Question

Some processors have two sets of registers, one for integer operations and one for floating point (fp) operations. The registers used in integer operations are always saved and restored during every context switch whether it is preemptive multithreading or non-preemptive multithreading(since they are also used for calculations for instructions, such as memory address calculation, branch target calculation etc.) However, for most machines it is not very clear what will happen to the floating-point registers. Below you are given 5 different processor architectures. As a programmer, for which architectures do you need to save and restore floating point registers? state your reasons. a) (5 pts) Machine uses non-preemptive multithreading and only allows only a single thread that can use the floating-point registers. There is a separate stack for integer and floating-point operations. b) (5 pts) Machine uses preemptive multithreading and only allows only a single thread that can use the floating-point registers. There is a separate stack for integer and floating-point operations. c) (5 pts) Machine uses non-preemptive multithreading and allows multiple threads that can use floating point registers. There is a separate stack for integer and floating-point operations. d) (5 pts) Machine uses preemptive multithreading and allows multiple threads that can use floating point registers. There is a separate stack for integer and floating-point operations. e) (5 pts) Machine uses non-preemptive multithreading and you know that the kernel is designed so that it also supports saving of floating-point registers. However, there is a unified stack for integer and floating-point operations.

Answer 1

* Which architectures do you need be save and restore floating Point registers? Answer: 6) (5 pts) Machine uses preemptive multithreading and Only allows Only a single thread that can use the Hoating -point registers. There is a separate stack for integer and loating - Point Operations. Reasons :- & The Floating point register bank contains thirty -two 32 bit registers. 64 bit floating Point registers. * It's need to execute the instrubion [: (FP) 0-31] * It's like so .... 331 represents. * Every instruction that interprets the contents of a Floating point register as a Floating - Poind value uses the double precision Floating point format for This interpretation. & 64 bit of Laba without conversion between storage and a floabing toind register in the floating Prinde Processor, * Load Single instructions conversta storage Single floating format value to the same value in double Floating format and that value transfer into - a floating point register. 17

* So bo Sis are called Saved registers. So if a Tuubion. À calle a function & Function À must save the contente of these registers, [ on the stack]. * Before calling Function & because these registers can be cheinged by the Funebion call. [ return result]. * Store single instructions do the opposite, Converting valid Single Precision values in a Floating Point register into a single floating formal values, Prior bo storage.