Next Generation Emulation banner
Cancel
Create thread New Forums

Benchmarking Framebuffer Access

Jump to Latest
21 - 40 of 48 Posts
Discussion starter · #21 ·
Originally posted by Dark Watcher
good idea....only good for those that have Xeno though ..hmm...

By the way...our D3D results were crap becuase we had 4XS enables *slaps forehead*...bah..it worked like crap anyways :)
Click to expand...
For now, but I'll do some tests with other games when I have time.

[]s Badaro
 
#22 ·
I think you must have pbs with your card... or the AGP bus (?) or some hidden apps running in background.

I allready got very poor perfs with that card first time I bought it (I would say at 1st try I wanted to play that game and some others).
And in fact I discovered that a 'hidden window' of my favorite mass downloader was the one creating that unexpected slowdowns.

Since that time I deacivate all the uneeded processes (services), all that junks on the screen and all that pgms running in the taskbar and belive me or not, I never had that kind of problems or slowdowns.

FYI: My AIW8500DV is overclocked to 285/415 Mhz
 
#23 ·
Badaro, can you upload your save state somewhere so I can try out Xenogears and see how it works on my PC? I remember that when I played through that bit it was running at close to full speed but I can't remember if that was because I was using FVP or not.
 
Discussion starter · #24 · (Edited)
Originally posted by chp
I think you must have pbs with your card... or the AGP bus (?) or some hidden apps running in background.
Click to expand...
I can tell you for sure there are no applications running in the background, this is a 2-day-old clean install of Windows XP. But it's quite possible some driver or settings combination are the source of my problems.

What OS and drivers are you running? Have you used any tweakers for your card?

Originally posted by Phoenix Flame
Badaro, can you upload your save state somewhere so I can try out Xenogears and see how it works on my PC? I remember that when I played through that bit it was running at close to full speed but I can't remember if that was because I was using FVP or not.
Click to expand...
PM me with your e-mail addess I'll send you a link to the savestate.

[]s Badaro
 
#25 ·
Ok, I've tried the SS out and it comes out in between 37 and 44FPS, that's using the config you posted. It just seems that PC graphics hardware isn't up to the task of rendering complex FB effects like the PSX and PS2 are capable of. Either that or it's down to (a) crap driver support or (b) the coding of the GPu plugin, in which case it's up to Pete.

In fact I'm interested to know how exactly Pete's FB code works, and whether or not it's optimized for specific APIs/graphics hardware. Pete, are you reading this?
 
#26 ·
hello!

I could seem stupid (as sometimes I'm :)), but I couldn't find the Pete's benchmark utility on his site and I wasn't there last w/e to follow the "testing protocol" (I mean, to make test w/ my gpu, a SiS 650). But, anyway, I'd like to make the tests, so could anyone tell me where I can find (or attach...) the Pete's benchmark utility?

Thx, bye
 
#28 ·

Pete's texture/framebuffer speed test
Mail: BlackDove@addcom.de
Homepage: http://home.t-online.de/home/PeteBernert/

Vendor: ATI Technologies Inc.
Version: 1.3.3561 WinXP Release
Renderer: Radeon 9700 Pro x86/MMX/3DNow!/SSE

---------------------------------
OpenGL extensions:
GL_ARB_multitexture GL_EXT_texture_env_add GL_EXT_compiled_vertex_array GL_S3_s3tc GL_ARB_depth_texture GL_ARB_fragment_program GL_ARB_multisample GL_ARB_point_parameters GL_ARB_shadow GL_ARB_shadow_ambient GL_ARB_texture_border_clamp GL_ARB_texture_compression GL_ARB_texture_cube_map GL_ARB_texture_env_add GL_ARB_texture_env_combine GL_ARB_texture_env_crossbar GL_ARB_texture_env_dot3 GL_ARB_texture_mirrored_repeat GL_ARB_transpose_matrix GL_ARB_vertex_blend GL_ARB_vertex_program GL_ARB_window_pos GL_ATI_draw_buffers GL_ATI_element_array GL_ATI_envmap_bumpmap GL_ATI_fragment_shader GL_ATI_map_object_buffer GL_ATI_separate_stencil GL_ATI_texture_env_combine3 GL_ATI_texture_float GL_ATI_texture_mirror_once GL_ATI_vertex_array_object GL_ATI_vertex_attrib_array_object GL_ATI_vertex_streams GL_ATIX_texture_env_combine3 GL_ATIX_texture_env_route GL_ATIX_vertex_shader_output_point_size GL_EXT_abgr GL_EXT_bgra GL_EXT_blend_color GL_EXT_blend_func_separate GL_EXT_blend_minmax GL_EXT_blend_subtract GL_EXT_clip_volume_hint GL_EXT_draw_range_elements GL_EXT_fog_coord GL_EXT_multi_draw_arrays GL_EXT_packed_pixels GL_EXT_point_parameters GL_EXT_rescale_normal GL_EXT_secondary_color GL_EXT_separate_specular_color GL_EXT_stencil_wrap GL_EXT_texgen_reflection GL_EXT_texture3D GL_EXT_texture_compression_s3tc GL_EXT_texture_cube_map GL_EXT_texture_edge_clamp GL_EXT_texture_env_combine GL_EXT_texture_env_dot3 GL_EXT_texture_filter_anisotropic GL_EXT_texture_lod_bias GL_EXT_texture_object GL_EXT_texture_rectangle GL_EXT_vertex_array GL_EXT_vertex_shader GL_HP_occlusion_test GL_KTX_buffer_region GL_NV_texgen_reflection GL_NV_blend_square GL_NV_occlusion_query GL_SGI_color_matrix GL_SGI_texture_edge_clamp GL_SGIS_texture_border_clamp GL_SGIS_texture_lod GL_SGIS_generate_mipmap GL_SGIS_multitexture GL_SUN_multi_draw_arrays GL_WIN_swap_hint WGL_EXT_extensions_string WGL_EXT_swap_control
---------------------------------

