<yuq825>
seems we don't really need that much of VS execution and output memory if we do well in command stream
<yuq825>
yes
<yuq825>
one call with glDrawElements(GL_TRIANGLES, sizeof(index)/sizeof(GLushort), GL_UNSIGNED_SHORT, index);
<yuq825>
GLushort index[] = {0, 1000, 2}
<rellla>
what about {0, 10, 2} ? when do we know, that splitting is better than doing all executions?
<yuq825>
yeah, we need more experiments to find the threshold, maybe we can do split when max_index - min_index + 1 > info->count
Net147 has quit [Ping timeout: 268 seconds]
<yuq825>
seems there's really hidden pattern for index draw that we can do VS_CMD_DRAW(3, info->index_size) instead of VS_CMD_DRAW(1001, info->index_size)
dddddd has joined #lima
adjtm has joined #lima
adjtm_ has quit [Ping timeout: 258 seconds]
yuq825 has quit [Remote host closed the connection]
Net147 has joined #lima
_whitelogger has joined #lima
yuq825 has joined #lima
yuq825 has quit [Quit: Leaving.]
Elpaulo has quit [Quit: Elpaulo]
chewitt has joined #lima
adjtm has quit [Ping timeout: 265 seconds]
chewitt has quit [Quit: Zzz..]
<rellla>
i've uploaded all dumps from the deqp tests done with mali450-r6p2-wayland and current mali-syscall-tracker for reference here:
<rellla>
ah, i stripped out some mali_memory_dump() and mali_mmu_dump() out of mali_syscall_tracker to save some space ... but the interesting things should be there