Hello Jarrett,
On Mon, Mar 16, 2009 at 3:29 PM, BCS <n...@anon.com> wrote:
I'm actually kind of shocked that given the prevalence of memory
block copy operations that more CPUs haven't implemented it as a
basic instruction. Yes, RISC is nice, but geez, this seems like a
no-brainer.
How about memory to memory DMA, Why even make the CPU wait for it to
finish?
Sure, then you have to worry about waiting for the *DMA* to finish. If
it's a copy whose result is largely unimportant to the
immediately-following code (copying a backbuffer into a frontbuffer,
for example), it doesn't matter. But for copying large value types,
I'd think it's pretty important that the copy is semantically atomic.
As long as either the CPU can voluntarily block or the MMU can block access
until the DMA is done then it's a free gain in that the CPU can do /something/
that it wouldn't be able to otherwise.