Texture uploading speed
---------------------------------

Format: 01 - Size: R=8 G=8 B=8 A=8
Speed: 105.6 Mtex/s 422.401 MB/s

Format: 02 - Size: R=8 G=8 B=8 A=8
Speed: 108.309 Mtex/s 433.237 MB/s

Format: 03 - Size: R=8 G=8 B=8 A=8
Speed: 107.648 Mtex/s 430.591 MB/s

Format: 04 - Size: R=8 G=8 B=8 A=8
Speed: 93.3888 Mtex/s 373.555 MB/s

Format: 05 - Size: R=8 G=8 B=8 A=8
Speed: 92.5368 Mtex/s 370.147 MB/s

Format: 06 - Size: R=8 G=8 B=8 A=8
Speed: 93.9131 Mtex/s 375.652 MB/s

Format: 07 - Size: R=8 G=8 B=8 A=8
Speed: 93.7602 Mtex/s 375.041 MB/s

Format: 08 - Size: R=8 G=8 B=8 A=8
Speed: 93.6728 Mtex/s 374.691 MB/s

Format: 09 - Size: R=8 G=8 B=8 A=8
Speed: 93.3888 Mtex/s 373.555 MB/s

Format: 10 - Size: R=5 G=5 B=5 A=1
Speed: 133.693 Mtex/s 534.774 MB/s

Format: 11 - Size: R=5 G=5 B=5 A=1
Speed: 300.67 Mtex/s 601.341 MB/s

Format: 12 - Size: R=4 G=4 B=4 A=4
Speed: 141.536 Mtex/s 566.144 MB/s

Format: 13 - Size: R=4 G=4 B=4 A=4
Speed: 9.80855 Mtex/s 19.6171 MB/s

Format: 14 - Size: R=8 G=8 B=8 A=8
Speed: 97.8452 Mtex/s 293.536 MB/s



Framebuffer texture speed (back)
---------------------------------

Format: 01 - Size: R=8 G=8 B=8 A=8
Speed: -572.501 Mtex/s -2290 MB/s

Format: 02 - Size: R=8 G=8 B=8 A=8
Speed: -574.773 Mtex/s -2299.09 MB/s

Format: 03 - Size: R=8 G=8 B=8 A=8
Speed: -574.445 Mtex/s -2297.78 MB/s

Format: 04 - Size: R=8 G=8 B=8 A=8
Speed: -574.489 Mtex/s -2297.95 MB/s

Format: 05 - Size: R=8 G=8 B=8 A=8
Speed: -574.947 Mtex/s -2299.79 MB/s

Format: 06 - Size: R=8 G=8 B=8 A=8
Speed: -574.51 Mtex/s -2298.04 MB/s

Format: 07 - Size: R=8 G=8 B=8 A=8
Speed: -579.535 Mtex/s -2318.14 MB/s

Format: 08 - Size: R=8 G=8 B=8 A=8
Speed: -575.253 Mtex/s -2301.01 MB/s

Format: 09 - Size: R=8 G=8 B=8 A=8
Speed: -579.032 Mtex/s -2316.13 MB/s

Format: 10 - Size: R=5 G=5 B=5 A=1
Speed: -445.099 Mtex/s -1780.39 MB/s

Format: 11 - Size: R=5 G=5 B=5 A=1
Speed: -450.036 Mtex/s -900.071 MB/s

Format: 12 - Size: R=4 G=4 B=4 A=4
Speed: -445.383 Mtex/s -1781.53 MB/s

Format: 13 - Size: R=4 G=4 B=4 A=4
Speed: -452.177 Mtex/s -904.353 MB/s

Format: 14 - Size: R=8 G=8 B=8 A=8
Speed: -574.379 Mtex/s -1723.14 MB/s



Framebuffer texture speed (front)
---------------------------------

Format: 01 - Size: R=8 G=8 B=8 A=8
Speed: -590.152 Mtex/s -2360.61 MB/s

Format: 02 - Size: R=8 G=8 B=8 A=8
Speed: -585.324 Mtex/s -2341.3 MB/s

Format: 03 - Size: R=8 G=8 B=8 A=8
Speed: -585.739 Mtex/s -2342.96 MB/s

Format: 04 - Size: R=8 G=8 B=8 A=8
Speed: -585.171 Mtex/s -2340.68 MB/s

Format: 05 - Size: R=8 G=8 B=8 A=8
Speed: -585.63 Mtex/s -2342.52 MB/s

Format: 06 - Size: R=8 G=8 B=8 A=8
Speed: -589.627 Mtex/s -2358.51 MB/s

Format: 07 - Size: R=8 G=8 B=8 A=8
Speed: -585.171 Mtex/s -2340.68 MB/s

Format: 08 - Size: R=8 G=8 B=8 A=8
Speed: -585.477 Mtex/s -2341.91 MB/s

Format: 09 - Size: R=8 G=8 B=8 A=8
Speed: -589.671 Mtex/s -2358.68 MB/s

Format: 10 - Size: R=5 G=5 B=5 A=1
Speed: -464.956 Mtex/s -1859.82 MB/s

Format: 11 - Size: R=5 G=5 B=5 A=1
Speed: -464.454 Mtex/s -928.907 MB/s

Format: 12 - Size: R=4 G=4 B=4 A=4
Speed: -469.849 Mtex/s -1879.4 MB/s

Format: 13 - Size: R=4 G=4 B=4 A=4
Speed: -461.789 Mtex/s -923.577 MB/s

Format: 14 - Size: R=8 G=8 B=8 A=8
Speed: -584.821 Mtex/s -1754.46 MB/s
 
#29 · (Edited)
Originally posted by Badaro


I can tell you for sure there are no applications running in the background, this is a 2-day-old clean install of Windows XP. But it's quite possible some driver or settings combination are the source of my problems.
Click to expand...
I use WinXP Pro (French) with lot of "anoying" services deactivated (specialy all these reporting & recovery services).
My ATI drivers are the latests Catalyst 3.1 (6292).
For the tweaker I use Rage3D tweaker 3.8 availaible at www.rage3d.com.
All that based on an MSI KT4 Ultra..
 
