AMD finally launches its FX processors by VR-Zone.com
Companies are generally plagued with delays when it comes to launching products, but AMD's CPU business has had a terrible year when it comes to delays. The new FX-series of CPUs have finally launched and the anticipation of what AMD will bring to the table to give Intel a run for its money is high. Although this is only a quick look at things from AMD's perspective, we'll have reviews up of the new CPUs as well where you can make a much more educated decision if the FX-series is for you.
AMD briefed us about the FX-series of CPUs at its TFE event last week and it was a fairly quick rundown of its new processors. Let's cut to the chase and have a look at the new models to start with. As you can see from the diagram below, we're looking at a total of seven new models, although only three of them will be available as of today. At the top we have the FX-8150 which is followed by the FX-8120 and FX-8100. The FX-8100 won't be available at launch; instead the six core FX-6100 will be one of the three launch models.
The three quad cores are somewhat peculiar, as in the FX-4170 for starters only has a single turbo frequency as you can see from the table above. This is due to the already high clock speed of the chip and it peaks out at 4.3GHz. The FX-B4150 is part of AMD's Business series of CPUs and as such it's unlikely that it'll be available to consumers. Finally the FX-4100 is the most basic model for now.
It's worth noting that there are some differences between the models, with the FX-8100, FX-6100 and FX-4100 having a 200MHz slower chipset interconnect. The 95W version of the FX-8120 won't be available in retail box for the time being at least, as AMD is targeting system integrators with this part as they don't like 125W TDP CPUs. With regard to CPU Turbo Core and CPU Max Turbo in the table, the first option is with all cores enabled whereas the Max Turbo is with half of the cores in sleep mode. All models have 8MB of L3 cache and 1MB of L2 cache per core. All of the FX CPUs are of course multiplier unlocked, although that 125W TDP might hold back the overclocking potential a little bit when it comes to air coolers.
That said, the Bulldozer architecture is quite different from AMD's traditional multi-core CPUs and as such we're looking at pairs of cores in "Bulldozer modules" where each pair is connected to 2MB of shared L2 cache and a shared FPU. We'll have to wait and see how well this works out in real world applications, but AMD seems to think this is the way forward for its CPUs and APUs.
AMD is making some interesting comparisons when it comes to performance, as it's only pitching its high-end FX-8150 as a Core-i5 2500K beater, whereas they figure it plays pretty evenly with Intel's Core i7-2600K with some benchmark wins and some benchmark losses. Of course this will depend quite a lot on the benchmark in question and according to AMD's own internal tests the FX-8150 doesn't really stand out in most of the tests and it's quite badly beaten in others.