The ARM Cortex-A9 Can Beat Out The Intel Atom

[zeealpal]

NEON doesn't provide IEEE-compliant FP, only single precision FP ops are available (and don't support all rounding modes for instance). You'll have to wait for ARMv8 processors to get SP/DP IEEE-compliant FP.

You can compare SSE vs NEON with FFmpeg by choosing a format that has been optimised for both ISA.

It should also be noted that I know someone working in a company doing post-production who told me they are using some nVidia farm that is based on ARM CPUs, they are probably using this: http://blogs.nvidia.com/2011/12/meet...velopment-kit/

Thanks for the reply Told me what I wanted to know That nVidia board (or future generations) could be quite useful for GPU only rendering, where the ARM core(s) would only be commanding the GPU.

[Darkseider]

Well the benchmark is a bit useless because a atom n450 usually runs with 1.66 ghz. so the cortex cpu needs first a higher clockspeed to match the atom. btw. there are also quad core atoms out there.

OK. That's all fine and well but even when equally clocked the dual core Cortex A9 does a fine job of squashing the Atom. It too was underclocked to 1 Ghz from its' native 1.2 Ghz. With that said there are 2 Ghz quad core Cortex A9 system out now like the ODROID-X board which can be used to compare against todays Atom processors. I would be willing to bet that it would run neck and neck if not faster than the Atom in most tests at a fraction of the cost and power consumption. Point is that from this test one can easily conclude that a high end dual core Cortex A9 and/or a quad core Cortex A9 based system could easily provide the performance needed for consumers looking for a netbook or nettop device.

[gururise]

Maybe Intel Atom can get popular on smartphones, tablets and embedded devices anyways.
All the graphics on ARM are closed source. Then comes Intel with open source Ivy Bridge graphics on Intel Atom? Me like!

The only problem is that the Intel ATOM series chips don't use Open-Source graphics. Instead we are stuck with the proprietary PowerVR Poulsbo crap. So really, there is no advantage to going x86 on the phone/tablet platforms.

[bachinchi]

The only problem is that the Intel ATOM series chips don't use Open-Source graphics. Instead we are stuck with the proprietary PowerVR Poulsbo crap. So really, there is no advantage to going x86 on the phone/tablet platforms.

That's gonna change with Intel's Valleyview.

[ldesnogu]

That's gonna change with Intel's Valleyview.

Valley View won't be available before Q4 2013 and seems to target tablets and nettops, not smartphones. I guess Intel will announce another chip based on Silvermont core (the core in Valley View) for smartphones, but they'd better have it earlier, because it doesn't seem Medfiled gained significant traction in the market.

[CFWhitman]

My question is, where's my Cortex A15 based smartbook with open source Mali graphics drivers for Linux? Heck, I'd probably be willing to settle for a Cortex A9 as long as it was unlocked and fully supported under Linux. Well, I can dream anyway.

[uid313]

My question is, where's my Cortex A15 based smartbook with open source Mali graphics drivers for Linux? Heck, I'd probably be willing to settle for a Cortex A9 as long as it was unlocked and fully supported under Linux. Well, I can dream anyway.

Imagine it with full hardware documentation, bill of materials list, flowcharts, diagrams, pin out lists, schematics/schemata, EDA/ECAD files, open source firmware, open source hardware, everything...

[gururise]

My question is, where's my Cortex A15 based smartbook with open source Mali graphics drivers for Linux? Heck, I'd probably be willing to settle for a Cortex A9 as long as it was unlocked and fully supported under Linux. Well, I can dream anyway.

ODROID-X is a quad core Cortex A9 board (Samsung Exynos) that is shipping now for $129. It might be what you are looking for.

[ciplogic]

Anyways back to the tests, anyone else find it weird that Atom under 64 bit overall did worse? Any particular reason for that?

Atom CPUs are in-order designs, which makes as its biggest issues that when it has to access memory, it has to wait for it and it cannot execute speculative other instructions (like other out-of-order designs), so the question most likely is: why on 64 bit it happens to access more memory than on 32 bit.
AMD64 architecture has more registers, but the address pointers are bigger. So: if the program did not fit well in the CPU registers and the CPU was taxed by the lag of waiting for memory, is most likely that was slower than the same CPU which would not wait for memory, because all data fits in either L1 cache (32 KB) or L2 cache (512 KB).
This is also why most likely A9 was faster than Atom on those specific programs: A9 is out-of-order design but it has a slow(er) memory bus than Atom has. So most likely the speculative execution will make Atom to advance doing computations, even the CPU has a crippled memory bandwidth, but it could still make computations that Intel CPU simply had to wait.

[gipsyblues]

As much as, one - for academic purposes- would love to see or read about how different hardware platforms perform at a an arbitrary operating point, it would have been a preferable and beneficial exercise if all these different hardware had been compared using the standard form in which each platform is sold on the market.

This would have been a more useful exercise for someone needing to make a choice of hardware platform for a pending project