Amd athlon identification: AMD Athlon SlotA type identification

Athlon MP — AMD — WikiChip

Athlon MP
Athlon MP logo

Developer AMD
Manufacturer AMD
Type Microprocessors
Introduction June 5, 2001 (announced)
June 5, 2001 (launch)
Architecture Server x86 multiprocessors
ISA x86
µarch K7
Word size 32 bit

4 octets
8 nibbles

Process 180 nm

0.18 μm
1.8e-4 mm

, 130 nm

0.13 μm
1.3e-4 mm

Technology CMOS
Clock 1,000 MHz-2,130 MHz
Package CPGA-453
Socket Socket A
Athlon Opteron

Athlon MP (Athlon Multiprocessor) was a family of 32-bit x86 server multiprocessors designed by AMD specifically for the server and workstations market. Athlon MP was AMD’s first multiprocessing-capable platform.


  • 1 Overview
    • 1.1 SmartMP Technology
    • 1.2 Modding Athlon XP
  • 2 Chip Identification
  • 3 Models
    • 3.1 Palomino Core
    • 3.2 Thoroughbred Core
    • 3.3 Barton Core
  • 4 Documents
    • 4.1 Datasheets
    • 4.2 Others
  • 5 Artwork


See also: K7 Microarchitecture

Tyan S2462 motherboard for dual-socket Athlon MP processors.

AMD announced their first multiprocessing-capable platform at Computex Taipei on June 5th, 2001. As with all the other Athlon families, the Athlon MP is also based on the K7 microarchitecture. The platform includes the Athlon MP processors as well as the AMD-760MP northbridge chipset. AMD-760MP supports one- and two-way setups and Double Data Rate (DDR) memory operating at 133 MHz. At the time, AMD’s vice president for their servers group stated Athlon MP processor delivers up to 38% higher performance over their competition (presumably referring to Xeon).

SmartMP Technology[edit]

Main article: SmartMP Technology
This section is empty; you can help add the missing info by editing this page.

Modding Athlon XP[edit]

In early 2002 published an article detailing the ability to modify a stock Athlon XP processor (which shares an identical core to the Athlon MPs) to allow it to recognize dual-socket configuration effectively turning it into an expensive Athlon MP processor. The modification is a trivial bridge on the L5 circuit on the top of the Athlon XP package (see article for detail). Various degrees of success has been reported.

Chip Identification[edit]

Athlon MP processor along with the AMD-760MP chipset (north and south bridge)

A HX 1200 A M S 3 C  
A MS N 1200 D K T 3 B  
B — 100 MHz (200 MT/s)
C — 133 MHz (266 MT/s)
                L2$ Size:
3 — 256 KiB
4 — 512 KiB
S — 95 °C
T — 90 °C
V — 85 °C

L — 1. 5 V H — 1.55 V
U — 1.6 V K — 1.65 V
P — 1.7 V M — 1.75 V
N — 1.80 V
        Speed (MHz)
      Max Power:
N — 60 W
HX — High-Performance Multiprocessors
MP — High-Performance Multiprocessors
MS — High-Performance Multiprocessors with QuantiSpeed
A — Athlon-based (K7)

Palomino Core[edit]

Palomino-based microprocessors (i.e. Model 6) were manufactured on AMD’s mature 180 nm process copper interconnect technology at Fab 30 foundry in Dresden, Germany. The core implements an exclusive 256 KiB L2$ and a 128 KiB L1$. As with all Socket A processors (EV6 system bus), Athlon MP operate on a 133 MHz FSB DDR (double data rate) yielding an effective 266 MT/s transfer rate (note that ‘B’ models operated at a lower FSB of 100 MHz). These processors support MMX, SSE, Enhanced 3DNow!, and SmartMP Technology. AMD came short with Palomino by not supporting SSE2 which came out in the various Pentium 4 models that were released around the same time.

Palomino-based Athlon MP Microprocessors
Model OPN Price Launched L2$ Freq Multiplier TDP VCORE
Athlon MP 1000 AHX1000AMS3C $ 215.00

€ 193.50
£ 174.15
¥ 22,215.95

5 June 2001 256 KiB

0.25 MiB
262,144 B
2.441406e-4 GiB

1 GHz

1,000 MHz
1,000,000 kHz

7.5 46.1 W

46,100 mW
0.0618 hp
0.0461 kW

1. 75 V

17.5 dV
175 cV
1,750 mV

Athlon MP 1200 AHX1200DHS3C $ 265.00

€ 238.50
£ 214.65
¥ 27,382.45

5 June 2001 256 KiB

0.25 MiB
262,144 B
2.441406e-4 GiB

1.2 GHz

1,200 MHz
1,200,000 kHz

9 1.55 V

15.5 dV
155 cV
1,550 mV

Athlon MP 1200 AHX1200ANS3B $ 265.00

€ 238.50
£ 214.65
¥ 27,382.45

5 June 2001 256 KiB

0.25 MiB
262,144 B
2.441406e-4 GiB

1.2 GHz

1,200 MHz
1,200,000 kHz

12 1.8 V

18 dV
180 cV
1,800 mV

Athlon MP 1500+ AMP1500DMS3C $ 180.00

€ 162.00
£ 145.80
¥ 18,599.40

15 October 2001 256 KiB

0.25 MiB
262,144 B
2.441406e-4 GiB

1.333 GHz

1,333 MHz
1,333,000 kHz

10 60 W

60,000 mW
0. 0805 hp
0.06 kW

1.75 V

17.5 dV
175 cV
1,750 mV

Athlon MP 1600+ AMP1600DMS3C $ 210.00

€ 189.00
£ 170.10
¥ 21,699.30

15 October 2001 256 KiB

0.25 MiB
262,144 B
2.441406e-4 GiB

1.4 GHz

1,400 MHz
1,400,000 kHz

10.5 62.8 W

62,800 mW
0.0842 hp
0.0628 kW

1.75 V

17.5 dV
175 cV
1,750 mV

Athlon MP 1800+ AMP1800DMS3C $ 302.00

€ 271.80
£ 244.62
¥ 31,205.66

15 October 2001 256 KiB

0.25 MiB
262,144 B
2.441406e-4 GiB

1.533 GHz

1,533 MHz
1,533,000 kHz

11.5 66 W

66,000 mW
0.0885 hp
0.066 kW

1.75 V

17.5 dV
175 cV
1,750 mV

Athlon MP 1900+ AMP1900DMS3C $ 319.00

€ 287. 10
£ 258.39
¥ 32,962.27

12 December 2001 256 KiB

0.25 MiB
262,144 B
2.441406e-4 GiB

1.6 GHz

1,600 MHz
1,600,000 kHz

12 66 W

66,000 mW
0.0885 hp
0.066 kW

1.75 V

17.5 dV
175 cV
1,750 mV

Athlon MP 2000+ AMP2000DMS3C $ 319.00

€ 287.10
£ 258.39
¥ 32,962.27

13 March 2002 256 KiB

0.25 MiB
262,144 B
2.441406e-4 GiB

1.667 GHz

1,667 MHz
1,667,000 kHz

12.5 66 W

66,000 mW
0.0885 hp
0.066 kW

1.75 V

17.5 dV
175 cV
1,750 mV

Athlon MP 2100+ AMP2100DMS3C $ 262.00

€ 235.80
£ 212.22
¥ 27,072.46

19 June 2002 256 KiB

0.25 MiB
262,144 B
2.441406e-4 GiB

1.733 GHz

1,733 MHz
1,733,000 kHz

13 66 W

66,000 mW
0. 0885 hp
0.066 kW

1.75 V

17.5 dV
175 cV
1,750 mV

Count: 9

Thoroughbred Core[edit]

AMD introduced Thoroughbred-based processors (i.e. Model 8) in late 2002. Those chips were manufactured on a newer 130 nm process which allowed them to be clocked at higher frequencies. The new process also allowed AMD to rearrange their design which allowed them to shave off roughly 300,000 transistors. These processors support MMX, SSE, Enhanced 3DNow!, and SmartMP Technology.

Thoroughbred-based Athlon MP Microprocessors
Model OPN Price Launched L2$ Freq Multiplier TDP VCORE
Athlon MP 2000+ AMSN2000DKT3C 27 August 2002 256 KiB

0.25 MiB
262,144 B
2.441406e-4 GiB

1.667 GHz

1,667 MHz
1,667,000 kHz

12. 5 1.65 V

16.5 dV
165 cV
1,650 mV

Athlon MP 2000+ AMSN2000DUT3C 27 August 2002 256 KiB

0.25 MiB
262,144 B
2.441406e-4 GiB

1.667 GHz

1,667 MHz
1,667,000 kHz

12.5 58.2 W

58,200 mW
0.078 hp
0.0582 kW

1.6 V

16 dV
160 cV
1,600 mV

Athlon MP 2200+ AMSN2200DKT3C $ 224.00

€ 201.60
£ 181.44
¥ 23,145.92

27 August 2002 256 KiB

0.25 MiB
262,144 B
2.441406e-4 GiB

1.8 GHz

1,800 MHz
1,800,000 kHz

13.5 60 W

60,000 mW
0.0805 hp
0.06 kW

1.65 V

16.5 dV
165 cV
1,650 mV

Athlon MP 2400+ AMSN2400DUT3C $ 228.00

€ 205.20
£ 184.68
¥ 23,559.24

10 December 2002 256 KiB

0. 25 MiB
262,144 B
2.441406e-4 GiB

2 GHz

2,000 MHz
2,000,000 kHz

15 1.6 V

16 dV
160 cV
1,600 mV

Athlon MP 2600+ AMSN2600DKT3C $ 273.00

€ 245.70
£ 221.13
¥ 28,209.09

4 February 2003 256 KiB

0.25 MiB
262,144 B
2.441406e-4 GiB

2.133 GHz

2,133 MHz
2,133,000 kHz

16 60 W

60,000 mW
0.0805 hp
0.06 kW

1.65 V

16.5 dV
165 cV
1,650 mV

Count: 5

Barton Core[edit]

The last chip in the Athlon MP series was introduced in early 2003. The Barton-based processor (i.e. model 10), which was also manufactured on a 130 nm process, doubled the amount of level 2 cache (to 512 KiB). Barton-based processors sold for significantly lower price than the newer Opteron models which made them attractive for entry-level servers and workstations. These processors support MMX, SSE, Enhanced 3DNow!, and SmartMP Technology.

