@Caoehill
As i aware, everything which is "proto/include_4/interfaces/etc" should go to SDK/include_h. And only those which is not specific os4 ones comes to local/ etc. I can be wrong, but that how it was always with everything. That of course not necesarity and everything can be put anywhere else, but just to have clean SDK, etc..

@kas1e

MiniGL uses SDK/local/common https://os4depot.net/?function=showcon ... river/graphics/minigl.lha

I don't remember how OGLES2 SDK looks officially but at least I have installed it in SDK/local/common.

Next one!
Sapphire RX 560
14 CU
56 units

radeon_gpu_info_probe: BEGIN installed-card report
Board and provider:
graphics.library 54.256
board index 0
board label Radeon RX Polaris11
board driver PCIGraphics.card 53.18
chip driver RadeonRX.chip 2.12
RM resource RadeonRX_RM.resource
RM resource/API 1 / 3
RM chipset 0x0000004f (POLARIS11)
PCI identity:
BDF 0:1:0.0
vendor:device 1002:67ef
subsystem 1da2:e348
revision / class e5 / 030000
expansion.library 54.1
GPU topology:
decoded family POLARIS11
raw GFX IP 8.1
internal/external revision 1 / 0x5b
active CUs 14
derived shader lanes 896 (64 per GCN CU)
derived texture units 56 (4 per GCN CU)
active render backends 4
derived ROPs 16 (4 per backend)
SE / SA-per-SE / max-CU-SA 2 / 1 / 8
RB-per-SE / tile-pipes / TCC 2 / 4 / 4
active CU mask SE0/SA0 0x0000007f (7 CUs)
active CU mask SE1/SA0 0x0000007f (7 CUs)
enabled RB mask 0x00000000:0000000f
Memory:
GPU-available memory 0x00000000:80000000 bytes (2048 MiB)
CPU-visible bitmap memory 0x00000000:0f0e1400 bytes (240 MiB)
current bitmap memory free 0x00000000:0c49bd00 bytes (196 MiB)
largest current free block 0x00000000:0c1f9100 bytes (193 MiB)
VRAM technology unavailable (not exposed)
memory-bus width unavailable (not exposed)
Power management (read-only):
RadeonRX DPM version 1
coarse policy 0 (provider value)
current state 1 (Performance)
available named states 2
state[0] Boot
state[1] Performance
live/max engine clock unavailable (DPM v1 has no sensor)
live/max memory clock unavailable (DPM v1 has no sensor)
state changes not performed
AtomBIOS and clocks:
option ROM bytes 59392
ATOM format/content 1.1
FirmwareInfo format/content 2.2 (108 bytes)
GFX_Info format/content 2.3 (24 bytes)
firmware revision 0x0f320201
default engine clock 214 MHz (Atom boot/default)
default memory clock 300 MHz (Atom boot/default)
reference clock 100 MHz
GPU timestamp crystal 25000 kHz
Register and queue facts:
MC_ARB_RAMCFG 0x00006062
GB_ADDR_CONFIG 0x12011002
PCIE_EFUSE4 0x15e67ef1
SMC CLKPIN CNTL / CNTL_2 0x00000003 / 0x00400009
visible rings 10
GFX / compute / UVD / other 1 / 8 / 1 / 0
radeon_gpu_info_probe: PASS no BO, command, submission, or rendering


