Day 4 Flashcards
Main Memory Organization
- .stack –>memory used for non-static, local variables
- Free space –> reserved for memory for the .stack and .heap to grow in
- .heap –> dynamically allocated memory; uninitialized at first
- .bss –>uninitialized and zero initialized global and static variables
- .data –>holds non-zero initialized global and static variables, exist throughout life of the program.
- .text –> stores the compiled code of the currently running program.
Software vulnerabilities
Weaknesses allowing an attacker to violate the security of the software system.
Buffer
Variable that holds more than one value.
Buffer Overflow
Occurs when too many values are written to the buffer, causes data to be written outside the buffer’s allocated memory space.
Stack Overflow
Goal is to overwrite the function return address or function pointer in an attempt to redirect the program to malicious code.
Multiplexor
Used to select the desired output. Cont selects the bitwise input or the adder input.
X86
example of Cisc processor architecture
Advanced Risk Machine (ARM)
RISC based processor architecture
Big Endian:
most significant byte first
Little Endian:
least significant byte first
Memory-to-Register:
architecture that supports data processing operations on both registers and memory locations. (cisc)
Variable length instructions:
A.
CISC processors use. Each instruction contains an opcode and either one or two operands. More complex to decode and do not perform as well as RISC processors when pipelining instructions. Save storage space.
Fixed length instructions:
RISC processors use this. Each instruction contains an opcode and three operands.
Principle of equivalence:
any task done by the software can be done by hardware, and any operation performed directly by hardware can be done using software. Hardware is faster, but software is more acceptable because it costs less and easier to modify.
Microprogrammed:
software programmed on top of hardware to implement the ISA. CU uses 6-16 registers in a single set.
Hardwired:
complete hardware solution to implement the ISA.
Which of the following is the implementation of multithreading used by Intel in x86 CPUs?
Hyperthreading
Using information sheet 4-8 x86 Processor spec-sheet, how many cores and threads does the Intel i7-3930K processor have?
6,12
Pipelining:
allows overlapping execution of multiple instructions in a single fetch-decode-execute data path.
Thermal Design Power:
tells cooling system designers how much heat must be dissipated from the CPU.
Multi-threaded Processing:
is intelligent scheduling of instructions from two fetch-decode data paths to a single execution stage. (Constantly processing)
Using Information sheet 4-8 x86 Processor spec-sheet, what is the TDP of the Intel Core i7-3930K processor?
130W
Superscalar Processors:
use a single fetch-decode data path leading to multiple execute stages allowing for more than one instruction per clock cycle. Must check for data dependencies at run time, to prevent executing an instruction that needs data from a previous instruction that is still executing.
Instruction- Level Parallelism:
is a result of each stage of a pipeline having an instruction.
Clustered Systems
In clustered systems, each computer in the group is a node, and is a complete processing unit. Unlike a network, the entire group of computers appears as a single unit to the user.
What purposes can GPUs fill?
- Graphics
- Mathematics
- Cryptography
What is short vector processing
an implementation of SIMD that allows execution of an instruction against all elements of a vector simultaneously.
What is Standard Performance Evaluation Corporation?
uses a tailored benchmark suite to test the specific functionality for which the system was designed.
Shared memory systems
all CPUs have access to a single memory space, communicate via the system bus/memory, and must lock memory prior to accessing it.
Multicore Processing:
consists of two or more processing units on a single chip.
Field-programmable gate array
is an integrated circuit configured by the customer
Asymmetric multiprocessing:
software known as hardware description language (HDL).
assigns one CPU as the master, which manages the system and controls all resource scheduling.
MIPS or Million instructions per second:
uses a mixture of floating-point and integer arithmetic and logical operations to make a general-purpose performance measurement of a computer system, ignoring system design concerns.
What provides true multitasking?
Multicores
Distributed shared memory systems
each CPU has independently allocated memory within the shared memory space, and communicate via a dedicated interconnect.
What does symmetric multiprocessing do?
gives all processors equal access to the OS and all system resources.
Multiple Instruction Multiple Data (MIMD:
is the computer architecture classification in which multiple processing units fetch individual instructions and perform their operations on individual pieces of data.
