What is
a chip?
A chip is
a microprocessor and is sometimes referred to as the CPU (Central
Processing Unit). The chip is the heart and soul of the machine
and almost everything has to pass through it.
Chip Manufacturers
The two most popular
chip manufactures are Intel & Advanced Micro Devices (AMD) . Other chip manufacturers are Cyrix, Chips and Technologies (C&T), IBM, NEC and Motorola.
Speed
There are three measurements used to determine how fast a chip will
operate. Basically
think of speed in terms of miles per hour when driving a car.
In the similar manner where your car may be capable of going 80
mph, but traffic or road conditions may not allow you to do so,
the same thing is true of processor operations. Thus the purpose
of the three different measurements.
CPU Clock
Speed
This represents
a relative indicator of performance from one chip to another. The CPU Clock Speed is normally the advertised speed for
a chip because it is the fastest of the three measurements.
Memory Bus
Speed
In simplest
terms, this is how fast the chip can talk to memory. Some times
this is represented as a value expressed as part of the chip
model number. For example in older models, the number two in the model number
486DX2/66 represents the memory bus speed. What it means is
the memory bus speed is 33 (66/2) and the real speed is doubled
to get to the clock speed of 66. With newer chip models
there is no reference to this speed.
Expansion
Bus Speed
This represents
the speed at which your PCI or VESA bus can run. In simple terms,
this represents how fast the components in your system can communicate
with the chip. The majority of the time, the Expansion Bus Speed
is the same as the Memory Bus Speed.
Voltage
It is obvious
that chips require power in order to operate, but not all chips
use the same amount of voltage. A lower voltage value is good
for portables because it will consume less battery life. The voltage
has no effect on the execution speed. Of special note is considering
the capability of upgrading to a different chip in the future. You can not easily upgrade to a chip with a different voltage
rating unless you replace the motherboard.
Data Bus
Width
This represents
how large a bite of cache memory can be accessed in a single read
or write. The larger the number, the bigger the chunks, therefore
the faster the performance.
Floating
Point
This option
speeds up numeric computations, only if the software application
supports floating-point. Of note is that most applications
do not use floating-point operations.
Super-scalar
This feature
allows a chip to execute more than one instruction in a single
clock cycle. Therefore Super-scalar chips are faster.
Cache Design
This specifies
the size of the on-board cache and storage capabilities. A larger
cache value minimizes the time consuming accesses to use off-chip
or main memory. Write-back (wb) cache can store both read and
write results, whereas write-through (wt) can only retain reads.
Thus write-back translates into faster operations.
Upgrade Concerns
If you are thinking
of upgrading to a newer chip, keep in mind that even though faster
chips are available, if your other computer components are not updated
to match the new speed, you may not gain much, if any improvement.
Also, depending on the age and style of your motherboard, your machine
may not be compatible with some of the newer chips.
Of course all
of this new technology means that the other computer components and software applications will
have to speed up in order to prevent a bottleneck. What does
it mean? Dont expect to buy a computer today that will last
for five years. It could easily be outdated within one year.
top of page
Current Status
Last update 10/27/06
Performance Dual-Core
If you're editing movies and music, or simply using your system to watch a bunch of video streams, the Performance Dual-Core processors will enable your applications to avoid those annoying, unintentional pauses that are sometimes a prelude to a lock-up.
AMD Athlon 64 FX-62
Introduced in May 2006 with dual-core chips and integrated memory controller. Putting the controller alongside the two CPUs, rather than in a separate area of silicon, enables faster memory access since data doesn't have to traverse a traditional front-side bus. The FX-62 uses the AM2 socket, which upgrades the integrated memory controller to work with faster DDR2 RAM. The socket also brings support for AMD's virtualization technology to the desktop. All of AMD's dual-cores use HyperTransport interconnect to communicate between the processor cores and I/O subsystems. Thermal rating of 125w.