=======================================================
glmark2 2023.01
=======================================================
OpenGL Information
GL_VENDOR: AMD
GL_RENDERER: POLARIS11 (radeonsi, polaris11, ACO, DRM 0.0, 54.57 (29.7.2023) AmigaOne X5000 release)
GL_VERSION: 4.6 (Compatibility Profile) Mesa 26.1.5 (git-2e7650ee6e)
Surface Config: buf=32 r=8 g=8 b=8 a=8 depth=24 stencil=0 samples=0
Surface Size: 800x600 windowed
=======================================================
[build] use-vbo=false: FPS: 280 FrameTime: 3.578 ms
[build] use-vbo=true: FPS: 4793 FrameTime: 0.209 ms
[texture] texture-filter=nearest: FPS: 4432 FrameTime: 0.226 ms
[texture] texture-filter=linear: FPS: 4468 FrameTime: 0.224 ms
[texture] texture-filter=mipmap: FPS: 4416 FrameTime: 0.226 ms
[shading] shading=gouraud: FPS: 3979 FrameTime: 0.251 ms
[shading] shading=blinn-phong-inf: FPS: 3977 FrameTime: 0.251 ms
[shading] shading=phong: FPS: 3986 FrameTime: 0.251 ms
[shading] shading=cel: FPS: 3982 FrameTime: 0.251 ms
[bump] bump-render=high-poly: FPS: 3809 FrameTime: 0.263 ms
[bump] bump-render=normals: FPS: 4721 FrameTime: 0.212 ms
[bump] bump-render=height: FPS: 4765 FrameTime: 0.210 ms
[effect2d] kernel=0,1,0;1,-4,1;0,1,0;: FPS: 3545 FrameTime: 0.282 ms
[effect2d] kernel=1,1,1,1,1;1,1,1,1,1;1,1,1,1,1;: FPS: 3251 FrameTime: 0.308 ms
[pulsar] light=false:quads=5:texture=false: FPS: 3088 FrameTime: 0.324 ms
[desktop] blur-radius=5:effect=blur:passes=1:separable=true:windows=4: FPS: 1868 FrameTime: 0.536 ms
[desktop] effect=shadow:windows=4: FPS: 1832 FrameTime: 0.546 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 140 FrameTime: 7.143 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=subdata: FPS: 135 FrameTime: 7.455 ms
[buffer] columns=200:interleave=true:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 216 FrameTime: 4.640 ms
[ideas] speed=duration: FPS: 1094 FrameTime: 0.915 ms
[jellyfish] <default>: FPS: 1587 FrameTime: 0.630 ms
[terrain] <default>: FPS: 684 FrameTime: 1.463 ms
[shadow] <default>: FPS: 3623 FrameTime: 0.276 ms
[refract] <default>: FPS: 1035 FrameTime: 0.967 ms
[conditionals] fragment-steps=0:vertex-steps=0: FPS: 3815 FrameTime: 0.262 ms
[conditionals] fragment-steps=5:vertex-steps=0: FPS: 3812 FrameTime: 0.262 ms
[conditionals] fragment-steps=0:vertex-steps=5: FPS: 3844 FrameTime: 0.260 ms
[function] fragment-complexity=low:fragment-steps=5: FPS: 3844 FrameTime: 0.260 ms
[function] fragment-complexity=medium:fragment-steps=5: FPS: 3853 FrameTime: 0.260 ms
[loop] fragment-loop=false:fragment-steps=5:vertex-steps=5: FPS: 3843 FrameTime: 0.260 ms
[loop] fragment-steps=5:fragment-uniform=false:vertex-steps=5: FPS: 3857 FrameTime: 0.259 ms
[loop] fragment-steps=5:fragment-uniform=true:vertex-steps=5: FPS: 3827 FrameTime: 0.261 ms
=======================================================
glmark2 Score: 3041
=======================================================

@McFly
Thanks a bunch, but for now there is no easy way for, but that not necessary, i more happy if we all test it and find bugs :) But thanks anyway!

@Capehill
Quote:

Yeah, but rest in SDK/include because those are usual "system" headers, i.e. proto/ interface/ inline_h, etc ? Or i miss something improtant ?:) At least that how i understand SDK for all last 15 years: SDK/local/common : for libs in includes which shared between any crt, SDK/local/newlib and SDK/local/clib2/4 only for specific to them. Then SDK/include are those include_h / interfaces / etc and all the proto/mesa.h , proto/warp3d.h, inline4/mesa.h, or inline4/wb.h , all there.

@kas1e

Quote:

I guess I need to deal with these using preprocessor (USE_MINIGL_HEADERS, USE_OGLES2_HEADERS) or something.

Quote:

Khronos headers are in SDK/local/ but the rest are in SDK/include.

@kas1e
Quick question ? how can I sponsor you?
I want to send you a bit of money for a coffee, a beer or something else :)

@ddni
Quote:

What i mean is to try with some game which show FPS and working via ogles2.library/warp3dnova, both your rx550, and your rx590. Just to see if rx590 start to be slower and buggy as hardware indeed.

