[REFLECTION BREAKING] Modify the internals of the map type to increase performance

This commit is contained in:
gingerBill
2020-09-07 11:41:42 +01:00
parent 7e08bccc9a
commit 7f48cf8405
7 changed files with 106 additions and 95 deletions

View File

@@ -210,11 +210,7 @@ marshal_arg :: proc(b: ^strings.Builder, v: any) -> Marshal_Error {
data := uintptr(entries.data) + uintptr(i*entry_size);
header := cast(^Map_Entry_Header)data;
if reflect.is_string(info.key) {
marshal_arg(b, header.key.str);
} else {
marshal_arg(b, any{rawptr(&header.key.hash), info.key.id});
}
marshal_arg(b, any{rawptr(&header.key.key.val), info.key.id});
write_string(b, ": ");

View File

@@ -1501,10 +1501,10 @@ fmt_value :: proc(fi: ^Info, v: any, verb: rune) {
header := cast(^runtime.Map_Entry_Header)data;
if reflect.is_string(info.key) {
strings.write_string(fi.buf, header.key.str);
strings.write_string(fi.buf, header.key.key.str);
} else {
fi := Info{buf = fi.buf};
fmt_arg(&fi, any{rawptr(&header.key.hash), info.key.id}, 'v');
fmt_arg(&fi, any{rawptr(&header.key.key.val), info.key.id}, 'v');
}
strings.write_string(fi.buf, "=");

View File

@@ -349,7 +349,17 @@ INITIAL_MAP_CAP :: 16;
Map_Key :: struct {
hash: u64,
str: string,
/* NOTE(bill)
size_of(Map_Key) == 16 Bytes on 32-bit systems
size_of(Map_Key) == 24 Bytes on 64-bit systems
This does mean that an extra word is wasted for each map when a string is not used on 64-bit systems
however, this is probably not a huge problem in terms of memory usage
*/
key: struct #raw_union {
str: string,
val: u64,
},
}
Map_Find_Result :: struct {
@@ -1407,28 +1417,34 @@ __get_map_key :: proc "contextless" (k: $K) -> Map_Key {
T :: intrinsics.type_core_type(K);
when intrinsics.type_is_integer(T) {
map_key.hash = default_hash_ptr(&key, size_of(T));
sz :: 8*size_of(T);
when sz == 8 do map_key.hash = u64(( ^u8)(&key)^);
else when sz == 16 do map_key.hash = u64((^u16)(&key)^);
else when sz == 32 do map_key.hash = u64((^u32)(&key)^);
else when sz == 64 do map_key.hash = u64((^u64)(&key)^);
else do #assert(false, "Unhandled integer size");
when sz == 8 do map_key.key.val = u64(( ^u8)(&key)^);
else when sz == 16 do map_key.key.val = u64((^u16)(&key)^);
else when sz == 32 do map_key.key.val = u64((^u32)(&key)^);
