Skip to content

Support for shader_explicit_arithmetic_types and shader_16bit_storage #1001

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

Merged
bjornbytes merged 1 commit into from
Aug 9, 2026
Merged

Support for shader_explicit_arithmetic_types and shader_16bit_storage #1001

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
13 changes: 13 additions & 0 deletions src/api/l_graphics.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -101,9 +101,17 @@ StringEntry lovrDrawStyle[] = {
};

StringEntry lovrDataType[] = {
[TYPE_I8] = ENTRY("i8"),
[TYPE_I8x2] = ENTRY("i8x2"),
[TYPE_I8x4] = ENTRY("i8x4"),
[TYPE_U8] = ENTRY("u8"),
[TYPE_U8x2] = ENTRY("u8x2"),
[TYPE_U8x4] = ENTRY("u8x4"),
[TYPE_SN8] = ENTRY("sn8"),
[TYPE_SN8x2] = ENTRY("sn8x2"),
[TYPE_SN8x4] = ENTRY("sn8x4"),
[TYPE_UN8] = ENTRY("un8"),
[TYPE_UN8x2] = ENTRY("un8x2"),
[TYPE_UN8x4] = ENTRY("un8x4"),
[TYPE_SN10x3] = ENTRY("sn10x3"),
[TYPE_UN10x3] = ENTRY("un10x3"),
Expand All @@ -113,8 +121,10 @@ StringEntry lovrDataType[] = {
[TYPE_U16] = ENTRY("u16"),
[TYPE_U16x2] = ENTRY("u16x2"),
[TYPE_U16x4] = ENTRY("u16x4"),
[TYPE_SN16] = ENTRY("sn16"),
[TYPE_SN16x2] = ENTRY("sn16x2"),
[TYPE_SN16x4] = ENTRY("sn16x4"),
[TYPE_UN16] = ENTRY("un16"),
[TYPE_UN16x2] = ENTRY("un16x2"),
[TYPE_UN16x4] = ENTRY("un16x4"),
[TYPE_I32] = ENTRY("i32"),
Expand All @@ -125,6 +135,7 @@ StringEntry lovrDataType[] = {
[TYPE_U32x2] = ENTRY("u32x2"),
[TYPE_U32x3] = ENTRY("u32x3"),
[TYPE_U32x4] = ENTRY("u32x4"),
[TYPE_F16] = ENTRY("f16"),
[TYPE_F16x2] = ENTRY("f16x2"),
[TYPE_F16x4] = ENTRY("f16x4"),
[TYPE_F32] = ENTRY("f32"),
Expand Down Expand Up @@ -582,8 +593,10 @@ static int l_lovrGraphicsGetFeatures(lua_State* L) {
lua_pushboolean(L, features.indirectDrawFirstInstance), lua_setfield(L, -2, "indirectDrawFirstInstance");
lua_pushboolean(L, features.packedBuffers), lua_setfield(L, -2, "packedBuffers");
lua_pushboolean(L, features.float64), lua_setfield(L, -2, "float64");
lua_pushboolean(L, features.float16), lua_setfield(L, -2, "float16");
lua_pushboolean(L, features.int64), lua_setfield(L, -2, "int64");
lua_pushboolean(L, features.int16), lua_setfield(L, -2, "int16");
lua_pushboolean(L, features.int8), lua_setfield(L, -2, "int8");
return 1;
}

Expand Down
24 changes: 23 additions & 1 deletion src/api/l_graphics_buffer.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,9 +7,17 @@
#include <string.h>