@McFly
Looks good for us, all what more than 4000 is more or less ok at least for now for our machines :) But of course all of this far from being perfect, because for reference there some examples i find in the net:


Ryzen 1800X + R9 Fury: glmark2 Score: 9618
Ryzen 3 3100 + RX580 XFX: glmark2 Score: 14778
Ryzen 5 5500 + RX6600 : glmark2 Score: 14365
Ryzen 7 7700x + RX 7900xt + 48GB RAM : glmark2 Score: 27955

Dell Inspiron 5577 (i5-7300HQ, 8GB of RAM and a 4GB dell oem GTX 1050) : glmark2 Score: 2329

i5-3470 + NVIDIA GeForce GTX 1060 6GB: glmark2 Score: 10783

So having at least ~10.000 can be better :)

@Capehill
Breaking of MiniGL of course expected, as it different gl includes (even when i use ogles2 i had to rename/move out minigl includes), so there were issue between two and expected that mesa includes broke minigl ones. What not expected that it broke ogles2 builds, as ogles2 use kronos includes as well ..

Quote:

But they already in this place ? i.e. sdk/local/common/include/ and sdk/local/common/lib/

X5000/20 - RX580
=======================================================
glmark2 2023.01
=======================================================
OpenGL Information
GL_VENDOR: AMD
GL_RENDERER: POLARIS10 (radeonsi, polaris10, ACO, DRM 0.0, 54.57 (29.7.2023) AmigaOne X5000 release)
GL_VERSION: 4.6 (Compatibility Profile) Mesa 26.1.5 (git-2e7650ee6e)
Surface Config: buf=32 r=8 g=8 b=8 a=8 depth=24 stencil=0 samples=0
Surface Size: 800x600 windowed
=======================================================
[build] use-vbo=false: FPS: 274 FrameTime: 3.654 ms
[build] use-vbo=true: FPS: 6417 FrameTime: 0.156 ms
[texture] texture-filter=nearest: FPS: 5915 FrameTime: 0.169 ms
[texture] texture-filter=linear: FPS: 5875 FrameTime: 0.170 ms
[texture] texture-filter=mipmap: FPS: 5975 FrameTime: 0.167 ms
[shading] shading=gouraud: FPS: 5677 FrameTime: 0.176 ms
[shading] shading=blinn-phong-inf: FPS: 5732 FrameTime: 0.174 ms
[shading] shading=phong: FPS: 5707 FrameTime: 0.175 ms
[shading] shading=cel: FPS: 5722 FrameTime: 0.175 ms
[bump] bump-render=high-poly: FPS: 5592 FrameTime: 0.179 ms
[bump] bump-render=normals: FPS: 6454 FrameTime: 0.155 ms
[bump] bump-render=height: FPS: 6291 FrameTime: 0.159 ms
[effect2d] kernel=0,1,0;1,-4,1;0,1,0;: FPS: 4583 FrameTime: 0.218 ms
[effect2d] kernel=1,1,1,1,1;1,1,1,1,1;1,1,1,1,1;: FPS: 4458 FrameTime: 0.224 ms
[pulsar] light=false:quads=5:texture=false: FPS: 4218 FrameTime: 0.237 ms
[desktop] blur-radius=5:effect=blur:passes=1:separable=true:windows=4: FPS: 1745 FrameTime: 0.573 ms
[desktop] effect=shadow:windows=4: FPS: 1747 FrameTime: 0.573 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 137 FrameTime: 7.303 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=subdata: FPS: 129 FrameTime: 7.802 ms
[buffer] columns=200:interleave=true:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 187 FrameTime: 5.353 ms
[ideas] speed=duration: FPS: 1002 FrameTime: 0.999 ms
[jellyfish] <default>: FPS: 2403 FrameTime: 0.416 ms
[terrain] <default>: FPS: 843 FrameTime: 1.187 ms
[shadow] <default>: FPS: 3906 FrameTime: 0.256 ms
[refract] <default>: FPS: 1982 FrameTime: 0.505 ms
[conditionals] fragment-steps=0:vertex-steps=0: FPS: 5403 FrameTime: 0.185 ms
[conditionals] fragment-steps=5:vertex-steps=0: FPS: 5359 FrameTime: 0.187 ms
[conditionals] fragment-steps=0:vertex-steps=5: FPS: 5421 FrameTime: 0.184 ms
[function] fragment-complexity=low:fragment-steps=5: FPS: 5377 FrameTime: 0.186 ms
[function] fragment-complexity=medium:fragment-steps=5: FPS: 5367 FrameTime: 0.186 ms
[loop] fragment-loop=false:fragment-steps=5:vertex-steps=5: FPS: 5335 FrameTime: 0.187 ms
[loop] fragment-steps=5:fragment-uniform=false:vertex-steps=5: FPS: 5342 FrameTime: 0.187 ms
[loop] fragment-steps=5:fragment-uniform=true:vertex-steps=5: FPS: 5317 FrameTime: 0.188 ms
=======================================================
glmark2 Score: 4116
=======================================================

