mirrors/ZLUDA
1
0
Fork 0
mirror of https://github.com/vosen/ZLUDA.git synced 2025-04-20 00:14:45 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Fix more bugs in implicit conversions

Browse source
This commit is contained in:
Andrzej Janik 2020-09-20 20:43:09 +02:00
parent b11ce92d2c
commit 5a7860d9f1
9 changed files with 243 additions and 72 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

25
ptx/src/test/spirv_run/b64tof64.ptx Normal file
View file

@ -0,0 +1,25 @@
.version 6.5
.target sm_30
.address_size 64
.visible .entry b64tof64(
.param .u64 input,
.param .u64 output
)
{
.reg .f64 in_addr_f;
.reg .b64 in_addr;
.reg .u64 out_addr;
.reg.u64 temp;
ld.param.f64 in_addr_f, [input];
ld.param.u64 out_addr, [output];
mov.b64 in_addr, in_addr_f;
ld.u64 temp, [in_addr];
st.u64 [out_addr], temp;
ret;
}

50
ptx/src/test/spirv_run/b64tof64.spvtxt Normal file
View file

@ -0,0 +1,50 @@
OpCapability GenericPointer
OpCapability Linkage
OpCapability Addresses
OpCapability Kernel
OpCapability Int64
OpCapability Int8
OpCapability Float64
%26 = OpExtInstImport "OpenCL.std"
OpMemoryModel Physical64 OpenCL
OpEntryPoint Kernel %1 "b64tof64"
%void = OpTypeVoid
%ulong = OpTypeInt 64 0
%29 = OpTypeFunction %void %ulong %ulong
%_ptr_Function_ulong = OpTypePointer Function %ulong
%double = OpTypeFloat 64
%_ptr_Function_double = OpTypePointer Function %double
%_ptr_Generic_ulong = OpTypePointer Generic %ulong
%1 = OpFunction %void None %29
%8 = OpFunctionParameter %ulong
%9 = OpFunctionParameter %ulong
%24 = OpLabel
%2 = OpVariable %_ptr_Function_ulong Function
%3 = OpVariable %_ptr_Function_ulong Function
%4 = OpVariable %_ptr_Function_double Function
%5 = OpVariable %_ptr_Function_ulong Function
%6 = OpVariable %_ptr_Function_ulong Function
%7 = OpVariable %_ptr_Function_ulong Function
OpStore %2 %8
OpStore %3 %9
%11 = OpLoad %ulong %2
%20 = OpBitcast %double %11
%10 = OpCopyObject %double %20
OpStore %4 %10
%13 = OpLoad %ulong %3
%12 = OpCopyObject %ulong %13
OpStore %6 %12
%15 = OpLoad %double %4
%21 = OpBitcast %ulong %15
%14 = OpCopyObject %ulong %21
OpStore %5 %14
%17 = OpLoad %ulong %5
%22 = OpConvertUToPtr %_ptr_Generic_ulong %17
%16 = OpLoad %ulong %22
OpStore %7 %16
%18 = OpLoad %ulong %6
%19 = OpLoad %ulong %7
%23 = OpConvertUToPtr %_ptr_Generic_ulong %18
OpStore %23 %19
OpReturn
OpFunctionEnd

24
ptx/src/test/spirv_run/implicit_param.ptx Normal file
View file

@ -0,0 +1,24 @@
.version 6.5
.target sm_30
.address_size 64
.visible .entry implicit_param(
.param .u64 input,
.param .u64 output
)
{
.reg .u64 in_addr;
.reg .u64 out_addr;
.reg .f32 temp;
.param .b32 temp_param;
ld.param.u64 in_addr, [input];
ld.param.u64 out_addr, [output];
ld.global.f32 temp, [in_addr];
st.param.f32 [temp_param], temp;
ld.param.f32 temp, [temp_param];
st.global.f32 [out_addr], temp;
ret;
}