#30 ·
ok, found Pete's benchmarks... here's my results, as u can see, I don't have a powerful gpu, in fact I don't use Pete's OGL, but the D3D (DX7) one w/ good results (for me)...

=====
Pete's texture/framebuffer speed test
Mail: BlackDove@addcom.de
Homepage: http://home.t-online.de/home/PeteBernert/

Vendor: SiS
Version: 1.2.1
Renderer: 650/M650 VGA / MMX/SSE2 /AGP

---------------------------------
OpenGL extensions:
GL_ARB_depth_texture GL_ARB_multitexture GL_ARB_point_parameters GL_ARB_shadow GL_ARB_texture_border_clamp GL_ARB_texture_compression GL_ARB_texture_cube_map GL_ARB_texture_env_add GL_ARB_texture_env_combine GL_ARB_texture_env_crossbar GL_ARB_texture_env_dot3 GL_ARB_texture_mirrored_repeat GL_ARB_transpose_matrix GL_ARB_vertex_program GL_ARB_window_pos GL_EXT_abgr GL_EXT_bgra GL_EXT_blend_color GL_EXT_blend_minmax GL_EXT_blend_subtract GL_EXT_compiled_vertex_array GL_EXT_draw_range_elements GL_EXT_fog_coord GL_EXT_multi_draw_arrays GL_EXT_packed_pixels GL_EXT_polygon_offset GL_EXT_rescale_normal GL_EXT_secondary_color GL_EXT_separate_specular_color GL_EXT_shadow_funcs GL_EXT_stencil_wrap GL_EXT_swap_control GL_EXT_texture3D GL_EXT_texture_edge_clamp GL_EXT_texture_env_add GL_EXT_texture_env_combine GL_EXT_texture_env_dot3 GL_EXT_texture_lod_bias GL_EXT_texture_object GL_EXT_vertex_array GL_KTX_buffer_region GL_NV_blend_square GL_SGIS_multitexture GL_SGIS_texture_edge_clamp GL_SGIS_texture_lod GL_SGIX_depth_texture GL_SGIX_shadow GL_SUN_multi_draw_arrays GL_WIN_swap_hint WGL_ARB_extensions_string WGL_EXT_extensions_string WGL_ARB_pixel_format WGL_EXT_pixel_format WGL_EXT_swap_control
---------------------------------

