Question

Question # 3 Consider a demand-paged system where the page table for each process resides in main memory. In addition, there is a fast associative memory (also known as TLB which stands for Translation Look-aside Buffer) to speed up the translation process. Each single memory access takes 1 microsecond while each TLB access takes 0.2 microseconds. Assume that 2% of the page requests lead to page faults, while 98% are hits. On the average, page fault time is 20 milliseconds (includes everything: TLB/memory/disc access time and transfer, and any context switch overhead). Out of the 9890 page hits, 80 % of the accesses are found in the TLB and the rest, 20%, are TLB misses. Calculate the effective memory access time for the system

Answer 1

We know that:

Effective Memory Access Time = TLB hit ratio * (TLB access time + Main memory access time) +

(1 – hit ratio) * (TLB access time + 2 * main memory access time)

TLB hit ratio = (TLB hits) / (Number of queries into TLB)

= 80/98 = 0.816 (approx.)

So, the Effective Memory Access Time is

= (80/98)*(0.2+1)+(1-(80/98))*(0.2+2*1) Microseconds

= (0.816*1.2)+(0.184*2.2) Microseconds

= (0.9792 + 0.4048) Microseconds

= 1.384 Microseconds

Hope this helps.