Updated C wrapper wrt. Torch v1.10 by stemann · Pull Request #61 · FluxML/Torch.jl

stemann · 2023-08-11T14:59:32Z

Updates the C wrapper based on ocaml-torch @ 0.14 - matching Torch v1.10 (current JLL-build)

Contributes to #54 - follow-up for #56

Notable included changes:

Updated C wrapper generator based on ocaml-torch @ 0.14 - matching Torch v1.10.
Updated C wrapper (generated part)
Updated C wrapper (manual part), torch_api.{cpp, h}
- Re-implemented change of return type of functions to status code etc.
Renamed C wrapper from doeye_caml to torch_c_api
Restored Julia-specific additions
Conditional CUDA build - C wrapper can be built without CUDA.
Added ~~buildkite steps~~GitHub Actions workflow for building C wrapper
Dev. container: Added support for CUDA and CUDNN

~~The last two changes could be moved to a separate PR (to reduce number of changes in this PR).~~

To-do:

Clean-up the automatic part (remove comments - fix indentation)
Adapt the manual part, torch_api.{cpp, h}
The commit Changed torch_api.cpp to reduce diff should be removed before merging: It is meant to reduce the diff when reviewing - it's only a bunch of indentation changes etc. to make the diff smaller.

zsz00 · 2023-10-21T10:14:01Z

What's the status now?

stemann · 2023-10-22T11:48:27Z

It's been a idle for a while, but status is summarised by these comments:

stemann · 2023-11-11T09:44:33Z

See #54 for status.

Copied from https://github.com/LaurentMazare/ocaml-torch/tree/a6499811f40282a071179d4306afbbb6023dcc4a/src/gen/gen.ml Also updated dune-project accordingly.

…a torch wrapper

A. Non-void, non-* methods. Search/replace: 1. torch_api.h: ^(?!void)(\w+) (at.+)$ -> int $2($1 *, 2. torch_api.h: , $;$ -> ); 3. torch_api.cpp: ^(?!void)(\w+) (at.+)$ -> int $2($1 *out__, 4. torch_api.cpp: , $ \{$ -> ) { B. void-methods. Search/replace 1. torch_api.{h,cpp}: ^void (at.+)$ -> int $1( C. *-methods. Search/replace 1. torch.api.h: ^(\w+ \*)(at.+)\( -> int $2($1*, 2. torch_api.cpp: ^(\w+ \*)(at.+)\( -> int $2($1*out__, D. Implemented return status code Replaced ``` ^(\s*) return new (.+) \s*$ \s*return nullptr; ``` with ``` $1 out__[0] = new $2 $1 return 0; $1) $1return 1; ``` E. Implemented return status code 1. Replaced ``` ^(\s*)PROTECT$return new (.+)$ \s*return nullptr; ``` with ``` $1PROTECT( $1 out__[0] = new $2 $1 return 0; $1) $1return 1; ``` F. Implemented return status code 1. Replaced ``` ^(\s*)PROTECT$return (.+)$ ``` with ``` $1PROTECT( $1 out__[0] = $2 $1 return 0; $1) ``` G. Implemented return status code 1. Replaced ``` ^(\s*) return (.+) \s*\) \s*return nullptr; ``` with ``` $1 out__[0] = $2 $1 return 0; $1) $1return 1; ``` H. Restored error handling Handled caml_failwith by search/replace: Replaced: ``` $ ^(\s*) caml_failwith$(.+) ``` with: ``` { $1 myerr = strdup($2 $1 return 1; $1} ``` I. Replaced ``` ^(\s+)PROTECT\( return (.+) $ ``` with ``` $1PROTECT( $1 out__[0] = $2 $1 return 0; $1) ``` J. Manual implement return status code K. Changed return code from -1 to 1 to reduce diff L. Fixed a couple of warnings

Also, made CUDA build optional. int at_empty_cache(); int at_no_grad(int flag); int at_sync(); int at_from_blob(tensor *, void *data, int64_t *dims, int ndims, int64_t *strides, int nstrides, int dev);

Also: * Dev. container: Updated for Torch 1.10.2 * Added /build to .gitignore

Factored out scripts - for re-use in CI and dev. container.

