There, I said it. I mean, OK, I’ve said it before. A number of times. But now I’m not the only one saying it. Engadget’s Joanna Stern just published a roundup of 4 laptops with NVIDIA ION graphics and Intel Atom processors. And while they blow away your typical netbook with integrated graphics when it comes to HD video playback and 3D gaming graphics, here’s the dirty little secret: When you’re performing tasks that don’t feature GPU acceleration, most of these computers actually feel slower than your average netbook.
My best guess as to what’s going on here is that there are 70% more pixels in a 1366 x 768 pixel display than a 1024 x 600 pixel screen. And that means that when you’re loading web pages or performing other tasks, the CPU has to work a lot harder to pump out 2D graphics.
That’s why I noticed laggy performance when I reviewed the HP Mini 311 laptop, the Gigabyte TouchNote T1028X tablet, and the Asus Eee Top ET1602 all-in-one desktop. All of these computers have single core Atom processors and 1366 x 768 pixel displays.
Now, the difference isn’t huge. That’s especially true now that Flash Player 10.1 beta supports GPU acceleration for computers NVIDIA ION chips or Broadcom Crystal HD video decoder cards. Because to be honest, I probably wouldn’t have noticed this problem early on if it weren’t for the fact earlier machines I tested struggled to play Flash video in full screen mode. The HP Mini 311 had no such problem with standard or high definition Flash video thanks to the ION chipset.
But some apps do take a bit longer to load, and your computer might feel a bit less responsive if you have a higher resolution display.
Now for the big caveat: This is true for every machine I’ve tested with an Intel Atom n270 or N280 chip and GMA 950 graphics. I have not yet tested a computer with an Atom N450 chip, GMA 3150 graphics and a 1366 x 768 pixel display. I’ll be curious to see if there’s any improvement with the new chips.
It’s also worth noting that Joanna found that the Asus 1201N performed about as well as a typical netbook on non-GPU accelerated tasks thanks to its dual core Atom 330 processor. Unfortunately, that CPU takes its toll on battery life. You’ll be lucky to get two and a half hours of run time while watching video.
Of course, Atom ain’t the only game in town anymore. While the chipset works well enough for 10 inch netbooks with 1024 x 600 pixel screens, Intel has a higher priced and more powerful CULV chipset that plays nicely with higher resolution displays. AMD and VIA also have their own solutions.
I think the performance problem is operating system based…
Would like to see the same test with GNU/Linux – e.g. Ubuntu 9.10, can you take a look see and post the results? Thanks!
Also, I’ve dugg this article via:
https://digg.com/tech_news/HD_displays_Atom_Diamondville_processors_slower_netbooks
Brad I know this is just a pet theory based on observations and you don’t have time to tinker with these devices…but.
What happens when you lower the resolution of the machines to 1024 x 600 resolution? I assume the video can be downgraded so what happens then? Do they perk up? Does it have no effect?
Like I said I’m sure you have better things to do then play with them, but it would be interesting. Thanks for you efforts so far!
Actually, I just don’t happen to have any of these devices on hand at the
moment. But I’ll forward the suggestion to Joanna.
Of course the problem is that these machines tend to look awful when you use
anything but the native resolution.
A new year, a new me, at least for a day.
I am going to agree with everyone today.
Brad has pointed out a significant relationship between display pixels
and something.
As others have mentioned, the “something” would be nice to narrow down.
MonkeyKing is on the right track with: “can we compare these machines
with **only** a change to the display driven”.
A good way to do that would be to just unplug the two displays and exchange
them between the two machines being compared.
But we can’t expect Brad to do that, even if he had the time and inclination,
these review machines are “loaners” – not his to tear apart physically.
My suggestion:
Brad, get two external, brand name, monitors. Two that differ only in the
number of pixels. You probably already have those on hand.
Add a comparison column: “When driving the 1024×768 vs aaaaXbbb monitor”
to your collection of reported tests.
_Almost_ as good as swapping out the internal displays of the two machines.
Wow. That seems like a horribly complicated way of testing this. Why don’t you just you know… change the resolution. Set it to 1024×600 manually and see if anything improves. It won’t though, as I am quite certain this resolution to speed conclusion is bogus.
Just like bbbl67 said near the top, resolution means pretty much nothing. I’m not sure what the slow down is, but I can bet you it isn’t the screen.
my 900ha is no slower driving a 19×12 panel than it is just running at 10×6. this whole discussion is bizarre. as someone mentioned, 2d has been fully accelerated for many years now, and display resolution has nothing to do with app load times.