Texture uploading speed
---------------------------------

Format: 01 - Size: R=8 G=8 B=8 A=8
Speed: 43.4871 Mtex/s 173.948 MB/s

Format: 02 - Size: R=8 G=8 B=8 A=8
Speed: 42.9562 Mtex/s 171.825 MB/s

Format: 03 - Size: R=8 G=8 B=8 A=8
Speed: 43.9161 Mtex/s 175.664 MB/s

Format: 04 - Size: R=8 G=8 B=8 A=8
Speed: 5.67034 Mtex/s 22.6813 MB/s

Format: 05 - Size: R=8 G=8 B=8 A=8
Speed: 5.60677 Mtex/s 22.4271 MB/s

Format: 06 - Size: R=8 G=8 B=8 A=8
Speed: 5.69404 Mtex/s 22.7762 MB/s

Format: 07 - Size: R=8 G=8 B=8 A=8
Speed: 5.62672 Mtex/s 22.5069 MB/s

Format: 08 - Size: R=8 G=8 B=8 A=8
Speed: 5.58496 Mtex/s 22.3398 MB/s

Format: 09 - Size: R=8 G=8 B=8 A=8
Speed: 5.60677 Mtex/s 22.4271 MB/s

Format: 10 - Size: R=8 G=8 B=8 A=8
Speed: 43.0509 Mtex/s 172.204 MB/s

Format: 11 - Size: R=8 G=8 B=8 A=8
Speed: 5.38861 Mtex/s 10.7772 MB/s

Format: 12 - Size: R=4 G=4 B=4 A=4
Speed: 46.2068 Mtex/s 184.827 MB/s

Format: 13 - Size: R=4 G=4 B=4 A=4
Speed: 5.80312 Mtex/s 11.6062 MB/s

Format: 14 - Size: R=8 G=8 B=8 A=0
Speed: 52.516 Mtex/s 157.548 MB/s



Framebuffer texture speed (back)
---------------------------------

Format: 01 - Size: R=8 G=8 B=8 A=8
Speed: 1.98528 Mtex/s 7.94111 MB/s

Format: 02 - Size: R=8 G=8 B=8 A=8
Speed: 1.99382 Mtex/s 7.97528 MB/s

Format: 03 - Size: R=8 G=8 B=8 A=8
Speed: 1.99382 Mtex/s 7.97528 MB/s

Format: 04 - Size: R=8 G=8 B=8 A=8
Speed: 1.97215 Mtex/s 7.88859 MB/s

Format: 05 - Size: R=8 G=8 B=8 A=8
Speed: 1.98462 Mtex/s 7.93847 MB/s

Format: 06 - Size: R=8 G=8 B=8 A=8
Speed: 1.97869 Mtex/s 7.91477 MB/s

Format: 07 - Size: R=8 G=8 B=8 A=8
Speed: 1.98528 Mtex/s 7.94111 MB/s

Format: 08 - Size: R=8 G=8 B=8 A=8
Speed: 1.99316 Mtex/s 7.97264 MB/s

Format: 09 - Size: R=8 G=8 B=8 A=8
Speed: 1.97869 Mtex/s 7.91477 MB/s

Format: 10 - Size: R=8 G=8 B=8 A=8
Speed: 1.98528 Mtex/s 7.94111 MB/s

Format: 11 - Size: R=8 G=8 B=8 A=8
Speed: 1.98462 Mtex/s 3.96924 MB/s

Format: 12 - Size: R=4 G=4 B=4 A=4
Speed: 2.04392 Mtex/s 8.17569 MB/s

Format: 13 - Size: R=4 G=4 B=4 A=4
Speed: 2.04392 Mtex/s 4.08785 MB/s

Format: 14 - Size: R=8 G=8 B=8 A=0
Speed: 2.12315 Mtex/s 6.36945 MB/s



Framebuffer texture speed (front)
---------------------------------

Format: 01 - Size: R=8 G=8 B=8 A=8
Speed: 1.97869 Mtex/s 7.91477 MB/s

Format: 02 - Size: R=8 G=8 B=8 A=8
Speed: 1.98528 Mtex/s 7.94111 MB/s

Format: 03 - Size: R=8 G=8 B=8 A=8
Speed: 1.97804 Mtex/s 7.91214 MB/s

Format: 04 - Size: R=8 G=8 B=8 A=8
Speed: 1.97215 Mtex/s 7.88859 MB/s

Format: 05 - Size: R=8 G=8 B=8 A=8
Speed: 1.97869 Mtex/s 7.91477 MB/s

Format: 06 - Size: R=8 G=8 B=8 A=8
Speed: 1.97804 Mtex/s 7.91214 MB/s

Format: 07 - Size: R=8 G=8 B=8 A=8
Speed: 1.96565 Mtex/s 7.86259 MB/s

Format: 08 - Size: R=8 G=8 B=8 A=8
Speed: 1.97869 Mtex/s 7.91477 MB/s

Format: 09 - Size: R=8 G=8 B=8 A=8
Speed: 1.965 Mtex/s 7.86 MB/s

Format: 10 - Size: R=8 G=8 B=8 A=8
Speed: 1.97215 Mtex/s 7.88859 MB/s

Format: 11 - Size: R=8 G=8 B=8 A=8
Speed: 1.98528 Mtex/s 3.97056 MB/s

Format: 12 - Size: R=4 G=4 B=4 A=4
Speed: 2.03649 Mtex/s 8.14596 MB/s

Format: 13 - Size: R=4 G=4 B=4 A=4
Speed: 2.02891 Mtex/s 4.05782 MB/s

Format: 14 - Size: R=8 G=8 B=8 A=0
Speed: 2.12385 Mtex/s 6.37156 MB/s
=====

Thx, bye
 
#32 ·
to chp: just a little question, have u read my sign? I've a notebook (cannot change my gpu at all) and anyway I can play well w/ all the games I've at 640x480 (even at 800x600, but I prefer 640x480)! ;)

Here my specs:
=====
Plugin: Pete's D3D Driver 1.1.68
Author: Pete Bernert
GFX card: SiS M650_651_650_740

Resolution/Color:
- 640x480 Window mode

Textures:
- R5G5B5A1
- Filtering: 6
- Hi-Res textures: on
- Palettized tex windows: off
- Garbage collection: on
- VRam size: 0 MBytes - 102 textures usable

Framerate:
- FPS limit: on
- Frame skipping: off
- FPS limit: Auto

Compatibility:
- Offscreen drawing: 3
- Framebuffer texture: 3
- Framebuffer access: 3
- Alpha multipass: on
- Mask bit: on
- Advanced blending: hardware

Misc:
- Scanlines: off [0]
- Unfiltered FB: off
- Dithering: on
- Screen smoothing: off
- Full vram: off
- Game fixes: off [00000000]
=====

BTW, some shots...

Thx, bye

PS this post of mines is NO OFFENSIVE at all, chp, I mean, if I could choose my GPU, I'd like an ATI 9800 :D, but I can't and I "survive" w/ a good (for the work it does) SiS 650 ;)
 
Discussion starter · #33 ·
Originally posted by chp


I use WinXP Pro (French) with lot of "anoying" services deactivated (specialy all these reporting & recovery services).
My ATI drivers are the latests Catalyst 3.1 (6292).
For the tweaker I use Rage3D tweaker 3.8 availaible at www.rage3d.com.
All that based on an MSI KT4 Ultra..
Click to expand...
Thanks. Have you set any tweaks or are just using it for OC'ing?

[]s Badaro
 
#34 ·
here are mine
Pete's texture/framebuffer speed test
Mail: BlackDove@addcom.de
Homepage: http://home.t-online.de/home/PeteBernert/

Vendor: NVIDIA Corporation
Version: 1.4.0
Renderer: GeForce4 Ti 4200/AGP/3DNOW!

