To get a well-rounded breakdown of where Linux is going, and where it trumps (or fails against) Windows, AnandTech took the two largest 64-bit Linux distributions, their 32-bit counterparts, and the Windows XP 64-bit public beta for a test drive.Anandtech has just put up a review of SuSE, Fedora and Windows XP in 32 and 64 bit modes. For all its methodological flaws, the article brings some much-needed information about the two OSes working in 64-bit mode and how they compare to their 32-bit counterparts. Those that have made the jump to a 64-bit AMD system will recognize some familiar problems, among other things the lack of stable drivers and uneven performance gains in 64-bit mode.
AMD and Linux: Reaching for the 64-bit Trophy
Submitted by Andrew 2004-07-12 Benchmarks 14 Comments
well maybe not really unfair (not all apps had 64-bit versions already) but RAW. I mean, let’s probably wait till the release candidates for windows is out. or better yet, let’s wait till the final 64-bit version is out for windows. and the drivers also play a great part, especially in the gaming department.
>especially in the gaming department.
The only test that in favor of Windows 64-bit was the gaming part.
Lets face it windows is for games
If you want to do serious stuff you better grap a penguin. I am not sure how troughout those tests where and if the person who has run the tests is very qualified but Linux is doing a great job in letting a multi-billion dollar company blush.
>let’s wait till the final 64-bit version is out for windows
Yeah lets wait and wait…and if its not allright then lets wait again for SP1 or SP2.
I guess they couldn’t get one of the main guys to handle this, but they should have.
They used old video drivers for Linux… very bad considering the trouble the older drivers have with the 2.6 kernel. The new drivers have been out for some time, but I guess they don’t check the nVidia site that often.
Then they had that inexplicable problem with video encoding where FC2 was apparently half the speed. I have not seen this myself on FC2 on my AMD64. That they didn’t bother to find the trouble tells you a lot about the review. None of the other benchmarks had that wide a difference. It should have clued them to the fact that they were doing something wrong.
Also, SuSE 9.1 COMPLETE can be had from their ftp site. You don’t HAVE to pay $90 for it. There are also people who legally sell SuSE 9.1 Pro DVDs with no manuals for a lot less than $90. If you don’t need the manuals, but want the CDs or DVDs, get it that way and save some bucks.
both Suse and Fedora have almost the same kernel. What else do you need for fast encoding ? You need a disk, but evey OS has to go through this bottleneck.
Maybe this “out of the box testing” resulted in a benchmark, which has been distorted by prelink of updatedb cronjobs.
In addition they do not compile the software for fairness reasons. Where is the fairness ? Maybe the downloaded mplayer has been compiled on a Suse and is thus optimized for it ?
I don’t think, that these results are representable.
> Linux is doing a great job in letting a multi-billion dollar company blush.
Yeah, like that’s a big accomplishment when you’re giving your product away for FREE. Whooppie doooo.
I’m tired of you Linux zealots. You’re all predictable.
Linux takes a serious backseat to *any* BSD product…which you all don’t recognize.
And on the desktop, it ain’t even in the same car as Windows.
Am I on a lucky “troll” strike or do I smell a sore loser around here?
Just try to imagine what this thread would look like if Windows had taken most of the benchmarks. We would have 300 posts instead of only 7.
well as the reviewers said, they compared stuff out of the box. they didn’t bother looking up FAQs, updating stuff and all.
so don’t blame the reviewer that they reviewed the old drivers. maybe you can ask them to do another review wherein they would tweak the boxes for maximum performance though.
That’s the problem – they’re doing a half-ass review of cutting-edge 64bit software. Is that an oxymoron or what? If you’re doing a review of cutting-edge OSes, you should put at least a LITTLE effort into it. My guess on the drivers is the review is probably a few weeks old and they are only now putting it up on their site.
To be honest, I think benchmarking default configurations is the fairest way to go about things.
“My guess on the drivers is the review is probably a few weeks old”
They are also using a Beta version of windows XP 64 from February. I thought the review was pretty good, at least they found something new to benchmark.
>Linux takes a serious backseat to *any* BSD product…which
>you all don’t recognize.
Its not about BSD or LInux its about Linux 64-bit versus Windows 64-bits. BSD is a very fine OS but it nowhere near Linux in the “real” world, i am not talking about being an alterantive for Windows for businesses. RedHat and Suse are companies how are supporting the customers and their software you do not have that with BSD as you have with RedHat or Novell Linux. So stop making a fool of yourself..
Its not about Linux versus BSD!
>i am not talking
must be: i am talking
ps. Eugenia: could it not be possible to login in here to post
and have the pos. to edit your posts?
Within the constraints of the test (‘out-of-the-box’) I guess this is a fair comparison. However,I have this little linux-zealot voice in my head telling me the concept of this test is unfairly stacked against Linux.
Today, if one buys a 64-bit machine it is to get maximum performance. Let’s face it, these are not your typical home/email/wordprocessing machines (yet…). So in this context, one would update all drivers (and kernels) to the latest ones available. Of course this IS an option with Linux (lots of people tweaking it), but not with the beta-Win64 (fresh out of the MS labs). Limiting the Suse9.1/Fedora Core setups to their out of the box components is unfairly holding these back and giving Win64 an unfair catch-up advantage.
…. but like I said, that is just this little linux-zealot voice talking… It still is an interesting comparison.
I’ll admit this review was fair if you consider it being taken as “out-of-the-box”. But, who runs linux out of the box? Not anyone I know at the university I work for. If you’re running a server or even a regular workstation, one of the first things you’ll do is recompile the kernel. Even for windows, most ppl update the drivers to the latest.
I think that the reviewer should have rather taken this review from the perspective of a mainstream user, not out of the box. This way, it would be more realistic and relevant to the users.
Additionally, 64-bit ain’t a mainstream product… yet. Everyone I know that has a athlon64 got it for the raw speed, not 64-bit. Well, except for the labs, they use it for 3-d rendering and spatial analysis. Personally, I drool over the possibilities of seeing the performance of ecryption and on-the-fly rendering.