else when sz == 64 do map_key.key.val = u64((^u64)(&key)^);
else do #panic("Unhandled integer size");
} else when intrinsics.type_is_rune(T) {
map_key.hash = u64((^rune)(&key)^);
map_key.hash = default_hash_ptr(&key, size_of(T));
map_key.key.val = u64((^rune)(&key)^);
} else when intrinsics.type_is_pointer(T) {
map_key.hash = u64(uintptr((^rawptr)(&key)^));
map_key.hash = default_hash_ptr(&key, size_of(T));
map_key.key.val = u64(uintptr((^rawptr)(&key)^));
} else when intrinsics.type_is_float(T) {
map_key.hash = default_hash_ptr(&key, size_of(T));
sz :: 8*size_of(T);
when sz == 32 do map_key.hash = u64((^u32)(&key)^);
else when sz == 64 do map_key.hash = u64((^u64)(&key)^);
else do #assert(false, "Unhandled float size");
when sz == 32 do map_key.key.val = u64((^u32)(&key)^);
else when sz == 64 do map_key.key.val = u64((^u64)(&key)^);
else do #panic("Unhandled float size");
} else when intrinsics.type_is_string(T) {
#assert(T == string);
str := (^string)(&key)^;
map_key.hash = default_hash_string(str);
map_key.str = str;
map_key.key.str = str;
} else {
#assert(false, "Unhandled map key type");
#panic("Unhandled map key type");
}
return map_key;
@@ -1443,10 +1459,18 @@ _fnv64a :: proc "contextless" (data: []byte, seed: u64 = 0xcbf29ce484222325) ->
}
default_hash :: proc "contextless" (data: []byte) -> u64 {
default_hash :: inline proc "contextless" (data: []byte) -> u64 {
context = default_context();
os.write_string(os.stdout, "here - default_hash\n");
return _fnv64a(data);
}
default_hash_string :: proc "contextless" (s: string) -> u64 do return default_hash(transmute([]byte)(s));
default_hash_string :: inline proc "contextless" (s: string) -> u64 {
return default_hash(transmute([]byte)(s));
}
default_hash_ptr :: inline proc "contextless" (data: rawptr, size: int) -> u64 {
s := Raw_Slice{data, size};
return default_hash(transmute([]byte)(s));
}
source_code_location_hash :: proc(s: Source_Code_Location) -> u64 {
@@ -1582,7 +1606,11 @@ __dynamic_map_full :: inline proc(using h: Map_Header) -> bool {
__dynamic_map_hash_equal :: proc(h: Map_Header, a, b: Map_Key) -> bool {
if a.hash == b.hash {
if h.is_key_string do return a.str == b.str;
if h.is_key_string {
return a.key.str == b.key.str;
} else {
return a.key.val == b.key.val;
}
return true;
}
return false;

