Transcript CPU and memory

CPU and memory
Unit 2 Computer Systems
HND in Computing and Systems
Development
CPU
• Highly integrated
• 4.31 billion transistors in 16
core Xeon
• Instruction pipelining (specific
circuit for each part of a
process)
• Superscalar (circuits in
parallel)
• Speed boosting (Turbo boost)
• Caches
Caches
• A CPU cache is used to reduce the amount
of time required to access data from main
memory
• When the CPU needs to read from or write to a
location in main memory it:
• checks whether a copy of that data is in the cache.
• If it is, then the CPU immediately reads from or writes
to the cache,
• This is much faster than reading from or writing to
main memory.
Levels of cache
• Modern CPUs have at least three
independent caches:
• an instruction cache to speed up fetching
instructions
• a data cache to speed up fetching and storing
data
• a translation lookaside buffer (TLB) used to speed
up virtual-to-physical address translation
Levels of cache
• The data cache is
usually organized
as a hierarchy of
more cache levels
(L1, L2)
• This example is an
AMD Athlon 64
• Note the L2 cache
is shared with other
cores
Cache integration
• The smaller caches (L1, L2) are made from
SRAM
• very fast, non volatile, but relatively large area
• Larger caches (L3) use eDRAM (embedded)
• small area, slower, requires refreshing
Further integration
• Memory controller
• Graphics
• Using external memory
• Bus controllers
• PCIe
• Direct media interface (DMI)
to chipset
Other types of CPU
• Microcontrollers
•
•
•
•
•
Embedded in devices (washing machines etc)
processor core
Memory (small)
programmable input/output PIO
Flash memory (program storage)
Other types of CPU
• SoC (system on a
chip)
• Typically used in
smartphones, tablets,
home routers
Choosing CPUs
• What is the application?
•
Productivity, creativity, gaming, business, server,
• Performance
•
Speed, cache, architecture, cores, memory support
• Power consumption
•
Battery life, cooling
• Reliability
•
Intermittent usage, always on, being thrashed
• Graphics
•
Onboard, not needed
• I/O
•
PCI lanes required
Intel/AMD/ARM
• Intel/AMD
• High complexity
• High, but reducing power consumption
• Windows/OSX compatible
• ARM
• Simple RISC architecture
• Low power consumption
• Android/iOS compatibility
Speeds
• CPU clock speed
•
A multiplier of the FSB speed
• Front side bus speed
•
Runs at the same speed as the
memory
• Memory speed
•
Depends on manufacturer’s
specification
• Example
•
CPU 3.2 GHz, multiplier 8, FSB 400
MHz, Memory DDR2-400
Memory
• DDR3
•
•
•
Up to 2800 MHz
( PC3-22400 )
PC3 = DDR3
2240 = bandwidth MB/s
• Packaged in
•
•
DIMM (dual inline memory
module)
SODIMM (small outline
DIMM)
Multi-channel
• Single channel
• Reads from one DIMM
• Dual channel
• Reads from 2 DIMMS in parallel
• Doubles the transfer rate
• Quad channel
• Reads from 4 DIMMS in parallel
• Intel i7 Extreme and some Xeons
Storage
• Compared to memory
• Higher capacity
• Lower cost
• Non volatile
• Types
• Solid state disks (flash) – fastest, costliest, smallest
• Magnetic disks (HDD) – slower, high capacity 10TB,
affordable
• Optical – slower again, very cheap, Blue Ray 128GB
• Tape – slowest, non random, enormous 185TB
Memory/storage hierarchy
Activity – assignment related
• Select matching CPU and memory for:
• A business desktop computer
• A high volume web server
• Refer to the choosing CPUs slide
• Justify your choices