@kas1e

Major issue:

If I copy headers from Mesa archive, it breaks both MiniGL and OGLES2 builds. When including our legacy GL/gl.h or GLES2/gl2.h, they normally include corresponding MiniGL and OGLES2 headers.

Minor issue:

Maybe mesa.library headers should be under SDK:local/common?

Thanks. Tried several reboot and repeats, no change. No the RX590 offers no speed increase in any graphics subsystem. I tried ogles and warp3dnova programs.
I have tried this RX590 using all previous betas for RadeonRX.chip. Didn’t really test it much as the fans always ran at full speed so I abandoned it.

@ddni
Just for sake of tests try to reboot and repeat. And also check something via old ogles2.library/warp3dnova in terms of speed : how it bettween rx550 and your rx590, is there same strange slow-fest issue with rx590, or it happens only with mesa ? Maybe beta of RadeonRX.chip 2.23 behave somehow .. Thanks!

@kas1e

Yay! It worked! I now have more questions. If I want to do more complex OpenGL programming, do I need to use a loader (like glad or glew)? If I can't use one of those,
do I need to access the gl functions manually? If I have to access manually, can I use SDL_GL_GetProcAddress? How have you been doing it?

Thank you!

Kevin

Wayfarer is built with Objective-C, with a newer version of MUI that supports modules built with that language. Both of them are not available on AmigaOS 4 right now, and I doubt MUI will be updated in the near future, or even support objective-C for us.

I do not believe that Jaca would care to make a port for OS4 either.

But why do you mention all these? You are not happy with the work we already do to update the browsers on our side?

@328gts

Yes, something is funky. Although the output from radeon_gpu_infor for my rx550 shows similar clocks...?

radeon_gpu_info_probe: BEGIN installed-card report
Board and provider:
graphics.library 54.256
board index 0
board label Radeon RX Polaris12
board driver PCIGraphics.card 53.18
chip driver RadeonRX.chip 2.23
RM resource RadeonRX_RM.resource
RM resource/API 1 / 3
RM chipset 0x00000050 (POLARIS12)
PCI identity:
BDF 0:2:0.0
vendor:device 1002:699f
subsystem 1da2:e367
revision / class c7 / 030000
expansion.library 54.1
GPU topology:
decoded family POLARIS12
raw GFX IP 8.1
internal/external revision 0 / 0x64
active CUs 8
derived shader lanes 512 (64 per GCN CU)
derived texture units 32 (4 per GCN CU)
active render backends 4
derived ROPs 16 (4 per backend)
SE / SA-per-SE / max-CU-SA 2 / 1 / 5
RB-per-SE / tile-pipes / TCC 2 / 4 / 4
active CU mask SE0/SA0 0x0000001e (4 CUs)
active CU mask SE1/SA0 0x0000001e (4 CUs)
enabled RB mask 0x00000000:0000000f
Memory:
GPU-available memory 0x00000000:80000000 bytes (2048 MiB)
CPU-visible bitmap memory 0x00000000:0f0e1400 bytes (240 MiB)
current bitmap memory free 0x00000000:0bf26600 bytes (191 MiB)
largest current free block 0x00000000:0bcb2500 bytes (188 MiB)
VRAM technology unavailable (not exposed)
memory-bus width unavailable (not exposed)
Power management (read-only):
RadeonRX DPM version 1
coarse policy 2 (provider value)
current state 1 (Performance)
available named states 2
state[0] Boot
state[1] Performance
live/max engine clock unavailable (DPM v1 has no sensor)
live/max memory clock unavailable (DPM v1 has no sensor)
state changes not performed
AtomBIOS and clocks:
option ROM bytes 59392
ATOM format/content 1.1
FirmwareInfo format/content 2.2 (108 bytes)
GFX_Info format/content 2.3 (24 bytes)
firmware revision 0x0f320001
default engine clock 214 MHz (Atom boot/default)
default memory clock 300 MHz (Atom boot/default)
reference clock 100 MHz
GPU timestamp crystal 25000 kHz
Register and queue facts:
MC_ARB_RAMCFG 0x000060a2
GB_ADDR_CONFIG 0x22011002
PCIE_EFUSE4 0x07c699f1
SMC CLKPIN CNTL / CNTL_2 0x00000003 / 0x00400009
visible rings 10
GFX / compute / UVD / other 1 / 8 / 1 / 0
radeon_gpu_info_probe: PASS no BO, command, submission, or rendering