AMD Athlon 64 X2 5000+
Introduced in May 2006, it also uses the AM2 socket. At peak operation HyperTransport can deliver up to 8GB/sec of total system bandwidth. On the downside, although it has a thermal rating of 89W, it seems to run hot. AMD's Cool'n'Quiet driver is available to downshift power usage when the extra juice isn't needed.
| Processor |
Clock Speed |
L2 Cache |
Bus Speed |
64 bit |
Socket |
Athlon 64 FX-62 |
2.8Ghz |
2x1MB |
2000MHz |
Yes |
AM2 |
Athlon 64 FX-60 |
2.8Ghz |
2x1MB |
2000MHz |
Yes |
939 |
Athlon 64 X2 5000+ |
2.6Ghz |
2x512KB |
2000MHz |
Yes |
AM2 |
Athlon 64 X2 5200+ |
2.6Ghz |
2x1MB |
2000MHz |
Yes |
AM2 Due in 4Q 2006 |
Intel Core 2 Extreme X6800
Introduced in July 2006. The X6800 appears to be more amenable than most CPUs to overclocking.
Intel Core 2 Duo E6700, E6600, E6400, E6300
Introduced in July 2006. The E models are slower than the Extreme X6800. The E6700 is only 9% slower than the X6800. The E6300 is 37% slower than the top of the line. All models have the same fast front-side bus and are amenable to overclocking.
| Processor |
Clock Speed |
L2 Cache |
Bus Speed |
64 bit |
Intel VT |
Fab Proc |
Core 2 Extreme X6800 |
2.93Ghz |
4MB (shared) |
1066MHz |
Yes |
Yes |
65nm |
| Core 2 Duo E6700 |
2.66Ghz |
4MB (shared) |
1066MHz |
Yes |
Yes |
65nm |
| Core 2 Duo E6600 |
2.40Ghz |
4MB (shared) |
1066MHz |
Yes |
Yes |
65nm |
| Core 2 Duo E6400 |
2.13Ghz |
4MB (shared) |
1066MHz |
Yes |
Yes |
65nm |
| Core 2 Duo E6300 |
1.86Ghz |
4MB (shared) |
1066MHz |
Yes |
Yes |
65nm |
Mainstream Dual-Core
Mainstream Dual-Core perform sufficiently for the vast majority of users who run office productivity applications along with the heavy Web surfing. The mainstream dual-cores run more than sufficiently fast to run Microsoft's Windows Vista operating system.
AMD Athlon 64 X2 4600+, 4200+
In 2005, the 4600+ was listed as a high-end offering. As of October 2006, the 4600+ and 4200+ remain as solid as ever, but now they fall firmly in the middle of the dual-core pack. In 2006 the two chips are being made in versions for the newer AM2 socket, which supports DDR2 memory. Older versions for the 939 socket have been "end-of-lifed" by AMD.
| Processor |
Clock Speed |
L2 Cache |
Bus Speed |
64 bit |
Socket |
Athlon 64 X2 4600+ |
2.4Ghz |
2x512KB |
2000MHz |
Yes |
AM2 |
| Athlon 64 X2 4200+ |
2.6Ghz |
2x1MB |
2000MHz |
Yes |
AM2 |
Intel Pentium Extreme Edition 965, 955
Introduced in January 2006 and based on the NetBurst architecture used in most Pentiums. Support has been added for Intel's Hyper-Threading which makes it easy to run multiple threads providing better multitasking performance. Also added is hardware-assisted Virtualization technologies which was largely unsupported in Intel's first dual-core chips. This lets users run whole operating systems and apps in separate partitions, turning one physical CPU into a couple of virtual processors. Uses Intel's advanced 65nm semiconductor fabrication process,which upgrades them from the 90nm used for the earlier 8XX line and putting them on par with the Core 2 Duos.