55
ptx/src/test/spirv_run/implicit_param.spvtxt Normal file
View file

@ -0,0 +1,55 @@
OpCapability GenericPointer
OpCapability Linkage
OpCapability Addresses
OpCapability Kernel
OpCapability Int64
OpCapability Int8
OpCapability Float64
%28 = OpExtInstImport "OpenCL.std"
OpMemoryModel Physical64 OpenCL
OpEntryPoint Kernel %1 "implicit_param"
%void = OpTypeVoid
%ulong = OpTypeInt 64 0
%31 = OpTypeFunction %void %ulong %ulong
%_ptr_Function_ulong = OpTypePointer Function %ulong
%float = OpTypeFloat 32
%_ptr_Function_float = OpTypePointer Function %float
%uint = OpTypeInt 32 0
%_ptr_Function_uint = OpTypePointer Function %uint
%_ptr_CrossWorkgroup_float = OpTypePointer CrossWorkgroup %float
%1 = OpFunction %void None %31
%8 = OpFunctionParameter %ulong
%9 = OpFunctionParameter %ulong
%26 = OpLabel
%2 = OpVariable %_ptr_Function_ulong Function
%3 = OpVariable %_ptr_Function_ulong Function
%4 = OpVariable %_ptr_Function_ulong Function
%5 = OpVariable %_ptr_Function_ulong Function
%6 = OpVariable %_ptr_Function_float Function
%7 = OpVariable %_ptr_Function_uint Function
OpStore %2 %8
OpStore %3 %9
%11 = OpLoad %ulong %2
%10 = OpCopyObject %ulong %11
OpStore %4 %10
%13 = OpLoad %ulong %3
%12 = OpCopyObject %ulong %13
OpStore %5 %12
%15 = OpLoad %ulong %4
%22 = OpConvertUToPtr %_ptr_CrossWorkgroup_float %15
%14 = OpLoad %float %22
OpStore %6 %14
%17 = OpLoad %float %6
%23 = OpCopyObject %float %17
%16 = OpBitcast %uint %23
OpStore %7 %16
%19 = OpLoad %uint %7
%24 = OpBitcast %float %19
%18 = OpCopyObject %float %24
OpStore %6 %18
%20 = OpLoad %ulong %5
%21 = OpLoad %float %6
%25 = OpConvertUToPtr %_ptr_CrossWorkgroup_float %20
OpStore %25 %21
OpReturn
OpFunctionEnd

11
ptx/src/test/spirv_run/ld_st_implicit.spvtxt
View file

@ -4,6 +4,7 @@
OpCapability Kernel
OpCapability Int64
OpCapability Int8
OpCapability Float64
%23 = OpExtInstImport "OpenCL.std"
OpMemoryModel Physical64 OpenCL
OpEntryPoint Kernel %1 "ld_st_implicit"
@ -34,14 +35,14 @@
%14 = OpLoad %ulong %4
%17 = OpConvertUToPtr %_ptr_CrossWorkgroup_float %14
%18 = OpLoad %float %17
%30 = OpBitcast %ulong %18
%32 = OpUConvert %uint %30
%13 = OpBitcast %uint %32
%31 = OpBitcast %uint %18
%13 = OpUConvert %ulong %31
OpStore %6 %13
%15 = OpLoad %ulong %5
%16 = OpLoad %ulong %6
%33 = OpBitcast %uint %16
%19 = OpUConvert %ulong %33
%32 = OpBitcast %ulong %16
%33 = OpUConvert %uint %32
%19 = OpBitcast %float %33
%20 = OpConvertUToPtr %_ptr_CrossWorkgroup_float %15
OpStore %20 %19
OpReturn

4
ptx/src/test/spirv_run/mod.rs
View file

@ -60,8 +60,10 @@ test_ptx!(call, [1u64], [2u64]);
test_ptx!(vector, [1u32, 2u32], [3u32, 3u32]);
test_ptx!(ld_st_offset, [1u32, 2u32], [2u32, 1u32]);
test_ptx!(ntid, [3u32], [4u32]);
test_ptx!(reg_local, [12u64], [12u64]);
test_ptx!(reg_local, [12u64], [13u64]);
test_ptx!(mov_address, [0xDEADu64], [0u64]);
test_ptx!(b64tof64, [111u64], [111u64]);
test_ptx!(implicit_param, [34u32], [34u32]);
struct DisplayError<T: Debug> {
err: T,

