I don\'t know what you\'re talking about, but polygons are totally independent of fillrates! If you\'re talking about rendering polygons, you must know that they could be of virtually any size; anywhere from taking up the whole screen and then some to less than one! Even if you were to "fix" their sizes for some benchmarking reason that may be attempting, did you ever think of hidden polygons that are behind one another? I think you\'re the one who needs research.
If you would have taken the time to look at what I wrote down, you would have seen that I took for 32 pixels for the size of one polygon. And we are comparing two systems, so in order to do that fairly, we have to compare them both using the same conditions. Now, I don\'t know what you know or where you do your research little guy, but speaking strictly of rendering polygons (you know the "drawing those polygons on screen") is
dependend on fillrate.
Now, off course a game could be drawn with smaller polygons (i.e. less pixels), but don\'t forget we want to compare both systems under the same conditions right. Not much point in Microsoft saying that the NV2a can draw 116.5 MTris/sec of 8 pixel Polygons (compared to the 75 MTris/sec of 32 pixel polygohns that the GS can render) right? Other people have done the maths and quoted a performance of 125 (now 116.5) MTris/sec using 32 Pixel Triangles. That was asuming a drawing fillrate of 4 GPixels/sec.
Now, anyone want the truth of how big Xbox\'s real fillrate is? Certainly not 4 GPixel/sec.
Xbox has got 4 pixelpipelines and is clocked at 233 MHz. That means, the NV2a can perform 4 pixels per clock cycle, so in 1 second, its:
Fillrate = 4 * 233 MHz
Fillrate =
0.932 GPixels/secThat number is the drawing fillrate of the system, thus how many pixels can be rendered per second. Want a link to back this up?
http://firingsquad.gamers.com/hardware/xboxtech/page4.aspThere! Confirmed source of Xbox having a fillrate of 0.932 GPixels/sec (or 932 MPixels/sec). Now, since this is a comparasment between the two (Xbox - PS2), we\'ll compare them both under the same conditions (using 32 Pixel polygons):
Triangles = 0.932 GPixels/sec / 32 Pixel Polygons
Triangles =
29.125 MTris/secPS2 has got 16 pixelpipelines and the GS is clocked at 150 MHz. So, 16 pixels done with each clock cycle. Fillrate = 16 * 150 MHz; Fillrate =
2.4 GPixels/secTriangles = 2.4 GPixels/sec / 32 Pixel Polygons
Triangles =
75 MTris/sec (the number Sony quoted)
Now this might seem that PS2 is much better at rendering polygons. While this is true to a certain extend, the above calculation shows one extreme witout any textures being rendered. While Xbox can do 2 texels (texel = textured pixel) per pass, it has a texelfillrate of 1.8 GTexels/sec.
The GS however needs 2 clock cycles to perform one texel, so the texelfillrate is 1.2 GTexels/sec. This basically means, that the Xbox can render up to 2 texture layers without a drop in polygons performance, while PS2\'s drops significantly. Using two texture layers, Xbox can do 29.125 MTris/sec (still using 32 Pixel polygons, same conditions) and PS2 drops down to 18 MTris/sec.
Now, as for the fill-rate, the Xbox does 1 Billion pixels per second with NO FSAA, but with 4X FSAA turned on, each pixel is rendered 4 times, which is where the 4 Billion pixels per second number comes from.""
Xbox still only has a drawing fillrate of 0.932 GPixels/sec.
Source:
http://www.gdconf.com/archives/proceedings/2001/abrash.ppt
Nothing! XBox has easy 3 times the graphics capabilities of PS2!
A statement that only a
fanboy would say. The above shows that PS2 has a edge from untextured polygons and with 1 texture layer. You\'d have to prove to me where Xbox is "easy" 3 times the graphics capabilities of PS2. That is simply laughable.
I\'ll do a research about it ! (don\'t flame me!)
Yep, using 8 pixel Polygons. It is possible, but for the comparasment above, we did use 32 pixel polygons to compare the 2 systems.
Really, what\'s there to argue??
I wasn\'t argueing - simply aswered Bizio\'s question above. You can\'t argue against facts.
