In my Archos 5 first impressions piece, I said that the Internet Media Tablet’s ARM Cortex A8 processor and Opera 9 browser create “the best ARM-based browsing experience around right now and [is] definitely the one to beat.” To qualify that statement, I’ve put the Archos 5 against the iPhone 3G and Nokia N810 (both commonly regarded as the best ARM-powered devices on the market) in a round of website load time tests.
Though page load time is just one of many ways to evaluate a web browsing experience, it is often the most important. Full Web 2.0 support is key, too, but most consumers tend to value faster over fuller, especially when it comes to mobile internet devices.
Parameters
I’ve done load time comparisons before (here, here, and here), so my standard disclaimer and testing parameters apply:
- This was not a scientific study or professional lab test.
- Each browser’s cache was cleared before each set of tests.
- Load times were measured from the same location (about 12 feet from a 802.11g wireless router) in a one-hour period from the click of the enter key or “Go” button in the browser’s address bar to the complete page load according to the progress bar.
- Load times vary by location, time of day, ad servers, content, etc., so your results may not be identical to mine.
Devices
Before we get into the results, let’s take a quick look at the gadgets.
- Archos 5: 600MHz ARM Cortex A8 (additional processor: 32-bit DSP @ 430MHz)
- iPhone 3G: 620MHz Samsung S5L8900 (ARM1176JZF) underclocked to 412MHz
- Nokia N810: 400MHz Texas Instruments OMAP2420
Load Times
| Archos 5 Opera 9 browser |
iPhone 3G Safari browser |
Nokia N810 MicroB browser |
|
| Amazon | 10 seconds | 26 seconds | 26 seconds |
| CNET | 24 seconds | 21 seconds | 32 seconds |
| Digg | 18 seconds | 29 seconds | 36 seconds |
| Engadget | 25 seconds | 27 seconds | 44 seconds |
| Google News | 4 seconds | 8 seconds | 11 seconds |
| Micro PC Talk | 13 seconds | 14 seconds | 20 seconds |
| MySpace | 11 seconds | 13 seconds | 16 seconds |
| NY Times | 19 seconds | 29 seconds | 36 seconds |
| Pocketables | 16 seconds | 31 seconds | 39 seconds |
| YouTube | 15 seconds | 22 seconds | 18 seconds |
Note: Nokia N810′s power-saving mode was enabled for these tests because it is the default setting.
Though the results aren’t consistent across the board, I think they speak for themselves. I’ll take a comprehensive look at web browsing on the Archos 5 soon, as there are definite shortcomings, but you can already see that the player is pretty close to the top of its game.
Related entries:
Another excellent article Jenn. I really like these common sense comparisons – I don’t see anyone else comparing devices that on paper do not seem to have much in common but in fact they are more similar than one would expect.
I will be interested to learn more about Archos’ browsing capability and the way sites render.
Many thanks,
Bruno
was the in-browser flash on or off on the N810? if it was on, and you bumped into a flash based ad, bb N810 load times…
archos and iphone on the other hand dont load flash that way…
also note, even with the latest updates, the N810′s browser have known issues with javascript speeds. this because nokia chose to freeze the code right before the mozilla gecko devs started on some serious optimizations of the javascript engine.
I forgot to mention this in the post, but everything is running the most current firmware.
Archos 5: Version 1.0.76 beta
iPhone 3G: Version 2.1 (5F136)
Nokia N810: OS2008 version 4.2008.36-5
@ Bruno: Browsing really is quite good. I’m actually finding myself reaching for the Archos over the iPhone when I want to look something up quickly now. I think I’ll do a video demo…
@ turn.self.off: Flash was on. The load times were actually slower than I expected. I knew they wouldn’t be identical to my previous tests, but they were several seconds slower across the board.
The thing is, though, that even if the connection was somehow slower this time around (all devices showed a full signal), the difference between all three sets of results is still the same. So rather than zero in on the actual numbers, I think it’s better to look at the differences between them.
Do you think there’s a “fairer” way to compare load times? I didn’t want to tinker with settings or anything because I wanted to keep everything in their default states.