Intel Pentium D 960, 950, 945, 920, 915
Ranging in clock speed from 2.8GHz to 3.6GHz. Support HyperThreading and Intel's Virtualization technology. To pare down its huge array of dual-core SKUs, Intel in mid-August 2006 issued a notice that it would stop selling the 3GHz 930 and the 3.2GHz 940 by the end of 2006. While both the Core 2 Duo and the 9XX use the same Intel 775 socket, the Core 2 Duos require a motherboard equipped with the proper Intel core-logic chipset and updated firmware.
| Processor |
Clock Speed |
L2 Cache |
Bus Speed |
64 bit |
Intel VT |
Fab Proc |
Pentium Extreme Edition 965 |
3.73Ghz |
2x2MB |
1066MHz |
Yes |
Yes |
65nm |
| Pentium Extreme Edition 955 |
3.46Ghz |
2x2MB |
1066MHz |
Yes |
Yes |
65nm |
| Pentium D 960 |
3.60Ghz |
2x2MB |
800MHz |
Yes |
Yes |
65nm |
| Pentium D 950 |
3.40Ghz |
2x2MB |
800MHz |
Yes |
Yes |
65nm |
| Pentium D 945 |
3.40Ghz |
2x2MB |
800MHz |
Yes |
Yes |
65nm |
| Pentium D 920 |
2.80Ghz |
2x2MB |
800MHz |
Yes |
Yes |
65nm |
| Pentium D 915 |
2.80Ghz |
2x2MB |
800MHz |
Yes |
Yes |
65nm |
Bargain Dual-Core
AMD Athlon 64 X2 3800+
The 3800+ has the same 2000MHz HyperTransport bus as its higher-end 5000+ sibling With a 2GHz clock and 2x512KB L2 cache. AMD's low-end dual-core isn't outmoded technologically, but it's positioned as an entry-level offering and priced accordingly. If you want to run Vista on a dual-core you can't get into the market any cheaper than with one of these CPUs.
| Processor |
Clock Speed |
L2 Cache |
Bus Speed |
64 bit |
Socket |
Athlon 64 X2 3800+ |
2.0Ghz |
2x512KB |
2000MHz |
Yes |
AM2 & 939 |
Intel Pentium D 805, 820
Both models are fabricated in Intel's older 90nm process technology and have smaller L2 caches than their 9XX cousins (2x1MB versus 2x2MB). Other than that there's no significant difference. Both are good on power consumption with a 95W thermal design power spec.
| Processor |
Clock Speed |
L2 Cache |
Bus Speed |
64 bit |
Intel VT |
Fab Proc |
Pentium D 820 |
2.80Ghz |
2x1MB |
800MHz |
Yes |
No |
90nm |
| Pentium D 805 |
2.66Ghz |
2x1MB |
533MHz |
Yes |
No |
90nm |
Faster Single-Core
AMD Athlon 64 3800+, 3500+, 3200+
Single-core Athlon 64 designs were among the first desktop processors to implement AMD's groundbreaking 64-bit architecture.
| Processor |
Clock Speed |
L2 Cache |
Bus Speed |
64 bit |
Socket |
Athlon 64 3800+ |
2.4Ghz |
512KB |
2000MHz |
Yes |
AM2 & 939 |
| Athlon 64 3500+ |
2.2Ghz |
512KB |
2000MHz |
Yes |
AM2 & 939 |
| Athlon 64 3200+ |
2.0Ghz |
512KB |
2000MHz |
Yes |
AM2 |
Intel Pentium 4 670, 661, 660, 651, 641, 631, 524, 521
This family still delivers solid performance, ranging from a 3.8GHz clock and 2MB L2 cache for the 670 down to a not-unimpressive 3.0GHz with the same cache for the 631. The 5XX series preceded the 6XX and is fabricated using older 90nm technology.
