skygpu
/
skynet
mirror of https://github.com/skygpu/skynet.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
281 Commits
5 Branches
10 Tags
4.5 MiB
Commit Graph

281 Commits (63c849a41e4dde8e9d4690e3c348c60f29827318)

Author SHA1 Message Date
Guillermo Rodriguez 63c849a41e
Fixes to tui and compatibility with frontend formated requests found while running worker 2025-02-18 16:07:57 -03:00
Guillermo Rodriguez f60e582ad5
Enable and fix all tests, make upscaler just use pipeline_for function, add upscaler mocker 2025-02-18 16:07:57 -03:00
Guillermo Rodriguez f4804b61d9
Increase ipfs startup timeout & update lock (new dockerctl v) 2025-02-18 16:07:57 -03:00
Guillermo Rodriguez 9eb46862ae
Add CI, finish structification, break smart contract table sync logic to its own class with context manager, add better test helpers 2025-02-18 16:07:57 -03:00
Guillermo Rodriguez 08b6b983a2
First pass adding msgspec structs 
Added nix-shell dev file
Switch to dev branch of py-leap for get_table row auto struct unpack
Add testing module with helpers like override global load config function
Add types for Queue V0
Rename test contract to gpu.scd to match current
Add new enqueue test
Update lock
 2025-02-18 16:07:57 -03:00
Guillermo Rodriguez 868e2489b3
Merge pull request #47 from skygpu/fomo_polish 
Fomo polish
 2025-02-18 15:57:28 -03:00
Guillermo Rodriguez a4e40ba662
Added auto-download through hf for the upscaler 2025-02-18 15:56:47 -03:00
Guillermo Rodriguez 8828fa13fc
Add table_index system in poller in order for daemon to be aware of stale data 2025-02-18 15:56:47 -03:00
Guillermo Rodriguez 30eaa6c194
Switch to non interator poller, NetConnector has wait_data_update() now 2025-02-18 15:56:47 -03:00
Guillermo Rodriguez 5d67b3cd60
Address last of fomo\'s comments, related to pinner logging, cleanup ipfs module 2025-02-18 15:56:46 -03:00
Guillermo Rodriguez 5edd7f4db4
Log information about non custom pipelines 2025-02-18 15:56:46 -03:00
Guillermo Rodriguez b365614194
Add real compute test and parametrize the mocker test 2025-02-18 15:56:46 -03:00
Guillermo Rodriguez 80633627da
Remove un-used util 2025-02-18 15:56:46 -03:00
Guillermo Rodriguez a652fdd781
Move skynet.utils to skynet.dgpu submodule 2025-02-18 15:56:46 -03:00
Guillermo Rodriguez cebbd1059e
Fix Upscaler hardcoding, add mockerpipeline and streamline dgpu tests 2025-02-18 15:56:46 -03:00
Guillermo Rodriguez d0e084b867
Update example config 2025-02-18 15:56:46 -03:00
Guillermo Rodriguez f39859943d
Update py-leap and use its new apis on testing, add a comment on fields of DgpuConfig struct, minor name changing and mechanics of open_worker, make logging configurable 2025-02-18 15:56:46 -03:00
Guillermo Rodriguez 149d9f9f33
Factor out WorkerDaemon, split into functions, made poller into an async gen and moved it to NetConnector as well as should_cancel 2025-02-18 15:56:46 -03:00
Guillermo Rodriguez ea3b35904c
Create msgspec struct for config 2025-02-18 15:56:46 -03:00
Guillermo Rodriguez 5a3a43b3c6
Refactoring tui to be functional style 2025-02-18 15:56:46 -03:00
Guillermo Rodriguez 12b32a7188
Refactor ModelMngr to be a context manager + function combo 2025-02-18 15:56:46 -03:00
Guillermo Rodriguez 8b45fb5979
Begin adding TUI 2025-02-18 15:56:45 -03:00
Guillermo Rodriguez e66f8d74fd
Fix asks -> httpx bug on get_ipfs_link response handling 2025-02-18 15:56:45 -03:00
Guillermo Rodriguez 4c9be4e63e
Fix cancellation system and provide a reason for the raise 2025-02-18 15:56:45 -03:00
Guillermo Rodriguez a5dbe5ab12
Further improvements in indentation and logic in daemons maybe_serve_one, also might of fixed a bug related to using id instead of request_id in the search existing results phase, and add way more logging 2025-02-18 15:56:45 -03:00
Guillermo Rodriguez cc7015eb03
Rework if statement to reduce indentation, add comment about logic 2025-02-18 15:56:45 -03:00
Guillermo Rodriguez e09652eaae
Drop old ipfs node docker managment 2025-02-18 15:56:45 -03:00
Guillermo Rodriguez eaad7d9112
Remove todo about hardcoded numbers in test setup 2025-02-18 15:56:45 -03:00
Guillermo Rodriguez 62f891c017
Rename my_results 2025-02-18 15:56:45 -03:00
Guillermo Rodriguez 1b437b761b
Change dgpu submodules classes name per fomos suggestion 2025-02-18 15:56:45 -03:00
Guillermo Rodriguez f0604f54fb
Fix some import related TODOs 2025-02-18 15:56:45 -03:00
Guillermo Rodriguez 5f5314cd35
Remove all shebangs 2025-02-18 15:56:45 -03:00
Guillermo Rodriguez f2a8f0367f
Switch failable to use outcome 2025-02-18 15:56:43 -03:00
Tyler Goodlet 336c0122eb
Tell vim about `uv.lock` toml-ness 2025-02-18 15:56:17 -03:00
Tyler Goodlet 18cdffa700
Doc `GPUConnector.get_full_queue_snapshot()` 2025-02-18 15:56:17 -03:00
Tyler Goodlet 1b528e1fa4
Fix non-f-string 2025-02-18 15:56:17 -03:00
Tyler Goodlet 286d49a7f5
Drop unused `as err`, suggest `logging.exception()` 2025-02-18 15:56:17 -03:00
Tyler Goodlet 2eb398bb8d
Bit more multi-line styling in `.cli` 
Such that i could more easily read the flag specs in a vsplit `vim`
buffer set ;)