---------------------------------
OpenGL extensions:
GL_ARB_depth_texture GL_ARB_imaging GL_ARB_multisample GL_ARB_multitexture GL_ARB_point_parameters GL_ARB_shadow GL_ARB_texture_border_clamp GL_ARB_texture_compression GL_ARB_texture_cube_map GL_ARB_texture_env_add GL_ARB_texture_env_combine GL_ARB_texture_env_dot3 GL_ARB_texture_mirrored_repeat GL_ARB_transpose_matrix GL_ARB_vertex_program GL_ARB_window_pos GL_S3_s3tc GL_EXT_abgr GL_EXT_bgra GL_EXT_blend_color GL_EXT_blend_minmax GL_EXT_blend_subtract GL_EXT_compiled_vertex_array GL_EXT_draw_range_elements GL_EXT_fog_coord GL_EXT_multi_draw_arrays GL_EXT_packed_pixels GL_EXT_paletted_texture GL_EXT_point_parameters GL_EXT_rescale_normal GL_EXT_secondary_color GL_EXT_separate_specular_color GL_EXT_shadow_funcs GL_EXT_shared_texture_palette GL_EXT_stencil_wrap GL_EXT_texture3D GL_EXT_texture_compression_s3tc GL_EXT_texture_cube_map GL_EXT_texture_edge_clamp GL_EXT_texture_env_add GL_EXT_texture_env_combine GL_EXT_texture_env_dot3 GL_EXT_texture_filter_anisotropic GL_EXT_texture_lod GL_EXT_texture_lod_bias GL_EXT_texture_object GL_EXT_vertex_array GL_HP_occlusion_test GL_IBM_texture_mirrored_repeat GL_KTX_buffer_region GL_NV_blend_square GL_NV_copy_depth_to_color GL_NV_depth_clamp GL_NV_fence GL_NV_fog_distance GL_NV_light_max_exponent GL_NV_multisample_filter_hint GL_NV_occlusion_query GL_NV_packed_depth_stencil GL_NV_pixel_data_range GL_NV_point_sprite GL_NV_register_combiners GL_NV_register_combiners2 GL_NV_texgen_reflection GL_NV_texture_compression_vtc GL_NV_texture_env_combine4 GL_NV_texture_rectangle GL_NV_texture_shader GL_NV_texture_shader2 GL_NV_texture_shader3 GL_NV_vertex_array_range GL_NV_vertex_array_range2 GL_NV_vertex_program GL_NV_vertex_program1_1 GL_NVX_ycrcb GL_SGIS_generate_mipmap GL_SGIS_multitexture GL_SGIS_texture_lod GL_SGIX_depth_texture GL_SGIX_shadow GL_WIN_swap_hint WGL_EXT_swap_control
---------------------------------

