AMD sneaks out new FX chips, hints at ARM deal
AMD's FX range just got bigger, while APU production has been sent to IBM to improve supplies.
Spotted by the guys over at CPU World, the three new processors include AMD's first production chip to feature a default clock speed above 4GHz. The FX-4170 is a quad-core part featuring 4.2GHz stock clock with Turbo Core functionality pushing it to 4.3GHz in single- or dual-core mode. 8MB of L3 cache is included, while the chip - in common with all the new FX series parts - has a thermal design profile (TDP) of 125W.
Joining the FX-4710 is the FX-6200, a six-core 3.8GHz part which hits 4.1GHz under Turbo Boost conditions. As with the FX-4170, 8MB of L3 cache is included and the chip is rated at a 125W TDP.
Finally, AMD has opened pre-ordering for stock of its FD8150FRGUWOX, also known as the FX-8150 WOX. Designed as an alternative to the standard FX-8150, the WOX edition has the same eight processing cores running at 3.6GHz and Turbo Core capabilities up to 4.1GHz but swaps the retail heatsink for a stock liquid cooling system.
Available for pre-order from AMD now, the new chips should start to appear in the UK supply channel soon. The FX-4710 is already showing as available from two retailers priced at around £120, while the FX-6200 and FX-8150 WOX have US pre-order prices of $188.48 and $398.29 (around £119 and £252 before taxes) respectively.
News of the new chips comes as AMD chief executive Rory Read quietly let slip that the company's accelerated processing unit (APU) product families are now produced by both spin-off foundry GlobalFoundries and computing giant IBM. Although no reason was given, it seems likely the move to IBM is as a result of poor yields for the chips that have led to certain models - in particular the A8-3850 - being near-impossible to buy.
Finally, an interview between AMD chief technology officer Mark Papermaster and Wired offered a sneak preview of the company's plan for taking on Intel at the low power end of the market.
While Intel is looking to fight ARM and its multifarious licensees head-on with its Atom system-on-chip (SoC) designs, AMD has a different tactic in mind. Asked whether AMD was looking to build ARM IP into its processors, Papermaster evasively stated: 'The answer is not no.'
Intel itself once held an ARM licence, which it used to produce its XScale range of low-power chips. Should AMD enter into an agreement with ARM, it's possible that its future range of Heterogeneous Systems Architecture (HSA) APUs could pack low-power ARM architecture cores for improved performance in mobile and tablet markets.
Tempted by a chip running at 4.2GHz stock, or are you more excited by the news of a possible AMD ARM processor? Share your thoughts over in the forums.
Previous Article
10 Comments
Discuss in the forums ReplyDont get me know im not a fanboy my recent build is based around a A6-3500 for my media PC and it works a treat. AMD just need to get it together at the mid to high end market as they have nothing worth buying. Intel equivalents are simply much much faster for the same price and bumping a few extra hundred mhz on poor performing instructions per-clock does nothing. Its like sprinkling glitter on crap.
Mobile cpu's are nice, but the market is just flooded with options to browse and play angry birds on the go.
I had a taste of it in my M14x with Intel and Nvidia optimus. Possibly the worst implementation and feature set to ruin a gaming laptop. Ever since then I've been wanting to move over completely to AMD.
I thought it was bad that Bulldozers performed so badly, but the fact that you basically can't buy one for cheaper than an Intel part at the same performance level is sad. At least AMD previously used to be cheaper for the same performance on the low/mid-level. Now they can't even claim that.
Seems like only Llano is worth anything, and that is only because of the pretty nice on die GPU...which if Intel is to be believed (baring driver issues), Ivy Bridge on the top level integrated GPU is going to be about on par or maybe even a little faster than current Llano iGPU bits...with CPU cores mustering the better part of double the performance.
Looking at similarly specced SKUs, yes IB is expected to be faster than Llano, but Trinity is supposed to perform better than IB (in 3D apps). So at the market segment that Llano is geared toward, Trinity will become the APU of choice. It doesn't have as powerful a CPU core as IB, but the iGPU is far more powerful.
This means that whichever is the faster will depend on what you want to use it for (much the same as current generation processors). According to AMD, Trinity will outperform IB in 3D applications by over 100% on similar products. Seems about normal to me, although I agree with everyone that wants to see them compete in the high-end market as well. Hopefully, working with IBM and ARM will help AMD get back into the high performance market.
That still only makes it as fast as a near 3yr old 955 Phenom 2 Quad Core @3.6Ghz (which BTW has 4 REAL cores) not 2 cores with AMD Hyperthreading, like these FX turkeys.
AMD Hyperthreading is one of the dumber terms I've heard to describe it, but I'm too tired to explain why. You're obviously half-trolling, so carry on.
From the sound of it, with all that came out from AMD's FAD, we might be near the end of AMD even releasing processors into this space. Trinity is aimed at the same market Llano is currently, while PileDriver may not even put an appearance in replacing BD on the consumer desktop.
For everyone still hating on AMD, will Intel look that much better when they stick you $500 for the equivalent of the 2500K in the generation following IB? Disappointing as BD has been, we're all going to be worse off if AMD gives up that end of the market entirely.
Since Trinity is not moving to a smaller process size and IS using the power thirsty Bulldozer cores, as well as supposedly a better GPU...I just do not see how AMD can be reducing power use of the chips (even if the 32nm node is doing better for them by the time Trinity comes around) with what they are attempting. IB on the other hand, even with the same TDP numbers in ULV sounds like they are likely in real world use to reduce power use a bit, combined with, lets call it, double the GPU performance and better CPU performance.
Looking at Llano benchmarks, some of the games even at the lower and mid level resolutions the chips are benchmarked at are CPU limited in part because of the poor CPU component to Llano, but also because of how the memory is shared between the GPU and CPU in Llano...which I don't believe Trinity is changing (but is using a hotter GPU, which'll use more memory bandwidth). So enter IB which has on par or better GPU performance to Llano, but has NO CPU bottlenecks nor memory bottlenecks that seem relevant up against Llano, which is CPU limited sometimes and at best as as good a GPU or Trinity which probably will use more power, might have a much better GPU, but unless bulldozer is improved is likely to have WORSE CPU than Llano, and isn't solving the memory bottle neck of AMDs APUs.
That sounds like a recipe of a really nice iGPU, that can't perform to its fullest because it is stuck with a piss poor CPU and insufficient memory bandwidth when sharing with the CPU.