Barton-based Athlon MP Microprocessors
Model OPN Price Launched L2$ Freq Multiplier TDP VCORE
Athlon MP 2600+ AMSN2600DUT4C 6 May 2003 512 KiB

0.5 MiB
524,288 B
4.882812e-4 GiB

2 GHz

2,000 MHz
2,000,000 kHz

15 60 W

60,000 mW
0.0805 hp
0.06 kW

1.6 V

16 dV
160 cV
1,600 mV

Athlon MP 2800+ AMSN2800DUT4C $ 275.00

€ 247.50
£ 222.75
¥ 28,415.75

6 May 2003 512 KiB

0.5 MiB
524,288 B
4.882812e-4 GiB

2.133 GHz

2,133 MHz
2,133,000 kHz

16 60 W

60,000 mW
0.0805 hp
0.06 kW

1.6 V

16 dV
160 cV
1,600 mV

Count: 2



  • AMD Athlon MP Processor Model 6 Data Sheet Multiprocessor-Capable for Workstation and Server Platforms; Publication # 24685; Rev. : B; Issue Date: June 2001.
  • AMD Athlon MP Processor Model 6 OPGA Data Sheet for Multiprocessor Platforms; Publication # 25480 Rev: D; Issue Date: June 2002.
  • AMD Athlon MP Processor Model 8 Data Sheet for Multiprocessor Platforms; Publication # 25722 Rev. E; Issue Date: March 2003.
  • AMD Athlon MP Processor Model 10 Data Sheet for Multiprocessor Platforms; Publication # 26426 Rev. C; Issue Date: October 2003.


  • System Considerations for Dual AMD Athlon MP Processors in Tower and 1U Form Factors; Publication # 25325; Rev: B; August 2002.
  • AMD-760 Chipset & DDR Memory Presentation; October 2000.


amd%20athlon%20ii%20pin%20layout%20voltage%20ground datasheet & applicatoin notes

Top Results (6)

Part ECAD Model Manufacturer Description Datasheet Download Buy Part

Texas Instruments


Texas Instruments


Texas Instruments


Texas Instruments


Renesas Electronics Corporation

Compact High Efficiency Low Power Buck-Boost Regulator

Renesas Electronics Corporation

Compact High Efficiency Low Power Buck-Boost Regulator

amd%20athlon%20ii%20pin%20layout%20voltage%20ground Datasheets Context Search

Catalog Datasheet MFG & Type PDF Document Tags

Abstract: irf 3502 AM27C010-150DC ADSP-2181KST-133 AM27C010-120PC AM26LS33PC AM27C040-150DC AM26LS32BPC AD823AN ADSP2181KS115

Text: No file text available



irf 3502
2000 — TSC 13003

Abstract: circuit diagrams retu 3.02 transistor EN 13003 A transistor EN 13003 H vid 13003 amd duron 1600 datasheet AMD xp datasheet amd duron 1100 AMD ATHLON 2400 AMD Athlon XP 2000

Text: No file text available



TSC 13003
circuit diagrams retu 3.02
transistor EN 13003 A
transistor EN 13003 H
vid 13003
amd duron 1600 datasheet
AMD xp datasheet
amd duron 1100
AMD Athlon XP 2000
2008 — AMD Athlon 64 X2 4200

Abstract: RS780E AMD ATHLON 64 X2 AM2 pin out RADEON e2400 AMD Athlon 64 X2 AMD turion 64 X2 AMD Athlon II X2 ATI Radeon X1250 AMD Phenom II AMD Turion 64 X2 thermal

Text: No file text available



AMD Athlon 64 X2 4200
AMD ATHLON 64 X2 AM2 pin out
RADEON e2400
AMD Athlon 64 X2
AMD turion 64 X2
AMD Athlon II X2
ATI Radeon X1250
AMD Phenom II
AMD Turion 64 X2 thermal
2002 — amd athlon 5400

Abstract: athlon pin configuration athlon 64 motherboard design Athlon 64 M AMD-766 AMD-762 AMD-760TM AMD-760 AMD Athlon 64 PC2100

Text: No file text available



amd athlon 5400
athlon pin configuration
athlon 64 motherboard design
Athlon 64 M
AMD Athlon 64

Abstract: ADV478KP50E AM27S19/BEA bra 94 4753 AM26LS31/BEA AM27S13A am2902apc AM27S27DC AM25LS2521PCB

Text: No file text available

OCR Scan


bra 94
2002 — «AMD Processors»

Abstract: AMD-Athlon4 Mobile Athlon

Text: No file text available



2002 — AMD xp datasheet

Abstract: AMD Athlon 64 pin diagram pentium4 instruction set AMD Athlon 64 athlon xp powernow ATHLON XP amd athlon datasheet Athlon 64 Thoroughbred athlon amd Athlon xp mobile

Text: No file text available



2001 — amd motherboard

Abstract: southbridge AMD ATHLON 64 X 2 AMD-766 AMD-762 AMD-760 Athlon MP datasheet Athlon 64 AMD-760TM AMD BGA

Text: No file text available



amd motherboard
Athlon MP datasheet
Athlon 64
1999 — Athlon

Abstract: AMD Athlon 64 athlon 64 motherboard design «AMD Processors» cache athlon 64 Athlon 64 datasheet Athlon 64 AMD athlon core datasheet amd athlon processor diagram

Text: No file text available



ATI Radeon

Abstract: ATI Radeon X1250 ATI RS690 AMD 690g AMD690G 690G amd northbridge chipset RS690 X1250 RADEON x1250

Text: No file text available




Abstract: AMD-768 AMD-762TM AMD Athlon 64 pin diagram PC2100 pin diagram AMD Athlon 64 AMD-762 AMD-760 AMD-760TM AMD Athlon 64

Text: No file text available



AMD Athlon 64 pin diagram
pin diagram AMD Athlon 64
AMD Athlon 64
2003 — AMD Thermal Design Guide

Abstract: AMD Processor Power and Thermal OSA244CCO5AH OPTERON 2200 OSA244CEP5AL AMD 940 Pin Package amd Power and Thermal AMD Athlon 64 and amd opteron OSA144CCO5AG OSA142CCO5AG

Text: No file text available



2011 — E4690

Abstract: AMD Phenom II x4 E6760 ati radeon bga Phenom II x4 965 amd RADEON e6760 TW3X4G1333C9A AMD Phenom II AMD Athlon II X4 athlon 64 x2

Text: No file text available



AMD Phenom II x4
ati radeon bga
Phenom II x4 965
amd RADEON e6760
AMD Phenom II
AMD Athlon II X4
athlon 64 x2
2001 — AMD datecode marking CODE

Abstract: AMD-762 AMD-762JLC AMD-766 AMD marking CODE amd part «marking» marking agp

Text: No file text available



AMD datecode marking CODE
AMD marking CODE
amd part «marking»
marking agp
2002 — sis 735

Abstract: SiS 7012 K7S5A soundmax integrated digital audio ecs k7s5a intel desktop board d845wn intel desktop board d845wn data sheet i815E Duron d845wn

Text: No file text available



2004 — amd athlon 64 3200 754 pin

Abstract: AMD Processor Power and Thermal cache «athlon 64» AMD FX Processor amd athlon 2800 amd athlon 64 3200 AMD Family 12h AMD Athlon 64 2800 amd athlon 64 3000 amd athlon 64 3000 754 pin

Text: No file text available



amd athlon 64 3200 754 pin
AMD Processor Power and Thermal
cache «athlon 64»
AMD FX Processor
amd athlon 2800
amd athlon 64 3200
AMD Family 12h
AMD Athlon 64 2800
amd athlon 64 3000
amd athlon 64 3000 754 pin




Infobox Computer Hardware Cpu
name = Athlon

caption = AMD Athlon logo
produced-start = mid 1999
produced-end = 2005
slowest = 500 | slow-unit = MHz
fastest = 2. 33 | fast-unit = GHz
fsb-slowest = 100 | fsb-slow-unit = MHz
fsb-fastest = 200 | fsb-fast-unit = MHz
size-from = 0.25
size-to = 0.13
manuf1 = AMD
core1 = K7 (Argon)
core2 = K75 (Pluto/Orion)
core3 = Thunderbird
core4 = Palomino
core5 = Thoroughbred A/B
core6 = Barton/Thorton
sock1 = Slot A
sock2 = Socket A
arch = x86

Athlon is the brand name applied to a series of different x86processors designed and manufactured by AMD. The original Athlon, or «Athlon Classic», was the first seventh-generation x86 processor and, in a first, retained the initial performance lead it had over Intel’s competing processors for a significant period of time. AMD has continued the «Athlon» name with the Athlon 64, an eighth-generation processor featuring x86-64 (later renamed AMD64) technology.

The Athlon made its debut on June 23, 1999. Athlon was the ancient Greek word for «Champion/trophy of the games».


AMD ex-CEO and founder Jerry Sanders developed strategic partnerships during the late 1990s to improve AMD’s presence in the PC market based on the success of the K6 architecture. One major partnership announced in 1998 paired AMD with semiconductor giant Motorola. [cite web|url= |title=Motorola Prepares to Manufacture AMD’s Upcoming K7 Chip |publisher=HP |date=1998-08-07] In the announcement, Sanders referred to the partnership as creating a «virtual gorilla» that would enable AMD to compete with Intel on fabrication capacity while limiting AMD’s financial outlay for new facilities. This partnership also helped to co-develop copper-based semiconductor technology, which would become a cornerstone of the K7 production process.

In August 1999, AMD released the Athlon (K7) processor. Notably, the design team was led by Dirk Meyer, one of the lead engineers on the DEC Alpha project. Jerry Sanders had approached many of the engineering staff to work for AMD as DEC wound the project down, and brought in a near-complete team of engineering experts. The balance of the Athlon design team comprised AMD K5 and K6 veterans.

By working with Motorola, AMD was able to refine copper interconnect manufacturing to the production stage about one year before Intel. The revised process permitted 180-nanometer processor production. The accompanying die-shrink resulted in lower power consumption, permitting AMD to increase Athlon clockspeeds to the 1 GHz range. [ [,,51_104_543_552~729,00.html AMD Announces First Revenue Shipments From Dresden «MEGAFAB»] , AMD Press Release, June 5, 2000.] AMD found processor yields on the new process exceeded expectations, and delivered high speed chips in volume in March 2000.

General architecture

Internally, the Athlon is a fully seventh generation x86 processor, the first of its kind. Like the AMD K5 and K6, the Athlon is a RISC microprocessor which decodes x86 instructions into its own internal instructions at runtime. The CPU is an out-of-order design, again like previous post-5×86 AMD CPUs. The Athlon utilizes the DEC Alpha EV6 bus architecture with double data rate (DDR) technology. This means that at 100 MHz the Athlon front side bus actually transfers at a rate similar to a 200 MHz single data rate bus (referred to as 200 MT/s), which was superior to the method used on Intel’s Pentium III (with SDR bus speeds of 100 MHz and 133 MHz).

AMD designed the CPU with more robust x86 instruction decoding capabilities than that of K6, to enhance its ability to keep more data in-flight at once. Athlon’s CISC to RISC decoder triplet could potentially decode 6 x86 operations per clock, although this was somewhat unlikely in real-world use.Hsieh, Paul. [ 7th Generation CPU Comparisons] .] The critical branch predictor unit, essential to keeping the pipeline busy, was enhanced compared to what was onboard the K6. Deeper pipelining with more stages allowed higher clock speeds to be attained. [De Gelas, Johan. [ The Secrets of High Performance CPUs, Part 1] , Ace’s Hardware, September 29, 1999.] Whereas the AMD K6-III+ topped out at 570 MHz due to its short pipeline, even when built on the 180 nm process, the Athlon was capable of going much higher.