Texture uploading speed
---------------------------------

Format: 01 - Size: R=8 G=8 B=8 A=8
Speed: 19.3728 Mtex/s 77.4913 MB/s

Format: 02 - Size: R=8 G=8 B=8 A=8
Speed: 22.0344 Mtex/s 88.1376 MB/s

Format: 03 - Size: R=8 G=8 B=8 A=8
Speed: 21.8526 Mtex/s 87.4104 MB/s

Format: 04 - Size: R=8 G=8 B=8 A=8
Speed: 22.318 Mtex/s 89.2721 MB/s

Format: 05 - Size: R=8 G=8 B=8 A=8
Speed: 22.4707 Mtex/s 89.8829 MB/s

Format: 06 - Size: R=8 G=8 B=8 A=8
Speed: 22.3324 Mtex/s 89.3296 MB/s

Format: 07 - Size: R=8 G=8 B=8 A=8
Speed: 24.4778 Mtex/s 97.9113 MB/s

Format: 08 - Size: R=8 G=8 B=8 A=8
Speed: 24.3469 Mtex/s 97.3877 MB/s

Format: 09 - Size: R=8 G=8 B=8 A=8
Speed: 24.3606 Mtex/s 97.4425 MB/s

Format: 10 - Size: R=5 G=5 B=5 A=1
Speed: 32.0044 Mtex/s 128.018 MB/s

Format: 11 - Size: R=5 G=5 B=5 A=1
Speed: 122.76 Mtex/s 245.52 MB/s

Format: 12 - Size: R=4 G=4 B=4 A=4
Speed: 31.9938 Mtex/s 127.975 MB/s

Format: 13 - Size: R=4 G=4 B=4 A=4
Speed: 122.127 Mtex/s 244.255 MB/s

Format: 14 - Size: R=8 G=8 B=8 A=0
Speed: 22.2308 Mtex/s 66.6923 MB/s

Framebuffer texture speed (back)
---------------------------------

Format: 01 - Size: R=8 G=8 B=8 A=8
Speed: 517.244 Mtex/s 2068.97 MB/s

Format: 02 - Size: R=8 G=8 B=8 A=8
Speed: 516.281 Mtex/s 2065.13 MB/s

Format: 03 - Size: R=8 G=8 B=8 A=8
Speed: 516.587 Mtex/s 2066.35 MB/s

Format: 04 - Size: R=8 G=8 B=8 A=8
Speed: 516.565 Mtex/s 2066.26 MB/s

Format: 05 - Size: R=8 G=8 B=8 A=8
Speed: 516.783 Mtex/s 2067.13 MB/s

Format: 06 - Size: R=8 G=8 B=8 A=8
Speed: 515.431 Mtex/s 2061.72 MB/s

Format: 07 - Size: R=8 G=8 B=8 A=8
Speed: 516.87 Mtex/s 2067.48 MB/s

Format: 08 - Size: R=8 G=8 B=8 A=8
Speed: 516.284 Mtex/s 2065.14 MB/s

Format: 09 - Size: R=8 G=8 B=8 A=8
Speed: 516.829 Mtex/s 2067.32 MB/s

Format: 10 - Size: R=5 G=5 B=5 A=1
Speed: 27.9155 Mtex/s 111.662 MB/s

Format: 11 - Size: R=5 G=5 B=5 A=1
Speed: 27.9684 Mtex/s 55.9369 MB/s

Format: 12 - Size: R=4 G=4 B=4 A=4
Speed: 28.0339 Mtex/s 112.135 MB/s

Format: 13 - Size: R=4 G=4 B=4 A=4
Speed: 27.9373 Mtex/s 55.8746 MB/s

Format: 14 - Size: R=8 G=8 B=8 A=0
Speed: 533.516 Mtex/s 1600.55 MB/s

Framebuffer texture speed (front)
---------------------------------

Format: 01 - Size: R=8 G=8 B=8 A=8
Speed: 515.823 Mtex/s 2063.29 MB/s

Format: 02 - Size: R=8 G=8 B=8 A=8
Speed: 516.26 Mtex/s 2065.04 MB/s

Format: 03 - Size: R=8 G=8 B=8 A=8
Speed: 516.63 Mtex/s 2066.52 MB/s

Format: 04 - Size: R=8 G=8 B=8 A=8
Speed: 516.521 Mtex/s 2066.09 MB/s

Format: 05 - Size: R=8 G=8 B=8 A=8
Speed: 514.867 Mtex/s 2059.47 MB/s

Format: 06 - Size: R=8 G=8 B=8 A=8
Speed: 516.502 Mtex/s 2066.01 MB/s

Format: 07 - Size: R=8 G=8 B=8 A=8
Speed: 515.782 Mtex/s 2063.13 MB/s

Format: 08 - Size: R=8 G=8 B=8 A=8
Speed: 516.543 Mtex/s 2066.17 MB/s

Format: 09 - Size: R=8 G=8 B=8 A=8
Speed: 515.605 Mtex/s 2062.42 MB/s

Format: 10 - Size: R=5 G=5 B=5 A=1
Speed: 27.9902 Mtex/s 111.961 MB/s

Format: 11 - Size: R=5 G=5 B=5 A=1
Speed: 27.9248 Mtex/s 55.8496 MB/s

Format: 12 - Size: R=4 G=4 B=4 A=4
Speed: 28.1336 Mtex/s 112.534 MB/s

Format: 13 - Size: R=4 G=4 B=4 A=4
Speed: 27.9466 Mtex/s 55.8932 MB/s

Format: 14 - Size: R=8 G=8 B=8 A=0
Speed: 534.083 Mtex/s 1602.25 MB/s
 
#35 ·
>In fact I'm interested to know how exactly Pete's
>FB code works, and whether or not it's optimized
>for specific APIs/graphics hardware. Pete, are
>you reading this?

yup, right now I am reading it :)

basically there are three types of vram operations in the plugins:
1) uploading texture data to vram (system memory -> gfx card memory operation)
2) copying part of the gfx's card framebuffer (front or backbuffer) to some gfx's card texture memory (gfx card memory -> gfx card memory operation)
3) reading the gfx's card framebuffer data (gfx card memory -> system memory operation).

The first type is needed, well, to upload the psx textures and the mdec data. That's also the first part of my benchmark output file.

The second type is my plugin's "framebuffer texture" option. The plugin detects if a game is reusing a displayed psx screen as a psx texture, and if yes, it will create a gfx card texture itself, and copy the framebuffer to it.

It's just one OpenGL function call to copy the framebuffer to texture (so hardly to optimize), the only thing which can make a speed difference is the chosen color format of the texture (because, if the color format is not well supported, the gfx card driver would need to download the framebuffer to system memory, convert the colors, and upload it to texture again... and that would be very slow).

That's why I made the benchmark: to see if some gfx cards are having an hard time with the tex color format I had used in the past, and if there are other formats with better performance.

The third operation (framebuffer -> system mem) is what I call my "Framebuffer access" option. Some psx games don't use the psx display as psx texture, but they download the display to the psx main memory, change the data, and upload it again for some fancy effects.

This third mode isn't covered by my benchmark application, since there I have even less influence about the speed (it's also just one OpenGL function call, but no textures are involved this time). I _do_ know that the latest nVidia drivers have a good speed with framebuffer reading. I _think_ that ATI drivers are not optimized for this yet (that's at least what I have seen in various developer forums).

As for the initial question: why is my ATI showing very good results in the benchmark app with framebuffer textures, but when I use them in the plugin, the speed drops dramatically?
Well, basically the benchmark app is doing the same as most psx games, with small differences, of course:
a) the benchmark app is just copying a 256x256 sized part of the framebuffer to texture, while in the plugin the size depends on the gpu window/fullscreen size.
If, for example, you play at a 1024x768 size, I have to create a 1024x1024 texture internally, and your gfx card has to copy a lot of data. To check out if that's the bottleneck, reduce the window soze to 640x480 as a test, and see if it performs better.

b) as I've said: the benchmark app is always copying 256x256 rects. A psx game will maybe copy just a part of the framebuffer, with some odd data alignment, for example a 241x227 rect. It could be well possible that your ATI gfx card driver doesn't like that, while a nVidia one has no problems.

c) a real psx game may do much more drawing commands (including many textures) before it will do a framebuffer texture. The benchmark app doesn't draw that much, as you may have noticed ;) So it could be possible, that the ATI gfx card's pipeline will stall much more in a real game situation.

Well, my initial intention was not to write a benchmark which showed exactly how good or bad the plugin's "Framebuffer texture" option was working, but to see if a certain texture color format has a chance to speed it up.

But there is a very easy way to check out the plugin's framebuffer texture speed in OpenGL, and if the speed depends on your gpu window size: simply run one of your games with "screen smoothing " enabled. That smoothing is doing a framebuffer texture as well each frame. If speed is still playable (>=60 fps), your card should be able to run the psx framebuffer textures at full speed as well.

and don't forget d): FSAA is slowing all framebuffer stuff down, and "framebuffer access" may also be very slow on your system, so turn it off (set it to "0") for a try.
 
#36 ·
Addendum: I just made a small check with Xenogears myself... in the opening sequence that game is using "Framebuffer textures" _and_ "Framebuffer access" as well.

So prolly the FBA is causing the slowdowns on ATI cards, not the FBT. But as I've said in the previous post: I don't think that I have the slightest chance to speed that up.

Just to give you some results: the opening sequence (at 640x480 resolution, 32 bit color depth) is running at around 140 fps with just framebuffer textures enabled (of course then the screen is mostly black), and 90 fps with FBT+FBA enabled (that's on my GF3).
 