| Processor |
Clock Speed |
L2 Cache |
Bus Speed |
64 bit |
Intel VT |
Fab Proc |
Pentium 4 670 |
3.80Ghz |
2MB |
800MHz |
Yes |
No |
90nm |
| Pentium 4 661 |
3.60Ghz |
2MB |
800MHz |
Yes |
No |
65nm |
| Pentium 4 660 |
3.60Ghz |
2MB |
800MHz |
Yes |
No |
90nm |
| Pentium 4 651 |
3.40Ghz |
2MB |
800MHz |
Yes |
No |
65nm |
| Pentium 4 641 |
3.20Ghz |
2MB |
800MHz |
Yes |
No |
65nm |
| Pentium 4 631 |
3.00Ghz |
2MB |
800MHz |
Yes |
No |
65nm |
| Pentium 4 524 |
3.06Ghz |
1MB |
800MHz |
Yes |
No |
90nm |
| Pentium 4 521 |
2.80Ghz |
1MB |
800MHz |
Yes |
No |
90nm |
Low-End Single-Core
AMD Sempron 3600+, 3500+, 3400+, 3200+, 3000+
The Semprons come in two different sockets: older 754 or AM2 supporting DDR2. The Semprons also boast a faster system bus and uses less power than the Celerons. All of AMD's current processors are fabricated using 90nm process technology. A 65nm process is being readied at the company's new Dresden facility.
| Processor |
Clock Speed |
L2 Cache |
Bus Speed |
64 bit |
Socket |
Sempron 3600+ |
2.0Ghz |
256KB |
1600MHz |
Yes |
AM2 |
| Sempron 3500+ |
2.0Ghz |
128KB |
1600MHz |
Yes |
AM2 |
| Sempron 3400+ |
1.8Ghz |
256KB |
2000MHz |
Yes |
AM2 |
| Sempron 3400+ |
2.0Ghz |
256KB |
2000MHz |
Yes |
754 |
| Sempron 3200+ |
1.8Ghz |
128KB |
1600MHz |
Yes |
AM2 |
| Sempron 3000+ |
1.6Ghz |
256KB |
800MHz |
Yes |
AM2 |
| Sempron 3000+ |
1.8Ghz |
128KB |
1600MHz |
Yes |
754 |
Intel Celeron D 356, 355, 352, 351, 350, 346, 345, 341, 340, 331, 326
The "D" after Celeron does not put these parts in the same class as the Pentium D. These are low-end, single-core processors. This family is divided up between Intel's 775 and 478 socket. Both types use a 533MHz front-side bus. The Celeron D 340, 345, and 350 use the 478 socket. They don't support Intel's 64-bit EM64T instruction set extensions. The Celeron D 326, 331, 341, 346, 351, 352, 355, and 356 are socket 775 and do support EM64T.
| Processor |
Clock Speed |
L2 Cache |
Bus Speed |
64 bit |
Intel VT |
Fab Proc |
Celeron D 356 |
3.33Ghz |
512KB |
533MHz |
Yes |
No |
65nm |
| Celeron D 355 |
3.33Ghz |
256KB |
533MHz |
Yes |
No |
90nm |
| Celeron D 352 |
3.20Ghz |
512KB |
533MHz |
Yes |
No |
65nm |
| Celeron D 351 |
3.20Ghz |
256KB |
533MHz |
Yes |
No |
90nm |
| Celeron D 350 |
3.20Ghz |
256KB |
533MHz |
No |
No |
90nm |
| Celeron D 346 |
3.06Ghz |
256KB |
533MHz |
Yes |
No |
90nm |
| Celeron D 345 |
3.06Ghz |
256KB |
533MHz |
No |
No |
90nm |
| Celeron D 341 |
2.93Ghz |
256KB |
533MHz |
Yes |
No |
90nm |
| Celeron D 340 |
2.93Ghz |
256KB |
533MHz |
No |
No |
90nm |
| Celeron D 331 |
2.66Ghz |
256KB |
533MHz |
Yes |
No |
90nm |
| Celeron D 326 |
2.53Ghz |
256KB |
533MHz |
Yes |
No |
90nm |
top of page
This
article won't help you cure any problems or provide any solutions.