AMD ended its long-time handicap with floating pointx87 performance by designing a super-pipelined, out-of-order, triple-issue floating point unit. Each of its 3 units were tailored to be able to calculate an optimal type of instructions with some redundancy. By having separate units, it was possible to operate on more than one floating point instruction at once. This FPU was a huge step forward for AMD. While the K6 FPU had looked anemic compared to the Intel P6 FPU, with Athlon this was no longer the case. [Pabst, Thomas. [ Performance-Showdown between Athlon and Pentium III] , Tom’s Hardware, August 23, 1999. ]

The 3DNow!floating pointSIMD technology, again present, received some revisions and a name change to «Enhanced 3DNow!». Additions included DSP instructions and an implementation of the extended MMX subset of Intel SSE. [Womack, Tom. [ Extensions to the x86 architecture] .]

CPU Caching onboard Athlon consisted of the typical two levels. Athlon was the first x86 processor with a 128 KB [BDprefix|p=b] split level 1 cache; a 2-way associative, later 16-way, cache separated into 2×64 KB for data and instructions (Harvard architecture). This cache was double the size of K6’s already large 2×32 KB cache, and quadruple the size of Pentium II and III’s 2×16 KB L1 cache. The initial Athlon (Slot A, later renamed Athlon Classic) used 512 KB of level 2 cache separate from the CPU, on the processor cartridge board, running at 50% to 33% of core speed. This was done because the 250 nm manufacturing processes was too large to allow for on-die cache while maintaining cost-effective die size. Later Athlon CPUs, afforded greater transistor budgets by smaller 180 nm and 130 nm process nodes, moved to on-die L2 cache at full CPU clock speed.


Athlon Classic

Athlon Classic launched on June 23, 1999. It showed superior performance compared to the reigning champion, Pentium III, in every benchmark. [Lal Shimpi, Anand. [ AMD Athlon] , August 9, 1999.]

Athlon Classic is a cartridge-based processor. The design, called Slot A, was quite similar to Intel’s Slot 1 cartridge used for Pentium II and Pentium III; actually it used mechanically the same slot part as competing Intel CPUs (allowing motherboard manufacturers to save on costs) but reversed «upside-down» to prevent users putting in wrong CPUs (as they were completely signal incompatible). The cartridge allowed use of higher speed cache memory than is possible to put on the motherboard. Like Pentium II and the «Katmai»-core Pentium III, Athlon Classic used a 512 KB secondary cache. This cache, again like its competitors, ran at a fraction of the core clock rate and had its own 64-bit bus, called a «backside bus» that allowed concurrent system front side bus and cache accesses. [De Gelas, Johan. [ Clash of Silicon, The Athlon 650] , Ace’s Hardware, September 29, 1999.] Initially the L2 cache was set for half of the CPU clock speed, on up to 700 MHz Athlon CPUs. Faster Slot-A processors were forced to compromise with cache clock speed and ran at 2/5 (up to 850 MHz) or 1/3 (up to 1 GHz). [Lal Shimpi, Anand. [ AMD Athlon 1 GHz, 950 MHz, 900 MHz] , Anandtech, March 6, 2000, p.2.] The SRAM available at the time was incapable of matching the Athlon’s clock scalability, due both to cache chip technology limitations and electrical/cache latency complications of running an external cache at such a high speed.

The Slot-A Athlons were the first multiplier-locked CPUs from AMD. This was partly done to hinder CPU remarking being done by questionable resellers around the globe. AMD’s older CPUs could simply be set to run at whatever clock speed the user chose on the motherboard, making it trivial to relabel a CPU and sell it as a faster grade than it was originally intended. These relabeled CPUs were not always stable, being overclocked and not tested properly, and this was damaging to AMD’s reputation. Although the Athlon was multiplier locked, crafty enthusiasts eventually discovered that a connector on the PCB of the cartridge could control the multiplier. Eventually a product called the «Goldfingers device» was created that could unlock the CPU, named after the gold connector pads on the processor board that it attached to. [Noonan, Jim and Rolfe, James. [ Athlon Gold-Finger Devices] ,, accessed August 24, 2006.]

In commercial terms, the Athlon Classic was an enormous success — not just because of its own merits, but also because the normally dependable Intel endured a series of major production, design, and quality control issues at this time. In particular, Intel’s transition to the 180 nm production process, starting in late 1999 and running through to mid-2000, suffered delays. There was a shortage of Pentium III parts. In contrast, AMD enjoyed a remarkably smooth process transition and had ample supplies available, causing Athlon sales to become quite strong.

* K7 «Argon» (250 nm)
* K75 «Pluto/Orion» (180 nm)
* L1-Cache: 64 + 64 KB (Data + Instructions)
* L2-Cache: 512 KB, external chips on CPU module with 50%, 40% or 33% of CPU speed
* MMX, 3DNow!
* Slot A (EV6)
* Front side bus: 200 MT/s (100 MHz double-pumped)
* VCore: 1.6 V (K7), 1.6–1.8 V (K75)
* First release: June 231999 (K7), November 291999 (K75)
* Clockrate: 500–700 MHz (K7), 550–1000 MHz (K75)

Thunderbird (T-Bird)

The second generation Athlon, the «Thunderbird», debuted on June 5, 2000. This version of the Athlon shipped in a more traditional pin-grid array (PGA) format that plugged into a socket («Socket A») on the motherboard (it also shipped in the slot A package). It was sold at speeds ranging from 600 MHz to 1400 MHz. The major difference, however, was cache design. Just as Intel had done when they replaced the old Katmai Pentium III with the much faster Coppermine P-III, AMD replaced the 512 KB external reduced-speed cache of the Athlon Classic with 256 KB of on-chip, full-speed exclusive cache. As a general rule, more cache improves performance, but faster cache improves it further still. [ K7 microarchitecture information] ,, accessed September 26, 2006.]

AMD changed cache design significantly with Thunderbird. With the older Athlon CPUs, the CPU caching was of an inclusive design where data from the L1 is duplicated in the L2 cache. Thunderbird moved to an exclusive design where the L1 cache’s contents are not duplicated in the L2. This increases total cache size of the processor and effectively makes caching behave as if there is a very large L1 cache with a slower region (the L2) and a very fast region (the L1). [Stokes, John. [ Inside AMD’s Hammer: the 64-bit architecture behind the Opteron and Athlon 64] , Ars Technica, February 1, 2005:p.9.] Because of Athlon’s very large L1 cache and the exclusive design which turns the L2 cache into basically a «victim cache», the need for high L2 performance and size was lessened. AMD kept the 64-bit L2 cache data bus from the older Athlons, as a result, and allowed it to have a relatively high latency. A simpler L2 cache reduced the possibility of the L2 cache causing clock scaling and yield issues. Still, instead of the 2-way associative scheme used in older Athlons, Thunderbird did move to a more efficient 16-way associative layout.

The Thunderbird was AMD’s most successful product since the Am386DX-40 ten years earlier. Mainboard designs had improved considerably by this time, and the initial trickle of Athlon mainboard makers had swollen to include every major manufacturer. Their new fab in Dresden came online, allowing further production increases, and the process technology was improved by a switch to copper interconnects. In October 2000 the Athlon «C» was introduced, raising the mainboard front side bus speed to 133 MHz (266 MT/s) and providing roughly 10% extra performance per clock over the «B» model Thunderbird.

* L1-Cache: 64 + 64 nmKB (Data + Instructions)
* L2-Cache: 256 KB, fullspeed
* MMX, 3DNow!
* Slot A & Socket A (EV6)
* Front side bus: 100 MHz (Slot-A, B-models), 133 MHz (C-models) (200 MT/s, 266 MT/s)
* VCore: 1.70–1.75 V
* First release: June 52000
* Clockrate:
** Slot A: 650–1000 MHz
** Socket A, 100 MHz FSB (B-models): 600–1400 MHz
** Socket A, 133 MHz FSB (C-models): 1000–1400 MHz

Athlon XP/MP

In performance terms, the Thunderbird had easily eclipsed the rival Pentium III, and the early Pentium 4 were a long way off the pace, but gradually clawed their way closer. The 1.7 GHz P4 (April 2001) served notice that the Thunderbird could not count on retaining performance leadership forever, and thermal and electricity-consumption issues with the Thunderbird design meant that it was not practical to take it past 1400 MHz (and even at that speed it was rather hot).


AMD released the third major Athlon version on October 9, 2001, code-named «Palomino», and named it «Athlon XP». The «Athlon XP» was marketed using a PR system, which compared its performance to an Athlon with the «Thunderbird» core. «Athlon XP» was introduced at speeds between 1333 MHz and 1533 MHz, with ratings from 1500+ to 1800+. At launch, the new core allowed AMD to take the x86 performance lead with the 1800+ model, and enhance that lead with the release of the 1600 MHz 1900+ less than a month later. [Wasson, Scott. [ AMD’s Athlon XP 1900+ processor: Pouring it on] , The Tech Report, November 5, 2001. ] The «XP» suffix is interpreted to mean «eXtreme Performance» and also as an unofficial reference to Windows XP. [Advanced Micro Devices, Inc. [ Introducing the AMD Athlon XP Processor] .]

Palomino was the first K7 core to include the full SSE instruction set from the Intel Pentium III as well as AMD’s 3DNow! Professional. It is roughly 10% faster than Thunderbird at the same clock speed, thanks in part to the new SIMD functionality and to several additional improvements. The core has enhancements to the K7’s TLB architecture and the addition of a hardware data prefetch mechanism to better take advantage of available memory bandwidth.Lal Shimpi, Anand. [ AMD Athlon 4 — The Palomino is Here] , Anandtech, May 14, 2001, p:4–5.]

Changes in core layout result in Palomino being more frugal with its electrical demands, consuming approximately 20% less power than its predecessor, and thus reducing heat output comparatively as well. [Wasson, Scott. [ AMD’s Athlon XP 1800+ processor: 1533 > 1800] , The Tech Report, October 9, 2001.] While Athlon «Thunderbird» was near its clock ceiling at 1400 MHz, changes to Palomino’s transistor layout and the reduction in power demands allowed it to continue increasing clock speed even at the same 180 nm manufacturing process node and core voltage.

The «Palomino» was actually first released as a mobile version, called the Mobile Athlon 4 (codenamed «Corvette»). Palomino was also available in a form that officially supports dual processing, known as Athlon MP. [Lal Shimpi, Anand. [ AMD 760MP & Athlon MP – Dual Processor Heaven] , Anandtech, June 5, 2001.]

* L1-Cache: 64 + 64 KB (Data + Instructions)
* L2-Cache: 256 KB, fullspeed
* MMX, 3DNow!, SSE
* Socket A (EV6)
* Front side bus: 133 MHz (266 MT/s)
* VCore: 1. 50 to 1.75 V
* Power consumption: 68 W
* First release: October 92001
* Clockrate:
** A4: 850–1400 MHz
** XP: 1333–1733 MHz (1500+ to 2100+)
** MP: 1000–1733 MHz

Thoroughbred (T-Bred)

The fourth-generation Athlon, the «Thoroughbred», was released 10 June2002 at 1.8 GHz, or 2200+ on the PR system. The «Thoroughbred» core marked AMD’s first production 130 nm silicon, resulting in a significant reduction in die size compared to its 180 nm predecessor.

