Scribd
Upload
5 views

assignment_7

The document outlines the differences between CPUs, GPUs, and TPUs, highlighting their specific optimizations for various tasks. It also discusses real-world applications of Recurrent Neural Networks (RNNs) in speech recognition, language translation, and time-series prediction, as well as the distinctions between FPGAs and microprocessors. Key takeaways emphasize the flexibility of CPUs, the parallel processing capabilities of GPUs, and the specialized nature of TPUs.

Uploaded by

biznes akaunt
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
5 views

assignment_7

The document outlines the differences between CPUs, GPUs, and TPUs, highlighting their specific optimizations for various tasks. It also discusses real-world applications of Recurrent Neural Networks (RNNs) in speech recognition, language translation, and time-series prediction, as well as the distinctions between FPGAs and microprocessors. Key takeaways emphasize the flexibility of CPUs, the parallel processing capabilities of GPUs, and the specialized nature of TPUs.

Uploaded by

biznes akaunt
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Computer Architecture

Assignment 07

preencoded.png
CPU, GPU, TPU: Key Differences
CPU GPU TPU

General processor optimized for Many small cores designed for parallel Specialized hardware for machine
sequential tasks. processing. learning acceleration.

• Few powerful cores • Excels at graphics and video • Matrix math optimized
• Handles diverse workloads • Optimized for SIMD tasks • Google-designed tensor cores
• Versatile and flexible • Supports large-scale parallelism • Speeds up neural-network
computations

preencoded.png
Real-world RNN
Applications
Speech Recognition Language Translation
Processes sequential audio to Maps input language sequences
convert to text. to another language.

Used in voice assistants and


dictation software. Enables online translation
services in real time.

Time-Series Prediction
Predicts future values from historical data.

Applied in stock market forecasts and weather models.

preencoded.png
FPGA vs Microprocessor
FPGA Microprocessor
Reconfigurable hardware via Fixed hardware with
coding. programmable software.

Custom logic tailored for Ideal for general-purpose


fast, specialized tasks. computing and easy
programming.

Key Difference
FPGA changes hardware architecture; microprocessor runs software
on fixed hardware.

preencoded.png
Summary & Key Takeaways
CPUs
Flexible, designed for varied tasks.

GPUs
Specialist in parallel processing and graphics.

TPUs
Optimized for fast deep learning operations.

RNNs
Enable speech-to-text, translation, and time-series forecasts.

FPGAs vs Microprocessors
FPGAs rewire hardware; microprocessors run software on fixed logic.

preencoded.png

You might also like