@ddni

something definitely not right since my HD7950 is more than 2x faster ?

RX590 seems not to be too happy....
Any thoughts?

==================================
glmark2 2023.01
==================================
OpenGL Information
GL_VENDOR: AMD
GL_RENDERER: POLARIS10 (radeonsi, polaris10, ACO, DRM 0.0, 54.57 (29.7.2023) AmigaOne X1000 release)
GL_VERSION: 4.6 (Compatibility Profile) Mesa 26.1.5 (git-2e7650ee6e)
Surface Config: buf=32 r=8 g=8 b=8 a=8 depth=24 stencil=0 samples=0
Surface Size: 800x600 windowed
=======================================================
[build] use-vbo=false: FPS: 310 FrameTime: 3.229 ms
[build] use-vbo=true: FPS: 1914 FrameTime: 0.523 ms
[texture] texture-filter=nearest: FPS: 1698 FrameTime: 0.589 ms
[texture] texture-filter=linear: FPS: 1684 FrameTime: 0.594 ms
[texture] texture-filter=mipmap: FPS: 1689 FrameTime: 0.592 ms
[shading] shading=gouraud: FPS: 1729 FrameTime: 0.578 ms
[shading] shading=blinn-phong-inf: FPS: 1728 FrameTime: 0.579 ms
[shading] shading=phong: FPS: 1703 FrameTime: 0.587 ms
[shading] shading=cel: FPS: 1709 FrameTime: 0.585 ms
[bump] bump-render=high-poly: FPS: 1587 FrameTime: 0.630 ms
[bump] bump-render=normals: FPS: 1854 FrameTime: 0.539 ms
[bump] bump-render=height: FPS: 1867 FrameTime: 0.536 ms
[effect2d] kernel=0,1,0;1,-4,1;0,1,0;: FPS: 1366 FrameTime: 0.732 ms
[effect2d] kernel=1,1,1,1,1;1,1,1,1,1;1,1,1,1,1;: FPS: 1340 FrameTime: 0.746 ms
[pulsar] light=false:quads=5:texture=false: FPS: 1399 FrameTime: 0.715 ms
[desktop] blur-radius=5:effect=blur:passes=1:separable=true:windows=4: FPS: 934 FrameTime: 1.071 ms
[desktop] effect=shadow:windows=4: FPS: 1165 FrameTime: 0.859 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 192 FrameTime: 5.227 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=subdata: FPS: 135 FrameTime: 7.427 ms
[buffer] columns=200:interleave=true:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 244 FrameTime: 4.104 ms
[ideas] speed=duration: FPS: 837 FrameTime: 1.196 ms
[jellyfish] <default>: FPS: 798 FrameTime: 1.255 ms
[terrain] <default>: FPS: 364 FrameTime: 2.752 ms
[shadow] <default>: FPS: 1476 FrameTime: 0.678 ms
[refract] <default>: FPS: 670 FrameTime: 1.493 ms
[conditionals] fragment-steps=0:vertex-steps=0: FPS: 1575 FrameTime: 0.635 ms
[conditionals] fragment-steps=5:vertex-steps=0: FPS: 1555 FrameTime: 0.643 ms
[conditionals] fragment-steps=0:vertex-steps=5: FPS: 1575 FrameTime: 0.635 ms
[function] fragment-complexity=low:fragment-steps=5: FPS: 1575 FrameTime: 0.635 ms
[function] fragment-complexity=medium:fragment-steps=5: FPS: 1562 FrameTime: 0.641 ms
[loop] fragment-loop=false:fragment-steps=5:vertex-steps=5: FPS: 1567 FrameTime: 0.638 ms
[loop] fragment-steps=5:fragment-uniform=false:vertex-steps=5: FPS: 1574 FrameTime: 0.635 ms
[loop] fragment-steps=5:fragment-uniform=true:vertex-steps=5: FPS: 1575 FrameTime: 0.635 ms
=======================================================
glmark2 Score: 1300
=======================================================

