Single Core Vs Multi Core Architectures
Uploaded by
Pon SaravananSingle Core Vs Multi Core Architectures
Uploaded by
Pon SaravananSi gle-Core v .
Multi-Core
Arc itecture
This presentation demystifies processor architectures, focusing on
single-core and multi-core designs. We'll explore the evolution from
sequential execution to parallel processing. This is targeted towards
developers, students, and tech enthusiasts.
U der ta di g Si gle-Core Arc itecture
Seque tial Executio Clock Speed I tructio Pipeli e
Single-core architecture executes Clock speed determines Fetch, decode, execute, write-
instructions sequentially. It uses execution speed. Example: 3 back stages for each instruction
one processing unit. GHz. are completed.
The Intel Pentium 4 and AMD Athlon are examples. Older architectures typically have an IPC of 1-2.
T e Ri e of Multi-Core
Arc itecture
Parallel Proce i g Core Cou t
Multi-core CPUs have multiple Dual-core, Quad-core, and
processing units. They Octa-core CPUs increase
distribute tasks across cores. processing units.
Overco i g Li itatio
Multi-core overcomes frequency scaling limits. It addresses heat
and power issues.
Intel launched its first dual-core processor, the Pentium D, in 2005.
Examples include Intel Core i7 and AMD Ryzen.
Adva tage of Multi-Core
E a ced Multita ki g
Smooth execution of multiple applications concurrently.
I crea ed T roug put
Higher processing capacity for demanding workloads.
I proved Re po ive e
Faster response times for user interactions.
Parallel Co puti g
Enables efficient execution of parallel algorithms.
Video editing software can leverage multiple cores for faster rendering.
C alle ge of Multi-Core
Software Opti izatio A da l' Law Re ource Co te tio
Parallel algorithms and multi- Limits speedup from parallelization. Cores compete for shared resources
threaded apps are required to see This is based on serial code portions. such as memory bandwidth.
benefits.
Poorly threaded applications may
only utilize one core.
Perfor a ce Metric
Si gle-Core
Clock Speed: GHz (higher is better).
1
Instructions Per Cycle (IPC).
Geekbench Single-Core.
Multi-Core
Aggregate Throughput.
2
Parallel Efficiency.
Cinebench, Geekbench Multi-Core.
Modern CPUs use Turbo Boost (Intel) or Precision Boost (AMD) to
adjust clock speeds. A quad-core could theoretically achieve 4x
performance.
Future Tre d
I crea i g Core Heteroge eou Co puti g
Processors with dozens or hundreds CPUs, GPUs, and specialized
of cores are coming. accelerators will be integrated.
3D Stacki g
C iplet
Vertically stacking processor
Modular design with interconnected
components will increase
chiplets will provide scalability.
performance.
Apple's M1 Ultra combines two M1 Max chips, representing a chiplet-based approach. Intel's Ponte Vecchio GPU utilizes a
multi-chiplet design.
Co clu io : E braci g
Paralleli
Multi-core i Software
fou datio al opti izatio
Multi-core architecture is Software optimization is
the foundation of modern crucial for maximizing
computing. multi-core performance.
Future Tre d
Future trends point towards greater parallelism and
specialization.
Understanding processor architecture empowers informed hardware
and software decisions.