static const uint32_t typeComponents[] = {
[TYPE_I8] = 1,
[TYPE_I8x2] = 2,
[TYPE_I8x4] = 4,
[TYPE_U8] = 1,
[TYPE_U8x2] = 2,
[TYPE_U8x4] = 4,
[TYPE_SN8] = 1,
[TYPE_SN8x2] = 2,
[TYPE_SN8x4] = 4,
[TYPE_UN8] = 1,
[TYPE_UN8x2] = 2,
[TYPE_UN8x4] = 4,
[TYPE_SN10x3] = 3,
[TYPE_UN10x3] = 3,
Expand All @@ -19,8 +27,10 @@ static const uint32_t typeComponents[] = {
[TYPE_U16] = 1,
[TYPE_U16x2] = 2,
[TYPE_U16x4] = 4,
[TYPE_SN16] = 1,
[TYPE_SN16x2] = 2,
[TYPE_SN16x4] = 4,
[TYPE_UN16] = 1,
[TYPE_UN16x2] = 2,
[TYPE_UN16x4] = 4,
[TYPE_I32] = 1,
Expand All @@ -31,6 +41,7 @@ static const uint32_t typeComponents[] = {
[TYPE_U32x2] = 2,
[TYPE_U32x3] = 3,
[TYPE_U32x4] = 4,
[TYPE_F16] = 1,
[TYPE_F16x2] = 2,
[TYPE_F16x4] = 4,
[TYPE_F32] = 1,
Expand Down Expand Up @@ -120,9 +131,17 @@ static bool luax_checkfieldn(lua_State* L, int index, const DataField* field, vo
luax_fieldcheck(L, lua_type(L, index + i) == LUA_TNUMBER, index + i, field, false);
double x = lua_tonumber(L, index + i);
switch (field->type) {
case TYPE_I8: p.i8[i] = (int8_t) x; break;
case TYPE_I8x2: p.i8[i] = (int8_t) x; break;
case TYPE_I8x4: p.i8[i] = (int8_t) x; break;
case TYPE_U8: p.u8[i] = (uint8_t) x; break;
case TYPE_U8x2: p.u8[i] = (uint8_t) x; break;
case TYPE_U8x4: p.u8[i] = (uint8_t) x; break;
case TYPE_SN8: p.i8[i] = (int8_t)(CLAMP(x, -1.f, 1.f) * INT8_MAX); break;
case TYPE_SN8x2: p.i8[i] = (int8_t)(CLAMP(x, -1.f, 1.f) * INT8_MAX); break;
case TYPE_SN8x4: p.i8[i] = (int8_t) (CLAMP(x, -1.f, 1.f) * INT8_MAX); break;
case TYPE_UN8: p.u8[i] = (uint8_t)(CLAMP(x, 0.f, 1.f) * UINT8_MAX); break;
case TYPE_UN8x2: p.u8[i] = (uint8_t)(CLAMP(x, 0.f, 1.f) * UINT8_MAX); break;
case TYPE_UN8x4: p.u8[i] = (uint8_t) (CLAMP(x, 0.f, 1.f) * UINT8_MAX); break;
case TYPE_SN10x3: p.u32[0] |= (((uint32_t) (int32_t) (CLAMP(x, -1.f, 1.f) * 511.f)) & 0x3ff) << (10 * i); break;
case TYPE_UN10x3: p.u32[0] |= (((uint32_t) (CLAMP(x, 0.f, 1.f) * 1023.f)) & 0x3ff) << (10 * i); break;
Expand All @@ -132,8 +151,10 @@ static bool luax_checkfieldn(lua_State* L, int index, const DataField* field, vo
case TYPE_U16: p.u16[i] = (uint16_t) x; break;
case TYPE_U16x2: p.u16[i] = (uint16_t) x; break;
case TYPE_U16x4: p.u16[i] = (uint16_t) x; break;
case TYPE_SN16: p.i16[i] = (int16_t)(CLAMP(x, -1.f, 1.f) * INT16_MAX); break;
case TYPE_SN16x2: p.i16[i] = (int16_t) (CLAMP(x, -1.f, 1.f) * INT16_MAX); break;
case TYPE_SN16x4: p.i16[i] = (int16_t) (CLAMP(x, -1.f, 1.f) * INT16_MAX); break;
case TYPE_UN16: p.u16[i] = (uint16_t)(CLAMP(x, 0.f, 1.f) * UINT16_MAX); break;
case TYPE_UN16x2: p.u16[i] = (uint16_t) (CLAMP(x, 0.f, 1.f) * UINT16_MAX); break;
case TYPE_UN16x4: p.u16[i] = (uint16_t) (CLAMP(x, 0.f, 1.f) * UINT16_MAX); break;
case TYPE_I32: p.i32[i] = (int32_t) x; break;
Expand All @@ -143,7 +164,8 @@ static bool luax_checkfieldn(lua_State* L, int index, const DataField* field, vo
case TYPE_U32: p.u32[i] = (uint32_t) x; break;
case TYPE_U32x2: p.u32[i] = (uint32_t) x; break;
case TYPE_U32x3: p.u32[i] = (uint32_t) x; break;
case TYPE_U32x4: p.i32[i] = (uint32_t) x; break;
case TYPE_U32x4: p.u32[i] = (uint32_t) x; break;
case TYPE_F16: p.u16[i] = float32to16(x); break;
case TYPE_F16x2: p.u16[i] = float32to16(x); break;
case TYPE_F16x4: p.u16[i] = float32to16(x); break;
case TYPE_F32: p.f32[i] = (float) x; break;
Expand Down
13 changes: 13 additions & 0 deletions src/core/gpu.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,9 +57,17 @@ void gpu_buffer_flush(gpu_buffer* buffer, uint32_t offset, uint32_t extent);
// Tree

