$4.(12$ points total$)$ Assume that we have a computer that uses virtual memory. It has a

TLB$,$ a $($L$1)$ cache, and a page table. Furthermore, we can assume the following:

$$ A TLB access requires $4$ ns

$$ A cache access requires $7$ ns

$$ A memory access requires $20$ ns

$$ A disk access requires $15$ ms $($this includes updating the page table, cache, and

TLB$)$

$$ The TLB hit ratio is $95\%$

$$ The cache hit rate is $98\%$

$$ The page fault rate is $\mathrm{.}005\%$

$$ On a page fault, the page is fetched from disk, and access is restarted. In other

words, you must re$-$do the access after the disk block is retrieved $($see flowchart

in Chapter $6).$

$$ All references are sequential $($no overlap, nothing done in parallel$)$

a$.(3$ points$)$ Assume that the CPU is trying to retrieve a virtual address. The

address is in a page that has just been retrieved from disk when the

preceding virtual address was processed. What is the time to access this new

address?

b$.(3$ points$)$ Assume that the CPU is trying to retrieve a virtual address. The

address is in a page that has already been retrieved from disk. However, this

retrieval was a relatively long time ago and neither the TLB nor the L$1$ cache

have any relevant info for this access. What is the time to access this

address?

c$.(3$ points$)$ Given the configuration above, give the effective access time for a

lookup in the page table $($this is just for the lookup $$ you are not concerned

about the cost of a page fault$).$

d$.(3$ points$)$ Assume that all caches have been disabled $($no TLB$,$ no data

caches, etc.$).$ The CPU now receives a virtual address. In this scenario, what

is the effective access time to retrieve the value associated with this virtual

address?