Why Dolphin progresses much faster than PCSX2 ?
#21
(09-13-2011, 02:53 AM)Darude Wrote: I approve of this thread.

I disapprove of this comment
<REDACTED>

Sponsored links

#22
(09-13-2011, 03:27 AM)nintendo85 Wrote: Gamecube GPU

162 MHz "Flipper" LSI (co-developed by Nintendo and ArtX, acquired by ATI)
180 nm NEC eDRAM-compatible process
8 GFLOPS
4 pixel pipelines with 1 texture unit each[14]
TEV "Texture EnVironment" engine (similar to Nvidia's GeForce-class "register combiners")
Fixed-function hardware transform and lighting (T&L), 20+ million polygons in-game[17]
648 megapixels/second (162 MHz × 4 pipelines), 648 megatexels/second (648 MP × 1 texture unit) (peak)
Peak triangle performance: 20,250,000 32-pixel triangles/s raw and with 1 texture and lit
337,500 triangles a frame at 60 FPS
675,000 triangles a frame at 30 FPS

8 texture layers per pass, texture compression, full scene anti-aliasing
8 simultaneous hardware light sources
Bilinear, trilinear, and anisotropic texture filtering
Multi-texturing, bump mapping, reflection mapping, 24-bit z-buffer
24-bit RGB/32-bit RGBA color depth
Hardware limitations sometimes require a 6r+6g+6b+6a mode (18-bit color), resulting in color banding.


PS2 GPU

Graphics processing unit: "Graphics Synthesizer" clocked at 147 MHz
Pixel pipelines: 16
Video output resolution: variable from 256x224 to 1280x1024 pixels
4 MB Embedded DRAM video memory bandwidth at 48 gigabytes per second (main system 32 MB can be dedicated into VRAM for off-screen materials)
Texture buffer bandwidth: 9.6 GB/s
Frame buffer bandwidth: 38.4 GB/s
DRAM Bus width: 2560-bit (composed of three independent buses: 1024-bit write, 1024-bit read, 512-bit read/write)
Pixel configuration: RGB: Alpha:Z Buffer (24:8, 15:1 for RGB, 16, 24, or 32-bit Z buffer)
Dedicated connection to: Main CPU and VU1
Overall pixel fillrate: 16x147 = 2.352 Gpixel/s (rounded to 2.4 Gpixel/s)
Pixel fillrate: with no texture, flat shaded 2.4 (75,000,000 32pixel raster triangles)
Pixel fillrate: with 1 full texture (Diffuse Map), Gouraud shaded 1.2 (37,750,000 32-bit pixel raster triangles)
Pixel fillrate: with 2 full textures (Diffuse map + specular or alpha or other), Gouraud shaded 0.6 (18,750,000 32-bit pixel raster triangles)

GS effects: AAx2 (poly sorting required),[47] Bilinear, Trilinear, Multi-pass, Palletizing (4-bit = 6:1 ratio, 8-bit = 3:1)
Multi-pass rendering ability
Four passes = 300 Mpixel/s (300 Mpixels/s divided by 32 pixels = 9,375,000 triangles/s lost every four passes

Just because the PS2 has a higher polygon/vertex count doesn't mean the number stays at 75,000,000 when all lighting and textures are applied to the gourad shaded polygons. In-game vertices are much lower after the GPU renders them into to the game; my understanding is that very few games reached 12,000,000 per second.

PS2 GPU < ATI's Gamecube GPU. Good day.

are you somewhat deficient at reading. the bold lines are higher on the GS.

Peak triangle performance: 20,250,000 32-pixel triangles/s raw and with 1 texture and lit
vs
Pixel fillrate: with 1 full texture (Diffuse Map), Gouraud shaded 1.2 (37,750,000 32-bit pixel raster triangles)

thats 20 million vs 37 million.

Quote:Not what I heard. The GPU and CPU were much more powerful on the gamecube. Why do character movements in FFX shake like jittery caffeine addicts yet they moved smoothly in Resident Evil 4?
on).

In your head.

Quote:Why did Soul Caliber 2 look and have a smoother/better framerate than the PS2 version?