2
ptx/src/test/spirv_run/reg_local.ptx
View file

@ -17,7 +17,7 @@
ld.param.u64 out_addr, [output];
ld.global.u64 temp, [in_addr];
st.u64 [local_x], temp;
st.u64 [local_x], temp + 1;
ld.u64 temp, [local_x];
st.global.u64 [out_addr], temp;
ret;

71
ptx/src/test/spirv_run/reg_local.spvtxt
View file

@ -4,43 +4,60 @@
OpCapability Kernel
OpCapability Int64
OpCapability Int8
%25 = OpExtInstImport "OpenCL.std"
%35 = OpExtInstImport "OpenCL.std"
OpMemoryModel Physical64 OpenCL
OpEntryPoint Kernel %1 "add"
OpEntryPoint Kernel %1 "reg_local"
OpDecorate %4 Alignment 8
%void = OpTypeVoid
%ulong = OpTypeInt 64 0
%28 = OpTypeFunction %void %ulong %ulong
%38 = OpTypeFunction %void %ulong %ulong
%_ptr_Function_ulong = OpTypePointer Function %ulong
%_ptr_Generic_ulong = OpTypePointer Generic %ulong
%uchar = OpTypeInt 8 0
%uint = OpTypeInt 32 0
%uint_8 = OpConstant %uint 8
%_arr_uchar_uint_8 = OpTypeArray %uchar %uint_8
%_ptr_Function__arr_uchar_uint_8 = OpTypePointer Function %_arr_uchar_uint_8
%_ptr_CrossWorkgroup_ulong = OpTypePointer CrossWorkgroup %ulong
%ulong_1 = OpConstant %ulong 1
%1 = OpFunction %void None %28
%8 = OpFunctionParameter %ulong
%1 = OpFunction %void None %38
%9 = OpFunctionParameter %ulong
%23 = OpLabel
%10 = OpFunctionParameter %ulong
%33 = OpLabel
%2 = OpVariable %_ptr_Function_ulong Function
%3 = OpVariable %_ptr_Function_ulong Function
%4 = OpVariable %_ptr_Function_ulong Function
%4 = OpVariable %_ptr_Function__arr_uchar_uint_8 Function
%5 = OpVariable %_ptr_Function_ulong Function
%6 = OpVariable %_ptr_Function_ulong Function
%7 = OpVariable %_ptr_Function_ulong Function
OpStore %2 %8
OpStore %3 %9
%11 = OpLoad %ulong %2
%10 = OpCopyObject %ulong %11
OpStore %4 %10
%13 = OpLoad %ulong %3
%12 = OpCopyObject %ulong %13
OpStore %5 %12
%15 = OpLoad %ulong %4
%21 = OpConvertUToPtr %_ptr_Generic_ulong %15
%14 = OpLoad %ulong %21
OpStore %6 %14
%17 = OpLoad %ulong %6
%16 = OpIAdd %ulong %17 %ulong_1
OpStore %7 %16
%18 = OpLoad %ulong %5
%19 = OpLoad %ulong %7
%22 = OpConvertUToPtr %_ptr_Generic_ulong %18
OpStore %22 %19
%8 = OpVariable %_ptr_Function_ulong Function
OpStore %2 %9
OpStore %3 %10
%12 = OpLoad %ulong %2
%11 = OpCopyObject %ulong %12
OpStore %5 %11
%14 = OpLoad %ulong %3
%13 = OpCopyObject %ulong %14
OpStore %6 %13
%16 = OpLoad %ulong %5
%25 = OpConvertUToPtr %_ptr_CrossWorkgroup_ulong %16
%26 = OpLoad %ulong %25
%15 = OpCopyObject %ulong %26
OpStore %7 %15
%18 = OpLoad %ulong %7
%27 = OpCopyObject %ulong %18
%24 = OpIAdd %ulong %27 %ulong_1
%28 = OpCopyObject %ulong %24
%17 = OpBitcast %ulong %28
OpStore %4 %17
%20 = OpLoad %_arr_uchar_uint_8 %4
%29 = OpBitcast %ulong %20
%30 = OpCopyObject %ulong %29
%19 = OpCopyObject %ulong %30
OpStore %7 %19
%21 = OpLoad %ulong %6
%22 = OpLoad %ulong %7
%31 = OpCopyObject %ulong %22
%32 = OpConvertUToPtr %_ptr_CrossWorkgroup_ulong %21
OpStore %32 %31
OpReturn
OpFunctionEnd