typedef enum {
GPU_TYPE_I8,
GPU_TYPE_I8x2,
GPU_TYPE_I8x4,
GPU_TYPE_U8,
GPU_TYPE_U8x2,
GPU_TYPE_U8x4,
GPU_TYPE_SN8,
GPU_TYPE_SN8x2,
GPU_TYPE_SN8x4,
GPU_TYPE_UN8,
GPU_TYPE_UN8x2,
GPU_TYPE_UN8x4,
GPU_TYPE_SN10x3,
GPU_TYPE_UN10x3,
Expand All @@ -69,8 +77,10 @@ typedef enum {
GPU_TYPE_U16,
GPU_TYPE_U16x2,
GPU_TYPE_U16x4,
GPU_TYPE_SN16,
GPU_TYPE_SN16x2,
GPU_TYPE_SN16x4,
GPU_TYPE_UN16,
GPU_TYPE_UN16x2,
GPU_TYPE_UN16x4,
GPU_TYPE_I32,
Expand All @@ -81,6 +91,7 @@ typedef enum {
GPU_TYPE_U32x2,
GPU_TYPE_U32x3,
GPU_TYPE_U32x4,
GPU_TYPE_F16,
GPU_TYPE_F16x2,
GPU_TYPE_F16x4,
GPU_TYPE_F32,
Expand Down Expand Up @@ -807,8 +818,10 @@ typedef struct {
bool subgroupQuad;
bool float32AtomicAdd;
bool float64;
bool float16;
bool int64;
bool int16;
bool int8;
} gpu_features;

typedef struct {
Expand Down
50 changes: 48 additions & 2 deletions src/core/gpu_vk.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -244,6 +244,9 @@ typedef struct {
bool hostImageCopy;
bool atomicFloat;
bool calibratedTimestamps;
bool float16Int8;
bool storage8;
bool storage16;
} gpu_extensions;

// State
Expand Down Expand Up @@ -3125,7 +3128,10 @@ bool gpu_init(gpu_config* config) {
{ "VK_EXT_pipeline_creation_cache_control", true, &state.extensions.pipelineCacheControl },
{ "VK_EXT_memory_budget", true, &state.extensions.memoryBudget },
{ "VK_EXT_host_image_copy", true, &state.extensions.hostImageCopy },
{ "VK_EXT_shader_atomic_float", true, &state.extensions.atomicFloat }
{ "VK_EXT_shader_atomic_float", true, &state.extensions.atomicFloat },
{ "VK_KHR_shader_float16_int8", true, &state.extensions.float16Int8 },
{ "VK_KHR_8bit_storage", true, &state.extensions.storage8 },
{ "VK_KHR_16bit_storage", true, &state.extensions.storage16 }
};

uint32_t extensionCount = 0;
Expand Down Expand Up @@ -3228,6 +3234,9 @@ bool gpu_init(gpu_config* config) {
VkPhysicalDeviceRayQueryFeaturesKHR rayQueryFeatures = { .sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_RAY_QUERY_FEATURES_KHR };
VkPhysicalDeviceHostImageCopyFeaturesEXT hostImageCopyFeatures = { .sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_HOST_IMAGE_COPY_FEATURES_EXT };
VkPhysicalDeviceShaderAtomicFloatFeaturesEXT atomicFloatFeatures = { .sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SHADER_ATOMIC_FLOAT_FEATURES_EXT };
VkPhysicalDeviceShaderFloat16Int8Features float16int8 = { .sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SHADER_FLOAT16_INT8_FEATURES };
VkPhysicalDevice8BitStorageFeatures storage8 = { .sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_8BIT_STORAGE_FEATURES };
VkPhysicalDevice16BitStorageFeatures storage16 = { .sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_16BIT_STORAGE_FEATURES };

if (state.extensions.foveation) {
CHAIN(supported, fragmentDensityMapFeatures);
Expand All @@ -3237,6 +3246,18 @@ bool gpu_init(gpu_config* config) {
CHAIN(supported, atomicFloatFeatures);
}

if (state.extensions.float16Int8) {
CHAIN(supported, float16int8);
}

if (state.extensions.storage8) {
CHAIN(supported, storage8);
}

if (state.extensions.storage16) {
CHAIN(supported, storage16);
}

vkGetPhysicalDeviceFeatures2(state.adapter, &supported);

