armel or armhf?

Ok, maybe I understand the difference between armel and armhf, in theory.. but what's the difference in "everyday use"? I mean, for a multimedia home server, I will get more or less using armel instead of my actual Debian armhf image? Thanks

 

The general rule will be that armhf is at least the same performance as armel, and often faster. It depends on how much use is made of floating point in the app in question as armhf speeds up floating point operations.

 

Well, Don't know if I need floating point or not. My needs are to have a multimedia server, with transmission, amule and oscam running 24/7

 

It all depends on exactly how those programs are encoded. Without feedback from the authors I do not know how you can tell if they use floating point. If they internally use floating point then you will get a performance gain from the amrhf version.

 

I doubt you'd be able to detect a difference in your stated Use Case.

I'd recommend you go with armel as that is the more stable at the moment.

 

ok, thank you, i will try

 

It depends on your use case, of course. I have some experience with the test code in the link below run on squeeze (Debian 6, armel) and wheezy (Debian 7, armhf) on a commercial SBC (not RPi, or BBB or Udoo).

viewtopic.php?f=2&t=470#p3807

Wheezy runs this code 5 times faster than squeeze. This is heavy math hitting the FPU; armhf cuts it (hard floats), armel doesn't (soft floats). Depends on your use case. My use case demands armhf!

Noel

 

I am noob on Linux, how to swap between armel and armhf on Udoo Linaro or ROS?

Excuse my english

 

In effect you swap by switching SD card, with the different versions on each (or if you have one SD, you reimage)

 

Thank you for all your posts, always helping. I hope can do it some day.

 

I tried out the debian wheezy armhf build and the ubuntu 12.04, and the armhf was definitely faster.

 

Does the new v2.1 ubuntu 12.04 build for the UDOO quad support armhf? I'm interested for future floating point applications development.