New vulkan stuff in amd driver = Might improve x4 ssr performance
Moderator: Moderators for English X Forum
New vulkan stuff in amd driver = Might improve x4 ssr performance
https://www.khronos.org/registry/vulkan ... int64.html
This was added to amd drivers the other day. Says minimal code changes for possible improved performance and is linked to space screen reflections, thought ego might fancy using it in x4.
Theres also 2 other things im not sure about.
VK_KHR_buffer_device_address
This extension is used to query the device address of a buffer to allow for shader access to that buffer’s storage via the SPV_KHR_physical_storage_buffer SPIRV extension.
VK_EXT_robustness2
This extension provides stricter restrictions for handling reads and writes that are out of bounds. It specifies that out-of-bounds reads must return zeros and out-of-bounds writes must be discarded. This extension also adds support for null descriptors.
VK_EXT_shader_image_atomic_int64
This extension extends existing 64-bit integer atomic support to images, which provides more efficient access than buffers. This allows applications to quickly improve their performance with minor changes to their code.
Figured id post here as not sure would be better elsewhere.
This was added to amd drivers the other day. Says minimal code changes for possible improved performance and is linked to space screen reflections, thought ego might fancy using it in x4.
Theres also 2 other things im not sure about.
VK_KHR_buffer_device_address
This extension is used to query the device address of a buffer to allow for shader access to that buffer’s storage via the SPV_KHR_physical_storage_buffer SPIRV extension.
VK_EXT_robustness2
This extension provides stricter restrictions for handling reads and writes that are out of bounds. It specifies that out-of-bounds reads must return zeros and out-of-bounds writes must be discarded. This extension also adds support for null descriptors.
VK_EXT_shader_image_atomic_int64
This extension extends existing 64-bit integer atomic support to images, which provides more efficient access than buffers. This allows applications to quickly improve their performance with minor changes to their code.
Figured id post here as not sure would be better elsewhere.
Re: New vulkan stuff in amd driver = Might improve x4 ssr performance
Hmm, unfortunately those won't really help.
SSR just needs a bit more love, atm I'm swamped with improving asset management, which is one of the blockers for improving stuff like ssr (we're getting overly tight on gpu-memory).
SSR just needs a bit more love, atm I'm swamped with improving asset management, which is one of the blockers for improving stuff like ssr (we're getting overly tight on gpu-memory).
Re: New vulkan stuff in amd driver = Might improve x4 ssr performance
Ur low on vram? Odd according to afterburner my 8gb is barely used for x4. Or do you use 4gb as the max limit for some reason? Unless i misunderstood when you ref gpu memory.
Shame the vulkan things cant help, oh well.
Gluck with the asset management.
Shame the vulkan things cant help, oh well.
Gluck with the asset management.
Re: New vulkan stuff in amd driver = Might improve x4 ssr performance
IDK about SSR but moving from W10 to Linux with MESA 20.3dev (latest with a lot of AMD related vulkan improvements) is a huge boost for both Vega 64 and 5700XT. Same settings (2560X1440, SSAA 2x, SSAO Normal, SSR Off, everything else on high/maxed out) W10 have V64 71fps, 5700XT 84fps and on Linux (Manjaro with MESA 20.3dev) 84fps and 100fps respectively on same scene on same system. 3900X with CPPC & CPPC preferred cores ON, with 3800C16 overclocked ram. No other changes.Skeeter wrote: ↑Wed, 30. Sep 20, 04:26https://www.khronos.org/registry/vulkan ... int64.html
This was added to amd drivers the other day. Says minimal code changes for possible improved performance and is linked to space screen reflections, thought ego might fancy using it in x4.
Theres also 2 other things im not sure about.
VK_KHR_buffer_device_address
This extension is used to query the device address of a buffer to allow for shader access to that buffer’s storage via the SPV_KHR_physical_storage_buffer SPIRV extension.
VK_EXT_robustness2
This extension provides stricter restrictions for handling reads and writes that are out of bounds. It specifies that out-of-bounds reads must return zeros and out-of-bounds writes must be discarded. This extension also adds support for null descriptors.
VK_EXT_shader_image_atomic_int64
This extension extends existing 64-bit integer atomic support to images, which provides more efficient access than buffers. This allows applications to quickly improve their performance with minor changes to their code.
Figured id post here as not sure would be better elsewhere.
Same applies to walking to stations FPS went up also from 40ish to mid 50s low 60s.
-
- Moderator (English)
- Posts: 4764
- Joined: Fri, 21. Dec 18, 18:23
Re: New vulkan stuff in amd driver = Might improve x4 ssr performance
Does the screen space stuff need any 64-bit integer atomic operations? I would imagine screen space stuff mostly using 32-bit access for looking at RGBA pixel information. Maybe this is intended for HDR pixel data?
X4 is built around using 3 GB of video memory, as listed in its minimum requirements. If it uses more than 3 GB of video memory then the official memory requirements would have to be raised and may cause many people who can currently play X4 to no longer be able to do so.
This is likely due to other changes that occur when moving between operating systems. For example scheduling algorithm differences or differences with OS call implementations. Even the way the operating system is setup can influence performance, such as if virtualization is enabled or not.Panos wrote: ↑Fri, 2. Oct 20, 22:46IDK about SSR but moving from W10 to Linux with MESA 20.3dev (latest with a lot of AMD related vulkan improvements) is a huge boost for both Vega 64 and 5700XT. Same settings (2560X1440, SSAA 2x, SSAO Normal, SSR Off, everything else on high/maxed out) W10 have V64 71fps, 5700XT 84fps and on Linux (Manjaro with MESA 20.3dev) 84fps and 100fps respectively on same scene on same system. 3900X with CPPC & CPPC preferred cores ON, with 3800C16 overclocked ram. No other changes.
Re: New vulkan stuff in amd driver = Might improve x4 ssr performance
Im not sure about the vram thing you said imperial.
Sure the min spec is a 3gb card that is expected for low settings. Ultra settings should require more as there should be more detail, yet i dont think really there is a difference between low and high textures in x4 is there which is a little strange. I mean u buy a 8gb card just for x4 so u can whack up the details especially textures but u dont get the benefits. Seems a little strange compared to other games. If its a one size vram for all then whats the point of options in the gfx menu i wonder.
Sure the min spec is a 3gb card that is expected for low settings. Ultra settings should require more as there should be more detail, yet i dont think really there is a difference between low and high textures in x4 is there which is a little strange. I mean u buy a 8gb card just for x4 so u can whack up the details especially textures but u dont get the benefits. Seems a little strange compared to other games. If its a one size vram for all then whats the point of options in the gfx menu i wonder.
Re: New vulkan stuff in amd driver = Might improve x4 ssr performance
Was this finished, the asset management i mean, for to test in 4.0 beta or is it more of a not yet but should make 4.0 final?