View File

@@ -17,6 +17,8 @@ HANDLE :: distinct LPVOID;
HINSTANCE :: HANDLE;
HMODULE :: distinct HINSTANCE;
HRESULT :: distinct LONG;
HWND :: distinct HANDLE;
HMONITOR :: distinct HANDLE;
BOOL :: distinct b32;
BYTE :: distinct u8;
BOOLEAN :: distinct b8;

View File

@@ -2794,6 +2794,8 @@ void check_map_type(CheckerContext *ctx, Type *type, Ast *node) {
if (is_type_string(key)) {
add_package_dependency(ctx, "runtime", "default_hash_string");
} else {
add_package_dependency(ctx, "runtime", "default_hash_ptr");
}

View File

@@ -3593,27 +3593,8 @@ irValue *ir_gen_map_key(irProcedure *proc, irValue *key, Type *key_type) {
irValue *v = ir_add_local_generated(proc, t_map_key, true);
Type *t = base_type(ir_type(key));
key = ir_emit_conv(proc, key, key_type);
if (is_type_integer(t)) {
ir_emit_store(proc, ir_emit_struct_ep(proc, v, 0), ir_emit_conv(proc, key, hash_type));
} else if (is_type_enum(t)) {
ir_emit_store(proc, ir_emit_struct_ep(proc, v, 0), ir_emit_conv(proc, key, hash_type));
} else if (is_type_typeid(t)) {
irValue *i = ir_emit_bitcast(proc, key, t_uint);
ir_emit_store(proc, ir_emit_struct_ep(proc, v, 0), ir_emit_conv(proc, i, hash_type));
} else if (is_type_pointer(t)) {
irValue *p = ir_emit_conv(proc, key, t_uintptr);
ir_emit_store(proc, ir_emit_struct_ep(proc, v, 0), ir_emit_conv(proc, p, hash_type));
} else if (is_type_float(t)) {
irValue *bits = nullptr;
i64 size = type_size_of(t);
switch (8*size) {
case 32: bits = ir_emit_transmute(proc, key, t_u32); break;
case 64: bits = ir_emit_transmute(proc, key, t_u64); break;
default: GB_PANIC("Unhandled float size: %lld bits", size); break;
}
ir_emit_store(proc, ir_emit_struct_ep(proc, v, 0), ir_emit_conv(proc, bits, hash_type));
} else if (is_type_string(t)) {
if (is_type_string(t)) {
irValue *str = ir_emit_conv(proc, key, t_string);
irValue *hashed_str = nullptr;
@@ -3628,9 +3609,27 @@ irValue *ir_gen_map_key(irProcedure *proc, irValue *key, Type *key_type) {
hashed_str = ir_emit_runtime_call(proc, "default_hash_string", args);
}
ir_emit_store(proc, ir_emit_struct_ep(proc, v, 0), hashed_str);
ir_emit_store(proc, ir_emit_struct_ep(proc, v, 1), str);
irValue *key_data = ir_emit_struct_ep(proc, v, 1);
key_data = ir_emit_conv(proc, key_data, alloc_type_pointer(key_type));
ir_emit_store(proc, key_data, str);
} else {
GB_PANIC("Unhandled map key type");
i64 sz = type_size_of(t);
GB_ASSERT(sz <= 8);
if (sz != 0) {
auto args = array_make<irValue *>(ir_allocator(), 2);
args[0] = ir_address_from_load_or_generate_local(proc, key);
args[1] = ir_const_int(sz);
irValue *hash = ir_emit_runtime_call(proc, "default_hash_ptr", args);
irValue *hash_ptr = ir_emit_struct_ep(proc, v, 0);
irValue *key_data = ir_emit_struct_ep(proc, v, 1);
key_data = ir_emit_conv(proc, key_data, alloc_type_pointer(key_type));
ir_emit_store(proc, hash_ptr, hash);
ir_emit_store(proc, key_data, key);
}
}
return ir_emit_load(proc, v);
@@ -9818,8 +9817,6 @@ void ir_build_range_indexed(irProcedure *proc, irValue *expr, Type *val_type, ir
break;
}
case Type_Map: {
irValue *key = ir_add_local_generated(proc, expr_type->Map.key, true);
irValue *entries = ir_map_entries_ptr(proc, expr);
irValue *elem = ir_emit_struct_ep(proc, entries, 0);
elem = ir_emit_load(proc, elem);
@@ -9827,15 +9824,9 @@ void ir_build_range_indexed(irProcedure *proc, irValue *expr, Type *val_type, ir
irValue *entry = ir_emit_ptr_offset(proc, elem, idx);
val = ir_emit_load(proc, ir_emit_struct_ep(proc, entry, 2));
irValue *hash = ir_emit_struct_ep(proc, entry, 0);
if (is_type_string(expr_type->Map.key)) {
irValue *str = ir_emit_struct_ep(proc, hash, 1);
ir_emit_store(proc, key, ir_emit_load(proc, str));
} else {
irValue *hash_ptr = ir_emit_struct_ep(proc, hash, 0);
hash_ptr = ir_emit_conv(proc, hash_ptr, ir_type(key));
ir_emit_store(proc, key, ir_emit_load(proc, hash_ptr));
}
irValue *key_raw = ir_emit_struct_ep(proc, entry, 0);
key_raw = ir_emit_struct_ep(proc, key_raw, 1);
irValue *key = ir_emit_conv(proc, key_raw, alloc_type_pointer(expr_type->Map.key));
idx = ir_emit_load(proc, key);