I don’t know how differently Archos handles flash content compared to Nokia, but there were times when a flash ad on Engadget caused the browser to hang. It would then be unable to load any page after that. A relaunch of the browser or a restart of the device would fix the problem, but something is definitely buggy. The player still has beta firmware on it, though, so I’m not surprised.
well it was not meant as any kind of attack, i just wanted to inform of the possible issue. this because cnet did a similar comparison between a wimax edition N810 and a iphone 3G to test the speed of the networks…
hell, people over at internettablettalk (itt) is a bit angry at nokia about the timing of that browser code freeze. and from the sound of it, nokia wont do anything about it until fremantle, and its starting to look like it will not run on “older” devices…
i wonder tho, will you be getting a pandora at some point? the preorders for the first run seems to have taken everyone by surprise.
ones that device gets a proper browser, comparisons could be interesting
Oh, no, I didn’t think it was an attack at all. I’m not sure if I sounded defensive, but I didn’t mean to if I did. Sorry about that!
I’m really glad you mentioned the issue because I wasn’t aware of it. I don’t keep a close eye on Nokia IT news, so I would never have known about the code freeze otherwise.
I don’t have any plans to buy the Pandora, but there’s always a chance I could be persuaded by user feedback and photos. Someone in the forum ordered one, I believe, so I’m looking forward to his thoughts.
Thanks for the great info! If you get a chance, I’d be interested to see the timings for at least one slow page with N810 power saving on vs. power saving off, just to get a rough idea of how that impacts load time.
Just for information : the Archos 5 dispayed on your screen is a high capacity version. The Archos 5 60Go has the same performance and almost the same
thickness than iPhone 3G…
@ B. Factor: Okay, just now Engadget loaded in 43 seconds with power-saving on and 55 seconds with power-saving off. Very strange, as latency is supposed to be significantly reduced at the expense of battery life. It’s tricky to find a heavy site whose content/ads don’t change with every refresh.
@ Arthur: Ah, thanks. I’ve been using the player so much lately that I completely forgot a slimmer version even existed!
Well, wouldn’t you know it. The day after I post this, Archos puts out new firmware. The new version (1.0.87) adds some new features (notably OGG/FLAC support and improved USB speed, which I’m testing now) and includes some bug fixes, but it doesn’t mention anything about web browsing or load times. I’m glad I don’t have to redo the tests!
Thanks for your tests, if you made the upgrade to the firmware 1.0.87 , can you give us some speed results ? (I saw it add some wifi enhancement)
Thanks in advance,
Yann
@ boosti: Browsing speeds are still the same. The new firmware fixed a wifi bug but did not improve load times.
Was fit to width checked on the N810? That slows things down a lot. Also, if you want the optimal speed on the N810, turn flash off, and uncheck javascript. I know if you want the full web experience that’s not the way to go, but to give you an idea how the numbers change.
8MB download speeds 768Kbs upload. Cache cleared, flash off, javascript off, fit to width unchecked, latest firmware
Amazon.com 13 seconds
Cnet 7 Seconds
Engadget 8 seconds
NY times 8 seconds
Youtube Flash on, Javascript on 12 seconds
The brand spanking new chip in the Archos 5 gives it better performance, but those numbers can be slashed on the N810 with a bit of tweaking, even if that does mean turning javascript and flash off.
I thought that Apple was supposed to be making the “brick”. Seriously, you think that having a fast browser makes up for it’s size? I doubt I could even fit the Archos 5 in my pocket.
Wish you could of thrown in a wireless G laptop in there for good measure….maybe when you have to redo the test.
With all the resources (Apple 35M US and Nokia 20M USD)spent on browser development since 2003 you think they would create a better performant browser than a small Norwegian company.
Instead of trying to control the web experience, they should be licensing a better product rather than trying to make their own…and giving their customers the best embedded web experience.
Both Apple and Nokia could have a better product by licencing Opera! At a cheaper cost than ammount of
engineer effort that they have thrown at their “closed source browsers” (yes both Apple and Nokia have closed source browsers with Open Source core libraries.)
This common sense review clearly shows large companies like Nokia and Apple don’t know when to buy a better component rather than spend a lot of effort and money internally to get less than better results for the end user.
In the end we (consumers) foot the bill for their control stratergies. The n95 browser is so bad you have to use Opera Mini!