Also includes the same for some (internal) `tuple`-imports.
 2025-02-18 15:56:17 -03:00
Tyler Goodlet bb4be24fac
Some more `import` fixes 
Removing unused imports in a few modules as well as converting a few
more tuple imports to multi-line style.

We should prolly consider a linter as part of pre-merge machinery;
I noticed a few pretty low hanging pep8 violations just spelunking rando
modules ;)
 2025-02-18 15:56:17 -03:00
Tyler Goodlet e0704e3787
Suggest `skynet.dgpu` docs, typing, pythonisms 
From the deep-ish dive drafting our first set of design/architecture
diagrams in https://github.com/skygpu/cyberdyne/pull/2, this adds
a buncha suggestions, typing, and styling adjustments.

Namely the code tweaks include,
- changing to multi-line import tuples where appropriate (since they're
  much handier to modify ;)
- adding typing in many spots where it wasn't clear to me the types
  being returned/operated-with in various (internal) methods.
- doc strings (in mostly random spots Xp ) where i had the need to
  remember the impl's purpose but didn't want to re-read the code in
  detail again.
- ALOT of TODOs surrounding various potential style changes,
  re-factorings, naming and in some cases "modernization" according to
  the latest python3.12 feats/spec/stdlib.
 2025-02-18 15:56:17 -03:00
Tyler Goodlet 01e6bfe252
Doc and type `skynet.dgpu` pkg mod 2025-02-18 15:56:17 -03:00
Guillermo Rodriguez 802af99c39
Add one more error to failable 2025-02-03 13:28:46 -03:00
Guillermo Rodriguez 417851c13e
Merge pull request #46 from skygpu/network-error-patch 
add httpx.ConnectTimeout and httpx.ReadTimeout to netowork failables
 2025-01-31 12:58:43 -03:00
Konstantine Tsafatinos bce2f1c8e4 add httpx.ConnectTimeout and httpx.ReadTimeout to netowork failables 2025-01-31 08:25:02 -05:00
Guillermo Rodriguez 2e50b1a542
Merge pull request #45 from skygpu/uv_migration 
Switch to uv package manager
 2025-01-22 14:42:59 -03:00
Guillermo Rodriguez 8f4429bd06
Switch to uv package manager 2025-01-22 01:18:21 -03:00
Guillermo Rodriguez b5f52b3b5b
Merge pull request #44 from skygpu/inpainting 
Add Inpainting to telos_testnet branch
 2025-01-20 20:54:17 -03:00
Guillermo Rodriguez 12d015f0b9
Fix resp none error and params on flux inpaint pipe 2025-01-20 00:50:39 -03:00
Guillermo Rodriguez 48a0d0db79
Add new leap error 2025-01-18 23:33:09 -03:00
Guillermo Rodriguez 67cff51085
Consolidate GPU abi location in constants and drop asks 2025-01-18 15:31:14 -03:00