VkPhysicalDeviceFeatures2 enabled = { .sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_FEATURES_2 };
Expand Down Expand Up @@ -3315,6 +3336,18 @@ bool gpu_init(gpu_config* config) {
CHAIN(enabled, atomicFloatFeatures);
}

if (state.extensions.float16Int8) {
CHAIN(enabled, float16int8);
}

if (state.extensions.storage8) {
CHAIN(enabled, storage8);
}

if (state.extensions.storage16) {
CHAIN(enabled, storage16);
}

if (config->features) {
config->features->textureBC = enabled.features.textureCompressionBC;
config->features->textureASTC = enabled.features.textureCompressionASTC_LDR;
Expand All @@ -3335,8 +3368,10 @@ bool gpu_init(gpu_config* config) {
config->features->subgroupQuad = subgroupProperties.supportedOperations & VK_SUBGROUP_FEATURE_QUAD_BIT;
config->features->float32AtomicAdd = atomicFloatFeatures.shaderBufferFloat32AtomicAdd && atomicFloatFeatures.shaderImageFloat32AtomicAdd;
config->features->float64 = enabled.features.shaderFloat64;
config->features->float16 = float16int8.shaderFloat16 && storage16.storageBuffer16BitAccess;
config->features->int64 = enabled.features.shaderInt64;
config->features->int16 = enabled.features.shaderInt16;
config->features->int16 = enabled.features.shaderInt16 && storage16.storageBuffer16BitAccess;
config->features->int8 = float16int8.shaderInt8 && storage8.storageBuffer8BitAccess;

// Formats
for (uint32_t i = 0; i < GPU_FORMAT_COUNT; i++) {
Expand Down Expand Up @@ -4349,9 +4384,17 @@ static VkFormat convertFormat(gpu_texture_format format, int colorspace) {

static VkFormat convertAttributeType(gpu_attribute_type type) {
static const VkFormat types[] = {
[GPU_TYPE_I8] = VK_FORMAT_R8_SINT,
[GPU_TYPE_I8x2] = VK_FORMAT_R8G8_SINT,
[GPU_TYPE_I8x4] = VK_FORMAT_R8G8B8A8_SINT,
[GPU_TYPE_U8] = VK_FORMAT_R8_UINT,
[GPU_TYPE_U8x2] = VK_FORMAT_R8G8_UINT,
[GPU_TYPE_U8x4] = VK_FORMAT_R8G8B8A8_UINT,
[GPU_TYPE_SN8] = VK_FORMAT_R8_SNORM,
[GPU_TYPE_SN8x2] = VK_FORMAT_R8G8_SNORM,
[GPU_TYPE_SN8x4] = VK_FORMAT_R8G8B8A8_SNORM,
[GPU_TYPE_UN8] = VK_FORMAT_R8_UNORM,
[GPU_TYPE_UN8x2] = VK_FORMAT_R8G8_UNORM,
[GPU_TYPE_UN8x4] = VK_FORMAT_R8G8B8A8_UNORM,
[GPU_TYPE_SN10x3] = VK_FORMAT_A2B10G10R10_SNORM_PACK32,
[GPU_TYPE_UN10x3] = VK_FORMAT_A2B10G10R10_UNORM_PACK32,
Expand All @@ -4361,8 +4404,10 @@ static VkFormat convertAttributeType(gpu_attribute_type type) {
[GPU_TYPE_U16] = VK_FORMAT_R16_UINT,
[GPU_TYPE_U16x2] = VK_FORMAT_R16G16_UINT,
[GPU_TYPE_U16x4] = VK_FORMAT_R16G16B16A16_UINT,
[GPU_TYPE_SN16] = VK_FORMAT_R16_SNORM,
[GPU_TYPE_SN16x2] = VK_FORMAT_R16G16_SNORM,
[GPU_TYPE_SN16x4] = VK_FORMAT_R16G16B16A16_SNORM,
[GPU_TYPE_UN16] = VK_FORMAT_R16_UNORM,
[GPU_TYPE_UN16x2] = VK_FORMAT_R16G16_UNORM,
[GPU_TYPE_UN16x4] = VK_FORMAT_R16G16B16A16_UNORM,
[GPU_TYPE_I32] = VK_FORMAT_R32_SINT,
Expand All @@ -4373,6 +4418,7 @@ static VkFormat convertAttributeType(gpu_attribute_type type) {
[GPU_TYPE_U32x2] = VK_FORMAT_R32G32_UINT,
[GPU_TYPE_U32x3] = VK_FORMAT_R32G32B32_UINT,
[GPU_TYPE_U32x4] = VK_FORMAT_R32G32B32A32_UINT,
[GPU_TYPE_F16] = VK_FORMAT_R16_SFLOAT,
[GPU_TYPE_F16x2] = VK_FORMAT_R16G16_SFLOAT,
[GPU_TYPE_F16x4] = VK_FORMAT_R16G16B16A16_SFLOAT,
[GPU_TYPE_F32] = VK_FORMAT_R32_SFLOAT,
Expand Down
13 changes: 13 additions & 0 deletions src/core/gpu_web.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -645,9 +645,17 @@ bool gpu_pipeline_init_graphics(gpu_pipeline* pipeline, gpu_pipeline_info* info,
};

