Commit Graph

71 Commits

Author SHA1 Message Date
Markus Wick 227db66e4f OGL: Use glBufferData on Mali.
tl;dr: This PR speedups dolphin on mobiles with the Mali GPU and ES 3.2
drivers by a factor of 10 by using the method with the biggest overhead.
Please keep care not to buy this shit!

The ARM driver team seems to care very well about their customers. But
bad luck, users and open source developers are *not* their customers. So
even device-independent feature requests are just ignored for *years*:

https://community.arm.com/graphics/f/discussions/4645/gl_ext_buffer_storage-support

The bad point, they neither implement any of the other common ways to
stream dynamic content in unextented GL:
- They just ignore the GL_MAP_UNSYNCHRONIZED_BIT flag
- They don't support on-device buffer updates and just stall with
glBufferSubData

It seems like no benchmark is using any dynamic content - and like no
customer cares about anything but benchmarks, or users...

We have a flag to disable the glBufferSubData way, this PR adds the flag
to also disable the unsychronized mapping way. The second one is
available since their ES 3.2 update, but slow as hell.

So how to continue? The last remaining technical way to stream dynamic
content at all is to alloc a new buffer per draw call with glBufferData.
This is very gross, but still a factor 10 speedup compared to stalling
the GPU. Small tests shows that you can expect another 3-5 times speedup
with EXT_buffer_data, so Mali would be on pair with Adreno here. So if
you have bought such a device unfortunately, please try to make noise on
your vendor forums/support and ask for this extension. If you are going
to buy a new mobile, I'd recormend to avoid *any* mobile with a Mali GPU
in it.
2018-02-25 17:12:36 +01:00
Stenzek 134daf3b00 Vulkan: Extend the NVIDIA MSAA bug to render-pass based clears
Calling vkCmdClearAttachments with a partial rect, or specifying a
render area in a render pass with the load op set to clear can cause the
GPU to lock up, or raise a bounds violation. This only occurs on MSAA
framebuffers, and it seems when there are multiple clears in a single
command buffer. Worked around by back to the slow path (drawing quads)
when MSAA is enabled.
2017-09-07 17:05:43 +10:00
Stenzek 7a4348b641 Merge pull request #5649 from JonnyH/WIP/disable-vk-clear-renderpass
Add DriverDetails::BUG to Disable "LoadOp" clear renderpass in vulkan
2017-09-03 13:49:51 +10:00
Jonathan Hamilton 658a4a6e29 Mark an ImgTec driver bug as fixed in 1.8@4693462
Now we correctly parse ImgTec's GL_VERSION string we can actually use
the BugInfo's version stuff correctly here
2017-09-02 14:05:16 -07:00
Jonathan Hamilton 62e8d25cd1 Add Bug to Disable "LoadOp" clear renderpass in vulkan
This optimisation doesn't work on PowerVR's Vulkan implementation. We
(incorrectly) disallow Framebuffer objects to be used with a different
load or store op than that which they were created with, despite the
spec allowing such.