73
ptx/src/translate.rs
View file

@ -299,6 +299,7 @@ fn emit_function_header<'a>(
pub fn to_spirv<'a>(ast: ast::Module<'a>) -> Result<Vec<u32>, TranslateError> {
let module = to_spirv_module(ast)?;
eprintln!("{}", module.disassemble());
Ok(module.assemble())
}
@ -309,6 +310,7 @@ fn emit_capabilities(builder: &mut dr::Builder) {
builder.capability(spirv::Capability::Kernel);
builder.capability(spirv::Capability::Int64);
builder.capability(spirv::Capability::Int8);
builder.capability(spirv::Capability::Float64);
}
fn emit_extensions(_: &mut dr::Builder) {}
@ -990,8 +992,13 @@ fn insert_implicit_conversions(
Statement::Call(call) => insert_implicit_bitcasts(&mut result, id_def, call)?,
Statement::Instruction(inst) => match inst {
ast::Instruction::Ld(ld, arg) => {
let pre_conv =
get_implicit_conversions_ld_src(id_def, ld.typ, ld.state_space, arg.src)?;
let pre_conv = get_implicit_conversions_ld_src(
id_def,
ld.typ,
ld.state_space,
arg.src,
false,
)?;
let post_conv = get_implicit_conversions_ld_dst(
id_def,
ld.typ,
@ -1024,8 +1031,11 @@ fn insert_implicit_conversions(
st.typ,
st.state_space.to_ld_ss(),
arg.src1,
true,
)?;
let (pre_conv_dest, post_conv) = if st.state_space == ast::StStateSpace::Param {
let (pre_conv_dest, post_conv) = if st.state_space == ast::StStateSpace::Param
|| st.state_space == ast::StStateSpace::Local
{
(Vec::new(), post_conv)
} else {
(post_conv, Vec::new())
@ -1667,7 +1677,7 @@ fn emit_implicit_conversion(
}
(TypeKind::Scalar, TypeKind::Scalar, ConversionKind::Default) => {
if from_parts.width == to_parts.width {
let dst_type = map.get_or_add(builder, SpirvType::from(cv.from));
let dst_type = map.get_or_add(builder, SpirvType::from(cv.to));
if from_parts.scalar_kind != ScalarKind::Float
&& to_parts.scalar_kind != ScalarKind::Float
{
@ -1714,7 +1724,7 @@ fn emit_implicit_conversion(
}
(TypeKind::Scalar, TypeKind::Scalar, ConversionKind::SignExtend) => todo!(),
(TypeKind::Vector, TypeKind::Scalar, ConversionKind::Default)
| (TypeKind::Scalar, TypeKind::Vector, ConversionKind::Default)
| (TypeKind::Scalar, TypeKind::Array, ConversionKind::Default)
| (TypeKind::Array, TypeKind::Scalar, ConversionKind::Default) => {
let into_type = map.get_or_add(builder, SpirvType::from(cv.to));
builder.bitcast(into_type, Some(cv.dst), cv.src)?;
@ -2409,7 +2419,8 @@ impl<T: ArgParamsEx> ast::Instruction<T> {
ast::Instruction::Call(_) => unreachable!(),
ast::Instruction::Ld(d, a) => {
let inst_type = d.typ;
let is_param = d.state_space == ast::LdStateSpace::Param;
let is_param = d.state_space == ast::LdStateSpace::Param
|| d.state_space == ast::LdStateSpace::Local;
ast::Instruction::Ld(d, a.map_ld(visitor, inst_type, is_param)?)
}
ast::Instruction::Mov(d, a) => {
