mirror of
https://chromium.googlesource.com/libyuv/libyuv
synced 2025-12-14 23:49:57 +08:00
a0632472af
Patch Set 2:
(2 comments)
Patch-set: 2
Attention: {"person_ident":"Gerrit User 1115898 \u003c1115898@3ce6091f-6c88-37e8-8c75-72f92ae8dfba\u003e","operation":"ADD","reason":"\u003cGERRIT_ACCOUNT_3355700\u003e replied on the change"}
Attention: {"person_ident":"Gerrit User 3355700 \u003c3355700@3ce6091f-6c88-37e8-8c75-72f92ae8dfba\u003e","operation":"REMOVE","reason":"\u003cGERRIT_ACCOUNT_3355700\u003e replied on the change"}
91 lines
2.9 KiB
Plaintext
91 lines
2.9 KiB
Plaintext
{
|
|
"comments": [
|
|
{
|
|
"unresolved": false,
|
|
"key": {
|
|
"uuid": "71fa882c_3119cbc3",
|
|
"filename": "/PATCHSET_LEVEL",
|
|
"patchSetId": 1
|
|
},
|
|
"lineNbr": 0,
|
|
"author": {
|
|
"id": 1115898
|
|
},
|
|
"writtenOn": "2024-03-19T19:21:44Z",
|
|
"side": 1,
|
|
"message": "Can this be done for row_neon.cc ?\nI need to run a benchmark, as the AGI for 2 loads will slow it down and I\u0027m not clear on why it would go faster, except on A510 where ST2 is slow.\nFor ST2 there should have been an ZIP+ST1 version of the code, as done in row_neon64.cc row functions.",
|
|
"revId": "46b862d24df18e14bd8fb22e28bc53108be8add7",
|
|
"serverId": "3ce6091f-6c88-37e8-8c75-72f92ae8dfba"
|
|
},
|
|
{
|
|
"unresolved": true,
|
|
"key": {
|
|
"uuid": "edb8bdef_c832e5f2",
|
|
"filename": "source/scale_neon64.cc",
|
|
"patchSetId": 1
|
|
},
|
|
"lineNbr": 1151,
|
|
"author": {
|
|
"id": 1115898
|
|
},
|
|
"writtenOn": "2024-03-19T19:21:44Z",
|
|
"side": 1,
|
|
"message": "this is an AGI on A55. Consider some alternatives\n1. offset for 2nd load and ADD instruction to src register after the loads\n2. 2 registers. src and src1. increment each 64. solves agi.\n3. only do 4 pixels per loop. a single ld2 and str",
|
|
"revId": "46b862d24df18e14bd8fb22e28bc53108be8add7",
|
|
"serverId": "3ce6091f-6c88-37e8-8c75-72f92ae8dfba"
|
|
},
|
|
{
|
|
"unresolved": false,
|
|
"key": {
|
|
"uuid": "df72eda2_a6887f74",
|
|
"filename": "source/scale_neon64.cc",
|
|
"patchSetId": 1
|
|
},
|
|
"lineNbr": 1151,
|
|
"author": {
|
|
"id": 3355700
|
|
},
|
|
"writtenOn": "2024-03-21T13:34:27Z",
|
|
"side": 1,
|
|
"message": "Option 2 seems good. Only doing 4 pixels per loop would be bad on little cores.",
|
|
"parentUuid": "edb8bdef_c832e5f2",
|
|
"revId": "46b862d24df18e14bd8fb22e28bc53108be8add7",
|
|
"serverId": "3ce6091f-6c88-37e8-8c75-72f92ae8dfba"
|
|
},
|
|
{
|
|
"unresolved": true,
|
|
"key": {
|
|
"uuid": "02d6093f_caf8c6e6",
|
|
"filename": "source/scale_neon64.cc",
|
|
"patchSetId": 1
|
|
},
|
|
"lineNbr": 1155,
|
|
"author": {
|
|
"id": 1115898
|
|
},
|
|
"writtenOn": "2024-03-19T19:21:44Z",
|
|
"side": 1,
|
|
"message": "can this be stp?",
|
|
"revId": "46b862d24df18e14bd8fb22e28bc53108be8add7",
|
|
"serverId": "3ce6091f-6c88-37e8-8c75-72f92ae8dfba"
|
|
},
|
|
{
|
|
"unresolved": false,
|
|
"key": {
|
|
"uuid": "c5e6dd1f_67cbcd20",
|
|
"filename": "source/scale_neon64.cc",
|
|
"patchSetId": 1
|
|
},
|
|
"lineNbr": 1155,
|
|
"author": {
|
|
"id": 3355700
|
|
},
|
|
"writtenOn": "2024-03-21T13:34:27Z",
|
|
"side": 1,
|
|
"message": "No difference.",
|
|
"parentUuid": "02d6093f_caf8c6e6",
|
|
"revId": "46b862d24df18e14bd8fb22e28bc53108be8add7",
|
|
"serverId": "3ce6091f-6c88-37e8-8c75-72f92ae8dfba"
|
|
}
|
|
]
|
|
} |