This fixes the windwaker intro "smearing"
2017-08-28 18:01:35 -07:00
Jonathan Hamilton 11057649ee Add VideoLogic (ImgTec) PCI ID 2017-08-28 18:01:16 -07:00
Stenzek 7f88711186 Vulkan: Add a driver bug for NV when MSAA is enabled
This causes the card to lock up when vkCmdClearAttachments is called,
when multisampling is enabled. Seems to be restricted to Maxwell and
newer?
2017-08-03 17:40:15 +10:00
Stenzek cd502990fa OGL: Uber shader support 2017-07-30 17:43:59 +10:00
Jonathan Hamilton 35caf48d7d Add a workaround for a PowerVR glsl compiler bug
It seems it doesn't like negating arguments for bitops, so suff like
"var <<= (-othervar);"
fails to compile
2017-05-03 19:33:56 -07:00
Matthew Parlane 85d74a506f Merge pull request #4951 from waddlesplash/haiku-2
Initial support for Haiku.
2017-03-28 17:19:35 +13:00
Augustin Cavalier 0831dad467 Initial support for Haiku. 2017-03-27 23:46:19 -04:00
z0z0z 005e6796b8 Disable pinned memory for AMD mesa drivers 2017-02-26 10:49:28 -05:00
Stenzek e2ddb3853e DriverDetails: Set BUG_PRIMITIVE_RESTART for AMD drivers 2017-02-18 15:07:35 +10:00
Jules Blok 086f839435 DriverDetails: Make the bug identifiers humanly readable. 2016-10-31 15:02:08 +01:00
Jules Blok d778400133 DriverDetails: Add OS X to the dual-source blending bug. 2016-10-30 00:47:02 +02:00
Jules Blok afe707bc18 DriverDetails: Disable dual-source blending on AMD OGL drivers. 2016-10-27 22:03:25 +02:00
Stenzek bac8c2d441 Vulkan: Work around indexed fragment output bug on AMD drivers 2016-10-01 02:40:02 +10:00
Stenzek 09638e714e VideoCommon: Extend DriverDetails to support both OpenGL and Vulkan 2016-10-01 01:09:12 +10:00
Jules Blok afa251af42 DriverDetails: Add bug for broken gl_ClipDistance on i965. 2016-08-15 13:11:28 +02:00
Anthony J. Bentley de6babb928 Add an OpenBSD driver identifier. 2016-07-17 04:13:06 -06:00
Pierre Bourdon 3570c7f03a Reformat all the things. Have fun with merge conflicts. 2016-06-24 10:43:46 +02:00
Stenzek 89e54fbd6c OGL: Work around slowdown of glMapBufferRange with SSBO on NVIDIA drivers
Using glMapBufferRange to read back the contents of the SSBO is extremely
slow on NVIDIA drivers. This is more noticeable at higher internal
resolutions. Using glGetBufferSubData instead does not seem to exhibit
this slowdown.
2016-05-19 21:24:09 +10:00
degasus bca0e06a95 OGL: Use coherent mapping on Qualcomm devices. 2016-05-11 23:55:28 +02:00
degasus 7517d126c8 DriverDetails: Drop BUG_BROKENALPHATEST.
This flag is not in use at all.
2016-05-11 21:22:09 +02:00
degasus 6219c39cf5 OGL: Drop QC ES3.1 workaround.
This was never tested well:

HdkR> The tester was most likely trying to load a stale shader cache or something
2016-05-11 20:45:07 +02:00
Ryan Houdek ff0aac82b5 Mark Mesa Intel Sandy Bridge geometry shaders not broken in Mesa 11.2 2016-03-04 10:45:13 -06:00
Ryan Houdek b2b0959f23 Geometry shader bug isn't fixed in 11.1.2 for Intel Sandy Bridge 2016-02-12 07:04:09 -06:00
Ryan Houdek cf6f9de350 Both Intel and Radeon Mesa geometry shader bugs are fixed in 11.1.2 2016-02-11 12:55:55 -06:00
Ryan Houdek 3dda36bc5b Blacklist Sandy Bridge on mesa from using geometry shaders. 2016-01-20 12:13:21 -06:00
Ryan Houdek b4eb5d8e3f Disable geometry shaders on mesa AMD/ATI drivers.
Causes misrenderings in games that uses them.
2016-01-09 15:09:37 -06:00
Ryan Houdek 721d319c7b Add a comment to DriverDetails about Qualcomm failing our alpha tests. 2015-12-31 10:14:14 -06:00
Ryan Houdek 9618738278 Remove all of our workarounds for Qualcomm devices we don't support anymore. 2015-09-04 23:45:35 -05:00
Ryan Houdek 01db003779 [Android] Workaround Mali driver issue on the Samsung Galaxy S6.
Samsung updated the video drivers on the SGS6 which introduced a bug when disabling vsync.
Both the driver versions are r5p0, but the md5sums of the blob differ.
To work around the issue, make sure to never disable vsync by calling eglSwapInterval.

We can't actually determine the driver version on Android yet.
So until the driver version lands that displays the driver version string in the GL_VERSION string
we will need to keep this workaround enabled at all times, which is a bit annoying.