#37 ·
Pete's texture/framebuffer speed test
Mail: BlackDove@addcom.de
Homepage: http://home.t-online.de/home/PeteBernert/

Vendor: NVIDIA Corporation
Version: 1.4.0

Geforce2MX400

Renderer: GeForce2 MX/AGP/SSE

---------------------------------
OpenGL extensions:
GL_ARB_imaging GL_ARB_multitexture
GL_ARB_point_parameters
GL_ARB_texture_compression
GL_ARB_texture_cube_map
GL_ARB_texture_env_add
GL_ARB_texture_env_combine
GL_ARB_texture_env_dot3
GL_ARB_texture_mirrored_repeat
GL_ARB_transpose_matrix
GL_ARB_vertex_program
GL_ARB_window_pos
GL_S3_s3tc GL_EXT_abgr
GL_EXT_bgra
GL_EXT_blend_color
GL_EXT_blend_minmax
GL_EXT_blend_subtract
GL_EXT_clip_volume_hint
GL_EXT_compiled_vertex_array
GL_EXT_draw_range_elements
GL_EXT_fog_coord
GL_EXT_multi_draw_arrays
GL_EXT_packed_pixels
GL_EXT_paletted_texture
GL_EXT_point_parameters
GL_EXT_rescale_normal
GL_EXT_secondary_color
GL_EXT_separate_specular_color
GL_EXT_shared_texture_palette
GL_EXT_stencil_wrap
GL_EXT_texture_compression_s3tc
GL_EXT_texture_cube_map
GL_EXT_texture_edge_clamp
GL_EXT_texture_env_add
GL_EXT_texture_env_combine
GL_EXT_texture_env_dot3
GL_EXT_texture_filter_anisotropic
GL_EXT_texture_lod
GL_EXT_texture_lod_bias
GL_EXT_texture_object
GL_EXT_vertex_array
GL_EXT_vertex_weighting
GL_IBM_texture_mirrored_repeat
GL_KTX_buffer_region
GL_NV_blend_square
GL_NV_fence
GL_NV_fog_distance
GL_NV_light_max_exponent
GL_NV_packed_depth_stencil
GL_NV_pixel_data_range
GL_NV_point_sprite
GL_NV_register_combiners
GL_NV_texgen_reflection
GL_NV_texture_env_combine4
GL_NV_texture_rectangle
GL_NV_vertex_array_range
GL_NV_vertex_array_range2
GL_NV_vertex_program
GL_NV_vertex_program1_1
GL_SGIS_generate_mipmap
GL_SGIS_multitexture
GL_SGIS_texture_lod
GL_WIN_swap_hint
WGL_EXT_swap_control
---------------------------------

