mirror of
https://chromium.googlesource.com/libyuv/libyuv
synced 2025-12-23 12:04:53 +08:00
658cf9586f
Patch Set 2: Code-Review+1 (1 comment) Patch-set: 2 Label: Code-Review=+1, 3e9b4c2b386960f2482c29d6218822353181a7a8
74 lines
2.7 KiB
Plaintext
74 lines
2.7 KiB
Plaintext
{
|
|
"comments": [
|
|
{
|
|
"unresolved": false,
|
|
"key": {
|
|
"uuid": "416705d1_c82a21f6",
|
|
"filename": "/PATCHSET_LEVEL",
|
|
"patchSetId": 1
|
|
},
|
|
"lineNbr": 0,
|
|
"author": {
|
|
"id": 1571352
|
|
},
|
|
"writtenOn": "2024-12-11T15:40:06Z",
|
|
"side": 1,
|
|
"message": "(this needs all of the earlier commits in the chain to be committed first)",
|
|
"revId": "b119232c3dcd49bc54f76c732ec21d0ed60617bc",
|
|
"serverId": "3ce6091f-6c88-37e8-8c75-72f92ae8dfba"
|
|
},
|
|
{
|
|
"unresolved": false,
|
|
"key": {
|
|
"uuid": "b26ed1c6_e79a978d",
|
|
"filename": "/PATCHSET_LEVEL",
|
|
"patchSetId": 1
|
|
},
|
|
"lineNbr": 0,
|
|
"author": {
|
|
"id": 1115898
|
|
},
|
|
"writtenOn": "2024-12-12T09:33:50Z",
|
|
"side": 1,
|
|
"message": "theres pros and cons to this approach, but the end result is roughly 4x less code, so thats a win.\nIts a little hard to see how the macros piece together into a row function because they are far away from the wrapper. If each macro is basically only used once, it may be easier to understand it if its just above the function?",
|
|
"revId": "b119232c3dcd49bc54f76c732ec21d0ed60617bc",
|
|
"serverId": "3ce6091f-6c88-37e8-8c75-72f92ae8dfba"
|
|
},
|
|
{
|
|
"unresolved": false,
|
|
"key": {
|
|
"uuid": "428cf28e_d73d5563",
|
|
"filename": "/PATCHSET_LEVEL",
|
|
"patchSetId": 1
|
|
},
|
|
"lineNbr": 0,
|
|
"author": {
|
|
"id": 1571352
|
|
},
|
|
"writtenOn": "2024-12-12T11:00:04Z",
|
|
"side": 1,
|
|
"message": "I\u0027ve moved a couple of macros above their usage, however most of the READ and conversion macros are used in multiple places so I think it makes more sense to keep them together since the register passing conventions between macros is pretty similar.",
|
|
"parentUuid": "b26ed1c6_e79a978d",
|
|
"revId": "b119232c3dcd49bc54f76c732ec21d0ed60617bc",
|
|
"serverId": "3ce6091f-6c88-37e8-8c75-72f92ae8dfba"
|
|
},
|
|
{
|
|
"unresolved": false,
|
|
"key": {
|
|
"uuid": "960e43ec_4c517a9a",
|
|
"filename": "/PATCHSET_LEVEL",
|
|
"patchSetId": 1
|
|
},
|
|
"lineNbr": 0,
|
|
"author": {
|
|
"id": 1115898
|
|
},
|
|
"writtenOn": "2024-12-12T11:07:48Z",
|
|
"side": 1,
|
|
"message": "Sounds good.\nFeels like this could be done cleaner but if we figure something out it can be revisited.\nFor actual sve we may want to do a full port, or back port, and then come up with some calling code logic for when sve is faster than neon... e.g. if vl \u003e 128",
|
|
"parentUuid": "428cf28e_d73d5563",
|
|
"revId": "b119232c3dcd49bc54f76c732ec21d0ed60617bc",
|
|
"serverId": "3ce6091f-6c88-37e8-8c75-72f92ae8dfba"
|
|
}
|
|
]
|
|
} |