It is for those of you who enjoy an occasional look at the past. I did not intend
to slight other chip manufacturers, but instead tried to present those
models that made a difference in the usage and popularity of computers. Hope
you enjoy.
| 1968 - 1970 |
Datapoint 2200 |
| In 1968 Phil
Ray and Gus Roche founded a firm called Computer Terminal Corporation (CTC).
In 1969 CTC wanted
to build a "smart" terminal which could mimic the terminals
of all the major computer vendors. To do this it would have to have
its own microprocessor. They hired Jack Frassanito to help them with the design. The original design was on printed circuit
boards but size constraints created heating problems. CTC contacted
both Intel and Texas Instruments about reducing the circuitry to a
single chip but neither Intel or TI completed the chip before CTC
unveiled its Datapoint 2200 terminal in June of 1970.
When CTC contacted
their customers to find out how things were going, they discovered
that a number of their customers were not using them as terminals
but programming the microprocessor themselves in machine language! They were using them as PCs! (Jack Frassanito's
name is on the patent which was issued on July 25, 1972.)
Thanks
to Rick Gaffney for providing the above information. |
| 1971 |
Intel 4004 |
| The world's first
general purpose microprocessor. If consisted of 2,300 transistors
and supported only 45 instructions. It ran at an amazing speed of
1 MHz using 4 bit architecture. It was created by the team of Ted
Hoff, Stan Mazor, Federico Faggin and Masatoshi Shima. |
| 1972 |
Intel 8008 |
| Intel finally
came back to CTC with their design of the first 8 bit microprocessor but by then CTC was no longer interested.
Intel designated their chip the 8008 and put it in their catalogue
of chips. This is the first
8-bit model and it contained 3,500 transistors. The 8th bit provided
the ability to manage alpha-numeric data. There were two clock speeds of 500 & 800Khz and supported up to 16KB of memory. |
| 1974 |
Intel 8080 |
| This chip allowed
16 bit addressing, 6,000 transistors and a whopping speed of 2 MHz.
This is the chip that was used in the MITS Altair computer. The Altair
was the third microcomputer. Bill Gates and Paul Allen started Microsoft and wrote their first BASIC interpreter for this chip and the Altair
computer.
The following
is a correction provided by Chris Titus of Milford, MA to my original
statement that the Altair was the first microcomputer.
The Altair computer
actually appeared first on the cover of the January 1975 issue of
Popular Electronics magazine. According to Les Solomon, of Popular
Electronics, the cover photo taken in the Fall, was made using a nonworking
mock up of the Altair. The Altair continued to experience difficulties
right up through shipping the first systems.
In July 1974, Radio-Electronics magazine published a construction article (and offered
a booklet of instructions, tutorial experiments,schematics and PCB
diagrams) for the Mark-8 computer. A company in New Jersey sold circuit
boards, and a mail-order company in Texas sold kits of parts. The
Mark-8, not the Altair, is the second home computer. It
beat the Altair by six months. And it worked. Larry Steckler,
the editor of Radio-Electronics saw a working model--the one on the
cover--in February or March of 1974. The original Mark-8 is now
in the Smithsonian's permanent Information Age exhibit. |
| Motorola 6800 |
| The 6800 contained
4,000 transistors and was designed by Chuck Peddle and Charlie Melear.
It was mainly used for automotive controls and small business computers. |
| 1975 |
Zilog Z80 |
| The Z80 was mainly
responsible for bringing PCs into business. It was designed by the Faggin and Shima team, contained 8,500 transistors and ran at 2.5 MHz. It
hosted the CP/M operating system and was used by early pioneer computer
makers such as Osborne and Kaypro. Names long forgotten. |
| 1976 |
Intel 8085 |
| This chip had a single 5V power supply, 8 bits and up to 8Mhz clock speed. |
| MOS 6502 |
| Chuck Peddle left
Motorola and under his new company, MOS Technologies, created the
6502. This is the first chip that Steve Wozniak used to create the Apple II design. It was also used in the Commodore and Atari machines.