DhairyaLGandhi · 2024-04-29T12:28:28Z

Perhaps it would make sense to start merging some of the excellent changes here?

stemann · 2024-04-29T12:39:22Z

Yes! :-)

Please give torch_api.{cpp,h} a thorough review - I have made some changes in an effort to make things a bit more consistent - e.g. wrt. always returning an int status code.

Edit: I'll try to go over it as well and try to make a re-cap of the changes.

stemann · 2024-04-29T12:47:35Z

This is the current main diff which covers the hand-written part (torch_api.{cpp,h}): https://github.com/FluxML/Torch.jl/compare/7828132d..ece96546

stemann · 2024-04-29T13:05:43Z

The overall aim was to to update for Torch v1.10.2 - but also to make it easier to apply a diff of changes for subsequent version updates...

stemann · 2024-04-29T13:22:46Z

Recap:

General

Removed argument names from function definitions in torch_api.h.
Definitions in torch_api.h were updated to match the implementation in torch_api.cpp - e.g. wrt. return value in the first argument, and return code as return value.

Modified function definitions

int at_float_vec(double *values, int value_len, int type);

int at_int_vec(int64_t *values, int value_len, int type);

int at_grad_set_enabled(int);

int at_int64_value_at_indexes(double *i, tensor, int *indexes, int indexes_len);

tensor at_load(char *filename);

int ato_adam(optimizer *, double learning_rate,
                   double beta1,
                   double beta2,
                   double weight_decay);

int atm_load(char *, module *);

Added function definitions

int at_is_sparse(int *, tensor)
int at_device(int *, tensor)

int at_stride(tensor, int *)

int at_autocast_clear_cache();
int at_autocast_decrement_nesting(int *);
int at_autocast_increment_nesting(int *);
int at_autocast_is_enabled(int *);
int at_autocast_set_enabled(int *, int b);

int at_to_string(char **, tensor, int line_size)

int at_get_num_threads(int *);
int at_set_num_threads(int n_threads);

int ati_none(ivalue *);
int ati_bool(ivalue *, int);
int ati_string(ivalue *, char *);
int ati_tuple(ivalue *, ivalue *, int);
int ati_generic_list(ivalue *, ivalue *, int);
int ati_generic_dict(ivalue *, ivalue *, int);
int ati_int_list(ivalue *, int64_t *, int);
int ati_double_list(ivalue *, double *, int);
int ati_bool_list(ivalue *, char *, int);
int ati_string_list(ivalue *, char **, int);
int ati_tensor_list(ivalue *, tensor *, int);

int ati_to_string(char **, ivalue);
int ati_to_bool(int *, ivalue);
int ati_length(int *, ivalue);
int ati_to_generic_list(ivalue, ivalue *, int);
int ati_to_generic_dict(ivalue, ivalue *, int);
int ati_to_int_list(ivalue, int64_t *, int);
int ati_to_double_list(ivalue, double *, int);
int ati_to_bool_list(ivalue, char *, int);
int ati_to_tensor_list(ivalue, tensor *, int);

stemann · 2024-07-09T11:24:57Z

@DhairyaLGandhi Do you know if anyone is available for reviewing these changes?

I'm at JuliaCon, FYI.

stemann force-pushed the stemann/ocaml_torch_1.10 branch from acb50ff to 6d3200a Compare August 11, 2023 15:05

stemann mentioned this pull request Nov 10, 2023

Fixed compat to enable tests on CI #60

Merged

stemann force-pushed the stemann/ocaml_torch_1.10 branch from 6d3200a to 6afc28e Compare November 11, 2023 09:36

stemann mentioned this pull request Nov 11, 2023

Update wrapper #54

Closed

stemann force-pushed the stemann/ocaml_torch_1.10 branch from 6afc28e to f407437 Compare November 12, 2023 11:44

stemann added 3 commits November 12, 2023 12:45

Added wrapper generator from ocaml-torch 0.14

28550c8

Copied from https://github.com/LaurentMazare/ocaml-torch/tree/a6499811f40282a071179d4306afbbb6023dcc4a/src/gen/gen.ml Also updated dune-project accordingly.