Developer discretion

Quote:If anything, the Gamcube was indeed capable of reaching near-Xbox levels in terms of graphics. Having a 1.5GB limit on games didn't help.

Not even in the slightest. the Gamecube has a fixed function gpu that has to switch back to CPU preparation for certain texture updates that would otherwise be handled in a programmable shader, the Xbox had a Shader model 1.0 class Geforce 3 device.

the GS and Geforce 3 can output pure 32bit graphics, while the Cube is limited to a 18bit framebuffer depth most of the time.
#23
(09-13-2011, 04:03 AM)Squall Leonhart Wrote: are you somewhat deficient at reading. the bold lines are higher on the GS.

check his nick, what u expect? xD, now hes gonna say that the wii was a rly 7gen console xD
#24
ooh yeah. tons of knowledge about emulating that ***** on x86 and directx hardware. Laugh

ATI's flipper got a fixed function hardware T&L vertex pipeline that is closer to x86 (means easier to emulate on a x86 PC - and more or less better in hardware) while VUs are completely different general purpose vertex unit that are in NO WAY x86 compliant.

this is the only and biggest difference. and the reason why the ***** is so slow sometimes. but the pcsx2 devs deserve more than a lil respect to make that beast playable on x86 hardware... forced to compute in software.

I don't wanna say anything more about that topic.
#25
I'm told Flipper's TEV unit isn't much different from the GF3's register combiners aka shaders but it didn't see much use.
The GS has a pixel fillrate higher than a G4Ti but the 4MB of eDRAM and no S3TC meant you were stuck with 8bit textures and <640x480 res. Particle effects are nice but I spend more time looking at textures. EE does the geometry for GS so triangle counts need an asterisk.
IMO 640x528x18>512x448x24
(09-12-2011, 11:10 PM)Fezzer Wrote: I was always so furious that Capcom released a playstation series title on just GC buuut now after playing it theres no way the game could run on a PS2.
Isn't REmake just some 3D models on pre rendered backgrounds? The DC could handle that.
#26
(09-12-2011, 11:23 PM)rama Wrote: What the GC has though is a PC like graphic unit, which must be a godsend for emulation Wink

IF that was True, then Where are THE XBOX 1 EMULATORS
#27
ask blushogun.
Quote:I'm told Flipper's TEV unit isn't much different from the GF3's register combiners aka shaders but it didn't see much use.

Register combiners aren't shaders.

Shader supporting cards had register combiners (for texture setup), till the gpu became fully programmable.
#28
(09-13-2011, 03:27 AM)nintendo85 Wrote: Not what I heard. The GPU and CPU were much more powerful on the gamecube. Why did Soul Caliber 2 look and have a smoother/better framerate than the PS2 version? Why do character movements in FFX shake like jittery caffeine addicts yet they moved smoothly in Resident Evil 4?
on).

If anything, the Gamcube was indeed capable of reaching near-Xbox levels in terms of graphics. Having a 1.5GB limit on games didn't help.

Gamecube GPU

162 MHz "Flipper" LSI (co-developed by Nintendo and ArtX, acquired by ATI)
180 nm NEC eDRAM-compatible process
8 GFLOPS
4 pixel pipelines with 1 texture unit each[14]
TEV "Texture EnVironment" engine (similar to Nvidia's GeForce-class "register combiners")
Fixed-function hardware transform and lighting (T&L), 20+ million polygons in-game[17]
648 megapixels/second (162 MHz × 4 pipelines), 648 megatexels/second (648 MP × 1 texture unit) (peak)
Peak triangle performance: 20,250,000 32-pixel triangles/s raw and with 1 texture and lit
337,500 triangles a frame at 60 FPS
675,000 triangles a frame at 30 FPS

8 texture layers per pass, texture compression, full scene anti-aliasing
8 simultaneous hardware light sources
Bilinear, trilinear, and anisotropic texture filtering
Multi-texturing, bump mapping, reflection mapping, 24-bit z-buffer
24-bit RGB/32-bit RGBA color depth
Hardware limitations sometimes require a 6r+6g+6b+6a mode (18-bit color), resulting in color banding.