static const WGPU_VERTEX_FORMAT attributeTypes[] = {
[GPU_TYPE_I8] = WGPU_VERTEX_FORMAT_SINT8,
[GPU_TYPE_I8x2] = WGPU_VERTEX_FORMAT_SINT8X2,
[GPU_TYPE_I8x4] = WGPU_VERTEX_FORMAT_SINT8X4,
[GPU_TYPE_U8] = WGPU_VERTEX_FORMAT_UINT8,
[GPU_TYPE_U8x2] = WGPU_VERTEX_FORMAT_UINT8X2,
[GPU_TYPE_U8x4] = WGPU_VERTEX_FORMAT_UINT8X4,
[GPU_TYPE_SN8] = WGPU_VERTEX_FORMAT_SNORM8,
[GPU_TYPE_SN8x2] = WGPU_VERTEX_FORMAT_SNORM8X2,
[GPU_TYPE_SN8x4] = WGPU_VERTEX_FORMAT_SNORM8X4,
[GPU_TYPE_UN8] = WGPU_VERTEX_FORMAT_UNORM8,
[GPU_TYPE_UN8x2] = WGPU_VERTEX_FORMAT_UNORM8X2,
[GPU_TYPE_UN8x4] = WGPU_VERTEX_FORMAT_UNORM8X4,
[GPU_TYPE_SN10x3] = WGPU_VERTEX_FORMAT_UNORM10_10_10_2, // TODO
[GPU_TYPE_UN10x3] = WGPU_VERTEX_FORMAT_UNORM10_10_10_2,
Expand All @@ -657,8 +665,10 @@ bool gpu_pipeline_init_graphics(gpu_pipeline* pipeline, gpu_pipeline_info* info,
[GPU_TYPE_U16] = WGPU_VERTEX_FORMAT_UINT16,
[GPU_TYPE_U16x2] = WGPU_VERTEX_FORMAT_UINT16X2,
[GPU_TYPE_U16x4] = WGPU_VERTEX_FORMAT_UINT16X4,
[GPU_TYPE_SN16] = WGPU_VERTEX_FORMAT_SNORM16,
[GPU_TYPE_SN16x2] = WGPU_VERTEX_FORMAT_SNORM16X2,
[GPU_TYPE_SN16x4] = WGPU_VERTEX_FORMAT_SNORM16X4,
[GPU_TYPE_UN16] = WGPU_VERTEX_FORMAT_UNORM16,
[GPU_TYPE_UN16x2] = WGPU_VERTEX_FORMAT_UNORM16X2,
[GPU_TYPE_UN16x4] = WGPU_VERTEX_FORMAT_UNORM16X4,
[GPU_TYPE_I32] = WGPU_VERTEX_FORMAT_SINT32,
Expand All @@ -669,6 +679,7 @@ bool gpu_pipeline_init_graphics(gpu_pipeline* pipeline, gpu_pipeline_info* info,
[GPU_TYPE_U32x2] = WGPU_VERTEX_FORMAT_UINT32X2,
[GPU_TYPE_U32x3] = WGPU_VERTEX_FORMAT_UINT32X3,
[GPU_TYPE_U32x4] = WGPU_VERTEX_FORMAT_UINT32X4,
[GPU_TYPE_F16] = WGPU_VERTEX_FORMAT_FLOAT16,
[GPU_TYPE_F16x2] = WGPU_VERTEX_FORMAT_FLOAT16X2,
[GPU_TYPE_F16x4] = WGPU_VERTEX_FORMAT_FLOAT16X4,
[GPU_TYPE_F32] = WGPU_VERTEX_FORMAT_FLOAT32,
Expand Down Expand Up @@ -1486,8 +1497,10 @@ bool gpu_init(gpu_config* config) {
config->features->subgroupQuad = false;
config->features->float32AtomicAdd = false;
config->features->float64 = false;
config->features->float16 = wgpu_device_supports_feature(state.device, WGPU_FEATURE_SHADER_F16);
config->features->int64 = false;
config->features->int16 = false;
config->features->int8 = false;

