So, the folks at linux-mag have bench marked Gentoo(x86_64) compiled with march=core2 and -O2,-O3 or -Os and compared it with Ubuntu 9.04. While Ubuntu 9.10 is already out, the software used on Gentoo (seems that the used the stable branch of Gentoo) is closer to 9.04 than 9.10. And what do the results tell us? Exactly what we already knew. Gentoo kicks ass. We already knew that, didn’t we. However, what is interesting is that when most of the people have been harping that optimizing for a particular CPU (which I believe is the primary reason for the differences that we see) is not useful anymore, it seems that the this is really not the case. In my own experience, I concur. My Gentoo system has been far more responsive than my Ubuntu or even Arch systems ever were.

Read the full article here: http://www.linux-mag.com/id/7574/

Could this be real? Mac OS X better than Linux on IO performance. A benchmark was hurled towards my general direction. A benchmark proving that Mac OS X (snow leopard) is better than Ubuntu Intrepid. What was especially interesting to me was the IO performance benchmarks. It seemed like Platypus was hammering right, left and center while good old penguin was sitting pretty on the side-walk.

Contrary to general belief that Mac performs better on the desktop, Linux performing better on servers with clear-cut edge in IO performance, these tests proved that Mac OS X was not only better in many Graphics tests. It could easily out-run Linux in IO tests by large margin.

By running some of the IO tests on whatever machine available, it was clear that Ubuntu Intrepid does not that perform badly in any of the tests as shown. Surprisingly, Ubuntu Hardy performs better than Intrepid.

Secondly looking at hardware specs was also a significant revelation.

Notice the processor used for Mac OS X has a Intel Core 2 Duo and the processor for Ubuntu tests has Intel Core 2. By further looking at Mac Mini Wikipedia page here, it is clear that 3rd generation Mac Mini hardware has been used for all Mac OS X tests while 2nd generation Mini for all Ubuntu related tests.

How much does this effect performance?
Quoting from comparison done between Core 2 and Core 2 Duo at tech.yahoo.com “Core 2 Duo computers outperform Core Duo computers on tasks across the board, from rendering graphics and video to spell-checking documents. The speed improvement varies widely, but on average it’s about 30 percent. That’s significant in an industry that thrives on tiny, incremental improvement.” That is in my opinion a huge difference in hardware for drawing an authoritative benchmark comparing two different operating systems.

Without questioning the intent for using two different hardware, looking at the benchmark again one can easily conclude that Linux performed better given Mac OS X considering Mac had the 30% processor advantage.

O Phoronix, site voltado à testes de hardware e software em sistemas opensource (especialmente Linux), realizou um benchmark com sua ferramenta de benchmark (pacote deb disponível no site. Estará nos repositórios do novo Jaunty), entre a mais nova versão do Kubuntu e o PC-BSD, ambos utilizando o KDE 4. Confira aqui os resultados, mas já vou adiantar que o PC-BSD terminou massacrado no teste…

A Phoronix article suggests that Dell should update the Ubuntu on its Inspiron netbook. Michael Larabel wrote:

Besides running faster, the newer Ubuntu releases have better hardware support, various package updates, many improvements to GNOME, and all sorts of other features. Ubuntu 8.04 is nice for its Long-Term Support with Canonical continuing to push out security updates for years to come, but a much better experience can be had on Ubuntu 9.04 with its faster performance, updated packages, and newer features.

I think this view is very wrong. It is really a rather geeky kind of way of looking at computers. Take alone the fact where they are basing their suggestion “For our Ubuntu 9.04 testing we used a development snapshot from 2009-04-10“. Jaunty also introduces two new design decisions/technologies like the new notification system and also the extended FUSA applet. Both cant be called stable and have caused some protest already. both technologies will either be removed or much better in the next LTS release.

The whole point of LTS is trading stability and predictability in favour of newer and quicker technologies. So Michael misses the core point of why LTS was invented. It also removes the need for administrators to look at updates besides the security updates. if the users running around with the newest non-LTS Ubuntu the admins will have to look on every update – this takes much more time on a current Ubuntu than on a LTS release.

It is still up to the decision of a user who bought such a netbook to update if he thinks Ubuntu Jaunty is better for him – if he desperately needs a quicker boot process and so on. So I am not arguing against progress but rather like to brake the very idea that newer is always better and that progress alone is an argument for updating. I know that this is a very common view in the Linux world, but outside of that it is not always a smart thing to do.

What was good some years ago does not become bad overnight.

I’ve been eschewing the virtues (no pun intended) of desktop OS virtualization for a few years now, and many people have found the same reasons liberating: ability to multiplex various OS combinations into one hardware system. Testing, development, and deployment are now rapid due to virtualization, and the ability to disconnect the OS from PC hardware.

Well, in that period of time that manny of us have been enjoying products such as VMWare, VirtualPC, VirtualBox, or Parallels, there have been a *couple* of really minor nagging things we’ve wanted. Things such as the ability to boot a real partition, not just virtual disk images. Or how about 3D support in the guest? You’d be surprised at how much modern OS’s actually use 3D support for standard functions, not just playing games. And lastly, we want performance. I mean, near-hardware level performance. Who cares if I’m running three guests at once? I want them *ALL* to feel like they’re the only OS installed on the thing.

The good news is: Most of the products mentioned above have taken strides to address ALL, if not most, of the items I’ve listed below, since the user community has been extremely vocal about those very issues. Well, guess what? At least 3 of the four products DO address almost all of the requests in their latest iterations, but ONE stands out. Which one? VirtualBox.

