We would like to launch a matrix multiplication kernel to multiply an $100*100$ matrix A with a $100*100$ matrix B with the simple matrix multiplication kernel using $16*16$ thread blocks. Answer the following question: How many blocks will be launched if each thread is responsible for computing one output element?

$64$

$49$

$81$

$36$

Question $4(3$ points$)$

Consider the following code in a CUDA kernel function.

$-\_$global$\_\_$ void do$\_$work$($int i$,$ int $*$A$)$

oad.

int temp $=0$;

if $($i $<10)$ temp $=$ threadlax.x;

Althread$|$dx$.$x$)=$ temp;