Texture uploading speed
---------------------------------

Format: 01 - Size: R=4 G=4 B=4 A=4
Speed: 3.17882 Mtex/s 12.7153 MB/s

Format: 02 - Size: R=4 G=4 B=4 A=4
Speed: 3.46637 Mtex/s 13.8655 MB/s

Format: 03 - Size: R=8 G=8 B=8 A=8
Speed: 1.83195 Mtex/s 7.32782 MB/s

Format: 04 - Size: R=4 G=4 B=4 A=4
Speed: 3.05835 Mtex/s 12.2334 MB/s

Format: 05 - Size: R=4 G=4 B=4 A=4
Speed: 3.44124 Mtex/s 13.765 MB/s

Format: 06 - Size: R=8 G=8 B=8 A=8
Speed: 1.97411 Mtex/s 7.89643 MB/s

Format: 07 - Size: R=4 G=4 B=4 A=4
Speed: 3.66391 Mtex/s 14.6556 MB/s

Format: 08 - Size: R=4 G=4 B=4 A=4
Speed: 3.51008 Mtex/s 14.0403 MB/s

Format: 09 - Size: R=8 G=8 B=8 A=8
Speed: 2.03986 Mtex/s 8.15945 MB/s

Format: 10 - Size: R=5 G=5 B=5 A=1
Speed: 3.39655 Mtex/s 13.5862 MB/s

Format: 11 - Size: R=5 G=5 B=5 A=1
Speed: 3.78972 Mtex/s 7.57944 MB/s

Format: 12 - Size: R=4 G=4 B=4 A=4
Speed: 3.62029 Mtex/s 14.4812 MB/s

Format: 13 - Size: R=4 G=4 B=4 A=4
Speed: 3.93216 Mtex/s 7.86432 MB/s

Format: 14 - Size: R=5 G=6 B=5 A=0
Speed: 3.98574 Mtex/s 11.9572 MB/s



Framebuffer texture speed (back)
---------------------------------

Format: 01 - Size: R=4 G=4 B=4 A=4
Speed: 3.79476 Mtex/s 15.1791 MB/s

Format: 02 - Size: R=4 G=4 B=4 A=4
Speed: 3.77798 Mtex/s 15.1119 MB/s

Format: 03 - Size: R=8 G=8 B=8 A=8
Speed: 213.478 Mtex/s 853.91 MB/s

Format: 04 - Size: R=4 G=4 B=4 A=4
Speed: 3.7281 Mtex/s 14.9124 MB/s

Format: 05 - Size: R=4 G=4 B=4 A=4
Speed: 3.73746 Mtex/s 14.9499 MB/s

Format: 06 - Size: R=8 G=8 B=8 A=8
Speed: 213.276 Mtex/s 853.102 MB/s

Format: 07 - Size: R=4 G=4 B=4 A=4
Speed: 3.882 Mtex/s 15.528 MB/s

Format: 08 - Size: R=4 G=4 B=4 A=4
Speed: 3.86912 Mtex/s 15.4765 MB/s

Format: 09 - Size: R=8 G=8 B=8 A=8
Speed: 213.254 Mtex/s 853.017 MB/s

Format: 10 - Size: R=5 G=5 B=5 A=1
Speed: 3.88847 Mtex/s 15.5539 MB/s

Format: 11 - Size: R=5 G=5 B=5 A=1
Speed: 3.87555 Mtex/s 7.7511 MB/s

Format: 12 - Size: R=4 G=4 B=4 A=4
Speed: 3.88847 Mtex/s 15.5539 MB/s

Format: 13 - Size: R=4 G=4 B=4 A=4
Speed: 3.88458 Mtex/s 7.76917 MB/s

Format: 14 - Size: R=5 G=6 B=5 A=0
Speed: 228.055 Mtex/s 684.164 MB/s



Framebuffer texture speed (front)
---------------------------------

Format: 01 - Size: R=4 G=4 B=4 A=4
Speed: 3.84866 Mtex/s 15.3947 MB/s

Format: 02 - Size: R=4 G=4 B=4 A=4
Speed: 3.86019 Mtex/s 15.4408 MB/s

Format: 03 - Size: R=8 G=8 B=8 A=8
Speed: 212.774 Mtex/s 851.094 MB/s

Format: 04 - Size: R=4 G=4 B=4 A=4
Speed: 3.6796 Mtex/s 14.7184 MB/s

Format: 05 - Size: R=4 G=4 B=4 A=4
Speed: 3.65782 Mtex/s 14.6313 MB/s

Format: 06 - Size: R=8 G=8 B=8 A=8
Speed: 213.014 Mtex/s 852.055 MB/s

Format: 07 - Size: R=4 G=4 B=4 A=4
Speed: 3.85378 Mtex/s 15.4151 MB/s

Format: 08 - Size: R=4 G=4 B=4 A=4
Speed: 3.82185 Mtex/s 15.2874 MB/s

Format: 09 - Size: R=8 G=8 B=8 A=8
Speed: 213.898 Mtex/s 855.592 MB/s

Format: 10 - Size: R=5 G=5 B=5 A=1
Speed: 3.86276 Mtex/s 15.451 MB/s

Format: 11 - Size: R=5 G=5 B=5 A=1
Speed: 3.82293 Mtex/s 7.64587 MB/s

Format: 12 - Size: R=4 G=4 B=4 A=4
Speed: 3.83328 Mtex/s 15.3331 MB/s

Format: 13 - Size: R=4 G=4 B=4 A=4
Speed: 3.81426 Mtex/s 7.62853 MB/s

Format: 14 - Size: R=5 G=6 B=5 A=0
Speed: 228.699 Mtex/s 686.096 MB/s


When I see others scores, I think: GF2MX sux
 
Save
Reply Quote Like Like
#38 ·
Originally posted by Pete Bernert
Just to give you some results: the opening sequence (at 640x480 resolution, 32 bit color depth) is running at around 140 fps with just framebuffer textures enabled (of course then the screen is mostly black), and 90 fps with FBT+FBA enabled (that's on my GF3).
Click to expand...
OK, that clears a lot of things up, thanks.

One other thing though - what exactly is the advantage of having FBT running using the vram buffer and using software processing at the same time (option 3)? Is this to get around the fact that there's a limitation in what either the graphics card or the API can do with FBTs? One thing I noticed was that if I used FBT with option 2 then the Xenogears screen runs at around 90FPS, but if I put it on option 3 then it runs slow, somewhere around 49FPS (on my GeForce4 Ti 4200).

Another thing which takes my interest is that FBT or FBA seem to run slower when Keep PSX Aspect Ratio is disabled. When this is disabled the plugin has to resize every frame to fit the output resolution, right? So then if the game calls for FBTs then it has to resize the screen again and upload it to the vram and this is causing the slowdown? I lose about 15FPS when Keep PSX Aspect Ratio is disabled - it doesn't make the game unplayable, it's still around 70FPS but it's a considerable drop for something which isn't supposed to do a lot.
 
#39 ·
>what exactly is the advantage of having FBT
>running using the vram buffer and using software
>processing at the same time (option 3)?

That mode has been developed by Lewpy, since in Glide all the framebuffer stuff seems to be slow. Mode 3 just handles the gfx card framebuffer one time (at the beginning of a special effect), and switches then to full software rendering (FVP) until the effect is done. yeah, in my experience it's faster on Geforce cards to do everything in hardware (mode 2), but sometimes mode 3 has a better compatibility (for example it can show the battle transitions of Legend of Dragoons, which isn't possible in mode 2).

>Another thing which takes my interest is that FBT
>or FBA seem to run slower when Keep PSX
>Aspect Ratio is disabled.

well, the only thing I can think off: when the option is enabled, the screen display usually gets smaller. Therefore the gpu plugin has to copy/read less data when doing FBT/FBA.
 
Discussion starter · #40 ·
Originally posted by Pete Bernert
Addendum: I just made a small check with Xenogears myself... in the opening sequence that game is using "Framebuffer textures" _and_ "Framebuffer access" as well.

So prolly the FBA is causing the slowdowns on ATI cards, not the FBT. But as I've said in the previous post: I don't think that I have the slightest chance to speed that up.

Just to give you some results: the opening sequence (at 640x480 resolution, 32 bit color depth) is running at around 140 fps with just framebuffer textures enabled (of course then the screen is mostly black), and 90 fps with FBT+FBA enabled (that's on my GF3).
Click to expand...
Pete,

Seems like you're right on the spot here. FBA, in particular the "GFX Card Buffer Moves" option is the cause of the problem. I did some extra benchmarking for Xenogears in my 8500:

Xenogears Opening Scene, 1024x768,Texture Format 8888, FBT set to "GFX Card Buffer":

FBA = Emulated VRAM -> 40-50 FPS
FBA = GFX Card Buffer Reads -> 40-50 FPS
FBA = GFX Card Buffer Moves -> 03-05 FPS
FBA = GFX Card Buffer Reads & Moves -> 03-05 FPS
FBA = Full VRAM Primitives -> 25-30 FPS

[]s Badaro
 
21 - 40 of 48 Posts
About this Discussion
47 Replies
13 Participants
Last post:
chp
Next Generation Emulation
posts
2.2M
members
465K
Since
2001
A forum community dedicated to all emulation enthusiasts. Come Join discussion on all platforms from Nintendo, Microsoft Xbox, Sony Playstation, to PC. Coding, tips, builds, specs, tricks and more.
Full Forum Listing
Explore Our Forums
Open Discussion Game/Console Discussion ePSXe Problems with Games No$GBA