I have HP Mini 311 and installed both Windows 7 Professional and Ubuntu 9.10 Desktop Edition in dual-boot environment (without Wubi).
I think Ubuntu 9.10 performs better in general. But in order to achieve a fully usable system (Wi-Fi, HD video playback with VDPAU), I need to do some initial configuration, setup, and tweaking.
Or rather it might have nothing to do with the video size difference, and everything to with the fact that when you get Nvidia ION graphics, you are also using an Nvidia chipset for controlling the hard disk, RAM, USB, etc., rather than an Intel chipset. Nvidia chipsets are typically buggier and slower than Intel ones at these very basic, non-graphics tasks.
The CPU normally isn’t a factor in file I/O, which is what’s responsible for app startup times. Of course, an overloaded CPU may be part of the problem, but I think the culprit lies more buggy drivers. That, or somebody’s AV program is really acting up.
The only way to judge compare systems’ performance is by holding constant all (or most) software variables. You can’t have one system running Win7 and another running XP, each with its different load of installed bloatware, and compare anything meaningful.
blogeee clocked the 1201N’s battery life when playing video at 4 hours 11 min.
I wonder how their tests differ. Also, I wonder how the 1201N performs when its in the mid and low perf modes.
https://translate.google.com/translate?hl=en&sl=auto&tl=en&u=https%3A%2F%2Fwww.blogeee.net%2F2009%2F12%2Ftest-asus-eeepc-seashell-1201n-performances-et-autonomie%2F
I have been telling for months now what a crap processor the Atom is in comparison to a CULV.
Now the biggest secret. The biggest secret that Brad fails to tell. Of course, he fails to tell this unintentionally.
Here is the secret. The Atom becomes a snail once you use Windows Update and Update all the neccessary updates especially the important security updates.
It becomes even worse when you Install the relevant service packs for Windows.
It becomes super worse when you update your Office 2003 or 2007 and it lags so much once the Office Service Packs are installed.
This is what nearly ALL netbook reviewers fail to tell. Of course, the netbook with a N280 is relatively fast when it is just only an unupdated plain Windows inside. Then it seems quite fast.
When you go to the shop, in the showrooms, you play with the touchpad, you click here and there, it seems so fast. Of course it does because, there is just ONLY Windows inside and it has not been updated. There is no antivirus inside even.
This is the biggest SECRET. This is what ALL the websites, PC Magazines, reviewers fails to tell. They conduct all the benchmark tests, but, fail to conduct the benchmark tests when Windows is fully updated.
Even the blogs such as Brad and Sacha fail to do this.
Eh, I’ve been using an Asus Eee PC 1000H for well over a year and while it,
like any computer, gets a bit slower after a while due to all the crap you
install and uninstall, fragmenting of the hard drive, and other issues, it’s
not that much slower today than it was a year ago.
Sure, most of the netbooks I review are machines I only get to spend a few
weeks with at most. But in my experience, I haven’t seen anything to
demonstrate that Atom processors are any worse than other CPUs when it comes
to Windows updates.
Of course other CPUs slow down a lot after numerous updates, but they can handle that, as the CPU is originally fast to start of with. So, you just notice a slight decrease in speed.
The Atom CPU is slow to begin with. Once you update your machine, it crawls like a snail. Don’t even talk about installing and uninstalling crap over time. It just can’t handle all the updates Microsoft churns out for both Office and Windows. It gets even worse with your Security Suite installed, say, Norton Internet security.
I dare readers to do a fresh install of their netbooks and see how it flies. Now, go and update your Windows and Office.
Just see the big difference. It becomes a snail.
I only dread to think what will happen in months to come when Microsoft pumps out more security updates to install. It would further slow the machine down.
I wonder if the lower performance would still be an issue if you ran a Linux Compiz desktop, since the desktop is then managed as 3D and processed by the GPU.
AND with an ARM CPU… ARM is going to kick the Intel CPU’s butt…!
Can’t wait for CES to see what goodies are out there… and to see the new Pixel Qi screen in SHOW devices on the Floor.
Actually the size of the video resolution should have very little bearing on the performance of any computer these days. This is known as 2D graphics, and it has been accelerated by even the most basic graphics systems since the days of Windows 3.0. Even a crappy video system like Intel GMA950 graphics handles it without sweat, and without any help from the Atom processor. 2D graphics is pretty simple stuff like moving windows & text around on the screen, etc. 3D graphics is what is involved in playing video clips, DVDs, games, etc.