Current mali drivers return the video driver version in one of the EGL strings you can query.
The issue with that is that Android eats all of those strings, so we can't query it.
2015-08-28 09:02:46 -05:00
Ryan Houdek b4e4a4cef4 Disable OpenGL ES 3.1 on all Qualcomm Adreno devices.
Their new driver that supports GLES3.1 + AEP has issues with it.
At the very least they don't implement all of the geometry shader features fully which causes shader linker issues when we attempt to use them.
I don't have a device so I can't fully test, so until I do I'm going to blanket disable the whole thing.
2015-08-22 09:12:19 -05:00
degasus d3cf4034de DriverDetails: Disable GL_ARB_copy_image on mesa. 2015-08-06 19:41:36 +02:00
Ryan Houdek 012a9afdb6 Work around slow glBufferSubData on PowerVR.
As with all tiled renderers, glBufferSubData is terribly slow.
Use glBufferData as typical.
2015-07-20 22:02:58 -05:00
Lioncash daa205990f Use emplace() instead of insert() where applicable for maps. 2015-06-28 19:52:40 -04:00
Jeffrey Pfau 7085fcc8d6 Fix FreeBSD build 2015-06-13 21:52:47 -07:00
Tillmann Karras cefcb0ace9 Update license headers to GPLv2+ 2015-05-25 13:22:31 +02:00
Dwayne Slater ae83a1b821 Fix OpenGLES 3.0 on Qualcomm's crappy driver, it can't bitshift sometimes.
[fixed lint issues and grammar ~comex]
2015-04-23 16:33:12 -04:00
Jules Blok 3f763e6999 DriverDetails: Remove Intel Structs bug since it doesn't affect us anymore. 2014-12-28 23:28:00 +01:00
Jules Blok 2c0bee5da9 DriverDetails: Update Intel bug description. 2014-12-28 23:28:00 +01:00
Jules Blok bc3ed44050 OGL: Disable geometry shaders on Intel's Windows drivers due to broken interface blocks. 2014-12-19 23:56:02 +01:00
Ryan Houdek 8c0e26d969 Minor changes to DriverDetails
Fixes a typo where the official IMGTec drivers were said to be the OSS driver support.
Removes Mali GPU family detection just like I removed the Adreno family detection.
We don't support Mali Utgard anyway.
If we need family detection we can properly add it, right now it isn't needed.
2014-12-19 21:41:12 +00:00
Ryan Houdek 0fec69001b Fixes Adreno 400 slow performance.
Adreno 300 and 400 have the same video driver performance issues because they are very similar architectures which use basically the same thing with
everything.
There isn't any need to detect the family of the driver with Qualcomm anyway. If we ever need family specific bugs then we can implement real support
for that.
Performance issue on Adreno 400 series was due to us only detecting Adreno 300 series, and with Adreno 400 it wouldn't use the bugs, which would cause
it to use glBufferSubData, causing the huge performance hit.
2014-12-19 21:31:37 +00:00
Ryan Houdek 5c3bbf7409 Works around broken Intel Windows video drivers.
Just use regular boolean negation in our pixel shader's depth test everywhere except on Qualcomm.
This works around a bug in the Intel Windows driver where comparing a boolean value against true or false fails but boolean negation works fine.
Quite silly.

Should fix issues #7830 and #7899.
2014-12-03 00:33:42 -06:00
Ryan Houdek 9da7e6ae79 Adds a DriverDetails bug to track Qualcomm attributeless rendering.
This particular issue was fixed in the v66 (07-08-2014) development drivers from Qualcomm.
To make sure we cover all drivers that may or may not have the issue fixed, make sure to mandate v95 minimum to work around the issue.
The next commit is the actual work around for post processing for this.
2014-10-29 19:58:18 -05:00
Ryan Houdek daabcfd6fc Removes Qualcomm's rotated framebuffer bug from DriverDetails.
Due to changes in how we render to the final framebuffer we no longer encounter this bug.
With the change to post processing being enabled at all times and no longer using glBlitFramebuffer, Qualcomm no longer has the chance to rotate our
framebuffer underneath of us.
2014-10-29 19:57:51 -05:00
Ryan Houdek 9108a11af4 Change driver details to reflect Qualcomm's changes with their v95 driver.
They fixed their issues with dynamic UBO array member access.
There are many other issues though.
2014-10-18 02:50:57 -05:00
degasus 8f403696ea DriverDetails: mark intel buffer_storage bug as fixed 2014-10-16 22:51:32 +02:00