ASUS Eee PC 1201N Netbook On Linux Update

[thefirstm]

The performance drop caused by that "emulator" is negligible.

[cl333r]

For alot of Windows applications, only 32-bit executables are (easily) available. On Ubuntu, most applications have native 64-bit executables in the repo. Therefore, testing 32-bit Windows executables against 64-bit Linux executables is actually fair, since most people are not going to be downloading the source for their Windows applications and recompiling them 64-bit.

I Agree.
But not for this/my test, cause it creates a lot of pointers to ints on the heap and since a pointer (no matter to what) on a 32 bit OS is typically 4 bytes and on 64 bit OS - 8 bytes, the Ubuntu 64 executable needs/allocates quite a lot more memory and runs out of RAM and starts using the SWAP - hence a big degradation of performance, hence (at least) for this particular test I'd rather use actual 64 bit executables on both/all OSes.

[deanjo]

The performance drop caused by that "emulator" is negligible.

I beg to differ. My companies dbase suffers roughly about 15% when ran through wow64 vs native 32 bit windows.

[Apopas]

We have 64bit systems since 2003. If there are not 64bit binaries commonly availiable for windows it's their own problem. Fair and simple. We can not be limited in old ways because they are not able to offer a pure 64 bit system and we shouldn't neglect each OS's specific advantages.

[kraftman]

I have on my computer XP, win7, Ubuntu etc.
I wanted to benchmark these OSes but not use anything that would involve drivers (eg. graphics driver, cause a lot boils down to the quality of drivers).
So I decided to create a simple (C++) test that would use the CPU and the RAM (allocating and initialising a lot of ints on the heap and testing their values).
And guess what, Ubuntu came out as the winner.

Ubuntu 32bit - 8 seconds.
WinXp 32bit = about 1 minute
Win7 (64bit but the .exe test was 32bit) = 13 seconds.

Under windows I compiled in Visual Studio with /O2, and under Ubuntu with gcc and -O2.

Nice Could you try doing the same, but using exactly the same version of GCC on both systems, please?

[cl333r]

windows 7 (32 bit) = 12 seconds

(Just installed Windows 7 32 bit on same computer)

@Kraftman
I've never used/installed gcc on windows, I'll try but don't promise anything.

The small C++ test/benchmark I'm using I put at:

http://sites.google.com/site/r45d15/cpptest

[monraaf]

While benchmarks will come out soon that compare Windows to Linux

I certainly hope these benchmarks will not be as misleading as the recent 1080p on an atom article, as it doesn't help the credibility of Phoronix.

[cl333r]

I installed 2 gcc's on Win 7 32 bit and compiled with -O2:
gcc 3.4.5 = 15 to 16 seconds, executable size around 800KB
gcc 4.3.3 = 16 seconds, executable size over 3MB

Thus with GCC on windows the executables are a lot bigger and the runtime slower.

[kraftman]

I installed 2 gcc's on Win 7 32 bit and compiled with -O2:
gcc 3.4.5 = 15 to 16 seconds, executable size around 800KB
gcc 4.3.3 = 16 seconds, executable size over 3MB

Thus with GCC on windows the executables are a lot bigger and the runtime slower.

Thank you. It seems it's the Linux kernel that "kicks" such "high quality stuff" not the compiler. Btw. those executables sizes look strange.

[oliver]

Backlight issue

Micheal,

I think the backlight issue is mentioned here I belive. I think it's the same reason why it kicks in on my T42.

https://bugs.launchpad.net/ubuntu/+source/gnome-power-manager/+bug/498974

So report! and maybe someone will notice it