There are two versions of this core, commonly called A and B. The A version was introduced at 1800 MHz, and had some heat and design issues that held its clock scalability back. In fact, AMD wasn’t able to increase its clock above Palomino’s top grades. Because of this, it was only sold in versions from 1333 MHz to 1800 MHz, replacing the larger Palomino core. The B version of Thoroughbred has an additional metal layer to improve its ability to reach higher clock speeds. It launched at higher clock speeds.

Other than the new manufacturing process, the Thoroughbred design was largely the same as the «Palomino». The Thoroughbred line received an increased front side bus clock during its lifetime, up to 333 MT/s from 266 MT/s. This improved the processor’s memory and I/O access efficiency, and improved per-clock performance as a result. AMD shifted their PR rating scheme accordingly, making lower clock speeds equate to higher PR ratings.

* L1-Cache: 64 + 64 KB (Data + Instructions)
* L2-Cache: 256 KB, fullspeed
* MMX, 3DNow!, SSE
* Socket A (EV6)
* Front side bus: 133/166 MHz (266/333 MT/s)
* VCore: 1.50–1.65 V
* First release: June 102002 (A), August 212002 (B)
* Clockrate:
** T-Bred «A»: 1400–1800 MHz (1600+ to 2200+)
** T-Bred «B»: 1400–2250 MHz (1600+ to 2800+)
** 133 MHz FSB: 1400–2133 MHz (1600+ to 2600+)
** 166 MHz FSB: 2083–2250 MHz (2600+ to 2800+)

Barton and Thorton

Fifth-generation Athlon «Barton»-core processors released in early 2003 featured PR ratings of 2500+, 2600+, 2800+, 3000+, and 3200+. While not operating at higher clock rates than «Thoroughbred»-core processors, they earned their higher PR-rating by featuring a total of 512 KB L2 cache and, in some models, a faster 400 MT/s front side bus.De Gelas, Johan. [ Barton: 512 KB Athlon XP Reviewed] , Ace’s Hardware, February 10, 2003.] The «Thorton» core was a variant of the «Barton» with half of the L2 cache disabled and thus functionally identical to the «Thoroughbred» core.

By the time of Barton’s release, the «Northwood» Pentium 4 had become more than competitive with AMD’s processors.Lal Shimpi, Anand. [ AMD’s Athlon XP 3000+: Barton cuts it close] , AnandTech, February 10, 2003.] Unfortunately, due to the architecture of AMD’s processor caches, an L2 cache increase to 512 KB did not have nearly the same impact as it did to Intel’s line. Only an increase of several percent was gained in per-clock performance. The PR rating became somewhat inaccurate because some Barton models with lower clock rate weren’t consistently outperforming their higher-clocked Thoroughbred predecessors with lower ratings.

The other improvement, a higher 400 MT/s bus clock, helped Barton gain some more efficiency. However, it was clear by this time that Intel’s quad-pumped bus was scaling well above AMD’s double-pumped EV6 bus. The 800 MT/s Pentium 4 bus was well out of Athlon’s reach. In order to reach the same bandwidth levels, the Athlon bus would have to be clocked at levels simply unreachable.

The K7 architecture had scaled to its limit. Maintaining performance equivalence with Intel’s improving processors would require a significant redesign. AMD would soon launch Athlon 64.

Specifications: «Barton (130 nm)»
* L1-Cache: 64 + 64 KB (Data + Instructions)
* L2-Cache: 512 KB, fullspeed
* MMX, 3DNow!, SSE
* Socket A (EV6)
* Front side bus: 166/200 MHz (333/400 MT/s)
* VCore: 1. 65 V
* First release: February 102003
* Clockrate: 1833–2333 MHz (2500+ to 3200+)
** 166 MHz FSB: 1833–2333 MHz (2500+ to 3200+)
** 200 MHz FSB: 2100, 2200 MHz (3000+, 3200+)

«Thorton (130 nm)»
* L1-Cache: 64 + 64 KB (Data + Instructions)
* L2-Cache: 256 KB, fullspeed
* MMX, 3DNow!, SSE
* Socket A (EV6)
* Front side bus: 133/166/200 MHz (266/333/400 MT/s)
* VCore: 1.50–1.65 V
* First release: September 2003
* Clockrate: 1667–2200 MHz (2000+ to 3100+)
** 133 MHz FSB: 1600–2133 MHz (2000+ to 2600+)
** 166 MHz FSB: 2083 MHz (2600+)
** 200 MHz FSB: 2200 MHz (3100+)

Mobile Athlon XP

Mobile Athlon XPs («Athlon XP-M») are identical to normal Athlon XPs, apart from running at lower voltages, often lower bus speeds, and not being multiplier-locked. The lower Vcore rating caused the CPU to have lower power consumption (ideal for battery-powered laptops) and lower heat production. Athlon XP-M CPUs also have a higher-rated heat tolerance, a requirement of the tight conditions within a notebook PC.

The Athlon XP-M replaced the older Mobile Athlon 4. The Mobile Athlon 4 used the older «Palomino» core, while the Athlon XP-M used the newer «Thoroughbred» and «Barton» cores. Some specialized low-power Athlon XP-Ms utilize the microPGA socket 563 rather than the standard Socket A.

The CPUs, like their mobile K6+ predecessors, were also capable of dynamic clock adjustment for power optimization. When the system is idle, the CPU clocks itself down through a lower bus multiplier and also reduces its voltage. Then, when a program demands more computational resources, the CPU very quickly (there is some latency) returns to intermediate or maximum speed to meet the demand. This technology was marketed as «PowerNow!». It was similar to Intel’s SpeedStep power saving technique. The feature was controlled by the CPU, motherboard BIOS, and operating system. AMD later renamed the technology to Cool’n’Quiet, on their K8-based CPUs (Athlon 64, etc), and re-imagined it for use on desktop PCs as well.

Athlon XP-Ms were popular with desktop overclockers, as well as underclockers. The lower voltage requirement and higher heat rating resulted in CPUs that were basically «cherry picked» from the manufacturing line. Being the best of the cores off the line, the CPUs typically were more reliably overclocked than their desktop-headed counterparts. Also, the fact that they weren’t locked to a single multiplier was a significant simplification for the overclocking process. Some «Barton» core Athlon XP-Ms have been successfully overclocked to as high as 3.1 GHz.

As stated, the chips were also liked for their underclocking ability. Underclocking is a process of determining the lowest Vcore at which a CPU can remain stable at for a given clock speed. The Athlon XP-M CPUs were capable of running lower voltages per clock rate compared to their desktop siblings. As such, the chips were used in home theater PC systems due to their high performance and low heat output at low Vcore settings.

Athlon competitors

* Intel Pentium III, Pentium 4, and Celeron
* VIA C3 and C7
* Transmeta Efficeon


The fastest supercomputers based on AthlonMP:

*Rutgers University, Department of Physics & Astronomy. Machine: NOW Cluster — AMD Athlon. CPU: 512 AthlonMP (1.65 GHz). Rmax: 794 GFLOPS.

ee also

* List of AMD Athlon 64 microprocessors
* List of AMD Athlon microprocessors
* List of AMD Athlon XP microprocessors
* List of AMD Sempron microprocessors


External links

* [] AMD Athlon processor images and descriptions
* [] AMD Athlon/Duron/Sempron CPU identification and OPN breakdown
* [ pdf AMD’s Technical Specifications] for 7th generation CPUs (.pdf)
* [ Easy identification with Interactive AMD product ID]
* [ AMD Athlon technical specifications]
* [ Xbit Labs EV6 vs GTL+ System Bus]

AMD’s Athlon 64 FX-51 | HotHardware

The Athlon 64 FX-51 Processor
AMD Drops the Hammer, On Your Desktop!

Marco Chiappetta
Dave Altavilla
September 23, 2003


It has been about two
years, since AMD first divulged information about their «K8»
architecture, also known as the «Hammer», at the
Microprocessor Forum in 2001.   At the time, AMD was
having much success with their «K7» line of processors. 
Enthusiasts and industry analyst were eager to see just what
AMD could do with their next generation processor
architecture.  AMD was no longer following in Intel’s
footsteps.  They were introducing new technology in an
effort to become an industry leader and innovator, rather
than just a «me too» player.  AMD’s break-away
technology initiative resulted in the Athlon, which as you
probably know, was AMD’s most successful line of
microprocessors to date.  In the early days of the
Athlon, who would have thought AMD could make such a
significant dent in Intel’s market share?  Home and
Enterprise level consumers rejoiced.  Finally, there
was real competition for Intel’s Pentium.  This rivalry
could only result in better technology, at faster design
cycles, with lower prices.  The future was bright for
Personal and Enterprise computing and it’s still getting
brighter, here in late 2003.

The «K8» architecture, which
has evolved into the Opteron and now the Athlon 64 line of
CPUs, is a significantly more radical departure from
traditional x86 architectures.   Opterons, Athlon 64s
and Athlon 64 FXs would be AMD’s first microprocessors built
using .13 micron SOI (Silicon-on-Insulator) technology,
which ideally would allow for higher clock speeds with lower
thermal characteristics.  AMD also planned on pulling
the memory controller out of the Northbridge block and
incorporating it into the processors die, to reduce latency,
which in turn would increase performance even further. 
Of course, then AMD decided to execute the boldest move the
industry has seen to date, in x86 computing.  As the
Athlon 64’s branding suggests, AMD’s new Athlon would be
designed from the ground up as a native 64-bit machine with
the capability to also run in 32-bit mode.  Around the
time AMD introduced the Opteron, Intel since scoffed at the
idea, stating that 64-bit computing will not be required for
at least a year down the roadmap.  However, AMD decided
to make 64-bit computing a reality, today for the Desktop
PC, with the introduction of the Athlon 64 and Athlon 64

A host of other enhancements
were implemented as well, culminating in the product we’ll
be looking at today, AMD’s new flagship desktop CPU, the
Athlon 64 FX-51.  The Athlon 64 FX-51 is a 2.2GHz
processor, targeted squarely at gamers and enthusiasts, who
need the absolute fastest machine available, at almost any
cost.  The mainstream Athlon 64 3200+ also debuts today
at 2.0GHz, with a price tag that will put it within reach of
a much larger audience.



Features & Specifications of the AMD Athlon 64 FX
and Athlon 64

Source: AMD


When utilizing the
AMD64 Instruction Set Architecture, 64-bit mode is
designed to offer:

  • Support for 64-bit
    operating systems to provide full, transparent, and
    simultaneous 32-bit and 64-bit platform application
  • A physical address
    space that can support systems with up to one
    terabyte of installed RAM, shattering the 4 gigabyte
    RAM barrier present on all current x86
  • Sixteen 64-bit
    general-purpose integer registers that quadruple the
    general purpose register space available to
    applications and device drivers.
  • Sixteen 128-bit XMM
    registers for enhanced multimedia performance to
    double the register space of any current SSE/SSE2

Integrated DDR memory controller:

  • Allows for a
    reduction in memory latency, thereby increasing
    overall system performance.

advanced HyperTransport link:

  • This feature
    dramatically improves the I/O bandwidth, enabling
    much faster access to peripherals such as hard
    drives, USB 2.0, and Gigabit Ethernet cards.
  • HyperTransport
    technology enables higher performance due to a
    reduced I/O interface throttle.

