2M1FN:f16:f32, {-1.0000,-0.8667,-0.7333,-0.6000,-0.4667,-0.3333,-0.2000,-0.0667,0.0667,0.2000,0.3333,0.4667,0.6000,0.7333,0.8667,1.0000}:2.0e+2> | ||
| func.func @parse() -> !qalias { | ||
| %0 = "foo"() : () -> !qalias | ||
| return %0 : !qalias | ||
| } | ||
|
|
||
| // ----- | ||
| // Expressed type: f32 | ||
| // CHECK: !quant.quantile<u4:f16:f32, {-1.000000e+00,-8.667000e-01,-7.333000e-01,-6.000000e-01,-4.667000e-01,-3.333000e-01,-2.000000e-01,-0.066699999999999995,0.066699999999999995,2.000000e-01,3.333000e-01,4.667000e-01,6.000000e-01,7.333000e-01,8.667000e-01,1.000000e+00}:2.000000e+02> | ||
|
|
||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -100,6 +100,16 @@ | |
| // expected-error@+1 {{illegal storage type minimum: -500}} | ||
| !qalias = !quant.uniform<f8E4M3FN<-500:448>:f32, 0.99872:127> | ||
|
|
||
| // ----- | ||
| // Illegal storage min/max: max > defaultMax | ||
| // expected-error@+1 {{illegal storage type maximum: 10}} | ||
| !qalias = !quant.uniform<f4E2M1FN<-6:10>:f32, 0.99872:127> | ||
|
|
||
| // ----- | ||
| // Illegal storage min/max: min < defaultMin | ||
| // expected-error@+1 {{illegal storage type minimum: -10}} | ||
| !qalias = !quant.uniform<f4E2M1FN<-10:6>:f32, 0.99872:127> | ||
|
|
||
| // ----- | ||
| // Illegal uniform params: invalid scale | ||
| // expected-error@+1 {{expected floating point literal}} | ||
|
|
||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -46,6 +46,15 @@ func.func @parse() -> !qalias { | |
| return %0 : !qalias | ||
| } | ||
|
|
||
| // ----- | ||
| // Default min/max value optimization for f4E2M1FN. | ||
| // CHECK: !quant.uniform<f4E2M1FN:f32, 9.987200e-01:127> | ||
| !qalias = !quant.uniform<f4E2M1FN<-6:6>:f32, 0.99872:127 > | ||
| func.func @parse() -> !qalias { | ||
| %0 = "foo"() : () -> !qalias | ||
| return %0 : !qalias | ||
| } | ||
|
|
||
| // ----- | ||
| // Required per-layer params specified: | ||
| // [unsigned] storageType, expressedType, scale | ||
|
|
@@ -92,6 +101,15 @@ func.func @parse() -> !qalias { | |
| return %0 : !qalias | ||
| } | ||
|
|
||
| // ----- | ||
| // Storage type: f4E2M1FN | ||
| // CHECK: !quant.uniform<f4E2M1FN:f32, 2.000000e+02> | ||
| !qalias = !quant.uniform<f4E2M1FN:f32, 2.0e+2> | ||
| func.func @parse() -> !qalias { | ||
| %0 = "foo"() : () -> !qalias | ||
| return %0 : !qalias | ||
| } | ||
|
|
||
| // ----- | ||
| // Storage type: i16 | ||
| // CHECK: !quant.uniform<i16:f32, 2.000000e+02> | ||
|
|
||
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this looks like a separate bug fix. why is it required?