It's too bad nvidia decided to kill Vulkan on Fermi cards.
Announcement
Collapse
No announcement yet.
Early OpenGL vs. Vulkan Linux Benchmarks With Talos Principle
Collapse
X
-
Originally posted by haagch View PostSo when the readme https://cgit.freedesktop.org/mesa/me...n.txt?h=vulkan says
Unsupported Features: - Tesselation shaders - Push constants in GS and VS on HSW and prior - Sparse resources - Input attachments
Is Vulkan still feature complete in the driver?
//EDIT: Just looked at your link:
First-wave conformance has been submitted for Broadwell, Sky Lake, and
Cherryview. They all pass 100% of the mustpass tests as of January 30, 2016.
- Likes 1
Comment
-
Originally posted by V10lator View Post^ This. According to users the Intel driver is completely broken so I wonder how it could pass the test. Did any owner of a Intel GPU test what happens if you try to validate manually ( https://github.com/KhronosGroup/Vulk...rnal/vulkancts ) ?
//EDIT: Just looked at your link:
So how can that be?
Depending on what it's doing (as I don't have an Intel GPU to test with), it might be the version mismatch. If an application requests 1.0.3 and the implementation only provides 1.0.2, the Vulkan ICD will say the implementation is unavailable. The driver probably works fine in the right environment.
EDIT: For what it's worth, the application can check for version by asking for version 1.0.0, querying the physical device for version supported, then recreating the instance with the version supported. Kind of a pain but it's reasonable.
I take that back: http://www.libretro.com/index.php/vu...l-impressions/Last edited by computerquip; 21 February 2016, 11:36 AM.
Comment
-
Originally posted by computerquip View PostI take that back: http://www.libretro.com/index.php/vu...l-impressions/What remains to be done? 1 – Waiting on better drivers, obviously. There is only so much stuff that can currently be worked around with the Ivy Bridge/Haswell drivers.
Well, that was not what we were hoping for.
Why even waste time on support for broken drivers? Just make it work on conformant drivers and complain to intel to fix their drivers.
Do we have any ETA for intel drivers? Hours? Days? Weeks?
Comment
-
This performance doesn't shock me at all.
Remember, when using anything much above a core i5 CPU, games are GPU bottlenecked. The primary advantage of Vulkan/DX12 is to theoretically improve CPU performance. Run the same tests with a Core i3/FX-4xxx series CPU, and you'll see performance improvements. But for every non-CPU bottlenecked CPU, I wouldn't expect any real performance advantage.
Secondly, people seem to be forgetting why APIs were developed that abstracted out memory management: Because Compilers are better at it then developers are. Willing to be a lot of performance loss is due to sub-optimal programming.
Comment
-
-
Originally posted by gamerk2 View PostThis performance doesn't shock me at all.
Remember, when using anything much above a core i5 CPU, games are GPU bottlenecked. The primary advantage of Vulkan/DX12 is to theoretically improve CPU performance. Run the same tests with a Core i3/FX-4xxx series CPU, and you'll see performance improvements. But for every non-CPU bottlenecked CPU, I wouldn't expect any real performance advantage.
Secondly, people seem to be forgetting why APIs were developed that abstracted out memory management: Because Compilers are better at it then developers are. Willing to be a lot of performance loss is due to sub-optimal programming.
Comment
Comment