config->features->formats[GPU_FORMAT_R8][0] = GPU_FEATURE_SAMPLE | GPU_FEATURE_RENDER | GPU_FEATURE_BLIT;
config->features->formats[GPU_FORMAT_RG8][0] = GPU_FEATURE_SAMPLE | GPU_FEATURE_RENDER | GPU_FEATURE_BLIT;
Expand Down
40 changes: 37 additions & 3 deletions src/core/spv.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -733,7 +733,13 @@ static spv_result spv_parse_field(spv_context* spv, const uint32_t* word, spv_fi
}
}

if (OP_CODE(word) == 22 && word[2] == 32) { // OpTypeFloat
if (OP_CODE(word) == 22 && word[2] == 16) { // OpTypeFloat 16-bit
if (columnCount == 1 && componentCount <= 4 && componentCount != 3) {
field->type = SPV_F16 + (componentCount / 2);
} else {
return SPV_UNSUPPORTED_DATA_TYPE;
}
} else if (OP_CODE(word) == 22 && word[2] == 32) { // OpTypeFloat 32-bit
if (columnCount >= 2 && columnCount <= 4 && componentCount >= 2 && componentCount <= 4) {
field->type = SPV_MAT2x2 + (columnCount - 2) * 3 + (componentCount - 2);
} else if (columnCount == 1 && componentCount >= 2 && componentCount <= 4) {
Expand All @@ -743,7 +749,35 @@ static spv_result spv_parse_field(spv_context* spv, const uint32_t* word, spv_fi
} else {
return SPV_UNSUPPORTED_DATA_TYPE;
}
} else if (OP_CODE(word) == 21 && word[2] == 32) { // OpTypeInteger
} else if (OP_CODE(word) == 21 && word[2] == 8) { // OpTypeInteger 8-bit
if (word[3] > 0) { // signed
if (columnCount == 1 && componentCount <= 4 && componentCount != 3) {
field->type = SPV_I8 + (componentCount / 2);
} else {
return SPV_UNSUPPORTED_DATA_TYPE;
}
} else {
if (columnCount == 1 && componentCount <= 4 && componentCount != 3) {
field->type = SPV_U8 + (componentCount / 2);
} else {
return SPV_UNSUPPORTED_DATA_TYPE;
}
}
} else if (OP_CODE(word) == 21 && word[2] == 16) { // OpTypeInteger 16-bit
if (word[3] > 0) { // signed
if (columnCount == 1 && componentCount <= 4 && componentCount != 3) {
field->type = SPV_I16 + (componentCount / 2);
} else {
return SPV_UNSUPPORTED_DATA_TYPE;
}
} else {
if (columnCount == 1 && componentCount <= 4 && componentCount != 3) {
field->type = SPV_U16 + (componentCount / 2);
} else {
return SPV_UNSUPPORTED_DATA_TYPE;
}
}
} else if (OP_CODE(word) == 21 && word[2] == 32) { // OpTypeInteger 32-bit
if (word[3] > 0) { // signed
if (columnCount == 1 && componentCount >= 2 && componentCount <= 4) {
field->type = SPV_I32x2 + componentCount - 2;
Expand All @@ -767,7 +801,7 @@ static spv_result spv_parse_field(spv_context* spv, const uint32_t* word, spv_fi
return SPV_UNSUPPORTED_DATA_TYPE;
}

field->elementSize = 4 * componentCount * columnCount;
field->elementSize = (word[2] / 8) * componentCount * columnCount;

return SPV_OK;
}
Expand Down
Loading