Re: flickering skybox (newb question)



Oooops, good catch. Phill is of course correct.

--
Zman
http://www.thezbuffer.com - News and information for Managed DirectX



"punkrock" <punkrock@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:6671D9E8-C60E-4398-813F-1B61F633E132@xxxxxxxxxxxxxxxx
> Check your math, ZMan. 1000 / 50 = 20, not 200. And 50ms is usually more
> than
> enough to draw a frame (and do some AI, etc). Most games run ~30fps which
> is
> only 33ms per frame (aren't video cards great?), shared with AI, network
> code, and the rest of the system, of course.
>
> Now if we're talking about a simple low-framerate app that runs in
> windowed
> mode, isn't built around the tightest possible loop, and has to share the
> CPU
> with a few web pages, a media player, five to ten games of Minesweeper,
> etc,
> then obviously playing nice and giving the rest of the system as much CPU
> as
> it can is more important than frame rates, in which case I totally agree
> with
> what you say about the Timer, it's a great way to go about doing this
> since
> it allows Windows to throttle your app at the message queue level, and in
> that case I wouldn't set the timer's resolution any lower than 100ms.
>
> Phill


.



Relevant Pages

  • freebsd panic on HP Proliant DL360
    ... CPU: IntelXeonCPU 2.80GHz ... <ACPI PCI bus> on pcib0 ... OHCI version 1.0, legacy support ... #12 0xc0838127 in syscall (frame= ...
    (freebsd-stable)
  • [PATCH 4/10] kdump: save registers early (inline functions)
    ... frame is pushed on to the stack and then cpu register state is captured. ... Later this frame is popped and new frames are pushed. ... crash_kexecis to capture the register states. ... be responsible for capturing the register states as well as to do some ...
    (Linux-Kernel)
  • Re: PII vs PIII
    ... >> gigabit NIC, the other CPU's free to run the app receiving the data, so ... > During data transfer one CPU or the other was at greater 90% CPU ... TCP/IP layer on Ethernet interface for QL got an interesting problem. ... demands acknowledge after each received frame. ...
    (comp.os.linux.hardware)
  • Re: Call ::EndScene as far ahead of calling ::Present
    ... I would say, though, that suggestions of calling EndScene as far ahead of Present as possible are indeed due to the blocking behavior of Present without this flag. ... Or maybe better stated, that you should delay calling Present by doing other useful tasks - when you call EndScene is usually not the issue, it's logical to call EndScene as soon as you are finished rendering, so it's more a matter of when you call Present. ... Calling Present immediately after EndScene increases the likelihood that the CPU would block while waiting for the scene to finish rendering. ... Calling Present later decreases the time that would be spent blocking, and if you can get useful work done during that time it would mean you can start rendering the next frame sooner. ...
    (microsoft.public.win32.programmer.directx.graphics)
  • Re: A reliable way to detect ARM type & speed from within BASIC.
    ... machines so you need to call the right routines for the right CPU? ... why do you need to skip some VSync? ... I've got many CPU cycles available after doing my stuff until the ... VSync comes at frame F, I know if at frame F+1 I'll be able to tidy my ...
    (comp.sys.acorn.programmer)