Seems stuck with low clocks?
radeon_gpu_info_probe: BEGIN installed-card report
Board and provider:
graphics.library 54.256
board index 0
board label Radeon RX Polaris10
board driver PCIGraphics.card 53.18
chip driver RadeonRX.chip 2.23
RM resource RadeonRX_RM.resource
RM resource/API 1 / 3
RM chipset 0x0000004e (POLARIS10)
PCI identity:
BDF 0:2:0.0
vendor:device 1002:67df
subsystem 148c:2391
revision / class e1 / 030000
expansion.library 54.1
GPU topology:
decoded family POLARIS10
raw GFX IP 8.1
internal/external revision 1 / 0x51
active CUs 36
derived shader lanes 2304 (64 per GCN CU)
derived texture units 144 (4 per GCN CU)
active render backends 8
derived ROPs 32 (4 per backend)
SE / SA-per-SE / max-CU-SA 4 / 1 / 9
RB-per-SE / tile-pipes / TCC 2 / 8 / 8
active CU mask SE0/SA0 0x000001ff (9 CUs)
active CU mask SE1/SA0 0x000001ff (9 CUs)
active CU mask SE2/SA0 0x000001ff (9 CUs)
active CU mask SE3/SA0 0x000001ff (9 CUs)
enabled RB mask 0x00000000:000000ff
Memory:
GPU-available memory 0x00000002:00000000 bytes (8192 MiB)
CPU-visible bitmap memory 0x00000000:0e6d1400 bytes (230 MiB)
current bitmap memory free 0x00000000:0952f800 bytes (149 MiB)
largest current free block 0x00000000:08fd2f00 bytes (143 MiB)
VRAM technology unavailable (not exposed)
memory-bus width unavailable (not exposed)
Power management (read-only):
RadeonRX DPM version 1
coarse policy 2 (provider value)
current state 1 (Performance)
available named states 2
state[0] Boot
state[1] Performance
live/max engine clock unavailable (DPM v1 has no sensor)
live/max memory clock unavailable (DPM v1 has no sensor)
state changes not performed
AtomBIOS and clocks:
option ROM bytes 59392
ATOM format/content 1.1
FirmwareInfo format/content 2.2 (108 bytes)
GFX_Info format/content 2.3 (24 bytes)
firmware revision 0x0f320300
default engine clock 300 MHz (Atom boot/default)
default memory clock 400 MHz (Atom boot/default)
reference clock 100 MHz
GPU timestamp crystal 25000 kHz
Register and queue facts:
MC_ARB_RAMCFG 0x000060a2
GB_ADDR_CONFIG 0x22011003
PCIE_EFUSE4 0x11e67df1
SMC CLKPIN CNTL / CNTL_2 0x00000003 / 0x00400009
visible rings 10
GFX / compute / UVD / other 1 / 8 / 1 / 0
radeon_gpu_info_probe: PASS no BO, command, submission, or rendering

@amiga_os
You don't need to post mesa changelogs, i of course know about and follow it closely : ) When there will be needs i will update to newer version, but so far there no big reasons for right now while current port its not stable enough.