@ -2432,7 +2443,9 @@ impl<T: ArgParamsEx> ast::Instruction<T> {
let inst_type = d.typ;
ast::Instruction::SetpBool(d, a.map(visitor, ast::Type::Scalar(inst_type))?)
}
ast::Instruction::Not(t, a) => ast::Instruction::Not(t, a.map(visitor, false, t.to_type())?),
ast::Instruction::Not(t, a) => {
ast::Instruction::Not(t, a.map(visitor, false, t.to_type())?)
}
ast::Instruction::Cvt(d, a) => {
let (dst_t, src_t) = match &d {
ast::CvtDetails::FloatFromFloat(desc) => (
@ -2459,7 +2472,8 @@ impl<T: ArgParamsEx> ast::Instruction<T> {
}
ast::Instruction::St(d, a) => {
let inst_type = d.typ;
let is_param = d.state_space == ast::StStateSpace::Param;
let is_param = d.state_space == ast::StStateSpace::Param
|| d.state_space == ast::StStateSpace::Local;
ast::Instruction::St(d, a.map(visitor, inst_type, is_param)?)
}
ast::Instruction::Bra(d, a) => ast::Instruction::Bra(d, a.map(visitor, None)?),
@ -3419,8 +3433,8 @@ fn get_implicit_conversions_ld_dst<
Ok(Some(ImplicitConversion {
src: u32::max_value(),
dst: u32::max_value(),
from: if !in_reverse { dst_type } else { instr_type },
to: if !in_reverse { instr_type } else { dst_type },
from: if !in_reverse { instr_type } else { dst_type },
to: if !in_reverse { dst_type } else { instr_type },
kind: conv,
}))
} else {
@ -3433,6 +3447,7 @@ fn get_implicit_conversions_ld_src(
instr_type: ast::Type,
state_space: ast::LdStateSpace,
src: spirv::Word,
in_reverse_param_local: bool,
) -> Result<Vec<ImplicitConversion>, TranslateError> {
let src_type = id_def.get_typed(src)?;
match state_space {
@ -3442,8 +3457,16 @@ fn get_implicit_conversions_ld_src(
ImplicitConversion {
src: u32::max_value(),
dst: u32::max_value(),
from: src_type,
to: instr_type,
from: if !in_reverse_param_local {
src_type
} else {
instr_type
},
to: if !in_reverse_param_local {
instr_type
} else {
src_type
},
kind: ConversionKind::Default,
};
1
@ -3512,32 +3535,6 @@ fn insert_conversion_src(
temp_src
}
/*
fn insert_with_implicit_conversion_dst<
T,
ShouldConvert: FnOnce(ast::StateSpace, ast::Type, ast::Type) -> Option<ConversionKind>,
Setter: Fn(&mut T) -> &mut spirv::Word,
ToInstruction: FnOnce(T) -> ast::Instruction<ExpandedArgParams>,
>(
func: &mut Vec<ExpandedStatement>,
instr_type: ast::Type,
id_def: &mut NumericIdResolver,
should_convert: ShouldConvert,
mut t: T,
setter: Setter,
to_inst: ToInstruction,
) {
let dst = setter(&mut t);
let dst_type = id_def.get_type(*dst);
let dst_coercion = should_convert(dst_type.unwrap(), instr_type)
.map(|conv| get_conversion_dst(id_def, dst, instr_type, dst_type.unwrap(), conv));
func.push(Statement::Instruction(to_inst(t)));
if let Some(conv) = dst_coercion {
func.push(conv);
}
}
*/
#[must_use]
fn get_conversion_dst(
id_def: &mut MutableNumericIdResolver,