PS2 GPU

Graphics processing unit: "Graphics Synthesizer" clocked at 147 MHz
Pixel pipelines: 16
Video output resolution: variable from 256x224 to 1280x1024 pixels
4 MB Embedded DRAM video memory bandwidth at 48 gigabytes per second (main system 32 MB can be dedicated into VRAM for off-screen materials)
Texture buffer bandwidth: 9.6 GB/s
Frame buffer bandwidth: 38.4 GB/s
DRAM Bus width: 2560-bit (composed of three independent buses: 1024-bit write, 1024-bit read, 512-bit read/write)
Pixel configuration: RGB: Alpha:Z Buffer (24:8, 15:1 for RGB, 16, 24, or 32-bit Z buffer)
Dedicated connection to: Main CPU and VU1
Overall pixel fillrate: 16x147 = 2.352 Gpixel/s (rounded to 2.4 Gpixel/s)
Pixel fillrate: with no texture, flat shaded 2.4 (75,000,000 32pixel raster triangles)
Pixel fillrate: with 1 full texture (Diffuse Map), Gouraud shaded 1.2 (37,750,000 32-bit pixel raster triangles)
Pixel fillrate: with 2 full textures (Diffuse map + specular or alpha or other), Gouraud shaded 0.6 (18,750,000 32-bit pixel raster triangles)

GS effects: AAx2 (poly sorting required),[47] Bilinear, Trilinear, Multi-pass, Palletizing (4-bit = 6:1 ratio, 8-bit = 3:1)
Multi-pass rendering ability
Four passes = 300 Mpixel/s (300 Mpixels/s divided by 32 pixels = 9,375,000 triangles/s lost every four passes

Just because the PS2 has a higher polygon/vertex count doesn't mean the number stays at 75,000,000 when all lighting and textures are applied to the gourad shaded polygons. In-game vertices are much lower after the GPU renders them into to the game; my understanding is that very few games reached 12,000,000 per second.

PS2 GPU < ATI's Gamecube GPU. Good day.

Wow, thats what everyone does to defend their console, post specifications and technical stuff, and numbers. If programming for a console only were that easy. I mean the Wii has equal if not greater specs than the old xbox but why do many of its games look crappy?
Consoles are really really really really really complicated. They are more than just numbers. Programmers were used to pc programming so the xbox was different and was easy to utilize to the fullest.

Here's a video explaining ps2 and gamecube architecture in layman's terms. It explains how ALL developers were probably only using 50-60% of the gamecube and ps2 power with citations and sources.

http://www.youtube.com/watch?v=4TBct24_ZNQ

The gamecube was stronger than the ps2 but not by much.
Also thats the first I heard of soul calibur 2 being smoother on the GC. I read that namco arcade machines architecture are extremely similar to the ps2. That's why some people were able to hack soul calibur 3 arcade edition and actually port it to ps2 format; all the new stages, characters and movesets were added to the existing SC3 iso. And SC3 was graphically superior to SC2. So yeah...

Here's link to Soul calibur 3 arcade edition hacked and ported to the ps2.
**removed**

In the end it depends on programmers to exploit a console:
The GC may be stronger but I yet to see it perform normal mapping, advanced bump mapping like the ps2 did in path of neo.
Link: http://www.youtube.com/watch?v=Kh4H51-6n2k
Or a racing game with 22 cars on the track at 60fps.
http://www.youtube.com/watch?v=VG5pIh3TKd8

The GC used simple height based bump maps in rogue squadron with level of detail removal. But rogue squadron did have huge areas and is probably the most impressive GC game imo. Height based bump maps on EVERYTHING is still hard to do kinda.

Each console has its strengths and weaknesses.
#29
Don't link any warez on this site.
Please read the rules before posting.
#30
Indeed Dolphin progresses much faster. It was in alpha stage in 2003, 2 years after the launch of the NGC, and had experimental Wii emulation in 2008, 2 years after the launch of the Wii. Now they can take a break while waiting for the upcoming Wii U and its documentations before starting a new project.

The PS3 came out in 2006 and after 5 years, PCSX2 team still stucks with the PS2. Why do you guys not abandon PCSX2 and shift to PCSX3 like you did before with PCSX1? PC hardware limitation or documentations unavailability?




Users browsing this thread: 1 Guest(s)