Large level one (L1) and level 2 (L2) on-die cache:

  • With 128 Kbytes of
    L1 cache and 1 Mbyte of L2 cache, the AMD Athlon 64
    processor is able to excel at performing matrix
    calculations on arrays.
  • Programs that use
    intensive large matrix calculations will benefit
    from fitting the entire matrix in the L2 cache.


  • A 64-bit address
    and data set enables the processor to process in the
    terabyte space.
  • Many applications
    improve performance due to the removal of the 32-bit

Processor core clock-for-clock improvements:

  • Including larger
    TLB (Translation Look-Aside Buffers) with reduced
    latencies and improved branch prediction through
    four times the number of bimodal counters in the
    global history counter, as compared to
    seventh-generation processors.
  • These features
    drive improvements to the IPC, by delivering a more
    efficient pipeline for CPU-intensive applications.
  • CPU-intensive games
    benefit from these core improvements.
  • Introduction of the
    SSE2 instruction set, which along with support of
    3DNow! Professional, (SSE and 3DNow! Enhanced)
    completes support for all industry standards.
  • 32-bit instruction
    set extensions.

Fab location: AMD’s Fab 30 wafer
fabrication facility in Dresden, Germany

Process Technology:

0.13 micron SOI (silicon-on-insulator) technology

Die Size: 193mm2

Transistor count: Approximately 105.9

Nominal Voltage: 1.50v

64 FX-51



AMD is taking the wraps of two new desktop processors, the
flagship Athlon 64 FX-51 and their new performance /
mainstream CPU, the Athlon 64 3200+.  The FX-51 debuts
at 2.2GHz, while the Athlon 64 3200+ arrives clocked at
2GHz.   The differences don’t stop there, however. 
As the chart above indicates, the Athlon 64 FX-51 uses a
940-pin package, similar to AMD’s Opteron, while the Athlon
64 3200+ uses a 754-pin package.  The Athlon 64 FX-51
also has a memory controller that is twice as «wide» as the
3200+; 128-bits vs. 64-bits respectively.  The Athlon
64 FX-51 also requires registered memory to function,
whereas the Athlon 64 3200+ can use standard unbuffered DDR
memory.  Registered memory uses an additional «buffer»
that isolates memory chip load from the memory controller,
which allows for the use of more DIMMS.  ECC memory has
extra bits of storage that help in the identification and
repairing of errors, hence «ECC» — Error Checking and
Correction.  Please don’t confuse registered memory
with ECC though.  ECC and registered memory types are
totally different animals.  It’s possible to buy memory
that is registered, but not ECC, or vice versa. 
Something the chart does not show is the packaging material
used for each CPU.   In its current form, the FX-51 is
housed is ceramic packaging material, ala the Thunderbird. 
The Athlon 64 3200+ is using organic packaging like the
current generation of Athlon XPs.  These processors do
share many features and enhancements, which is why you’re
here reading about their release today…


— 64-bit Processing:

The Athlon 64s, like the Opteron, have the ability to run
64-bit operating systems though the use of a new set of
extensions to the x86 ISA (Instruction Set Architecture). 
With the 64-bit Itanium, Intel introduced the IA-64 ISA,
which has its advantages, but one major caveat with
introducing a new ISA and microprocessors that use the new
instructions set, is that they are not natively compatible
with x86 code.  AMD took a much different approach to
64-bit computing.   They simply extended the x86 ISA to
support 64-bit memory addressability. This makes the Athlon
64 natively compatible with current x86 code, while giving
it support for 64-bit applications going forward.  Due
to the fact that the Athlon 64 can run two different types
of code, x86 and AMD64, the CPU operates in two different
modes dubbed «legacy mode» and «long mode».  In legacy
mode, the Athlon 64 natively runs all 16-bit or 32-bit x86
applications.  In long mode, which requires a 64-bit
AMD64 compliant operating system, the Athlon 64 will enjoy
all of the benefits of 64-bit computing.  Long mode
also has a compatibility sub-mode that allows the running of
32-bit applications with a 64-bit operating system. 
The Athlon 64’s ability to run all these different types of
code make it a very versatile processor.

Integrated DDR Memory Controller:

One of the Athlon 64’s major new
features performance enhancing features is its integrated
memory controller.  With most current processors, the
Northbridge houses the memory controller, which communicates
with the CPU via the Front Side Bus (FSB).   With the
Athlon 64, the memory controller is now on the processor’s
die, which means memory traffic no longer has to travel out
of the CPU to chipset and back.  Being that the memory
controller is now integrated into the CPU, it will run at
the same speed as the host processor.  This type of
configuration drastically reduces latency, which should
yield significant performance gains.  One negative to
having the memory controller integrated into the processor’s
die is that to support emerging memory technologies, like
DDR2 for example, the controller has to be redesigned and
the processor needs to be replaced.

Advanced HyperTransport Link:
AMD has also
replaced aging chip-to-chip
interconnects with their HyperTransport technology. Today’s
fastest desktop processors interface with the motherboard’s
chipset, and subsequently the memory and AGP bus, etc,
through the FSB at 200MHz (400MHz effective with the Athlon
XP — 800MHz effective with the Pentium 4).   The Athlon
64s, however, are equipped with a HyperTransport link that
operates at up to 800MHz DDR (1600MHz effective).  When
operating at top-speed, a single HyperTransport link offers
a maximum of 6.4GB/s of bandwidth.

L1 & L2 On-Die Cache:

In February of this year, AMD released Athlon XPs based on
the «Barton» core, with double the amount of on-die L2 cache
as the older «Thoroughbred» core.  The Bartons have
512KB of full-speed L2 cache versus the Thoroughbred’s 256K. 
The Athlon 64s take things a step further with a full 1MB
(1024KB) of on-die L2 cache.  This added cache should
provide a boost in performance, especially in applications
where large amounts of data are being sent to the processor
and main system memory. With twice the L2 cache of the
Barton based Athlon XPs, the new Athlon 64 core can run a
larger chunk of code out of its on-chip cache resources,
versus having to fetch it from system memory.  A side
effect of having this much L2 cache is that the Athlon 64
now has a die size of 193mm2, almost twice the
size of the Athlon XP.   With a die this large, the
Athlon 64 is going to be expensive to produce.  AMD
claims that when they move to 90nm (.09-micron)
manufacturing process next year, the corresponding die
shrink will bring the die size on a comparable chip down to
a much more palatable 120mm2.

Larger TLBs, Better Branch Predicition, More Counters:

The Pentium 4 has taken a lot of flak because its deep
20-stage pipeline was less efficient than the Athlon XP’s
10-stage pipeline.  The deep pipeline is part of what
allowed the Pentium 4 to reach such high clock speeds, but
it Is also why an Athlon, clocked at a much lower clock
speed than a P4, can perform at similar levels. 
Clock-for-clock, that Athlon XP can handle more
instructions.  With the Athlon 64, AMD has deepened the
processor’s pipeline to 12-stages, which you’d think would
lower the core’s IPC (Instructions Per Clock). 
However, thanks to some core architectural improvements, it
hasn’t.  The Athlon 64 has larger Translation
Look-Aside Buffers (TLB), with improved latency and improved
branch prediction.   The Athlon 64 has quadruple the
number of bimodal counters in its global history counter,
when compared to the Athlon XP.  All this technical
jargon means that at similar clock speeds, even though it
has a deeper pipeline, an Athlon 64 should outperform an
Athlon XP in most circumstances.  Later on, you’ll see
we tested an Athlon XP 3200+, alongside the Athlon 64 FX-51,
and with both processors clocked at 2.2GHz, the FX-51 was
clearly a much faster chip.  AMD’s efforts to increase
the Athlon 64’s IPC seems to have paid dividends nicely.


Supporting Hardware & Chipsets

Security vulnerabilities found in Intel and AMD processors

Security researchers have discovered vulnerabilities in Intel and AMD processors that may lead to information disclosure.

Most Intel 10th, 11th and 12th generation processors are affected by a new vulnerability that the researchers have named ÆPIC Leak. The vulnerability is an architectural bug according to the researchers, which sets it apart from Spectre and Meltdown vulnerabilities that have haunted Intel and AMD in the past years.

AMD Zen 2 and 3 processors are affected by a security vulnerability that the researches named SQUID. It is a side channel attack that is targeting CPU schedulers.

The following paragraphs provide a high-level overview of both security issues. We provide links to the research papers and security advisories released by Intel and AMD.

Most home devices with affected processor models should be safe, as the attacks have certain requirements that make attacks on home systems unlikely.

ÆPIC Leak: important resources

  • Intel Security Advisory
  • Intel list of affected processors
  • Research Paper

Security researchers from Sapienza University or Rome, Graz University of Technology, Amazon Web Services, and CISPA Helmholtz Center for Information Security published the research paper ÆPIC Leak: Architecturally Leaking Uninitialized Data from the Microarchitecture recently.

The name is derived from the Advanced Programmable Interrupt Controller (APIC) and affects all Intel processors that are based on the Sunny Cove architecture. In particular, Ice Lake and Alder Lake processors are affected.

Attackers may exploit the vulnerability to retrieve data from the cache hierarchy. Without going into too many details — the research paper provides all the technical information needed — Æpicleak exploits a bug in Sunny-Cove based processors. When reading data on Sunny-Cove based CPUs, stale data from the superqueue is returned; this is not by design, as it should result in undefined behavior instead according to Intel.

The researchers note that the returned data is not restricted to security domains.

The uninitialized data returned from ÆPIC Leak is not restricted to any security domain, i.e., the origin can be userspace applications, the kernel, and, most importantly, SGX enclaves.

Experiments confirmed that the superqueue is used «as a temporary buffer for APIC requests». The superqueue contains recent memory loads and stores, and the APIC «only overwrites the architecturally-defined parts of the register and leaves the stale values in the reserved part».

In other words, attackers may exploit the bug to read data, including AES-NI keys from SGX enclaves.

Several different attack techniques are described in the research paper:

  • Leaking data and code pages — The most straightforward attack type combines «Enclave Shaking and Cache Line Freezing» to «leak data (and code) at rest of an SGX enclave.
  • Leaking register values — Attack targets a specific cache line to reconstruct the value of the register.

How to look up the processor generation on Windows


Windows users may do the following to check the processor generation of Intel processors:

  1. Open the Start Menu.
  2. Type System Information.
  3. Load the System Information result.
  4. Check the value of the processor entry, and there specifically the first or the first two digits after the dash, e. g., Intel Core i5-1035G1 is a 10th generation processors.

Sunny-Lake based processors are not vulnerable to Meltdown attacks.

