SSE4 instruction set
Started by braddiddy




4 posts in this topic
braddiddy
Member


0
4 posts 1 threads Joined: Dec 2017
12-21-2017, 01:51 PM -
#1
I'm sure this question depends on a lot of specifics, but given the GPU is not holding the a system back, and all else the same, does a similar processor that supports SSE4 instructions have a significant advantage in emulation over one that only supports SSE3? If so would it matter between SSE4.1 and SSE4.2?
Ani
Administrator
*******


16
4,389 posts 107 threads Joined: Aug 2017
12-21-2017, 03:36 PM -
#2
LLVM Recompiler automatically recompiles code to the best optimized path depending on what instruction sets your CPU has.
    Desktop: Ryzen 7 5800X,   Radeon RX 6800 XT, 2x8G DDR4 3600MHz, Manjaro Linux
     Laptop: Ryzen 9 5900HX,  Radeon RX 6700M,   2x8G DDR4 3200MHz, Manjaro Linux
Old Desktop: AMD FX-8350,     Radeon R9 280X,    2x4G DDR3 1600MHz, Manjaro Linux
braddiddy
Member


0
4 posts 1 threads Joined: Dec 2017
12-21-2017, 04:23 PM -
#3
But would there be a lot of instances where a SSE4 capable processor would significantly outperform one that has SSE3 but lacks SSE4?
Ani
Administrator
*******


16
4,389 posts 107 threads Joined: Aug 2017
12-21-2017, 04:29 PM -
#4
If your CPU doesn't support SSE4, forget about RPCS3, it's time to buy a new one

Just use LLVM CPU setting to compile to a CPU type that makes LLVM not use instructions from SSE4 and find out the performance difference, but I doubt it's noticeable if any
    Desktop: Ryzen 7 5800X,   Radeon RX 6800 XT, 2x8G DDR4 3600MHz, Manjaro Linux
     Laptop: Ryzen 9 5900HX,  Radeon RX 6700M,   2x8G DDR4 3200MHz, Manjaro Linux
Old Desktop: AMD FX-8350,     Radeon R9 280X,    2x4G DDR3 1600MHz, Manjaro Linux
braddiddy
Member


0
4 posts 1 threads Joined: Dec 2017
12-21-2017, 04:58 PM -
#5
Well I've got an old computer, It's a dell with a core 2 quad Q6600 / R9360 gpu I can get 30 frames on games I've tried on it. I'm thinking about overclocking this one, which is easy to do, with stock everything else 2.4ghz to 3.0ghz or upgrading it with one that has a stock speed of 3.0Ghz by the same method (333X9) as opposed to (266X9). The two would then be equivalent except for differences being 12MB vs 8MB L2 cache, and support for SSE4.1. I guess if it doesn't help much for RPCS3 then I'll overclock this one for now unless I come across another need for SSE4 instructions. Thank you for taking the time to answer my questions.


Forum Jump:


Users browsing this thread: 2 Guest(s)