Great article. Thanks.
Archos is a small french company, not Norwegian, isn’t it?
I’m guessing the opera speeds were only faster because of the way that opera uses a proxy to downscale the gfx first?
so if you are happy with trusting opera people with your secure data then it’s better!?
Heh, what use is looking at youtube load time on iphone
You definitely should compare page render speeds with both flash enabled and off to get a fair review. Even then it’s already clear that archos has some kickass hardware.
Just to clear things up for all you Americans:
Apple = US company
Nokia = Finnish company
Archos = French company
Keep up the good work Jenn. I noticed lots of negative comments about your testing on Engadget but I’m glad you do it. Can’t wait for the Aigo review.
Bah, those Engadgeters just didn’t like that the N810 wasn’t tweaked first. They don’t get that the point was to show default states if all three devices. Disabling this or changing that would have made the test result really skewed and unbalanced. Turning off flash and javascript on the N810 would have been really dumb. Even if it cuts load time, it compromises the test and undermines the browser.
The test was as real as it could have been. Don’t mind the haters. People just get cranky when the iPhone and N810 don’t come out on top. Insecure fanboys!
LTM: You can see in the pix that fit to width wasn’t used on the N810. Your load times are great but what’s the point of having a browser capable of flash and JS if you’re just gonna turn it off? I don’t get it.
It wouldnt be a fair test if you turned all the settings off on one device to speed it up.
Leaving all settings as defaults is fair in my opinion to be honest.
swerve,
My point was to have the option to turn it off changes the Nokia surfing experience from frustrating to very tolerable. When I’m mobile, I don’t want the full bells and whistles especially when all I’m trying to do is look up something or check webmail without anything slowing it down. I’m still getting 80% of the web experience while cutting webpage times to 1/3 of the bloat time. If this is the compromise that has to be made on a 400mhz processor until the next generation Nokia tablet comes out, then I don’t have a problem with that. All I’m saying is that by having the option to turn javascript and flash off, it will render speeds comparable to the archos 5. Granted, the Archos 5 has the high ground right now with giving you the maximum web experience without compromise due to the faster processor, but even if I owned an archos 5, I’d probably be looking for the option to turn flash and javascript off to speed things up even further.
To make the comparison fair, why don’t you setup a local web server on a desktop machine and make a static copy of those websites (ads and everything). This would eliminate server load/traffic/adnetwork from the equation.
Please add something like a HTC Touch Diamond running Opera Mobile 9.5 and a HTC Kaiser (AT&T Tilt) running SkyFire beta2
So you know, the iPhone, iPhone 3g and iPod touch all use a 620mhz processer that is underclocked to ~415MHz.
Interesing comparison! I think that the archos is running more bleeding edge code, while apple hasn’t yet finished incorperating the latest webkit changes into safari, and people have already mentioned the issue with the n810 code freezing. Flash support is really not needed on a mobile device (IMHO I never miss it) but fast javascript is. Opera vs mozilla vs webkit, a great war.
What about what has the most full featured, usable, and best experience when it comes to the browser? I get the impression the 810 would be the best.
I don’t understand why the comparison is with a 3G device such as the iPhone. Because my informal testing with the 1st gen iTouch is faster than the Archos page loads.
I think its bull-shit to call it “fair” out the box default testing. If the nokia tablet supports flash while the other two do not, how can you call it to be on the same playing field. Flash should have been turned off. NO that is not tweeking, it is called evening the odds. By the way loved the review, an i dont wanna be steppin on ya, but gotta put that into consideration.
@Joey,
The iPhone is the only one that does not support flash.
@Tim Dunne,
The Opera proxy for downscaling graphics is only used in Opera Mini, which is great for phones, but is a security issue as you say. However, Archos is using Opera Mobile, not not the Mini version.
@Jenn,
Interesting comparison. According to Engadget, the interface on the Archos can be an issue (touch screen and scrolling probs), have you had any probs? The Archos does sound like a nice portable surfing machine.
Gosh – I’ve had the N810 since Dec 07 and love it. Didn’t know that I can switch off flash and javascript to get more speed. (Not that it really bothered.) Now I can have best of both worlds! Thanks guys.
and the best of three is…. OMNIA !900..!!!
