Posts: 114
Threads: 14
Joined: Dec 2013
Reputation:
1
09-09-2017, 10:10 PM
(This post was last modified: 09-09-2017, 10:13 PM by jamesstuller.)
Request to ADD (Rewriting Vertex Processing) to GPU massive boost frame rate so even slow pc may run pcsx2 at full speed.
RPCS3 CEMU and Dophin ps3 has this Rewriting Vertex Processing to GPU and game now run very good low end pc.
Intel core i3-6100u
intel hd 520
12gb ram
drakengard 1&2 don't run full speed on PCSX2 on this laptop.
drakengard 3 run full speed in RPCS3 on this laptop.
This Rewriting Vertex Processing to GPU is last hope for pcsx2 to run games full speed so it WIN WIN for pcsx2.
Posts: 28.930
Threads: 16
Joined: Dec 2008
Reputation:
606
Location: 127.0.0.1
(09-09-2017, 10:10 PM)jamesstuller Wrote: Request to ADD (Rewriting Vertex Processing) to GPU massive boost frame rate so even slow pc may run pcsx2 at full speed.
RPCS3 CEMU and Dophin ps3 has this Rewriting Vertex Processing to GPU and game now run very good low end pc.
Intel core i3-6100u
intel hd 520
12gb ram
drakengard 1&2 don't run full speed on PCSX2 on this laptop.
drakengard 3 run full speed in RPCS3.
This Rewriting Vertex Processing to GPU is last hope for pcsx2 to run games full speed so it WIN WIN for pcsx2.
Can you explain how we can add it ?
CPU : AMD Ryzen 7 3800X
Mobo : Asus PRIME B450-PLUS
GPU : NVIDIA GeForce RTX 3070
RAM : 16 Go
Posts: 28.930
Threads: 16
Joined: Dec 2008
Reputation:
606
Location: 127.0.0.1
Are you sure the PS3 RSX GPU is similar to PS2 GPU ?
CPU : AMD Ryzen 7 3800X
Mobo : Asus PRIME B450-PLUS
GPU : NVIDIA GeForce RTX 3070
RAM : 16 Go
Posts: 12.837
Threads: 48
Joined: Oct 2013
Reputation:
252
Yeah, none of this would help.
Posts: 3.526
Threads: 6
Joined: Dec 2013
09-10-2017, 05:23 PM
(This post was last modified: 09-10-2017, 05:24 PM by dabore.)
ugh. the rsx is a regular gpu in terms of shaders. some input stream mapping isn't easily done with the gs and the problem with this whole vertex processing on the ps2 is the complexity of the vu code. you'd need explicit gpu machine code to recompile vu shaders to run on regular gpu not with the regular shaders and pipeline stages. you basicly have a typeless buffer and compute it into another basicly typeless buffer with a bit of gif stuff to glue it together and which you have to reinterpret later to get it sorted anyway. you can't convert any of this into a virtual buffer formats to transfer them. getting all those instruction to run in parallel to get a speed boost is even harder.