@all
Issue which 328gts had fixed : he's card have some stale option-rom checksum error in VBIOS, but my ATOM parser were strict about , so now tolerate it. This may help other cards whish may have this issue. While as tests shows, it was just one this card, but who know. Fix will be in next mesa update

HD7950-X1000
=======================================================
glmark2 2023.01
=======================================================
OpenGL Information
GL_VENDOR: AMD
GL_RENDERER: TAHITI (radeonsi, tahiti, ACO, RM 3.0, 54.57 (29.7.2023) AmigaOne X1000 release)
GL_VERSION: 4.5 (Compatibility Profile) Mesa 26.1.5 (git-2e7650ee6e)
Surface Config: buf=32 r=8 g=8 b=8 a=8 depth=24 stencil=0 samples=0
Surface Size: 800x600 windowed
=======================================================
[build] use-vbo=false: FPS: 310 FrameTime: 3.234 ms
[build] use-vbo=true: FPS: 5017 FrameTime: 0.199 ms
[texture] texture-filter=nearest: FPS: 4024 FrameTime: 0.249 ms
[texture] texture-filter=linear: FPS: 4038 FrameTime: 0.248 ms
[texture] texture-filter=mipmap: FPS: 3797 FrameTime: 0.263 ms
[shading] shading=gouraud: FPS: 4122 FrameTime: 0.243 ms
[shading] shading=blinn-phong-inf: FPS: 4069 FrameTime: 0.246 ms
[shading] shading=phong: FPS: 3971 FrameTime: 0.252 ms
[shading] shading=cel: FPS: 3962 FrameTime: 0.252 ms
[bump] bump-render=high-poly: FPS: 3280 FrameTime: 0.305 ms
[bump] bump-render=normals: FPS: 4804 FrameTime: 0.208 ms
[bump] bump-render=height: FPS: 4350 FrameTime: 0.230 ms
[effect2d] kernel=0,1,0;1,-4,1;0,1,0;: FPS: 3509 FrameTime: 0.285 ms
[effect2d] kernel=1,1,1,1,1;1,1,1,1,1;1,1,1,1,1;: FPS: 3506 FrameTime: 0.285 ms
[pulsar] light=false:quads=5:texture=false: FPS: 2583 FrameTime: 0.387 ms
[desktop] blur-radius=5:effect=blur:passes=1:separable=true:windows=4: FPS: 927 FrameTime: 1.079 ms
[desktop] effect=shadow:windows=4: FPS: 1102 FrameTime: 0.908 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 181 FrameTime: 5.552 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=subdata: FPS: 124 FrameTime: 8.117 ms
[buffer] columns=200:interleave=true:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 236 FrameTime: 4.238 ms
[ideas] speed=duration: FPS: 782 FrameTime: 1.280 ms
[jellyfish] <default>: FPS: 1557 FrameTime: 0.643 ms
[terrain] <default>: FPS: 366 FrameTime: 2.736 ms
[shadow] <default>: FPS: 2124 FrameTime: 0.471 ms
[refract] <default>: FPS: 823 FrameTime: 1.216 ms
[conditionals] fragment-steps=0:vertex-steps=0: FPS: 4086 FrameTime: 0.245 ms
[conditionals] fragment-steps=5:vertex-steps=0: FPS: 4057 FrameTime: 0.247 ms
[conditionals] fragment-steps=0:vertex-steps=5: FPS: 4086 FrameTime: 0.245 ms
[function] fragment-complexity=low:fragment-steps=5: FPS: 4062 FrameTime: 0.246 ms
[function] fragment-complexity=medium:fragment-steps=5: FPS: 4066 FrameTime: 0.246 ms
[loop] fragment-loop=false:fragment-steps=5:vertex-steps=5: FPS: 4052 FrameTime: 0.247 ms
[loop] fragment-steps=5:fragment-uniform=false:vertex-steps=5: FPS: 4063 FrameTime: 0.246 ms
[loop] fragment-steps=5:fragment-uniform=true:vertex-steps=5: FPS: 4043 FrameTime: 0.247 ms
=======================================================
glmark2 Score: 2910
=======================================================