42bs Posted June 3, 2022 Author Share Posted June 3, 2022 Corner numbers: 1800 frames, 75340 polygons of which ; Polygon sizes: ; 3 9754 ; 4 56289 ; 5 8349 ; 6 902 ; 7 46 runtime 33s (triple buffer, 60Hz) , 42s (double buffer) That's 2283 polygons/s or 42 polygons/frame. 2 Quote Link to comment Share on other sites More sharing options...
DEATH Posted June 3, 2022 Share Posted June 3, 2022 3 hours ago, 42bs said: Corner numbers: 1800 frames, 75340 polygons of which ; Polygon sizes: ; 3 9754 ; 4 56289 ; 5 8349 ; 6 902 ; 7 46 runtime 33s (triple buffer, 60Hz) , 42s (double buffer) That's 2283 polygons/s or 42 polygons/frame. what is the maximum polygons in 1 frame? Quote Link to comment Share on other sites More sharing options...
42bs Posted June 3, 2022 Author Share Posted June 3, 2022 57 minutes ago, DEATH said: what is the maximum polygons in 1 frame? Really? Quote Link to comment Share on other sites More sharing options...
DEATH Posted June 3, 2022 Share Posted June 3, 2022 33 minutes ago, 42bs said: Really? yes, why? did I say something stupid? Quote Link to comment Share on other sites More sharing options...
42bs Posted June 3, 2022 Author Share Posted June 3, 2022 (edited) 8 minutes ago, DEATH said: yes, why? did I say something stupid? Not stupid, but what is it good for? Anyway, "Use the Source Luke" Edited June 3, 2022 by 42bs Quote Link to comment Share on other sites More sharing options...
Cyprian Posted June 3, 2022 Share Posted June 3, 2022 6 hours ago, 42bs said: Corner numbers: ... 42 polygons/frame. to be honest, I'm a bit surprised by such a small number of polygons per frame Quote Link to comment Share on other sites More sharing options...
agradeneu Posted June 3, 2022 Share Posted June 3, 2022 (edited) 15 minutes ago, Cyprian said: to be honest, I'm a bit surprised by such a small number of polygons per frame 42 faces do not sound very high yes Edited June 3, 2022 by agradeneu Quote Link to comment Share on other sites More sharing options...
jguff Posted June 3, 2022 Share Posted June 3, 2022 51 minutes ago, agradeneu said: 42 faces do not sound very high yes If the challenge is to port someone elses code, and someone elses code specifies the number of polygons in each frame, then the number of polygons in each frame does not seem to have much meaning. Bob Quote Link to comment Share on other sites More sharing options...
42bs Posted June 3, 2022 Author Share Posted June 3, 2022 Acutally, I find the "flight" quiet impressive even with avg. 42 polys/s frame. A grep in scene.S revealed the max. is 164 polygons. 1 Quote Link to comment Share on other sites More sharing options...
agradeneu Posted June 3, 2022 Share Posted June 3, 2022 (edited) Yeah looks impressive, altough it's precalculated and thus a bit cheating involved Edited June 3, 2022 by agradeneu Quote Link to comment Share on other sites More sharing options...
42bs Posted June 3, 2022 Author Share Posted June 3, 2022 3 minutes ago, agradeneu said: Yeah looks impressive, altough it's precalculated and thus a bit cheating involved Still the original was on an Atari with 8MHz and is from 2000. Demos live from "cheating". It is all about the impression. And: It is a good benchmark for a poly routine and other stuff. Let's take it this way: If you want to make a 3D race with fixed perspective (never played wipe-out),a pre-calculated track seems to be a fair compromise. I mean, you also "pre-calculate" the different positions of a sprite. 1 Quote Link to comment Share on other sites More sharing options...
ggn Posted June 3, 2022 Share Posted June 3, 2022 ...just sayin'! 4 Quote Link to comment Share on other sites More sharing options...
42bs Posted June 5, 2022 Author Share Posted June 5, 2022 On 6/2/2022 at 11:09 AM, DEATH said: is there a new video on youtube? Triple buffer version: 6 Quote Link to comment Share on other sites More sharing options...
DEATH Posted June 6, 2022 Share Posted June 6, 2022 7 hours ago, 42bs said: Triple buffer version: 1800 frame in 30943 ms, almost perfect ! what is the rendering resolution? Quote Link to comment Share on other sites More sharing options...
DEATH Posted June 6, 2022 Share Posted June 6, 2022 oh i found, 256*200 Quote Link to comment Share on other sites More sharing options...
DEATH Posted June 6, 2022 Share Posted June 6, 2022 is the demo supposed to run at 50 or 60Hz? Quote Link to comment Share on other sites More sharing options...
42bs Posted June 6, 2022 Author Share Posted June 6, 2022 4 hours ago, DEATH said: is the demo supposed to run at 50 or 60Hz? Both Quote Link to comment Share on other sites More sharing options...
42bs Posted June 6, 2022 Author Share Posted June 6, 2022 Further optimization on github. Now it runs in 60FPS (runtime is 1800*0.017ms), but calculation takes only 17.6s, that's 4280 polys/s or 71 polys/frame. 4 Quote Link to comment Share on other sites More sharing options...
DEATH Posted June 6, 2022 Share Posted June 6, 2022 2 hours ago, 42bs said: Further optimization on github. Now it runs in 60FPS (runtime is 1800*0.017ms), but calculation takes only 17.6s, that's 4280 polys/s or 71 polys/frame. 2 things, on my Jaguar with the Skunkboard the screen background is not black (it makes a kind of white and pink grid) Could you make a version/option without Vsync? Quote Link to comment Share on other sites More sharing options...
DEATH Posted June 6, 2022 Share Posted June 6, 2022 also on my Jaguar there is no sound on the right channel Quote Link to comment Share on other sites More sharing options...
42bs Posted June 6, 2022 Author Share Posted June 6, 2022 30 minutes ago, DEATH said: version/option without Vsync Makes no sense for this scenario as drawing takes more time then the VBlank which means the screen gets scrambled. 27 minutes ago, DEATH said: also on my Jaguar there is no sound on the right channel Strange, but noticed this also on one of my SCART adapters. Maybe LSP mixing is wrong. 33 minutes ago, DEATH said: Skunkboard the screen background is not black (it makes a kind of white and pink grid I will try with my Skunk. Quote Link to comment Share on other sites More sharing options...
DEATH Posted June 6, 2022 Share Posted June 6, 2022 29 minutes ago, 42bs said: Makes no sense for this scenario as drawing takes more time then the VBlank which means the screen gets scrambled. for benchmark test, the goal is to know at what maximum speed the Jaguar is able to render all the 1800 frames (like the benchmark version of 3DO which is without Vsync) Quote Link to comment Share on other sites More sharing options...
42bs Posted June 6, 2022 Author Share Posted June 6, 2022 As I wrote: 17.6s for the pure rendering w/o V sync. Quote Link to comment Share on other sites More sharing options...
DEATH Posted June 6, 2022 Share Posted June 6, 2022 1 hour ago, 42bs said: As I wrote: 17.6s for the pure rendering w/o V sync. Realy ? avg FPS of +102. So it's better than 3DO (21.5 s AVG FPS: 83)? Quote Link to comment Share on other sites More sharing options...
42bs Posted June 6, 2022 Author Share Posted June 6, 2022 IMHO the new PC style thinking of FPS makes no sense, since redraw during displaying the same screen results in scrambled display. I tried to blit the final screen to the display buffer, but for this the bus is to slow. So, there is not just tearing as with modern gfx cards. Anyway, it outperforms 3DO ? Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.