Intel Ivy Bridge Continues To Grind Quickly On Linux

[phoronix]

Phoronix: Intel Ivy Bridge Continues To Grind Quickly On Linux

There's many more Ivy Bridge Linux tests coming out in the next few weeks, but here's some new results today for the Core i7 3770K processor looking at the overclocking results and effects of Turbo Boost and Hyper Threading...

http://www.phoronix.com/vr.php?view=MTA5NjI

[allquixotic]

How do you OC a 3770K on Ubuntu? Any instructions? I have my BIOS set to XMP profile, I have XMP-compatible RAM, but I don't know how to see if turbo mode is even working, or OC...

[Michael]

How do you OC a 3770K on Ubuntu? Any instructions? I have my BIOS set to XMP profile, I have XMP-compatible RAM, but I don't know how to see if turbo mode is even working, or OC...

All OCing needs to be done from the UEFI/BIOS. Unfortunately there isn't a way to see if Turbo Boost is working under Linux aside from toggling the feature and see -- same for seeing the actual frequency. There is a small Intel Linux utility available that should show the TB/frequency data, but I forgot the name of it nor do I know if it even works on IVB at this point... I wish such information could be integrated into cpufreq or similar and then exposed from the kernel over sysfs.....

[ChrisXY]

There is a small Intel Linux utility available that should show the TB/frequency data, but I forgot the name of it nor do I know if it even works on IVB at this point...

It's called i7z.

[Fenrin]

I can't find the comparison of sensors temperature readings (or temp measuring).

There are some rumors that Ivy Bridge die reaches very high temperatures when overclocked compared to the previous generation.
see here: http://www.overclockers.com/ivy-bridge-temperatures

[Michael]

I can't find the comparison of sensors temperature readings (or temp measuring).

It's in the main Intel Core i7 3770K article, along with the power consumption results....

[Fenrin]

[...]

ah ok thanks, that's good.

here is it: CPU Temperature monitor

[kukreknecmi]

I remember Fx-8150 with march bdver1 on GCC 4.7 getting as low as around 26 on c-ray(3690x was 25 or so). So how should i interpret IB on c-ray with around 40 score?

[curaga]

I remember Fx-8150 with march bdver1 on GCC 4.7 getting as low as around 26 on c-ray(3690x was 25 or so). So how should i interpret IB on c-ray with around 40 score?

These tests were with gcc default targets, ie march=generic. You can't compare the results, until we see how the ivy does with ivy optimizations