Fillrates
BlitzPlus Forums/BlitzPlus Programming/Fillrates
| ||
Hello, I am experiencing something strange. I put together this little benchmark thing: ; Drawimage Fillrate Test ; Set GFX Graphics 640,480,16,1 ; Set temporary image gfxBenchImage=CreateImage (640,480) SetBuffer ImageBuffer(gfxBenchImage) For t=1 To 1000 Color Rand(255), Rand(255), Rand(255) Oval Rand(600), Rand (440), Rand(40), Rand(40),1 Next ; Do the fillrate test SetBuffer BackBuffer() L_Timer#=MilliSecs() For t=1 To 20000 DrawImage gfxBenchImage,0,0 Next L_Time#=MilliSecs() - L_Timer# ; Print result Cls Result# = (((2000*64*48) / L_Time#)) Text 10,10,"DrawImage: " + Result# + " million pixels per second" Flip WaitKey() And my results are: P4 2.6GHz, Radeon 9700Pro, WinXP: 2117 mpix / s P4 2.53GHz, GeForce4, WinXP: 2282 mpix / s AMD 1.1GHz, GeForce2, WinXP: 1084 mpix / s P3 333MHz, TNT2, Win2K Server: 16340 mpix / s (<-- here is a problem??) As you can see, the P3 is flipping out, but I don't know why yet. Those of you who want, please run this code on your PC's and report "Processor, GFX Card: <result>". I am hoping to find a pattern. Maybe it's the OS... |
| ||
Athlon XP 2000+ (1.67ghz), Geforce 3 Ti 200, and only 1075.25 mpix/s oh, and winXP, might be cuz i need to reboot, i'll reboot my system, and test when I get a chance to reboot :P edit: also forgot to mention that this was ran in blitz2D, not plus |
| ||
AMD Athlon 1.2GHz, Voodoo3 3000, Windows 2000 467.29 mpix/s full screen mode 479.82 mpix/s windowed mode One thing: Since you do not use a fixed random seed, the code above will not be a very reliable benchmark, since the actual work performed on the test machines can vary greatly, depending on the random number sequence it gets assigned. |
| ||
lol. I get 271.0169 mpix/full screen I have a Dell GX240. 1.8 GHz P4 WinXP Pro 1 Gig Ram Radeon 7000 VE Family 32 MB Nice business computer but Dell SUCKS as a gaming machine. cb |
| ||
1091.49 on: AMD Athlon XP 2400+ GeForce 3 Ti200 512 MB RAM Windows XP Pro |
| ||
Thanks so far. I was advised to use Lockbuffer and Unlockbuffer just before and after blit loop, this greatly affected the speed my PCs blitted them, even though it shouldnt make that much of an impact. Those of you who got less than you expected would probably notice a big difference by adding that. I am trying to get some sort of an idea what graphics cards are capable of so I can do a "recommended" settings function in blitz, but maybe I should simply base it on measured MHz speed and video memory instead, as these results are so different from each other. Thanks, |
| ||
1049.75 mpix on full screen AMD 1GHz 512 Mb 98SE GeForce2 GTS |