Yes, you have to go indirect on a network. But if you were going over the
network, you probably weren't expecting any performance anyways.
>> Assume you want to achieve 10 Million lit, texture-mapped vertices per
>> second. How much memory bandwidth do you need? Well, you have to transfer
>> the vertex coordinates (3x4 =12 bytes), normal (3x4=12 bytes), and texture
>> coordinates (2x4=8 bytes), for a total of 32 bytes per vertex.
>
>So I put it in a mitshm object, it arrives with 0 copies, its blasted into
>the video card. Next question 8)
Umm... You still have to copy the data from the application, into the
shared memory and then out from the shared memory into the video card.
This sucks three times the system memory bandwidth of going out of the
application memory directly to the video card. Am I missing something?
nate
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu