I happen to find myself with a number of rather of interesting computers at the moment with a nice mix of the latest and greatest in low power, decent performance specs. So I spent much of the afternoon running benchmarks on the following computers:
- HP Mini 311 notebook with NVIDIA ION graphics and a 1.6GHz Atom N270 CPU
- ASRock ION 330 nettop with NVIDIA ION graphics and a dual core 1.6GHz Atom 330 processor
- Asus UL20A with a dual core Intel SU7300 Core 2 Duo CULV processor and integrated GMA 4500MHD graphics
I ran three different sets of benchmarks on each system. These benchmarks are designed to test CPU and GPU performance. First up, 3DMark06, a widely used graphics benchmark that tests performance for rendering 3D video gameplay. It’s worth pointing out that none of these computers got particularly high scores on this test, but as you can see from the chart above, the two notebooks with NVIDIA ION graphics did significantly better in this GPU-intensive test than the Asus UL20A. Higher scores are better.
Next up we have Cinebench. This test consists of three parts. In the first test, the utility tests OpenGL 3D graphics performance. Not surprisingly, the NVIDIA ION-powered machines are the winners again here.
Next, Cinebench attempts to render a high quality photo-realistic image using a single CPU. This takes forever on a notebook with an Intel Atom processor, although the Intel SU7300 processor is pretty sluggish too. The final test uses as many CPUs as the program can find to render the same image. This gives the dual core ASRock and Asus computers a big advantage over the HP Mini 311.
The final test is PassMark PerformanceTest 7. This utility runs a number of tasks to check a computer’s CPU performance as well as 2D and 3D graphics processing capabilities. As with the other tests, higher scores are better. It appears that PassMark PerformanceTest 7 doesn’t take advantage of all the power that the NVIDIA ION GPU has to offer, because the Asus UL20A trounced the HP and ASRock computers in every area of this test.
These benchmarks should give you a basic idea of how these computers stack up against one another. But I’m not a huge fan of these types of benchmarks which spit out an abstract number. I’ve also started running another set of tests to see how these computers stack up against others I’ve tested in everyday tasks. For example, how long does it take to transcode an audio or video file? Unfortunately, there’s not a particularly clearcut answer there either, because the NVIDIA ION platform allows the ASRock and HP computers to take advantage of NVIDIA’s CUDA platform.
What that means is that CUDA-enabled applications will run incredibly fast on these computers, even though they have fairly slow processors. And for that reason, I’m in the middle of adjusting my personal benchmarks. I had been using VirtualDub to transcode video files to XViD. But as far as I know, there’s no CUDA-enabled application that performs this task. So I’m starting to use MediaCoder to transcode video from uncompressed AVI to H.264 using a CUDA encoder.
What kind of difference does that make?
See those enormous blue bars? That’s how long it took to transcode a 4 and a half minute video to H.264 without using the CUDA encoder. The green bars show how long it took using MediaCoder’s CUDA encoder. The dual core Atom processor certainly helps speed up the task, but the GPU-enabled software made a much bigger difference. In other words, if you have an NVIDIA ION powered system and a choice of using GPU-enabled software, use it.
On the other hand, while these scores represent a relatively common task such as transcoding video, they don’t apply to all everyday situations. For instance, the dual core Atom 330 processor was able to transcode the video almost twice as fast as the single core Atom N270 CPU using the x.264 encoder. But the computer doesn’t feel twice as fast when surfing the web, running multiple programs, or performing other common tasks. It feels a little faster, but not twice as fast.
Could you benchmark CULV platform doing video encoding ? Could be interesting to compare encoding video performances against best Atom + ion platform.
Any chance of getting the HP dm3z included (or any system with AMD Neo X2 + RS780 IGP)?
Thanks Brad for this review. We appreicate your efforts.
Thanks for the comparison, and I’ll be looking forward to more in the future. This is the sort of thing I’ve been needing to help make sense of all the different processor and graphics combinations coming out now.
you should throw the MSI u210 in the mix, the Athlon neo is faster than the CULV processor according to notebookreview.com
Misleading. It’s faster than the single core CULV in their Dell Inspiron 11z (Intel Celeron 723, single core) and faster than the single CULV in their MSI x340 (Core 2 Solo SU3500). Looking at other comparisons, it loses to Core 2 Duo CULVs.
Here is my take on a comparison between ION and CULV. HP Mini 311 and Acer Aspire 1810TZ with Pentium Su4100 dual core. https://cdnn.wordpress.com/2009/12/12/atom-ion-culv/
After reading your take on the CULV vs ION issue, I fully agree with you that all these ION stuff is all not worth it compared to a good SU4100 or a SU7300.
People are just wasting their money on all these ION stuff. Video and grahics are always done on one’s primary PC in their house and not on their netbook.
Well, actually, I require my netbook to be video and graphics capable…that is the whole point of my search to replace my Aspire One. And the Aspire 1810 as described in my piece is capable of both. I am simply on the road too much to be reliant on my PC at home, and already carrying another laptop for work…thus the netbook form factor is very important to me. Your mileage may vary.
YMMV should probably be my site motto.
Every now and again people ask me which notebook is *best* or why I don’t provide rankings or scores with my reviews. The answer is because the things I’m looking for in a notebook may not be the same as what you’re looking for. Some people might *want* a netbook primarily for playing games on the go. For them, an ION system might actually be better than a CULV system, since it will do a better job with many 3D games.
Others want a well rounded system that is jack of all trades, master of none. A CULV or maybe an AMD Neo/ATI Radeon 3200 system might be a better choice (provided battery life isn’t as important).
Speak for yourself
All this ION stuff as you call does far more than waste money, if you ever tried watching Hulu, or playing just basic RTS game to pass the time away on a train ride or sitting at starbucks unwinding you would know that. There is a segment of the market that wants better multimedia capabilities in a netbook sized package. I would also state this is not a CULV vs ION discussion as those are two different categories of hardware, that would be more accurately stated as CULV vs ATOM issue.
Well said – good read. I got an Acer 1810T with the SU7300 – couldn’t tell you for the life of me which is better, the 7300 or the 4100, but from what I could see, the 7300 was in the lead.
I want to be able to do Photoshop, stream HD video, and minimal video editing. I had a Dell M1210, which was a unique powerhouse in it’s day – I wish someone would make an updated version of it. I can do PS and Adobe Premier Elements on it easy, and it’s OLD and full… it’s just too heavy and I don’t need the weight-suck of an optical drive, etc. Why is no one making a high-horsepower ultraportable like that Dell? Rrrrr..
the 7300 is higher rated on the Intel site and is newer tech so I am sure it is faster. No one had a 1810T in stock when I wanted it, so I got the TZ. Does everything I have asked of it so far, and I saved $50.
Awesome Comparo Brad!!
I think a big reason different spec’d computers don’t FEEL much faster is similar HD speeds. Video & gaming aside, the HD is often the performance bottleneck (somewhat mitigated by excess ram).
After thinking a bit on how best to test/benchmark the speed “feel”, I remembered a good piece on SSDs (https://anandtech.com/storage/showdoc.aspx?i=3631&p=20) Anand does essentially an application launching speed test.
I personally would nix the OS startup test as that can be impacted by bios splash screens, etc and really with sleep/hibernate – who cares if it takes 10 more secs to boot.
Anyway, with this free tool (https://blogs.howtogeek.com/mysticgeek/2008/10/30/benchmark-startup-times-of-software-applications-with-apptimer/) you could easily compile a “real world” benchmark of the total time to launch several common applications.
Knowing Brand A can on average launch Firefox, Photoshop, and Word (or whatever) 20% faster than Brand B is probably more useful than know it scored 200pts higher on the PCMark05…
Just a thought in the quest for the perfect set of benchmarks. 🙂
The netbook of my dreams has a 12″ screen, CULV Core 2 Duo processor, and ION or ION2 graphics. Why doesn’t this exist?
Partly because ION 2 doesn’t exist yet. It’s possible that we’ll start to
see something fitting this description after CES or when NVIDIA launches its
next-gen ION platform later next year.
In the meantime, Asus does have the UL30VT which has switchable discrete
graphics, but it obviously costs more than a UL30A with dedicated graphics
only. And it’s a 13 inch notebook. But it’s not too big a stretch to think
that someone could come out with a similar 12 inch model. Just expect the
price tag to be closer to $800 than to $500.
Interestingly enough I spent today installing all my software on, and copying files to, an Acer Aspire Timeline 1810TZ with dual core Pentium (SU4100) and Intel GM4500 graphics to test against an HP Mini 311 with Atom 270 and ION. So far, in real world use, iTunes, Chrome, even Lightroom, Corel VideoStudio 2X, full screen video on Hulu, the Acer is simply way faster and more stable than the HP. More through testing scheduled for tomorrow.
i dont see why you didnt test a CULV w/ ION… like asus 1201n or whatever its called… that will be best of both worlds
unfortunately there is no such thing quite yet
a) The Asus Eee PC 1201N has a dual core Atom 330 CPU and NVIDIA ION graphics, not a CULV processor. It’s a lot like the ASROCK computer, but in laptop form.
b) I don’t have an Eee PC 1201N on my desk yet, but I hope to soon.
Here is a bit of trivia that helps explain what you are experiencing…
As you probably well know, the human senses are logrithmetic responses.
But the bit of trivia: so are “subjective impressions”.
The brain only works on a log scale (yeah, I know, like a dead log sometimes).
So it takes approximately 10 times the procession power to give a subjective
response of “twice as fast”.
Which is just what you are reporting –
A very measurable numerical difference and a subjective “so what?”.
Suggestion: If you have time, plot those results on a log scale.
They should be close to your subjective results.
I think you need to establish a methodology for your benchmarking. Many of the sub categories in the benchmarks aren’t described, e.g. what’s a SM score, and what’s the diff between SM v2 and SM v3? The graphs also need labels as to which is better, longer or shorter bars, not just a mention in the text.
Secondly, video transcoding is an irrelevant benchmark for this category of devices, as people don’t do transcodes on netbooks or ultraportables (not sane ones anyway). CUDA-based uses are rare, and it is biased in favor of Nvidia chipset. One may say that it’s common enough for hard-core video transcoders, but those again would be done on a desktop with sufficient power. What you’re doing is giving Ion-based netbooks an arbitrary boost over other choices.
You’re kind of ignoring the fact that the Asus UL20A trounced the other two computers in two out of three Cinebench tests.
The point is, if you want better graphics performance, the ION will outperform the GMA 4500MHD. If you want better CPU performance and decent graphics for some applications, the CULV chipset is the way to go.
But this is just one set of benchmarks. I’ll be posting detailed reviews of the HP and ASRock computers and you can already find a detailed review of the UL20A in the reviews section of the site.
As for methodology, the Cinebench, 3DMark06, and PassMark tests aren’t mine. I downloaded and ran the tests and posed the results. You can find more information by clicking the links I’ve provided to their respective home pages. The video transcoding test is the one that I did decide to run on my own, as it’s a CPU-intensive task that people may decide to use if, say, they want to upload a video to YouTube from their portable laptop.
I’ve been performing similar benchmarks on a number of CULV systems. I haven’t bothered with Atom-based netbooks, because as you can see, without the GPU-acceleration, it’s slow as all get out on an Atom machine. So once I installed MediaCoder and used the CUDA encoder, I was able for the first time ever to find an Atom powered notebook that could outperform a CULV powered one on a video transcoding test. I don’t think that makes this a biased test. I think it makes it notable because it stands out from all the other tests I’ve run to date.
A methodology isn’t just about running any particular test, it’s about delineating the common functions that will be used on a particular device (or category of devices), and coming up with a suite of benchmarks that best test those functions. Ask yourself some questions: Why am I testing transcoding on netbooks/ultraportables? What does Cinebench test, and how is it relevant to netbook use? Ditto for PCMark and any other benchmark you care to use.
More importantly, it is not just for you to know, it is for your reader to know.
But I understand that this is a blog, which are suited for more off-the-cuff style commentary, which perhaps extends to reviews as well. If that’s the case, please state up front, so again your readers will know just how serious they should take your efforts.
“…and coming up with a suite of benchmarks that best test those functions.”
I’d be curious to hear what sort of magical benchmarks Brad should be using instead, as he seems to be using industry-standard choices.
Typically, the entire point of benchmarking is comparison. Without similar choices across the entire gamut of machines, comparison becomes pretty difficult.
I imagine many would-be purchasers are asking themselves questions such as, “How does this machine compare to Brand X 12″ UMPC, or Brand Y 15.6″ notebook?”. Using some industry-standard choices for benchmarking means at least some small measure of comparison can be made.
And finally – thank you, Brad. Your work is appreciated. I can’t recall an instance of thinking your thoughts or efforts were ever off-the-cuff.
Although, my brain benchmarks could be too low to perceive such perhaps?
In this comparison,
Highest score is better for all Tests performed.
It takes detailed eye and common sense to know which components of the test actually test the CPU and which ones test the GPU.
When people see this they’re gonna think the ULV is slower than ASUS ION then buy it. Then they’re really gonna regret it.
Bottom line is ATOM is meant to be cheap.
The ULV (SU7300) is supposed to give you battery life without sacrificing performance.
