By John L. Hennessy & David Patterson
Read Online or Download A solution manual to Computer Architecture: A Quantitative Approach 4E (John L. Hennessy & David Patterson) PDF
Similar nonfiction_6 books
Smooth hide: 134pgs
- Charles Matthews Article Series On Your Side
- Expedient Repair of Utility Systems Manual NAVFAC P-1100 - US Navy
- A Century of Rice Improvement in Burma
- Energy Release Per Fission in the Hanford Reactors [declassified]
- Explanations of Misfortune in the Buddha's Life: The Buddha's Misdeeds in his Former Human Lives and Their Remnants
- Bind, Torture, Kill The Inside SDoor
Additional resources for A solution manual to Computer Architecture: A Quantitative Approach 4E (John L. Hennessy & David Patterson)
B. Since read requests can go to either disk, we model this system as a single M/ M/2 queue. Arrival rate × Service time 165 requests/sec × 5 ms c. 2% N servers 2 2 d. 020 e. 18 a. For a write-only workload, the requests must go to both disks. Therefore, this system performs identically to the single disk system. b. 15. 19 If reliability is not a concern, then RAID 0 gives the best capacity and performance; with RAID 0, we waste no space for redundancy to recover from failures and each independent disk can be used to handle a random I/O request.
With offline reconstruction, the performance during reconstruction is 0. 14 To construct the missing blocks, one should first read all of the blocks from the four available disks (that fit in main memory) in order to obtain the best sequential read bandwidth. Then, only after all of the missing blocks have been reconstructed should those blocks be written out to the two repaired disks, to again obtain the best sequential write bandwidth. The XOR computations can be performed in the following order (note that other orderings are also possible): Disk 0, Block 2: Reconstruction using diagonal parity.
14. 34. c. There are two banks. Assuming a random distribution of addresses and a steady stream of accesses, each access has a 50% probability of conflicting with the previous access. 57 cycles. 4 a. With critical word first, the miss service would require 100 cycles. Without critical word first, it would require 100 cycles for the first 16 bytes and 16 cycles for each of the next 3 16-byte blocks, or 100 + (3 × 16) = 148 cycles. b. It depends on the contribution to AMAT of the level 1 and level 2 cache misses and the percent reduction in miss service times provided by critical word first and early restart.