Adapted wrapper_generator to produce equivalent code to existing juli…

57ded88

…a torch wrapper

Removed writing of OCaml stubs and wrapper

f374bf6

stemann force-pushed the stemann/ocaml_torch_1.10 branch 19 times, most recently from 6f18889 to beb24f2 Compare November 17, 2023 10:38

stemann force-pushed the stemann/ocaml_torch_1.10 branch 3 times, most recently from 41f53f2 to 1051fc8 Compare April 29, 2024 07:52

stemann added 3 commits April 29, 2024 10:25

deps/c_wrapper: Renamed doeye_caml to torch_c_api

f82525a

Updated generated wrapper for Torch v1.10

075ecf7

stemann force-pushed the stemann/ocaml_torch_1.10 branch 2 times, most recently from 090f09e to aa302ad Compare April 29, 2024 10:28

stemann added 2 commits April 29, 2024 12:29

Restored Julia-specific additions

ece9654

Also, made CUDA build optional. int at_empty_cache(); int at_no_grad(int flag); int at_sync(); int at_from_blob(tensor *, void *data, int64_t *dims, int ndims, int64_t *strides, int nstrides, int dev);

Dev. container: Added support for CUDA and CUDNN

c8f1e9c

Also: * Dev. container: Updated for Torch 1.10.2 * Added /build to .gitignore

stemann force-pushed the stemann/ocaml_torch_1.10 branch 9 times, most recently from 0ef32d1 to 6483d63 Compare April 29, 2024 11:46

stemann added 2 commits April 29, 2024 14:18

Added GitHub Actions workflow for building C Wrapper

f799133

Factored out scripts - for re-use in CI and dev. container.

Fixed C wrapper wrt. formatting

37c6056

stemann mentioned this pull request May 7, 2024

CUDA: Introduce arm64 subarchitecture. JuliaPackaging/Yggdrasil#8622

Merged

ToucheSir approved these changes Aug 8, 2024

View reviewed changes

stemann mentioned this pull request Nov 15, 2024

Updated for Torch v1.10.2, and NNlib >= v0.8 #64

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Updated C wrapper wrt. Torch v1.10#61

Updated C wrapper wrt. Torch v1.10#61
ToucheSir merged 12 commits intoFluxML:masterfrom
stemann:stemann/ocaml_torch_1.10

stemann commented Aug 11, 2023 •

edited

Loading

Uh oh!

zsz00 commented Oct 21, 2023

Uh oh!

stemann commented Oct 22, 2023

Uh oh!

stemann commented Nov 11, 2023

Uh oh!

DhairyaLGandhi commented Apr 29, 2024

Uh oh!

stemann commented Apr 29, 2024 •

edited

Loading

Uh oh!

stemann commented Apr 29, 2024

Uh oh!

stemann commented Apr 29, 2024

Uh oh!

stemann commented Apr 29, 2024 •

edited

Loading

Uh oh!

stemann commented Jul 9, 2024 •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Uh oh!

Conversation

stemann commented Aug 11, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

zsz00 commented Oct 21, 2023

Uh oh!

stemann commented Oct 22, 2023

Uh oh!

stemann commented Nov 11, 2023

Uh oh!

DhairyaLGandhi commented Apr 29, 2024

Uh oh!

stemann commented Apr 29, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

stemann commented Apr 29, 2024

Uh oh!

stemann commented Apr 29, 2024

Uh oh!

stemann commented Apr 29, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

General

Modified function definitions

Added function definitions

Uh oh!

stemann commented Jul 9, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

stemann commented Aug 11, 2023 •

edited

Loading

stemann commented Apr 29, 2024 •

edited

Loading

stemann commented Apr 29, 2024 •

edited

Loading

stemann commented Jul 9, 2024 •

edited

Loading