I have questions on start of exections of cmds
We all know that cmd2 can start executing only after cmd1 .
If cmd2 is recorded latter than cmd1
If there are 2 different subpasses in a renderpass
Can cmds in subpass 1 start before cmds in subpass 0 within that render pass?
Or they also have the implicit ordering?
I had a question, what all commands do pipeline bariers affect.
so if we have a command buffer
cmd1 cmd2 bar1 cmd3 cmd4 bar2 cmd5 cmd6
so can bar1 apply its synchronization scopes to cmd5 and cmd6??
and can bar2 apply its synchronizationscopes to cmd1 and cmd2??
I know that if the barrier is in a subpass of a renderpass then its scope of synchronization will not extend beyond the bounds of that subpass.
also why do we even need renderpasses when we have pipeline barriers?
are renderpasses more optimized in cache control
So I am observing one thing with this function vkacquirenextimagekhr.
If I do this
Fence begins signaled Vkresetfence(fence1,fence2,fence3);
Vkacquirenextimagekhr(signal semaphore1,signal fence1);
Vkqueuesubmit(wait on semaphore1,signal semaphore2,signal fence2);
Vkqueuepresent(wait on semphore2,signal fence3);
Vkwaitforfence(fence3);
// We can reset fence 3 and fence2 after this wait but when we try to reset fence1 after this wait I get the validation error saying fence1 is still in use?????
The queue present function I am use swapchain maintanace 1 extension to add extra signal functionality to vkqueuepresent function.
Why is the fence1 still in use but all the queuesubmit operations are truly done. Is there real exta work that acquirenextimagekhr is doing there after it has signaled the semaphore1 and before fence1???
I can only reuse fence1 only after waiting for it specifically only.