Mitigations and fixes

The vulnerability requires root or administrative level access to the machine to exploit the vulnerability. Most home systems should be safe because of that, but it is still recommended to install updates once they become available.

Æpic Leak requires a hardware fix according to the researchers. They assume that the fix should not be too complex, as older processors are not affected by the issue. The research paper lists several mitigation suggestions, ranging from disabling SGX to disabling caching for EPC.

Intel reveals on the 2022.2 IPU — Intel® Processor Advisory support page that customers should install the latest firmware versions provided by the system manufacturer to address the issue. Intel plans to release SGX SDK updates once the public embargo is lifted.

Intel has released microcode updates for affected processors that are already available on the company’s public GitHub repository.

AMD processors affected by SQUIP vulnerability

A new research paper by researchers from Lamarr Security Research, Graz University of Technology and Georgia Institute of Technology have discovered a new vulnerability affecting certain AMD processors.

Resource links:

  • Execution Unit Scheduler Contention Side-Channel Vulnerability on AMD Processors
  • SQUIP: Exploiting the Scheduler Queue Contention Side Channel

The linked research paper provides technical details on the vulnerability. Researchers discovered a vulnerability in CPU schedulers of affected AMD processors. SQUIP is the first side-channel attack on scheduler queues, according to the research paper.

The SQUIP attack observes the occupancy level from within the same hardware core and across SMT threads.

An attacker could extract sensitive data from a co-located victim in under 45 minutes, according to tests performed by the research team.

Hardware and software mitigations are suggested in the research paper. One of the easier options is to deactivate SMT or to prevent that processors from different security domains from running co-located on the same core.

The following processors are affected by the vulnerability:

  • AMD Ryzen 2000, 3000 and 5000 series
  • AMD Ryzen 4000 and 5000 with Radeon graphics series.
  • 2nd and 3rd generation AMD Ryzen Threadripper processors.
  • AMD Ryzen Threadripper PRO processors.
  • AMD Athlon 3000 mobile processors with Radeon graphics.
  • AMD Ryzen 2000 mobile processors.
  • AMD Ryzen 3000 mobile processors.
  • AMD Ryzen 3000, 4000 and 5000 processors with Radeon graphics.
  • AMD Athlon 3000 series with Radeon graphics. (Chromebook)
  • AMD Athlon mobile processors with Radeon graphics. (Chromebook)
  • AMD Ryzen 3000 series processors with mobile graphics. (Chromebook)
  • 1st, 2nd and 3rd generation AMD EPYC processors.

AMD users may use System Information to look up the processor. Other options include opening Settings on Windows 10 or 11 devices, and to select System > About to display the processor make and model.

AMD does not plan to release any kernel mitigations or microcode updates for affected processors. Instead, the company offers the following recommendation:

AMD recommends software developers employ existing best practices1,2, including constant-time algorithms and avoiding secret-dependent control flows where appropriate to help mitigate this potential vulnerability.



Computer Processor Types — iFixit

A few years ago, choosing a processor was pretty straightforward. AMD and Intel each produced two series of processors, a mainstream line and a budget line. Each company used only one processor socket, and there was a limited range of processor speeds available. If you wanted an Intel processor, you might have a dozen mainstream models and a half-dozen budget models to choose among. The same was true of AMD.

OEM Versus Retail-Boxed

To further confuse matters, most AMD and Intel processors are available in two types of packaging, called OEM and retail-boxed. OEM processor packages include only the bare processor and usually provide only a 90-day warranty. Retail-boxed processors include the processor, a compatible CPU cooler, and a longer warranty, typically three years.

A retail-boxed processor is usually the better deal. It typically costs only a few dollars more than the OEM version of the same processor, and the bundled CPU cooler is usually worth more than the price difference. But if you plan to install an after-market CPU cooler for example, because you are upgrading your system to be as quiet as possible it may make sense to buy the OEM processor.

Nowadays, choosing a processor isn’t as simple. AMD and Intel now make literally scores of different processor models. Each company now offers several lines of processors, which differ in clock speed, L2 cache, socket type, host-bus speed, special features supported, and other characteristics. Even the model names are confusing. AMD, for example, has offered at least five different processor models under the same name Athlon 64 3200+. An Intel Celeron model number that ends in J fits Socket 775, and the same model number without the J designates the same processor for Socket 478. A Pentium 4 processor model number that ends in J says nothing about the socket type it is designed for, but indicates that the processor supports the execute-disable bit feature. And so on.

AMD and Intel each offer the three categories of processors described in the following sections.

Budget processors give up a bit of performance in exchange for a lower price. At any given time, AMD or Intel’s fastest available budget processor is likely to have about 85% of the performance of their slowest mainstream model. Budget processors are more than sufficient for routine computing tasks. (After all, today’s budget processor was yesterday’s mainstream processor and last week’s performance processor. ) Budget processors are often the best choice for a system upgrade, because their lower clock speeds and power consumption make it more likely that they’ll be compatible with an older motherboard.

The various models of the AMD Sempron processor sell in the $50 to $125 range, and are targeted at the budget through low-end mainstream segment. The Sempron replaced the discontinued Socket A Duron processor in 2004, and the obsolescent Socket A Athlon XP processor in 2005. Various Sempron models are available in the obsolescent Socket A and in the same Socket 754 used by some Athlon 64 models.

AMD actually packages two different processors under the Sempron name. A Socket A Sempron, also called a K7 Sempron, is in fact a re-badged Athlon XP processor. A Socket 754 Sempron, shown in Figure 5-1 is also called a K8 Sempron, and is really a cut-down Athlon 64 model running at a lower clock speed with a smaller L2 cache and a single-channel memory controller rather than the dual-channel memory controller of the Athlon 64. Early Sempron models had no support for 64-bit processing. Recent Sempron models include 64-bit support, although the practicality of running 64bit software on a Sempron is questionable. Still, like the Athlon 64, the Sempron also runs 32-bit software very efficiently, so you can think of the 64-bit support as future-proofing.

Figure 5-1: AMD Sempron processor (image courtesy of AMD, Inc.)

If you have a Socket 462 (A) or Socket 754 motherboard in your system, the Sempron offers an excellent upgrade path. You’ll need to verify compatibility of your motherboard with the specific Sempron you intend to install, and you may need to upgrade the BIOS to recognize the Sempron.

For more information about Sempron processor models, visit

For many years, the Intel Celeron processor was the poor stepsister, offering too little performance at too high a price. Cynical observers believed that the only reason Intel sold any Celeron processors at all was that system makers wanted the Intel name on their boxes without having to pay the higher price for an Intel mainstream processor.

That all changed when Intel introduced their Celeron D models, which are now available for Socket 478 and Socket 775 motherboards. While Celeron D models are still slower than Semprons dollar-for-dollar, the disparity is nowhere near as large as in years past. Celeron D processors, which sell in the $60 to $125 range, are very credible upgrade processors for anyone who owns a Socket 478 or Socket 775 motherboard. Like the Sempron, Celeron models are available with 64-bit support, although again the practicality of running 64-bit software on an entry-level processor is questionable. Once again, it’s important to verify the compatibility of your motherboard with the specific Celeron you intend to install, and you may need to upgrade the BIOS to recognize the Celeron.


Celeron processors (without the «D») are based on the Northwood core and have only 128 KB of L2 cache. These processors have very poor performance, and unfortunately remain available for sale. The Celeron D models are based on the Prescott-core, and have 256 KB of L2 cache.

For more information about Celeron processor models, visit

Mainstream processors

Mainstream processors typically cost $125 to $250 although the fastest models sell for $500 or more and offer anything up to about twice the overall performance of the slowest budget processors. A mainstream processor may be a good upgrade choice if you need more performance than a budget processor offers and are willing to pay the additional cost.

However, depending on your motherboard, a mainstream processor may not be an option even if you are willing to pay the extra cost. Mainstream processors consume considerably more power than most budget processors, often too much to be used on older motherboards. Also, mainstream processors often use more recent cores, larger L2 caches, and other features that may or may not be compatible with an older motherboard. An older power supply may not provide enough power for a current mainstream processor, and the new processor may require faster memory than is currently installed. If you intend to upgrade to a mainstream processor, carefully verify compatibility of the processor, motherboard, power supply, and memory before you buy the processor.

The AMD Athlon 64 processor, shown in Figure 5-2, is available in Socket 754 and Socket 939 variants. As its name indicates, the Athlon 64 supports 64-bit software, although only a tiny percentage of Athlon 64 owners run 64-bit software. Fortunately, the Athlon 64 is equally at home running the 32-bit operating systems and applications software that most of us use.

Figure 5-2: AMD Athlon 64 processor (image courtesy of AMD, Inc.)

Like the Sempron, the Athlon 64 has a memory controller built onto the processor die, rather than depending on a memory controller that’s part of the chipset. The upside of this design decision is that Athlon 64 memory performance is excellent. The downside is that supporting a new type of memory, such as DDR2, requires a processor redesign. Socket 754 models have a single-channel PC3200 DDR-SDRAM memory controller versus the dual-channel controller in Socket 939 models, so Socket 939 models running at the same clock speed and with the same size L2 cache offer somewhat higher performance. For example, AMD designates a Socket 754 Newcastle-core Athlon 64 with 512 KB of L2 cache running at 2.2 GHz a 3200+ model, while the same processor in Socket 939 is designated an Athlon 64 3400+.


The model numbers of Athlon 64 and Sempron processors are scaled differently. For example, the Socket 754 Sempron 3100+ runs at 1800 MHz and has 256 KB of cache, and the Socket 754 Athlon 64 2800+ runs at the same clock speed and has twice as much cache. Despite the lower model number, the Athlon 64 2800+ is somewhat faster than the Sempron 3100+. Although AMD hotly denies it, most industry observers believe that AMD intends Athlon 64 model numbers to be compared with Pentium 4 clock speeds and Sempron model numbers with Celeron clock speeds. Of course, Intel also designates their recent processors by model number rather than clock speed, confusing matters even further.

For more information about Athlon 64 processor models, visit

The Pentium 4, shown in Figure 5-3, is Intel’s flagship processor, and is available in Socket 478 and Socket 775. Unlike AMD which sometimes uses the same Athlon 64 model number to designate four or more different processors with different clock speeds, L2 cache sizes, and sockets Intel uses a numbering scheme that identifies each model unambiguously.

Older Pentium 4 models, which are available only in Socket 478, are identified by clock speed and sometimes a supplemental letter to indicate FSB speed and/or core type. For example, a Socket 478 Northwood-core Pentium 4 processor operating at a core speed of 2.8 GHz with the 400 MHz FSB is designated a Pentium 4/2.8. The same processor with the 533 MHz FSB is designated a Pentium 4/2. 8B, and with the 800 MHz FSB it’s designated a Pentium 4/2.8C. A 2.8 GHz Prescott-core Pentium 4 processor is designated a Pentium 4/2.8E.