ATOM is gonna be in trouble when netbooks with beefed up Tegra 2 come out with Linux, probably offering same performance but more battery life in a smaller package.
Brad, awesome article. This is something I’ve been VERY interested in for some time now. Would you say, just mucking about that the dual core atom makes much of a difference over the Mini 311? I know I want something with at least minimal graphics (we have a application development suite at work that won’t run on GMA4500). Whether I can hold out until ION2 and CULV merge, or whether to get a Congo based platform like the stupid Ferrari or the MSI 230, Asus 1201n, or even a Mini 311 have been some somewhat agonizing decisions. I’m going to have to live with whatever I get for a few years. So I’ve been leery of the 311. I’ve been impressed, but not overly by the CULV lineup so far, but again the GMA4500 is a killer. I’m not overly certain HD3200 does what I need but I’m interested in that platform as well. Anyway I digress. I think I’m basically just interested in more personal insight into what your experience has been with the 3 systems (I know the ASRock isn’t a netbook, but I’d expect REASONABLY similar experiences from the Asus 1201n). My primary need is for something portable, because I’ll need to carry it in the field for 6-10 hours a day. So weight and to a limited extent small footprint are major requirements for me, otherwise I’d be looking at larger ‘more powerful’ systems in the 500-750 dollar bracket. Any feedback you can give me would be GREATLY appreciated.
I need to spend a bit more time with these machines, but from what I can
tell, if you go with ION you get great graphics, but sluggish everyday
performance. And I mean, potentially a bit more sluggish than a typical Atom
computer. My best guess is that the culprit is the high resolution display.
GPU-accelerated applications have no problem pumping out 1366 x 768 pixel
images and videos. But applications that rely on the CPU to do that? That’s
Good to know. Need to research the Congo Platform more I guess… Since that supposedly has better performance than Atom, and has some graphics support… If only the battery life was stronger.
So Brad, if you had your choice for for multi tasking (documents/internet/databases) and limited graphics, would you say the su 7300 duo core is preferable over the duo atom 330? I’m debating the asus 1201n over their ul20 …thoughts?
The conclusion of this benchmark seems biased.
What needed to be said was:
The better processor is: (Cinebench Multi CPU test is relevant)
SU7300 (3,148), 330 Dual-Atom (1,409), N270 Atom (795).
The better Graphics Card is: (Cinebench OpenGL)
ION (1,163), GMA 4500 (689).
I’d take better processor over better graphics any day. Why? Processor affects everything including graphics, Graphics only affect graphics.
Notice the was the only one with dedicated graphics card (the ION) compared to the other integrated ones which opt for battery-life not performance.
You NEED to TEST the BATTERY LIFE.
I have a Acer Aspire 4810TG= SU7300, integrated and dedicated graphic, HD 4330 graphics card, 4GB DDR3. It can do Crysis mediocre (35fps) with medium settings, can run “8+ hrs” (my video benchmark was 5hr 49min), is 1.5kg and 14″ size.
My dedicated video card destroys the ION, or ION2 hands-down. SO I’ve got the best of both worlds. As far as anyone knows, my laptop is top of the line / bleeding edge.
ie The most powerful Laptop you can get for its effeciency. I’d recommend my Laptop for anyone, except for Acer’s cheap build-quality (touchy keyboard!!, scratch prone lid)
Sorry, Atom is just not as effecient yesteryear-gen device.
To be honest, I now feel like “Windows benchmarking tool” is accurate enough, all it needs is a battery life included then everyone can buy the best without being confused. BTW computer retailers have now locked away this WinBenchmark on display models as it slows the sale of some stock, they want customers confused.
Let’s put it this way… I bought a UL20A and I’m pretty happy with it. But
I don’t use my computer for gaming, and the ul20a does cost a bit more than
Thank you for responding…I was debating…I don’t game either..I saw complaints on problems w/ wireless and the keyboard being flexible and the trackpad not sensitive…I have two eee, ones a 1000ha and ones a 1005ha and I have been very happy….I’m replacing a hp 12inch tablet which has been less than reliable.. the size is perfect though….
So no performance issues no quality issues?
All I can tell you is that it meets my needs in a portable computer at the
moment. It might not be the best choice for everybody. You can check out my
detailed review in the reviews section.
Thank you for this review first of all. I am planning to buy either an ASUS UL20A , or HP MINI 311. I have no interest in gaming. I just want to watch 720P or 1080P MKV movies with it. Which one is more capable for that?
Comments are closed.