Ubuntu 12.10 Is Faster With Intel Hardware

[Kano]

@AJSB

Basically the tests should be done crossplattform. The raw speed for intel on win is much higher which is really sad. The diff between linux+win is lowest with nvidia, amd performs up to opengl 3 really similar, but has got a problem with opengl 4 (only heaven compared with hd5670). intel however is a day/night experience - lots of apps do not even run on linux which do on win like heaven (with tesselation) or rage (via wine). Haswell, the IVB successor will get a much more powerful - i expect about twice the speed on certain benchmarks, maybe not all. Of course intel could optimize the drivers for a certain workload (like the source 2 engine) but most likely amd will be still faster - only compared the combined cpu+gpu solutions. As always the number of possible sales will impact drivers. It's all about making money.

[GreatEmerald]

GPU (or whatever it's called)

It is still called GPU, as it is a graphics processing unit. Just that it's not a graphics card, but a graphics chip.

[curaga]

I only wish there was at least one ARM vendor supporting open-source driver development like Intel but I digress.

I know you mean graphics, but you should still say so. Almost all ARM vendors do open-source drivers for the non-gpu-vpu parts.

[misGnomer]

Well, since Ivy Bridge is pretty much Sandy Bridge with a beefier GPU (or whatever it's called), most improvements still apply.

True. Here's hoping... I usually root for the underdog (e.g. AMD), but I must admit Intel has been doing a sterling job looking after their past platforms.

Now it's down to our dear devs to make those updates percolate down to us faithful masses.

I only wish there was at least one ARM vendor supporting open-source driver development like Intel but I digress.

[bug77]

Interesting bits of information, although since my gear is either Sandy Bridge or earlier I'm not sure how or if "Ubuntu 12.10 Is Faster With Intel Hardware" applies to my hardware.

Well, since Ivy Bridge is pretty much Sandy Bridge with a beefier GPU (or whatever it's called), most improvements still apply.

[misGnomer]

@darkseider

Hostility due to elementary reading miscomprehension? That might be somewhat embarrassing.

There was no 1:1 ARM vs Intel IB comparison, but a reference ? perhaps a little out of place ? of architectural improvements to both.


Interesting bits of information, although since my gear is either Sandy Bridge or earlier I'm not sure how or if "Ubuntu 12.10 Is Faster With Intel Hardware" applies to my hardware.

[AJSB]

If the numbers of that SoC are correct, it's clearer than distilled water, than a SteamBOX won't use a Intel architecture ....both from a console price, console power drain and from a console performance POV, it will be the best decision for Valve and the consumers.

It will be also from a strategic POV , Valve stay away from Intel architecture.

[vitiv]

"While benchmarks have already indicated ARM performance improvements in Ubuntu 12.10, early testing of this "Quantal Quetzal" release has also revealed that Intel hardware is benefiting too from performance optimizations for this Linux operating system update due out in October."

No kidding? Really? Who would have thought that a current gen dual core, hyperthreaded 64 bit processor that draws 17W and costs $300 would perform better than a quad core, single threaded 32 bit Cortex A9 SoC that draws around 4W - 5W and costs about $25? WOW! What an epiphany! I never knew that would be the case! Thank you so much Phoronix for stating the blatantly obvious!

/end sarcasm

Sorry, but isn't the article supposed to say that performance has improved on intel hardware compared to the same intel hardware in previous releases not ARM vs Intel?

Nonetheless, you are missing the point.

[GreatEmerald]

Interesting, I thought the updates (which don't affect the kernel or X server version, apparently) wouldn't make a difference, but apparently they do. Good to know, since I have Ivy Bridge hardware as well. Though it makes me wonder, why is something like libvpx encoding time impacted at all? How is it possible to get better results when the kernel itself, as well as the tested program, haven't changed?

Who would have thought that a current gen dual core, hyperthreaded 64 bit processor that draws 17W and costs $300 would perform better than a quad core, single threaded 32 bit Cortex A9 SoC that draws around 4W - 5W and costs about $25?

Uhm, you might want to read what you quoted again. It says that "latest Ubuntu has performance improvements that work for both ARM and x86". Not that "x86 > ARM".

[Darkseider]

Ubuntu 12.10 Is Faster With Intel Hardware

"While benchmarks have already indicated ARM performance improvements in Ubuntu 12.10, early testing of this "Quantal Quetzal" release has also revealed that Intel hardware is benefiting too from performance optimizations for this Linux operating system update due out in October."

No kidding? Really? Who would have thought that a current gen dual core, hyperthreaded 64 bit processor that draws 17W and costs $300 would perform better than a quad core, single threaded 32 bit Cortex A9 SoC that draws around 4W - 5W and costs about $25? WOW! What an epiphany! I never knew that would be the case! Thank you so much Phoronix for stating the blatantly obvious!

/end sarcasm