Figure 5-3: Intel Pentium 4 600 series processor (image courtesy of Intel Corporation)

Socket 775 Pentium 4 models belong to one of two series. All 500-series processors use the Prescott-core and have 1 MB of L2 cache. All 600-series processors use the Prescott 2M core and have 2 MB of L2 cache. Intel uses the second number of the model number to indicate relative clock speed. For example, a Pentium 4/530 has a clock speed of 3 GHz, as does a Pentium 4/630. The 540/640 models run at 3.2 GHz, the 550/650 models at 3.4 GHz, the 560/660 models at 3.6 GHz, and so on. A «J» following a 500-series model number (for example, 560J) indicates that the processor supports the XDB feature, but not EM64T 64-bit support. If a 500-series model number ends in 1 (for example, 571) that model supports both the XDB feature and EM64T 64-bit processing. All 600-series processors support both XDB and EM64T.

For more information about Pentium 4 processor models, visit

Extreme Processors

We classify the fastest, most expensive mainstream processors those that sell in the $400 to $500 range as performance processors, but AMD and Intel reserve that category for their top-of-the-line models, which sell for $800 to $1,200. These processors the AMD Athlon 64 FX, the Intel Pentium 4 Extreme Edition, and the Intel Pentium Extreme Edition are targeted at the gaming and enthusiast market, and offer at best marginally faster performance than the fastest mainstream models.

In fact, the performance bump is generally so small that we think anyone who buys one of these processors has more dollars than sense. If you’re considering buying one of these outrageously expensive processors, do yourself a favor. Buy a $400 or $500 high-end mainstream processor instead, and use part of the extra money for more memory, a better video card, a better display, better speakers, or some other component that will actually provide a noticeable benefit. Either that, or keep the extra money in the bank.

By early 2005, AMD and Intel had both pushed their processor cores to about the fastest possible speeds, and it had become clear that the only practical way to increase processor performance significantly was to use two processors. Although it’s possible to build systems with two physical processors, doing that introduces many complexities, not least a doubling of the already-high power consumption and heat production. AMD, later followed by Intel, chose to go dual-core.

Combining two cores in one processor isn’t exactly the same thing as doubling the speed of one processor. For one thing, there is overhead involved in managing the two cores that doesn’t exist for a single processor. Also, in a single-tasking environment, a program thread runs no faster on a dual-core processor than it would on a single-core processor, so doubling the number of cores by no means doubles application performance. But in a multitasking environment, where many programs and their threads are competing for processor time, the availability of a second processor core means that one thread can run on one core while a second thread runs on the second core.

The upshot is that a dual-core processor typically provides 25% to 75% higher performance than a similar single-core processor if you multitask heavily. Dual-core performance for a single application is essentially unchanged unless the application is designed to support threading, which many processor-intensive applications are. (For example, a web browser uses threading to keep the user interface responsive even when it’s performing a network operation.) Even if you were running only unthreaded applications, though, you’d see some performance benefit from a dual-core processor. This is true because an operating system, such as Windows XP, that supports dual-core processors automatically allocates different processes to each core.

The AMD Athlon 64 X2, shown in Figure 5-4, has several things going for it, including high performance, relatively low power requirements and heat production, and compatibility with most existing Socket 939 motherboards. Alas, while Intel has priced its least expensive dual-core processors in the sub-$250 range, the least expensive AMD dual-core models initially sold in the $800 range, which is out of the question for most upgraders. Fortunately, by late 2005 AMD had begun to ship more reasonably priced dual-core models, although availability is limited.

Figure 5-4: AMD Athlon 64 X2 processor (image courtesy of AMD, Inc.)

For more information about Athlon 64 X2 processor models, visit

The announcement of AMD’s Athlon 64 X2 dual-core processor caught Intel unprepared. Under the gun, Intel took a cruder approach to making a dual-core processor. Rather than build an integrated dual-core processor as AMD had with its Athlon 64 X2 processors, Intel essentially slapped two slower Pentium 4 cores on one substrate and called it the Pentium D dual-core processor.

The 800-series 90 nm Smithfield-core Pentium D, shown in Figure 5-5, is a stop-gap kludge for Intel, designed to counter the AMD Athlon 64 X2 until Intel can bring to market its real answer, the dual-core 65 nm Presler-core processor, which is likely to be designated the 900-series Pentium D. The Presler-based dual-core processors will be fully integrated, compatible with existing dual-core Intel-compatible motherboards, and feature reduced power consumption, lower heat output, twice as much L2 cache, and considerably higher performance.

Figure 5-5: Intel Pentium D dual-core processor (image courtesy of Intel Corporation)

Reading the foregoing, you might think we had only contempt for the 800-series Pentium D processors. In fact, nothing could be further from the truth. They’re a kludge, yes, but they’re a reasonably cheap, very effective kludge, assuming that you have a motherboard that supports them. We extensively tested an early sample of the least expensive 800-series Pentium D, the 820. The 820 runs at 2.8 GHz, and under light, mostly single-tasking use, the 820 «feels» pretty much like a 2.8 GHz Prescott-core Pentium 4. As we added more and more processes, the difference became clear. Instead of bogging down, as the single-core Prescott would have done, the Pentium D provided snappy response to the foreground process.

For more information about Pentium D processor models, visit….

Table 5-2 lists the important characteristics of current AMD processors, including the special features they support.

Table 5-2: Table 5-2. AMD processor summary

Table 5-3 lists the important characteristics of current Intel processors, including the special features they support.

Table 5-3: Intel processor summary


Special features are not always implemented across an entire line of processors. For example, we list the Pentium D 8XX-series processors as supporting EM64T, SSE3, EIST, and dual core. At the time we wrote this, three Pentium D 8XX models were available: the 2.8 GHz 820, the 3.0 GHz 830, and the 3.2 GHz 840. The 830 and 840 models support all of the special features listed. The 820 model supports EM64T, SSE3, and dual-core operation, but not EIST. If a special feature listed as being supported by a particular line of processors is important to you, verify that it is supported in the exact processor model you intend to buy.

More about Computer Processors

Obvious Probable or «The Star and Death of Athlone Thorobred»

This article was submitted to our second competition.

1. Say a word about the poor athlone.

The computer market consists of two parts — the market for finished computers and the market for components for self-assembly or modernization of computers. Today, the situation in the first market share is such that Athlons are bought only because of low prices. Many buyers still know only one brand of processor — Pentium — and are very surprised to hear about the existence of «non-Pentium» computers. Vendors who, on duty, have to know about Athlon and Duron, for some non-objective reasons, prefer Intel products, categorically recommending Celeron and Pentium4 to buyers. Naturally, in this case we cannot talk about any serious level of sales of AMD products.

There are problems in the enthusiast sector too. Even if a person wants to buy an Athlon, having assessed the possibilities and price, they will try to dissuade him ten times from «well-read» friends and negligent sellers, intimidating them with the possibility of splitting the crystal during assembly, promising terrible heat and, as a result, an unexpected failure of the computer from overheating , accompanied by a host of other horrors. If an enthusiast is adamant, he will practically be forced to buy a turbojet cooler like Volcano 6CU+, which will further overshadow the joy of buying and can still make a person go over to the camp of atlon-haters.

This is such a bleak situation for AMD in our market, and it needs to take serious measures to improve it. Because no one will buy an Athlon if its price is approximately equal to the price of a Pentium4 of the same rating. Not so long ago AMD released the Athlon XP 2800+ and 3000+ based on the Barton core. This core is a long-awaited successor to the Palomino architecture and, in my opinion, will gradually force the Thoroughbred core out of the «productive» market sector. Thus, the Athlon XP 2800+ model, which AMD failed to mass-produce on the Thoroughbred core, will now be produced on the Barton core. And everything would be fine, but the prices for older models are so high that their release is still only a marketing ploy designed to stop the fall in AMD’s share price, but is not capable of bringing real profit.

2. Things of the past.

The appearance of the Athlon XP brought back to life the practice of labeling processors with a performance rating instead of a real frequency, which was practiced some time ago. By the way, the official decoding of the XP suffix sounds like eXtra Performance, which means «extra performance». So, according to AMD, the rating in the new Athlons with the Palomino core was calculated relative to the previous core — Thunderbird — however, it was clear to everyone that AMD had to take such unpopular measures due to the need to maintain parity in «boa length» with Intel Pentium4.

Indeed, the most significant change in the new kernel was support for SSE instructions, and in tasks that do not use these instructions, the old and new Athlones turned out to be equal. However, Palomino with a frequency of 1400 MHz was labeled as 1600+, i.e. we were assured that it is 14% faster than Thunderbird of the same frequency. Of course, this was not the case, but the «advanced» part of the buyers, when choosing a processor, was guided not by the marking, but by the test results, while the rest of the buyers — by the «parrots / price» ratio. Both the first and the second criteria were excellent compared to the Pentium4 based on the Willamette core, so no one was offended, and the market quickly got used to the fact that the Athlon XP frequency did not match the marking.


However, when Intel mastered the release of the Pentium4 with the Northwood core, due to the finer manufacturing process, it received twice as much L2 cache, and due to copper connections (which AMD has been using for quite some time) also 50% — a new increase in the maximum frequency, AMD’s business went awry. She was unable to switch to the 0.13 micron process technology in time, and therefore she began to rapidly lose credibility in the eyes of buyers. When Intel switched its processors to the 133 MHz bus frequency, also practiced by AMD for a long time, Athlones lost all their advantages except for the most important one — the price. It was literally the final nail in Palomino’s coffin.

It should be noted that AMD has already been in a similar situation in the days of the first Athlon models, which were out of work after the release of the Pentium III Coppermine with an integrated full-speed L2 cache. Then AMD, tightening its belts, was waiting for the completion of the implementation of the 0.18 micron process technology, which allowed it to breathe deeply and win back a decent market share. History repeated itself, with one exception: AMD made a serious mistake in the design of the processor core, which did not allow the new Thoroughbred to noticeably exceed the maximum Palomino frequencies (although everything is relative, and revision A is capable of reaching frequencies above 2GHz). Re-designing, of course, took a considerable time, during which AMD was still rapidly losing ground. But in the end, revision B of the Thoroughbred core saw the light of day, and things seemed to be looking up.

In the meantime, Intel also did not sit idly by and improved production processes, as a result of which the Pentium4 reached a frequency of 3 GHz. But insidious Intel didn’t think that was enough, and in the older model Pentium4 3.06 it released a dormant genie for a long time — Hyper Threading technology, which makes it possible to increase the efficiency of the processor by executing commands «for two». Nevertheless, AMD’s assets still had a lower cost of processors, due to the smaller area of ​​the core die, and the frequencies of the second Thoroughbred revision could reach 2.2 GHz. Having made a long-awaited transition to a system bus frequency of 166 MHz, AMD announced Athlones with ratings of 2700 and even 2800. But the latter, due to the extremely low percentage of crystals suitable for operation at such a high frequency, could not appear on the market.

AMD’s last chance to compete with Pentium4 was the release of the Barton core, which also had twice as much L2 cache, which was done. This is how the Athlon XP 2500+, 2800+ and 3000+ appeared, and formally it restored parity with Intel. However, Intel has another trump card up its sleeve — the 800MHz Quad Pumped system bus, which will further increase the performance of the Pentium4. Of course, AMD can do the same, but it’s too early to say with certainty. The FSB frequency of 200 MHz is currently supported by only one commercially available chipset — nForce2 — and even that is semi-official.

