Orion_ Posted February 5, 2007 Share Posted February 5, 2007 Example: If you were blitting in 4 4k modules into the gpu every frame, thats 1,200,000 cycles a second wasted waiting for the blitter to load code in the local RAM. oh, by the way, real coders won't wait for the blitter, so, no cycles wasted. 1 Quote Link to comment Share on other sites More sharing options...
JagChris Posted February 5, 2007 Share Posted February 5, 2007 Example: If you were blitting in 4 4k modules into the gpu every frame, thats 1,200,000 cycles a second wasted waiting for the blitter to load code in the local RAM. oh, by the way, real coders won't wait for the blitter, so, no cycles wasted. Good for you boss! Glad you got it all figured out. Quote Link to comment Share on other sites More sharing options...
JagChris Posted February 5, 2007 Share Posted February 5, 2007 But you're saying that a compiler his better than human mind... Absolutely not. Nothing can beat hand written code for optimization or flexibility. No, i've never told that. There is a lot of good C coders, i talk with some everyday, i'm too bad in C, but how many of them think the ASM give faster, better result and smaller program than C. What I meant was, every C coder that decides to code on the Jag probably will not be a lousy coder. It would probably be best to use C for non time critical stuff and do the time critical stuff with assembly. It doesn't have to be completely one or the other. It just means that the option to use C is more plausible now. No one is forced to use C on the Jag. They have a choice. Don't we all like choices? Quote Link to comment Share on other sites More sharing options...
Gorf Posted September 17, 2007 Share Posted September 17, 2007 I know this thread has been dormant but the hits you take out in main are extremely insignificant compared to running the 68k at all! And it definitely better than flipping code in and out. the only time you run slower out in main is a DRAM page hit or refresh occurs and the indexed load/store instructions can take up to five cycles to complete. It's probably taking a mip or two away even in worst case senerios. Oh and you need to interleave code to reach 80% efficiency. 18 MIPs is about all she wrote. Fuggin cheapo fake dual port write backs. Quote Link to comment Share on other sites More sharing options...
JagChris Posted January 31, 2008 Share Posted January 31, 2008 'MORE THAN GAMES' link updated. Quote Link to comment Share on other sites More sharing options...
Gorf Posted March 2, 2008 Share Posted March 2, 2008 Example: If you were blitting in 4 4k modules into the gpu every frame, thats 1,200,000 cycles a second wasted waiting for the blitter to load code in the local RAM. oh, by the way, real coders won't wait for the blitter, so, no cycles wasted. A good coder should never make a processor sit around and wait, especially if it has interrupts and can be doing something else. this is also a BIG reason why a lot of the games on the Jag suck. The waste a lot of time waiting. Lazy coding is what that is. 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.