It was basically an enhanced version of the 6800 with 9,000 transistors,
provided faster graphic operations and sold for about $25. |
| 1978 |
Intel 8086 |
| The 8086 was the
start of the x86 family. It was a 16 bit chip
with 29,000 transistors and up to 10Mhz clock speed. |
| 1979 |
Intel 8088 |
| Almost identical to the 8086 but had an external 8 bit data bus. This chip is based
on the 8086 and was the model that IBM used in their first PCs. This
chip helped launch DOS and Lotus 123. |
| Motorola 68000 |
| The 68000 used
32 bits and 68,000 transistors. Apple's Lisa and Macintosh computers
were based on this chip, which also introduced the first successful
graphical user interface (GUI) systems. Many Unix based systems, such
as NCR and Motorola, used this chip for their Unix servers. |
| 1982 |
Intel 80286 |
| This model introduced
the concept of general protection and virtual memory. It supported
up to 16MB of physical memory, 134,000 transistors and a 16 bit bus. A blazing speed
of 8 to 25 MHz. IBM's AT bus was first used with this chip. |
| 1985 |
Intel 80386 |
| Third genereation chip that introduced linear addressing, used a 32 bit design and contained 275,000 transistors.
It could address up to 4GB of memory and clock speeds of 33Mhz. The first versions of Windows and OS/2 were introduced with this chip. This model also made users
more aware of the type of chip and operating system being used instead
of the actual brand computer. |
| 1986 |
MIPS R2000 |
| The first RISC
based chip was introduced containing 185,000 transistors. |
| 1987 |
Sun SPARC |
| Sun introduces
its first chip, with only 50,000 transistors, that spawned several
generations of RISC based workstations. These workstations were geared
for heavy graphic use. |
| 1989 |
Intel 80486 |
| It contained 1,200,000 transistors
and introduced the ability to include a floating-point unit and 8K
of internal RAM cache within the chip. This greatly improved the execute
speed for GUI interfaces, such as Windows. Speed was increased to
66 MHz using internal clocking. |
| 1993 |
Intel Pentium |
| After the longest
gap in the development of new models, Intel created the Pentium that
contains 3,100,000 transistors. It is capable of executing two instructions
at the same time using a dual-pipeline design. Initially ran very hot. This chip became the
defacto standard in it's time. |
| IBM/Motorola PowerPC
601 |
| From the results
of a joint effort, a new RISC chip was created using 2,800,000 transistors.
This chip was used in IBM's RS/6000 and Apple's Power Macs computers. |
| 1996 |
Intel Pentium Pro |
| The Pentium Pro
increased the ability to execute three instructions at once and contains
5,500,000 transistors. It contains a second chip that provides Level
2 caching. |
| 1997 |
Intel Pentium MMX |
Introduced in the early part of 1997 and labeled as model MMX. It was designed to improve the performance for multimedia and communications.
The estimates were a boost of performance of up to 60%. Speed
increased to 200 MHz.
How
did it work
To accomplish
the speed improvements, Intel added 57 new instructions to
the chip set. What used to require several dozen CPU steps was accomplished
in a single step or operation.
- 2Q
1997
- Pentium
II @233 and 266Mhz. Pentium II chips incorporate the MMX technology.
- Pentium
MMX @ 233Mhz.
- 3Q
1997
- Tillamook
- code name for Pentium MMX designed for notebooks.
|
| AMD K6 |
| A 32-bit microprocessor that had RISC86 core that was used in the Nexgen Nx586 and Nx686 microprocessors. Special decoder units translated complex x86 instructions into short RISC-like instructions. This design allowed execution of up to 6 RISC-like instructions per cycle. It had the ability to run at frequences higher then processors with CISC-based cores. It supported the MMX instruction set. Models were produced for both desktop and laptops. There were 2 clock speeds available:166 and 266Mhz. |
| |