As I already mentioned, Barton should replace Thoroughbred in older models, and Thornton is preparing a replacement for younger models. Thus, Thorobred the Glorious, who restored AMD’s reputation and gave people who were not burdened with excess money the opportunity to get more performance for less money, will have to retire. However, this is clearly not a matter of tomorrow, and so far the «torik» (as he was dubbed by the people) is alive and able to fight.

3. What about overclocking?

If you already have this question, I hasten to reassure you: the lyrical part is over, it’s time to practice. Namely, overclocking Athlon XP to study its immediate prospects. So, the Athlon AXDA1700DU I bought was subjected to, if not extreme, but very intense overclocking, after which, in fact, this article was born. Along the way, I will reveal some technical details that «surfaced» during the process.

How do you usually find out the overclocking potential of a processor? They mount the most powerful cooler available on the processor, raise the voltage on the core to indecent and begin to storm the frequency heights. However, this method is at least not optimal, it is unsafe for the processor (we all remember the recently sensational NSDS — Northwood Sudden Death Syndrome 🙂 and has dubious theoretical value. You can find out what your processor is capable of in another way, which I have been developing and testing for several years, but for the first time I am making it public.

The processor frequency is always limited by a certain ceiling, determined by the production technology, the quality of a particular crystal, temperature, and, finally, the core supply voltage. Moreover, the maximum frequency is a function of the supply voltage, sometimes linear, but more often power-law. Take a look at the graph from the above mentioned Athlon XP 1700+:

It is clearly seen that at a frequency slightly below 2000 MHz there is a break in the graph, i.e. after a frequency of 2 GHz, the effect of increasing the voltage decreases. The maximum operating frequency of my processor copy was 2100 MHz, after which the increase in voltage within reasonable limits did not allow Windows to be loaded. Actually, this is the ceiling above which this instance will not work under any sauce (cryogenic installations are not within the scope of my competence :-). Here are the exact numbers of the overclocking results (it was produced on the EPoX 8RDA motherboard, for convenience the multiplier was lowered to 10, at a given voltage the maximum frequency at which the processor allowed loading Windows 9 was recorded8 and engage in non-resource intensive tasks):

Please note that the results are recorded from the maximum frequency down. I deliberately changed the frequency and voltage from maximum to minimum in order to warm up the crystal more and get more life numbers. The actual processor frequency was determined using Frank Delattra’s CPU-Z, and the voltage was controlled using Alfredo Comparetti’s Speedfan. The motherboard did not allow to lower the voltage below 1.4V, but, using the approximation capabilities built into Excel, we will be able to view the picture both in the infrared and in the ultraviolet range 🙂

Please note that when the voltage reaches 2V, the frequency increase stops, so there is no reason to raise the voltage above 1. 9V. Increasing the voltage leads to a significant increase in processor heat, and this should always be remembered. If you use conventional air cooling, then an increase in heat dissipation will lead to an increase in processor temperature, which, in turn, will lead to an increase in internal resistance and, as a result, an increase in voltage drop across the circuit elements. In other words, an increase in voltage provokes its further increase and a vicious circle is obtained, which can be broken only by the use of special cooling systems. Therefore, you should not get carried away by increasing the voltage, especially since to determine the potential of your processor, it is enough to make several measurements in gentle conditions and build a graph. You can quickly figure out which of several processor instances is more overclockable, even without overclocking the processor, but simply by reducing the voltage at a constant frequency!

As for the older Athlon XP models operating at frequencies above 2GHz at a standard voltage of 1. 65V, they are selected from the same crystals precisely for their ability to operate at higher frequencies and the graph for them will be exactly the same, except for the values ​​along the ordinate axis . The difference in frequencies of seemingly the same crystals is explained by the purity of the semiconductor and microdefects in the crystal structure. The «cleaner» the crystal turned out, the higher the frequency it can operate at a given voltage. Of course, an increase in voltage will allow them to show better results than the younger crystals left after «screening out».

Now take another look at the above graph. Even such, by no means an outstanding example (the official Thoroughbred frequency limit is 2250 MHz) can operate at its native frequency of 1467 MHz already at a voltage of 1.25V! Of course, to achieve stability under load, this voltage will have to be slightly increased (for example, at 2 GHz, stability was achieved at a voltage of 1.73V), but in this case, the processor easily overcomes the 1800 MHz bar at default voltage, which corresponds to a rating of 2200+! So why is AMD releasing lower rated models? You will find the answer to this question below.

4. Where does the rating come from?

Before turning to the conclusions about the prospects of Athlon XP, I want to tell you about the rating system. Once upon a time, Intel added a CPUID identification command to processors. In early processors (i486DX4, Pentium), this function reported only model codes (Family, Model, Stepping) and manufacturer name (GenuineIntel). Later, the feature list grew to include cache and TLB information, as well as advanced vendor-specific features. One of these functions allows you to ask the processor for its name (and not codes, as before), which, among other things, includes the model number. However, if the Pentium4 returns the name that was hardwired into it at the factory (for example, Intel(R) Pentium(R) 4 CPU 1.80GHz), then the result returned by Athlon XP is calculated depending on the current processor frequency.

Once upon a time I became a happy owner of a new at that time Athlon XP 1500+ and, of course, could not resist the temptation to overclock it. And I was quite surprised to see that the above-mentioned identification string changes depending on the frequency of the processor. I was even more impressed by the transformation of my Athlon XP into an MP (on the first Athlon XP models, the ability to work in dual-processor boards was not blocked, and EPoX motherboards can detect this). And until recently, I was firmly convinced that the processor itself recognizes its mode of operation and presents itself depending on the situation. The first doubts crept into my head only when preparing this article, when I saw these «miracles»:

This table should be read like this: starting from 1270 MHz, the processor identifies itself as AMD Athlon(tm) XP 1500+ and continues to return this name up to 1380 MHz, when the name changes to AMD Athlon(tm) XP 1600+. At frequencies below 1270, as well as from 1630 to 1690 and from 1840 to 1940, the processor refused to give its rating. You can see more clearly all the originality of the rating calculation in the following diagram of the dependence of the rating on the frequency:

Unfortunately, I did not manage to see the inscription AthlonXP 2700+ — AthlonXP 2600+ kept up to a frequency of 2160 MHz, and my copy refused to start up higher. By the way, I also failed to set the multiplier required for older Athlon models above 12.5. Could it be that the 2700+ and 2800+ processors are not exactly identical to their smaller counterparts?

So we’re seeing an amazing fermentation of frequency ranges, the widest being the 2400+, 1500+ and 2600+ rating ranges, and the narrowest 2000+. Suspicions that «something is wrong here» made me start searching. And I found mentions that the identification string changes during overclocking, not for everyone! The rest was a matter of technique: I just checked how a similar processor would behave in another motherboard. Up to 1900+ inclusive the behavior was absolutely identical and I almost stopped the experiment. However, at a frequency of 1640 MHz, where my processor «lost» the rating, it was called 2000+! So, after all, the rating depends on the board, and not on the processor.

Even more convincing evidence was the «software» overclocking with Gigabyte Easy Tune. The rating remained the same as it was set when loading. Thus, the processor name for the CPUID function is programmed by the motherboard BIOS at the initialization stage. This is done through special status registers (MSR). I could not find out the numbers of these registers — the information is closed under the non-proliferation agreement (NDA). Of course, they can be recognized by digging through the BIOS code with a disassembler, but this is a titanic work that no one has dared to do yet.

5. Model range and prices.

Non-strict frequency dependence of the rating allows AMD to adjust it to various bus frequencies and multipliers, which is clearly seen in the example of such a 2600+ model released for both 266 and 333 MHz bus. Here is the frequency and rating table for Athlon XP with Thoroughbred core:

Hypothetical models are highlighted in gray in the table. Obviously, since the frequency of 2250 is difficult to achieve at the nominal voltage, Athlon will not be able to reach the frequency of 2267 MHz, but the younger versions with a bus frequency of 166 MHz deserve the most serious attention. In particular, the 2200+ model with a frequency of 1833 MHz would be a good competitor to the Pentium4 2.26 and, I believe, may well see the light of day.

So what does AMD need to do to succeed in the market, and what exactly is not desirable to do? AMD’s main trump card was and will remain the price, but sometimes it goes too far, sometimes giving away processors for almost nothing and losing possible profit, sometimes setting prices higher than those for Pentium4 and not making a profit at all. Therefore, first of all, it needs to adjust the prices of its products. Thus, a Pentium4 with a frequency of 1.80 GHz today costs about $150, and an Athlon of the same frequency (2200+) is only 106, significantly outperforming its competitor in performance. Reverse example — Athlon XP 2800+

Recently there was a mutual reduction in prices for processors, after which we can directly compare the prices of Pentium4 and Athlon XP.

As you can see, Intel is phasing out the production of junior Pentium4 models at an accelerated pace, while AMD continues to sell junior models at bargain prices! And this is despite the fact that the yield of crystals suitable for at least the 2000+ rating is huge!!! This is AMD’s second error to date. In a good way, AMD should be made the entry-level Athlon XP 2000+ with a price of about $100, while today’s price for this model is less than $85. If we bring all the price data into one table (retail prices were mainly taken from the price list of the F-Center), we will see the following:

Now it’s clear that the prices for Barton 2800+ and 3000+ (while this article was being written, the 2600+ model has fallen in price to an acceptable level) must be reduced by at least 10%, otherwise they will not be sold at all. Of course, I understand that the output of crystals at a frequency of more than 2 GHz is small, but it exists and such processors should be sold. And when their prices are comparable to Pentium4, for the reasons mentioned above, the buyer will prefer Pentium4. As the saying goes, better is a tit in the hand… AMD has set a very correct price for a 2400-rated processor, but the same mess begins below: the 2200+ model could have been perfectly sold at a price of about $130, but it only costs $106. And the reason AMD is so underpriced is Celeron.

Low-end Intel processors with only 128KB of cache memory are selling quite well today, and the exceptionally magical influence of the Intel brand helps them in this. The Athlon XP 2000+, of course, can easily deal with the Celeron 2.00GHz in most exercises, however, in order not to miss out on potential customers, AMD continues to release junior Athlon XP models, pricing them in line with Celeron prices:

So the low prices for the junior Athlon XP have quite good reasons. Given this fact, I see the prospects for the AMD processor line as follows:0004

Transferring the older Barton to the 200MHz bus will allow it to get the performance it lacks in Pentium4 3.06 due to faster memory and a slightly higher frequency (2200MHz versus 2167 at 166MHz FSB). A 10% price cut for all models older than 2600+ will increase their attractiveness in the eyes of the buyer, and the 2200+ model with a 166 MHz bus frequency will be in demand by the market and can bring real profit, unlike the younger Athlon XP, which AMD has to sell almost at cost .

That, in fact, is all that I wanted to bring to you with this long and confusing story :).

This article was submitted to our second contest.