Intel Atom: NVIDIA ION vs. Radeon HD 4330 Graphics
Phoronix: Intel Atom: NVIDIA ION vs. Radeon HD 4330 Graphics
Last week we featured a review on two MSI WindBox Atom 330 NetTops that we had purchased to add to our testing farm, which as you may now know went into our Phoromatic Ubuntu Tracker setup that is monitoring the performance of the latest Ubuntu development packages on a daily basis. Before devoting this hardware to the farm, we ran a few benchmarks comparing the performance of NVIDIA's ION GeForce 9400M graphics processor to the ATI Radeon HD 4330 graphics processor found on the MSI 6667BB-004US and several other Atom-powered devices.
Sorry, but I have to express my disappointment with this article -- it seems to be time wasted both from the part of Phoronix and any visitors that care to read it, and invites the question "why?"
The title implied a comparison, but that was sadly not the case, since it is obvious that the proprietary Nvidia drivers have no competition at the moment, especially from the very young and pre-alpha stage AMD open-source ones. I was expecting either nvidia-blob vs. amd-blob or nouveau vs. ati-radeon, at the least (preferably BOTH) -- it would've offered an insight into what people can expect from the respective platforms on Linux.
Furthermore, the reasons given for this weird comparison are quite superficial and basically amount to "didn't care enough" -- either driver can be made to work, even on Ubuntu, if there is time and will. If there is not, better spend the time on more useful endeavours, of which there are plenty...
Just my thoughts on the article, don't take them personally -- hopefully a follow-up will make things right.
I think the reason the comparison uses the OSS drivers for ATI is that in Ubuntu 10.04 the Catalyst driver is not supported at this time. But adding a plot for Ubuntu 9.10 with Catalyst would have helped.
In my machine (Radeon HD3450), Ubuntu 9.10 + Catalyst gives four times the fps you get with 10.04 and the open source stack (Padman @ 1920x1080). This would place the ATI results slightly higher than the ION overall.
I'm a member of the Phoronix Forums since (literally!) the very first day. My Phoronix Premium Subscription is about to be renewed these days, for the second time. That makes the year to come my third year of supporting Phoronix with actual cold, hard cash. I'm sorry to say that it will be my last, if article quality doesn't improve drastically. I'll give it another year, and then decide.
If you want to keep my as a paying member of this community, please establish a sensible modus operandi of what to benchmark, and how to benchmark it. This article makes about as much sense as running a crapload of 3D-rendering benchmarks on similar setups where, say, only the mainboard differs, and then declare a winnder because somehow, one setup came in 0.2% ahead of the other. I don't want to spend my time with crap like that, and it _always_ takes a little bit of my precious time away when an article pops up in my feed reader. Make quality a priority over quantity - if an article - like this one here I'm commenting - seems kinda moot, it's not worth writing or publishing it.
And finally make the fscking PTS graphing output more readable. Please. Pretty-please.