Leverage `#cir.zero` in large constant strings initializers #248

sitio-couto · 2023-08-30T17:12:42Z

The initializer for the null strings should leverage #cir.zero in some way. The reason for this is to improve performance when facing large zero/prefix-initialized strings. For example:

struct {
  char r[100000];
  char g[100000];
  char b[100000];
} image = {"", "1234", "\0"};

The resulting CIR code for this example consists of three #cir.const_arrays with thousands \00 hexadecimal chars each, making them unnecessarily large.

Originally posted by @sitio-couto in #244 (comment)

The text was updated successfully, but these errors were encountered:

This PR addresses #248 . Currently string literals are always lowered to a `cir.const_array` attribute even if the string literal only contains null bytes. This patch make the CodeGen emits `cir.zero` for these string literals.

This PR addresses llvm#248 . Currently string literals are always lowered to a `cir.const_array` attribute even if the string literal only contains null bytes. This patch make the CodeGen emits `cir.zero` for these string literals.

bcardosolopes · 2024-04-23T03:40:48Z

We have trailing_zeros these days, could be use here.

This PR addresses #248 . Currently string literals are always lowered to a `cir.const_array` attribute even if the string literal only contains null bytes. This patch make the CodeGen emits `cir.zero` for these string literals.

This PR addresses llvm#248 . Currently string literals are always lowered to a `cir.const_array` attribute even if the string literal only contains null bytes. This patch make the CodeGen emits `cir.zero` for these string literals.

This PR addresses #248 . Currently string literals are always lowered to a `cir.const_array` attribute even if the string literal only contains null bytes. This patch make the CodeGen emits `cir.zero` for these string literals.

ivanmurashko · 2024-05-13T19:05:08Z

I am going to take this task

The patch resolves [issue llvm#248](llvm#248). It can be considered a subsequent patch to [llvm#373](llvm#373), where the case of empty strings was processed. The new patch adds processing for non-empty strings that may contain trailing zeros, such as: ``` char big_string[100000] = "123"; ``` That is converted to ``` @big_string = #cir.const_array<"123" : !cir.array<!s8i x 3>, trailing_zeros> : !cir.array<!s8i x 100000> ``` It's worth noting that ordinary strings that are terminated with a single zero also get the trailing zeros, for instance ``` char string[] = "whatnow"; ``` Is converted to ``` @string = #cir.const_array<"whatnow" : !cir.array<!s8i x 7>, trailing_zeros> : !cir.array<!s8i x 8> ```

) The patch resolves [issue #248](#248). It can be considered a subsequent patch to [#373](#373), where the case of empty strings was processed. The new patch adds processing for non-empty strings that may contain trailing zeros, such as: ``` char big_string[100000] = "123"; ``` That is converted to ``` @big_string = #cir.const_array<"123" : !cir.array<!s8i x 3>, trailing_zeros> : !cir.array<!s8i x 100000> ```

…lvm#617) The patch resolves [issue llvm#248](llvm/clangir#248). It can be considered a subsequent patch to [llvm#373](llvm/clangir#373), where the case of empty strings was processed. The new patch adds processing for non-empty strings that may contain trailing zeros, such as: ``` char big_string[100000] = "123"; ``` That is converted to ``` @big_string = #cir.const_array<"123" : !cir.array<!s8i x 3>, trailing_zeros> : !cir.array<!s8i x 100000> ```

This PR addresses llvm#248 . Currently string literals are always lowered to a `cir.const_array` attribute even if the string literal only contains null bytes. This patch make the CodeGen emits `cir.zero` for these string literals.

…lvm#617) The patch resolves [issue llvm#248](llvm#248). It can be considered a subsequent patch to [llvm#373](llvm#373), where the case of empty strings was processed. The new patch adds processing for non-empty strings that may contain trailing zeros, such as: ``` char big_string[100000] = "123"; ``` That is converted to ``` @big_string = #cir.const_array<"123" : !cir.array<!s8i x 3>, trailing_zeros> : !cir.array<!s8i x 100000> ```

This PR addresses llvm#248 . Currently string literals are always lowered to a `cir.const_array` attribute even if the string literal only contains null bytes. This patch make the CodeGen emits `cir.zero` for these string literals.

…lvm#617) The patch resolves [issue llvm#248](llvm#248). It can be considered a subsequent patch to [llvm#373](llvm#373), where the case of empty strings was processed. The new patch adds processing for non-empty strings that may contain trailing zeros, such as: ``` char big_string[100000] = "123"; ``` That is converted to ``` @big_string = #cir.const_array<"123" : !cir.array<!s8i x 3>, trailing_zeros> : !cir.array<!s8i x 100000> ```

bcardosolopes added the good first issue Good for newcomers label Oct 23, 2023

Lancern mentioned this issue Dec 30, 2023

[CIR][CodeGen] emit cir.zero for constant string literals #373

Merged

bcardosolopes assigned ivanmurashko May 13, 2024

ivanmurashko mentioned this issue May 19, 2024

[CIR][CodeGen] Support trailing_zeros for constant string literals #617

Merged

bcardosolopes closed this as completed in #617 May 24, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Leverage `#cir.zero` in large constant strings initializers #248

Leverage `#cir.zero` in large constant strings initializers #248

sitio-couto commented Aug 30, 2023 •

edited

Loading

bcardosolopes commented Apr 23, 2024

ivanmurashko commented May 13, 2024 •

edited

Loading

Leverage #cir.zero in large constant strings initializers #248

Leverage #cir.zero in large constant strings initializers #248

Comments

sitio-couto commented Aug 30, 2023 • edited Loading

bcardosolopes commented Apr 23, 2024

ivanmurashko commented May 13, 2024 • edited Loading

Leverage `#cir.zero` in large constant strings initializers #248

Leverage `#cir.zero` in large constant strings initializers #248

sitio-couto commented Aug 30, 2023 •

edited

Loading

ivanmurashko commented May 13, 2024 •

edited

Loading