VirtualBox just recently announced v2.1 of its software, which boasts probably the best performance of all competing products, the ability to use competitors virtual-disk formats, as well as real disk partitions, and finally, something only VirtualBox can lay claim to: real Direct3D acceleration support for its guests.

Does that mean you can play games in a guest? You betcha! Does that mean you’re gonna get great frame rates in Crysis? I doubt that seriously. But at least the guts are there and working now. Tweaking and refinement begin here, and someday, Crysis running on a guest might actually be feasible.

Read More at Phoronix: VirtualBox Gets Accelerated 3D Support

Phoronix performed    interesting benchmarking analyses to determine the efficiency of Java on Windows Vista and Ubuntu platform.The results were quite obvious ,Ubuntu won hands-down.

What follow are few major extracts from their experiments:

• Windows Vista Home Premium SP1 and Ubuntu 8.10 were used .
• Each OS was left to its default settings, including the use of the standard desktop effects.
• All tests were executed through the Phoronix Test Suite.In Windows they were manually run .
• In Sunflow Rendering System,the performance between Ubuntu Linux and Windows Vista was quite close and  between the official Java
  and OpenJDK build on Ubuntu Linux the results were very close.
• In  Fast Fourier Transform (FFT) performance measure, Java on Ubuntu was about 30% faster than Windows Vista Premium.
• In The Successive Over Relaxation (SOR) performance  measure , Java on Ubuntu was about 33% faster than with Microsoft Windows.
• In  Monte Carlo performance measure , Ubuntu’s JVM was 3 times faster than what was tested  on Windows Vista.
• In composite performance measure , the best performance was exhibited using Java on Ubuntu. There was  over a 50% difference, that separated Windows Vista and Ubuntu 8.10.
• The j2dbench performance was significantly better on Windows Vista Premium, but that’s likely attributed to the Intel graphics stack on Linux having little in the way of performance optimizations.

Read the thorough analysis by Phoronix here .

Phortunate the Phoronix Phaggots knew what to do. ATI has had a horrible problem with their drivers in 64bit mode since day one. I’m working on a Dual Core Pentium 4 64bit with an ATI x600 Radeon card here. OpenSuSE 11.1 is a dream but has it’s share of bugs. Ah, the joys of operating system maintenence. Anyway, turns out that whatever you do with the ATI drivers, you end up with /usr/lib/dri. This is full of fail. You want to remove this directory and link ./usr/lib64/dri to here.

Special thanks to the Phoronix Crew.

La gente de Phoronix ha realizado numerosos benchmarks a Ubuntu, en uno de sus últimos artículos hacen una comparación de las varias de las versiones, desde 7.04 hasta 8.10 RC1 utilizando su propia plataforma de benchmarking, Phoronix Test Suit, los resultados, personalmente, me preocupan.

Miremos algunas gráficas.

Arriba se puede ver el resultado de un benchmark de lectura secuencial de memoria RAM, la cantidad de información (MB/s) que el sistema es capaz de leer se ha reducido drásticamente en las versiones más nuevas.

En el segundo gráfico se aprecia el tiempo (segundos) que tarda en codificar un archivo WAV a MP3, nuevamente se ve la caída de performance.

En esta se mide el tiempo que tarda en compilar Apache, vemos que este aumenta de manera progresiva a lo largo de la evolución de Ubuntu.

En Phoronix han realizado muchos test más, y también son preocupantes. Utilizo Ubuntu para mi familia y para mis estudios, y siempre estoy esperando a que se libere la próxima a que salga la próxima versión, pero encontrarme con este tipo de cosas me hacen cuestionarme ¿Será solo un caso particular de esta distribución?,¿No se supone que en el mundo GNU/Linux la siguiente versión es siempre más rápida?,¿Me paso a Mandriva o a Fedora o a Arch Linux?,¿Pollo o asado?.

Esperemos que las cosas mejoren para Ubuntu, noto que se esfuerzan por estar siempre a la vanguardia pero me parece que les falta trabajar un poquito más las cosas. En fin, nada es perfecto.

На днях вышел набор тестирования phoronix. Решил взглянуть на него мельком.
Скачать пакет для убанту phoronix-test-suite (deb)

Установить все тестики, можно командой phoronix-test-suite install-all

А можно подсмотреть список тестиков и установить только интересующие вас тесты.

К примеру, хотите тестик ffmpeg.  В терминале пишете:

phoronix-test-suite install ffmpeg

Запускаем свеже-скаченный тестик

phoronix-test-suite run ffmpeg

Покажется результат в секундах, что бы посмотреть по нагляднее, на вопрос Do you want to view the results in your web browser (y/N)? нужно ответить да (y) Тогда результат отобразиться в браузере.

#phoronix-test-suite # Расскажет о возможных командах для бенчмарка.

Ecco a voi la nuova versione del nuovo programma di messaggistica istantanea universale

Il changelog: http://developer.pidgin.im/wiki/ChangeLog
Il sito: http://www.pidgin.im/

Quote (da Phoronix.com):

“Pidgin 2.4.0 adds support for offline messages for AIM, Yahoo! Messenger 7.0+ file transfer support, 64-bit D-BUS fixes, auto-resizing of the chat input text area, theme support for conversation name colors through the GTK+ theme, and new bindings.”