- Asked 2/22
- 2079 Baisakh · 2+4 marks
- 2070 Asar · 6 marks
Performance balance and bus design elements
Short answer
Performance balance aligns processor, memory, and I/O speeds to prevent bottlenecks. It is achieved via cache, wider memory, DMA, and bus hierarchy. Bus design elements include dedicated or multiplexed lines, centralized or distributed arbitration, synchronous or asynchronous timing, and bus widths (address and data).
What is performance balance and why is it required? Explain the different elements of bus design.
Performance balance
Performance balance means adjusting the organization of the system so that the speeds of the processor, main memory and I/O devices match each other and no one part becomes a bottleneck. Processor speed has grown much faster than memory speed and I/O speed, so a fast processor often waits for data. Balance is needed so that the processor's power is actually used and the extra cost of a fast part is not wasted.
Ways to keep balance:
- Make memory wider (more bits per access) and use interleaved memory.
- Use a cache between processor and memory.
- Use a faster bus or a bus hierarchy.
- Use DMA and buffering for I/O.
- Reduce memory accesses with a better processor design (more registers, pipelining).
Elements of bus design
| Element | Choices |
|---|---|
| Bus type | Dedicated (separate address and data lines) or multiplexed (shared lines at different times) |
| Arbitration | Centralised (one arbiter) or distributed (all modules share control) |
| Timing | Synchronous (clock-controlled) or asynchronous (handshaking, no common clock) |
| Bus width | Number of address lines (memory capacity) and data lines (transfer size) |
| Data transfer type | Read, write, read-modify-write, read-after-write, block transfer |
- Dedicated bus: faster and simpler control, but needs more wires.
- Multiplexed bus: fewer wires and lower cost, but slower and needs more complex circuits.
- Synchronous timing: easy to implement and test, but all devices must work at the same clock.
- Asynchronous timing: each transfer uses request and acknowledge signals, so fast and slow devices can share the bus.
- Wider buses give more bandwidth but need more pins and wires.
✓ Answer checked by Claude Opus 5.5 (high effort)
Related questions
- Performance balance of processor and memory · Computer Organization and Architecture, chapter 1
- Driving factors for performance design · Computer Organization and Architecture, chapter 1
- Design goals and performance metrics · Computer Organization and Architecture, chapter 1
- Elements of cache design · Computer Organization and Architecture, chapter 6
- Need for bus hierarchy · Computer Organization and Architecture, chapter 1
- Bus hierarchy and bus arbitration · Computer Organization and Architecture, chapter 1
- Bus admittance matrix and its elements · Power System Analysis II, chapter 1
- Design: 200 m head plant, bus MVA levels · Power Plant Design, chapter 4 · asked 14×
- Design: 120 m head plant, bus MVA levels · Power Plant Design, chapter 4 · asked 8×
- Design: 150 m head plant, bus MVA levels · Power Plant Design, chapter 4 · asked 8×
All Computer Organization and Architecture chapter 1 questions →