My first encounter with G-Force happened in March this year. I was browsing through visualisation plugins on the Winamp's site, and when I came to G-Force's screenshot I thought to myself, Wow this really looks awesome. I was only a bit worried whether it would work on my computer since it's already getting old hardware-wise (Celeron 2 at 633 MHz).
I can't remember from which server I downloaded it, because the Winamp's page didn't redirect me anywhere. Anyhow, when I first started G-Force I was really impressed (to say the least) - no other visualisation has ever moved me so much. Also the frame rate was sufficient for very smooth animation.
Later on I found out I was using G-Force 2.4.3 in resolution 640x480 and that the frame rate varied between 20 and 23 frames/s. All this on a PC with Windows XP professional.
Then somewhere in April this year I tried a newer version which was downloaded from www.55ware.com , but it couldn't reach more than 15 frames/second + it didn't respond to music as good as the older one. Thus I returned to my beloved 2.4.3.
Yesterday I tried versions 2.6.7 and 2.6.8 beta. They respond to music as good as 2.4.3 but the frame rate doesn't exceed 12-13 f/s (in the 640x480 resolution).
Is that normal? If it is a necessary sacrifice in order to improve the effects in newer versions, is there a way to increase the frame rate for us who still have slower computers? I tried playing with the resolution but it doesn't look too promising.
Newer versions more hardware demanding?
Moderators: BTT, andy55, b.dwall, juxtiphi
Re: Newer versions more hardware demanding?
There's a lot of components into the frame-to-frame load of G-Force. Other than the necessary processing (that has been the same 2.4 and on), the oncoming flowfield calculations are fairly intensive. GF has to chip away at calculating the next flowfield, and over the versions, I've stepped up the size of the chunks it chomps through each frame (the payoff is that the next field is ready sooner). To see how much that's the culprit of your performance description, turn the flowfield slideshow off and then do frame rate tests (when the flowfield slideshow is off, GF will not be calculating the next flowfield).alphaone wrote:My first encounter with G-Force happened in March this year. I was browsing through visualisation plugins on the Winamp's site, and when I came to G-Force's screenshot I thought to myself, Wow this really looks awesome. I was only a bit worried whether it would work on my computer since it's already getting old hardware-wise (Celeron 2 at 633 MHz).
I can't remember from which server I downloaded it, because the Winamp's page didn't redirect me anywhere. Anyhow, when I first started G-Force I was really impressed (to say the least) - no other visualisation has ever moved me so much. Also the frame rate was sufficient for very smooth animation.
Later on I found out I was using G-Force 2.4.3 in resolution 640x480 and that the frame rate varied between 20 and 23 frames/s. All this on a PC with Windows XP professional.
Then somewhere in April this year I tried a newer version which was downloaded from www.55ware.com , but it couldn't reach more than 15 frames/second + it didn't respond to music as good as the older one. Thus I returned to my beloved 2.4.3.
Yesterday I tried versions 2.6.7 and 2.6.8 beta. They respond to music as good as 2.4.3 but the frame rate doesn't exceed 12-13 f/s (in the 640x480 resolution).
Is that normal? If it is a necessary sacrifice in order to improve the effects in newer versions, is there a way to increase the frame rate for us who still have slower computers? I tried playing with the resolution but it doesn't look too promising.
Another factor is that many recent waveshapes, colormaps, and flowfields have become increasingly complicated and math-intensive, contributating additional CPU load. This is more noticable, of course, on older machines.
the good news, however, is that we're only a few months away from me bringing an entirely new math engine on line. this new engine will turn the above two issues into only a trivial fraction of the CPU. We're looking at a 10-1000 time speedup (keep in mind that around half of the current frame load is stuff unaffected by the new engine).
as you can imagine, i'm really really excitied about this new engine, codenamed "VectorC" because it makes the g-force into the lightweight I want it to be. this is why i'm trying desperately hard to stabilize G-Force 2.6--so i can put it on a shelf and really dive into VectorC development.
thanks for your patience and support,
andy
Re: Newer versions more hardware demanding?
andy55 wrote: There's a lot of components into the frame-to-frame load of G-Force. Other than the necessary processing (that has been the same 2.4 and on), the oncoming flowfield calculations are fairly intensive. GF has to chip away at calculating the next flowfield, and over the versions, I've stepped up the size of the chunks it chomps through each frame (the payoff is that the next field is ready sooner). To see how much that's the culprit of your performance description, turn the flowfield slideshow off and then do frame rate tests (when the flowfield slideshow is off, GF will not be calculating the next flowfield).
Another factor is that many recent waveshapes, colormaps, and flowfields have become increasingly complicated and math-intensive, contributating additional CPU load. This is more noticable, of course, on older machines.
Andy, thanks for the reply.
I experimented quite a bit today. The conclusion is that whether the flowfield is on or off doesn't affect frame rate in any noticable way. However, lowering the resoultion did the job. I managed to achieve 20 f/s in 640x400 with the newest version.
Btw. the frame rate goes well down while "This is free version..." message is being displayed. It doesn't happen with other text messages, such as song titles etc. But maybe this was intentional?

That's great news, keep up the good job!the good news, however, is that we're only a few months away from me bringing an entirely new math engine on line. this new engine will turn the above two issues into only a trivial fraction of the CPU. We're looking at a 10-1000 time speedup (keep in mind that around half of the current frame load is stuff unaffected by the new engine).
as you can imagine, i'm really really excitied about this new engine, codenamed "VectorC" because it makes the g-force into the lightweight I want it to be. this is why i'm trying desperately hard to stabilize G-Force 2.6--so i can put it on a shelf and really dive into VectorC development.
thanks for your patience and support,
andy
