Already been discussed in other threads, please use the search function.
Hey everyone.
I've been gone for a long time now, and am willing to get back into CPU coding. But with the recent upgrades to E2 (for loops and such), I heard that E2 has surpassed (for a while I suppose) in terms of speed, or overhead as everyone is calling it. All I want to know is what exactly is overhead and what benefits CPU has over E2, especially since E2 can control hi-speed devices now. Is CPU faster/more flexible in certain aspects now? Or just something to use to brag about?
And feel free to move this thread, I really didn't know where to put it.
Already been discussed in other threads, please use the search function.
Indeed, ultimate conclusion was that CPU was more realistic and thus more fun, but E2 has both higher speed and can do more. Especially once Syranide's new patch is released.
I really dont see a use for CPU anymore considering E2 beats it in most ways, but its still fun to use every once in awhile. GPU is here to stay though, nothing can beat that.
Not even the EGP. It is nice, but extremely laggy.
.siht daer ot gniyrt emit detsaw ev'uoY
Computers. The realistic ones.Originally Posted by Drunkie
(I know, the computers themselves aren't of much use, but they are still fun to build and code)
My programs: BIOS - Alcyone - Calculator - Notepad - Movie Player
My tutorials: applyTorque - Quaternions - PID controllers
Some other things I made: FT Chip - RK4 Solar System
Black Phoenix says hes got some patches coming when he can get to them, which will make CPU comparable to E2 in calculation power and speed again. He also wants to tie E2 more closely so that CPU will have E2 opcode functionality, or be able to directly make calls to E2.
E2 is the cake as the god object of wire, but supposedly CPU will do screen rendering and large mathematical computations faster once these patches are committed. If they are committed.
Last edited by Anticept; 03-21-2010 at 05:51 PM.
"You can only tie the record for flying low."
The VM will be much faster than current one, although will be slightly slower than E2 (which means less instructions per second; it has to be slower to account for some features of CPU like self-modifying code), but it will have ability to call E2 as opcode whenever you want (as many times as you want, assuming your script fits well within tick limit); this requires slightly different E2 script than usual (because it will always pass you registers of the CPU, for write and read access).
The exact way of calling is not yet defined, but probably it's gonna be a special opcode.
This puts E2 on level of hardcoded logic chip, for two main purposes: put some repetetive hardcoded logic outside of the CPU (think power management, wireless support), and also put streaming math outside of CPU (think mathematical equations, etc). You can use self-aware functions of E2, but CPU only accepts and gives out numbers (you can access RAM if you wirelink it yourself).
E2 working in such mode can still be triggered by external stuff, but you don't wire it to anything, you only wire address bus to the E2 (this link is done over hispeed link, although no data is transferred).
If you put math into external E2, and leave logic in CPU, you'll be able to make things run faster than if it was just E2.
Bookmarks