Progressive RFX merge #50
Conversation
|
Happy to merge if it doesn't break any existing RFX functionality, if that would help. |
|
It's important to get this merged as a first step to integrating GFX, because otherwise you'll have build breaks in the progressive RFX part. Other than xorgxrdp, this is the only other dependency that needs updating to start that integration. I will commit to doing some testing of this branch to verify that regular RFX still works, that Progressive RFX still works with my latest mainline_merge branch, and that the performance isn't degraded in any meaningful way. In order to test with MSTSC and the Mac OS Client, I have to actually disable the H264 capability set manually to force the progressive mode, but other clients are easier such as Microsoft Remote Desktop on Windows and FreeRDP (where I can manually specify the protocol). |
|
Please do some regular testing. When you're happy with it, I'll merge it in. |
Nexarian
force-pushed
the
egfx_mainline_merge
branch
3 times, most recently
from
e92ab8e
to
c0e6938
Compare
metalefty
changed the title
|
Verified that it doesn't break vanilla RFX. That is, connecting with MSTSC on a 2560x1440 monitor to an Ubuntu 20.04 machine running in VirtualBox. That works. Next is to test: Dynamic resizing with MSTSC, RemoteFX with FreeRDP, and make sure this works correctly with Progressive RFX on the GFX branch(es) |
|
Verified that with Will test MSTSC dynamic resize and Progressive RFX tomorrow. |
Largely a consolidation of the work by jsorg71, with a few minor bug fixes. Verified that RFX Progressive works.
Nexarian
force-pushed
the
egfx_mainline_merge
branch
from
bf820d5
to
34d0513
Compare
|
@matt335672 This works with Progressive RFX now as well. Didn't test with MSTSC, but I don't think I need to. This also works with FreeRDP RFX. I recommend we merge this if @jsorg71 is OK with it. I've at least verified that it doesn't make anything worse. |
Nexarian
changed the title
|
Let's merge 'cause librfxcodec is a submodule. |
| @@ -53,7 +53,7 @@ rfx_encode_component_rlgr1_amd64_sse2(struct rfxencode *enc, const char *qtable, | |||
| { | |||
| return 1; | |||
| } | |||
| *size = rfx_encode_diff_rlgr1(enc->dwt_buffer1, buffer, buffer_size); | |||
| *size = rfx_encode_diff_rlgr1(enc->dwt_buffer1, buffer, buffer_size, 64); | |||
There was a problem hiding this comment.
Just curious, where is this 64 come from? How to calculate?
There was a problem hiding this comment.
Two steps are done, a diff and rlgr1. LL3 is the band that gets diffed. In the original RemoteFX, LL3 was 8x8. In Progressive RemoteFX, LL3 is 9x9. These should be defines.
There was a problem hiding this comment.
What should the defines be? Maybe a second pull request to clean it up?
|
Tested by MSTSC without Progressive RFX. Nothing's broken and it works well. Thanks! |
This is simply a consolidation of https://github.com/jsorg71/librfxcodec/commits/egfx rebased on the latest devel.
Haven't tested it yet.