We're celebrating our seven year anniversary!
Read about it on our website at
https://fex-emu.com/FEXiversary/Looks like we finally have a new SBC that can take a dGPU without kernel level hacks! Looks like we can finally retire our old aged NVIDIA Jetson Orin boards! Looking forward to playing with this board some more.
We've been tinkering with NVIDIA's Orin over the past couple of days. This looks like a really fun platform for testing games! Shame about those old CPU cores; If only NVIDIA would make a new Tegra for the modern day.
We didn't get to all the optimizations last month. So we came back this month with a new way to optimize and a ton of performance improvements! Be prepared to get nerdsniped in to writing optimizations!
Read more at:
https://fex-emu.com/FEX-2309/FEX 2309 Tagged!
Last month we hinted that we didn’t get all optimizations in that we wanted. There’s more of that this month but we have also had an entire month to
push optimizations in. This month was a whirlwind of optimizations improving performance all over the place because of one feature that landed;
Instruction Count continous integration! Let’s dive in to what this is.
Today we were investigating multiple minute startup times in The Witcher 3. We found out that the Intel XeSS library has some pedantic code that is materializing a ~261,432 byte object on the stack one byte at a time! We have no idea why it does this, but we will investigate ways to improve this performance.
We love seeing code melt away in our translation of instructions over to ARM64! FEX's new Instruction Count CI not only allows us to highly target instruction optimizations, but also ensure we don't break implementations! Thanks to all the developers that we nerdsniped this last month in to optimizing instructions!