You know, your timings of n800/n810 browser are incorrect. Not slightly – they are totally fake in fact. I don’t know what device have you tested but that was deffinetely not a n810. Heres my timings (with all same rules as described):
Amazon: 26s (26s)
CNet: 24s (32s)
Digg: 14s (36s?!)
Engadget: 36s (44s)
Google news: 11s (11s)
MicroPCTalk: 15s (20s)
Myspace: 11s (16s)
NYTimes: 13s (36s again?)
Pocketables: 21s (39s…)
Youtube: 19s (18s)
I think any comments are unnecessary…
two things, the N810 is the only device of the 3 that do inline flash, so even flash based ads are included.
the archos is able to play flash based videos in its video player.
the iphone i thing dont support flash at all.
and a little anecdote, i find myself using noscript in desktop firefox, it allows me to strip out some 90% of the ads, select what flash or other plugin elements i want to see, and do so on a page pr page basis.
now thats a feature i would love to see in a pocketable browser
@wazd: It clearly states in the article that load times vary by content, ads, and location. If an ad server on one site isn’t serving the ad quickly at a certain time, that is the ad server’s fault. This test is not scientific. It says that in the first bullet point. It’s just load times taken by one person during one day in one location. Try timing those sites again every day this week and you’ll get different times each time. Ad servers play a huge role in load times, especially when sites serve so many different kinds of ads from so many different companies.
As iPhone doesn’t support Flash at all, it should be turned off from all the devices for the numbers to be comparable. If Archos supports Flash like N810 does, there could be separate numbers for Flash enabled in those to see how much that affects the performance.
Note about N810: The speed could be indicated as 330/400Mhz. Depending on situation, N8x0 are run either at 400Mhz or at about 330Mhz (when DSP does heavy operations, its speed needs to raised which due to clock multipliers requires lowering the CPU speed. AFAIK DSP isn’t used for anything needed by the browser though).
Posted by Tim Dunne | 08:14 AM on Oct 18, 2008:
“I’m guessing the opera speeds were only faster because of the way that opera uses a proxy to downscale the gfx first?”
That’s Opera Mini, not Opera Mobile. Archos uses Opera Mobile.
Well, these results aren’t suprising, to me. The N8x0 devices have the oldest, slowest hardware out of the three. It’s the slowest.
I have an N800, and I wouldn’t trade it for anything other than a Pandora or the rumored N900. Those two devices will have more comparable hardware to the new Archos.
Anyway, my next device will probably be the Pandora. I had a GP2X, and it was fun. When I wanted a more full-featured device, I got a N800. Now the Pandora looks set to bring together the best of both devices.
I just did this test with my Nokia E51 and the Opera Mini browser, much much faster than in your test.
I would never believe that Archos is a French Company because the mobile’s design is like a BRICK or a fat pensioner in front of other models. But… it’s much faster runner…
Hmmm, intresting really. It confirms my view of the iPhone being the best such device.
I mean overal. Best form factor, and best display brightness as u can see from the pictures. The only pda I have used that is usable under direct sunlight… this is really big issue with all LCD display devices…
i loved your review on the speeds of these devices. I have been looking for alot of reviews of the archos and you did a great job comparing i for me… i am definatly going with it because they fix alot of things with updates thanks
Jenn, thanks for the fantastic comparison that just shows that you don’t need a X86-architecture to handle web browsers.
BTW, the Archos-5 uses Texas Instruments’ 3rd generation applications processor viz. OMAP3.
the iphone processor is a older generation ARM than the one in the archos…and it is not running at 620, it is running at 432mhz
im not sure where you got your info.
the upcoming iphone will likely have the ARM cortex a8..and will be running at 600mhz
I believe it’s a 620MHz CPU underclocked to 412MHz (not 432). I don’t think the underclocked speed was confirmed when the article was written 7 months ago, but I’ve added the info now that it’s a known fact.
i have an n810 and it goes the exact same speen as the archos 9 with tear.
I agree with Ash, I have the ipod touch and Nokia N810, and the Nokia goes twice as fast as the ipod touch!!!!!!
too bad they didnt use tear or midori on n810 when testing. n810 would have pwnd the iphone and archos