计算机组成原理代写 | ELEC S363F Advanced Computer Design

这个作业是完成一个计算机的设计
ELEC S363F Advanced Computer Design

Part 1: Answer ALL questions in this part.
(24 marks) Question 1: The specification of the Coffee Lake-based Core i7-9850H CPU is listed in the
following table.
(a) Determine the number of bits for the tag fields in the 38-bit memory address for:
(i) L1-D Cache (4 marks)
(ii) L2 Cache (4 marks)
(b)
(i) L1 is splited into I Cache and D Cache. Give a reason to support such design. (2 marks)
(ii) L3 is shared among cores rather separated and dedicated for each core. Give a reason to
support such design. (2 marks)
(iii) Which cache eviction policy would you recommend for the L3 cache? Explain your answer.
(2 marks)
(iv) The cache block size is larger than the maximum data size operated on the processor. Explain
whether this design would increase the hit rate or decrease the hit rate. (2 marks)
(c) Determine
(i) The generation number of this processor. (1 mark)
(ii) The number of logical processors. (1 mark)
(iii) The maximum core frequency when only half of the cores are active. (3 marks)
(iv) The maximum core frequency when only 1 core is active. (3 marks)
(14 Marks) Question 2: Consider a branch instruction Z has been executed several times in a program. The
branch Z had the following outcomes in sequence
T, T, N, N, T, N, T, T, T, T (T: branch taken; N: branch not taken)
There are three prediction schemes to handle the branches:
Scheme W: static branch-not-taken prediction
Scheme X: 1-bit dynamic branch prediction (with initial state as “N”) (please refer to the state
diagram in page 59 of the lecture notes 06)
Scheme Y: 2-bit dynamic branch prediction (with initial state as “SN”) (please refer to the state
diagram in page 65 of the lecture notes 06)
(a) Determine the accuracy of using scheme W to predict branch Z. (2 marks)
(b) Determine the accuracy of using scheme X to predict branch Z. Copy the below table to your answer
file and fill in your answer there.
Part 2: Answer any 1 question in this part.
(30 Marks) Question 5: Deep Learning Boost (DL Boost) is a new technology introduced by Intel since the Tiger
Lake-based (11th generation Core-i) CPUs. It extends AVX-512 with a new Vector Neural Network Instruction
(VNNI) that accelerates Artificial Intelligence deep learning use cases. In your own words, write a report in
about 400 to 500 words with appropriate figures to explain the operating principle of DL Boost, technical
specifications of VNNI, and how it is better than its previous generations that without the DL Boost technology.
Note: You may use the Internet as a source of information. However, you should digest the content and use your
own words to write the report. You are also required to indicate the total number of words and the sources of
information in your report.