View File

@@ -3077,8 +3077,6 @@ void lb_build_range_indexed(lbProcedure *p, lbValue expr, Type *val_type, lbValu
break;
}
case Type_Map: {
lbAddr key = lb_add_local_generated(p, expr_type->Map.key, true);
lbValue entries = lb_map_entries_ptr(p, expr);
lbValue elem = lb_emit_struct_ep(p, entries, 0);
elem = lb_emit_load(p, elem);
@@ -3086,17 +3084,11 @@ void lb_build_range_indexed(lbProcedure *p, lbValue expr, Type *val_type, lbValu
lbValue entry = lb_emit_ptr_offset(p, elem, idx);
val = lb_emit_load(p, lb_emit_struct_ep(p, entry, 2));
lbValue hash = lb_emit_struct_ep(p, entry, 0);
if (is_type_string(expr_type->Map.key)) {
lbValue str = lb_emit_struct_ep(p, hash, 1);
lb_addr_store(p, key, lb_emit_load(p, str));
} else {
lbValue hash_ptr = lb_emit_struct_ep(p, hash, 0);
hash_ptr = lb_emit_conv(p, hash_ptr, key.addr.type);
lb_addr_store(p, key, lb_emit_load(p, hash_ptr));
}
lbValue key_raw = lb_emit_struct_ep(p, entry, 0);
key_raw = lb_emit_struct_ep(p, key_raw, 1);
lbValue key = lb_emit_conv(p, key_raw, alloc_type_pointer(expr_type->Map.key));
idx = lb_addr_load(p, key);
idx = lb_emit_load(p, key);
break;
}
@@ -9655,45 +9647,45 @@ lbValue lb_gen_map_header(lbProcedure *p, lbValue map_val_ptr, Type *map_type) {
lbValue lb_gen_map_key(lbProcedure *p, lbValue key, Type *key_type) {
Type *hash_type = t_u64;
lbAddr v = lb_add_local_generated(p, t_map_key, true);
lbValue vp = lb_addr_get_ptr(p, v);
Type *t = base_type(key.type);
key = lb_emit_conv(p, key, key_type);
if (is_type_integer(t)) {
lb_emit_store(p, lb_emit_struct_ep(p, v.addr, 0), lb_emit_conv(p, key, hash_type));
} else if (is_type_enum(t)) {
lb_emit_store(p, lb_emit_struct_ep(p, v.addr, 0), lb_emit_conv(p, key, hash_type));
} else if (is_type_typeid(t)) {
lbValue i = lb_emit_transmute(p, key, t_uint);
lb_emit_store(p, lb_emit_struct_ep(p, v.addr, 0), lb_emit_conv(p, i, hash_type));
} else if (is_type_pointer(t)) {
lbValue ptr = lb_emit_conv(p, key, t_uintptr);
lb_emit_store(p, lb_emit_struct_ep(p, v.addr, 0), lb_emit_conv(p, ptr, hash_type));
} else if (is_type_float(t)) {
lbValue bits = {};
i64 size = type_size_of(t);
switch (8*size) {
case 32: bits = lb_emit_transmute(p, key, t_u32); break;
case 64: bits = lb_emit_transmute(p, key, t_u64); break;
default: GB_PANIC("Unhandled float size: %lld bits", size); break;
}
lb_emit_store(p, lb_emit_struct_ep(p, v.addr, 0), lb_emit_conv(p, bits, hash_type));
} else if (is_type_string(t)) {
if (is_type_string(t)) {
lbValue str = lb_emit_conv(p, key, t_string);
lbValue hashed_str = {};
if (false && lb_is_const(str)) {
String value = lb_get_const_string(p->module, str);
u64 hs = fnv64a(value.text, value.len);
hashed_str = lb_const_value(p->module, t_u64, exact_value_u64(hs));
if (lb_is_const(str)) {
String v = lb_get_const_string(p->module, str);
u64 hs = fnv64a(v.text, v.len);
hashed_str = lb_const_int(p->module, t_u64, hs);
} else {
auto args = array_make<lbValue>(heap_allocator(), 1);
args[0] = str;
hashed_str = lb_emit_runtime_call(p, "default_hash_string", args);
}
lb_emit_store(p, lb_emit_struct_ep(p, v.addr, 0), hashed_str);
lb_emit_store(p, lb_emit_struct_ep(p, v.addr, 1), str);
lb_emit_store(p, lb_emit_struct_ep(p, vp, 0), hashed_str);
lbValue key_data = lb_emit_struct_ep(p, vp, 1);
key_data = lb_emit_conv(p, key_data, alloc_type_pointer(key_type));
lb_emit_store(p, key_data, str);
} else {
GB_PANIC("Unhandled map key type");
i64 sz = type_size_of(t);
GB_ASSERT(sz <= 8);
if (sz != 0) {
auto args = array_make<lbValue>(heap_allocator(), 2);
args[0] = lb_address_from_load_or_generate_local(p, key);
args[1] = lb_const_int(p->module, t_int, sz);
lbValue hash = lb_emit_runtime_call(p, "default_hash_ptr", args);
lbValue hash_ptr = lb_emit_struct_ep(p, vp, 0);
lbValue key_data = lb_emit_struct_ep(p, vp, 1);
key_data = lb_emit_conv(p, key_data, alloc_type_pointer(key_type));
lb_emit_store(p, hash_ptr, hash);
lb_emit_store(p, key_data, key);
}
}
return lb_addr_load(p, v);