Language Selection

English French German Italian Portuguese Spanish

To 64 or Not to 64, That Was the Question

Filed under
Reviews

With my nice new motherboard & cpu, I've been quite anxious to test some of my new-found powers. My first thought after the hardware installation was: Gentoo 64-bit! By way of testing, I installed the 64-bit version of SuSE 10.1 rc1 yesterday and had planned on writing this wonderfully informative comparison article of it and the 32-bit version. I was expecting the 64-bit to smoke 32 and had even made preliminary reads into installing the 64-bit version of Gentoo. Since this was my first foray into the world of 64-bit, I expected to be lost and confused. Well, the former may not have come to fruition, but the latter certainly did.

The install went fine. No errors or problems were had. It went as smoothly as the 32-bit the day before. I chose all the same options, packages, etc as the other system and again, I was expecting the install to be quite faster.

Now here's my first mistake for the planned article: I didn't time the 32-bit install, so saying the 64-bit seemed faster is about as precise as I can be. In total from soup to nuts it took right at an hour. The best I can estimate for the 32-bit is little over an hour just for the packages. As much as I love all my readers, I just couldn't talk myself into a reinstall of the 32-bit system just to time it. So, no legitimate points in this category can be given to either system.

        

Ask any expert in the technology field and they'll tell you boot times and application open times are not an accurate means of testing the speed of an operating system. But you know, to the average Joe, this is exactly was speed means to us. So, I had big plans of putting this great comparison chart together that we could all ooo and aah over. I thought another great test would be compile times of a notoriously long build. As you can see from the chart below, our "notoriously long build" failed on the 64-bit and as a result I just skipped it on the 32-bit. But quite frankly, the other speed increases of the 64-bit applications are quite unimpressive. With the failed compile taking the wind outta my sails, this article idea almost withered on the vine. In fact, I'd not even be publishing this at all, even as a blog, if not for the really slow news day and my needing something to post. Big Grin

For the boot test, this was timed with a stopwatch from the moment of depressing the enter key on their entries in lilo 'til the KDE desktop was fully up and ready. Autologin was enabled for the default user. Fully up and ready meant all informative "pop-ups" had disappeared and "busy cursors" had stopped. I used the same X driver and the same options enabled for it as well as kde, and the only app to 'restore' was a two-tabbed konsole on each of the same approximate size. Suse 64 had been booted a few times prior to the test to allow for all of its pre-configuring and such to complete and the tested applications had been started and closed a few times. For the test, the opening of apps were the first for them after a fresh reboot.



32-bit     64-bit
Boot 85 secs 80 secs
Firefox blank 3 secs 3 secs
Firefox TM home 8 6
Firefox Compile failed skipped
OOowriter 6 5
Shutdown 30 27



There are two sides for every argument and there are masses of supporters for each side. But basically the 64-bit experience was a let down for me here. As I read the documentation for installing 64-bit Gentoo systems and saw references to about 20 other docs for special instructions in order to get common tools and basic apps to work, and considering the lackluster performance increase of the same binary distro and setup over it's 32-bit counterpart, I for one am not impressed with the progress of the 64-bit computing systems at this point.

In my Gentoo install I've changed my cflags for athlon64 optimizations with some supported use flags (and emerge'd -e world), am using a k8 kernel, and am rebuilding KDE --with-cflags=march=athlon64. But that's about all I'm interested in as far as Gentoo is concerned for now.

As far as other binary 64-distros? You betcha I'll be checking them out regularly.

Is the failing of one infamously stubborn package to compile enough to say the compiler don't work? Of course not. But no problems were had compiling Firefox 1.5.0.2 on my Gentoo system or my best friend's PCLOS system. As far as stability was concerned with SuSE 10.1 rc1 x86_64, there were no issues. It was as rock solid as any other SuSE install I've had over the passed year. It seemed as functional as it's counterpart with no application failure as far as I tested (which I admit was limited). If you want a binary based 64-bit distro and use only pre-compiled software for that distro, then why not. Go for it.

But am I going to build a new Gentoo 64-bit system? Nope. Maybe later.

More in Tux Machines

Fedora 21 Alpha to release on Tuesday

Today the Fedora Engineering Steering Commitee held a “Go/No Go” meeting regarding the Fedora 21 alpha, and it was agreed that the current release candidates for Fedora 21 met the release criteria. With this decision, this means that Fedora 21 will be released on Tuesday September 23, 2014. Read more

Teaching open source changed my life

Teaching open source has been a breath of fresh air for myself and for many of our students because with the open source way, there are no official tests. There is no official certification for the majority of open source projects. And, there are no prescribed textbooks. In open source, no employer worth working for will ask for official proof of your abilities. A good employer will look at what you’ve done and ask you to showcase what you can do. Yes, it still helps to have a Computer Science degree, but the lack of one is often no drawback. Read more

KDE Touchpad configuration ported to Frameworks 5

Fedora 20 with KDE SC 4.14 has been very stable, and after a while it gets…boring – especially when Plasma 5 is already released and you see screenshots everywhere. If you cannot hold the urge and feel sufficiently adventurous, Dan Vratil has built the Fedora 20 and 21 rpms here. If you install i386 versions, beware that baloo-widgets cannot be installed due to unmet dependencies. So, once you install dvratil’s copr repo, just do: Read more

Intel P-State vs. CPUFreq Benchmarks On The i7-5960X

Testing was done with the Intel Core i7 5960X at stock speeds. The Linux 3.17 Git kernel atop Ubuntu 14.10 was used for this round of testing. The Core i7 5960X has eight physical cores plus Hyper Threading, 3.0GHz base frequency, and 3.5GHz turbo frequency (that's the reported difference in the system information table due to P-State advertising the turbo frequency where as CPUFreq just reports the base frequency). Read more