AMD FX

One notable feature of the AMD FX microprocessors is that they are all unlocked and overclockable, a feature reserved for the high-end K suffix SKUs from Intel. This allows users to increase the clock speed of their CPU and to gain extra performance out of their product.[15] The personal world record for highest overclock was achieved on an FX-8350, which is clocked up to 8794.33 MHz.[16]

• 4 "Bulldozer cores" (dual-core modules) within FX-8 series, 3 in FX-6 series, and 2 in FX-4 series, with two integer clusters (seen as logical cores from OS) and a shared floating-point unit in each "Bulldozer core".
• All models manufactured from 8 logical cores with simple Orochi die production, in 938 pins µPGA package AM3+ socket.
• All models support up to 4 DIMMs of DDR3 memory.

Unlike the majority of their Intel counterparts, FX chips offer no integrated graphics, a feature reserved to AMD's APU line of processors. Both Zambezi and Vishera use a module design containing two cores on one module.

Upon launch, the FX series was met with criticism from reviewers.[17] Due to multiple cores sharing common resources, most tasks were substantially slower on the FX lineup than the Intel Sandy Bridge equivalent. In many single-threaded applications, it was worse than the previous generation of Phenom II microprocessors.[17] The power consumption of the lineup, while not as poor as the Phenom II generation, was also worse than what Intel was providing at the time.[17] The refresh to the processors that were brought about by the piledriver architecture in 2012 generally improved performance across the board by increasing clock speeds at similar power consumption levels,[18] but by then Intel's Ivy Bridge architecture was available and provided much better performance per watt and total performance to consumers.[19] With AMD only being responsible for 20% of consumer CPU sales in 2016,[20] Intel continued to gain market share in the industry during the lifespan of the FX series.[21]

The AMD FX line-up was generally worse performing than its Intel competitors during the lifespan of the range. The floating-point performance was relatively poor due to a single shared FPU per module and most games could not take advantage of the high core counts that the series provided. In processes that thrived in multithreaded applications, AMD SKUs typically pulled ahead. This came at a great cost, however, as thermal efficiency was often worse than the previous generation of processors.[22] Updates to the architecture that came with the Piledriver revision allowed for higher clock speeds. This led to better performance, but that came with the cost of even higher thermal output on the high end, which can be seen with the FX 9590, which outputs 220 Watts.[23]

In 2015 AMD was accused of falsely advertising the core count of its FX lineup. The claim stated that because the FX series' cores shared common resources such as the FPU, AMD was falsely advertising its high core counts.[24] The company had to pay out $12.1 million to California residents who bought a high-end FX chip.[25]

CPU features table

• AMD later re-used the FX designation for some processors in its socket FM2/FM2+ APU lineup.

• List of AMD FX microprocessors
• AMD Accelerated Processing Unit
• Bulldozer Architecture
• Piledriver Architecture
• AMD Ryzen
• AMD Phenom II
• AMD Athlon II