Wifi performance and the new iPad | Hacker News Reader

Wifi performance and the new iPad

workstuff.tumblr.com

58 pointsfields23 comments

Read to the end of the article, folks.

This is what happened: he tweaked settings galore, hoping to squeeze out more performance somehow. Nerds tend to do this; it's natural.

However by removing the automatic settings and forcing it into one or more configurations that may have been suboptimal, performance actually decreased.

When he returned the settings to "fully automatic" everything was fine and dandy and maximal.

Moral of the story (and I quote): "Resist the urge to tinker with these settings."

You can basically ignore everything except that last line—and if you're familiar with the excellent AirPort Extreme, you probably already knew that.

The problem is that there is no particular reason those configurations are suboptimal, except the empirical "it makes the router run slower." Enforcing 802.11n for instance. If settings that should be good choices end up having side effects, that's not a great design of the Airport Extreme.

I agree. Hence the warning.

Agreed, the settings should be more useful in general.

However, it's possible that the flexibility of the settings toward other cases (like using the router with older hardware, or less compatible or buggy devices) necessitate their presence and complexity. Seems likely to me that each of these settings has been placed due to a specific support case with the plethora of device data and testing cases that they must have from years of customer support.

I think, then, that the settings are not for the maximal optimal case, and the misunderstanding here is that the router is somehow not initialized for maximum performance. This may be true for other router brands, but in my experience, is not true for Apple.

In general, I agree that messing with random settings is not a great idea. But these particular settings are some that I would feel pretty confident messing with. Like setting a different 5Ghz network name is done for a particular reason, because you want to be able to choose 5Ghz or 2.4GHz for the connection, and you don't want it to switch back and forth as it wishes.

It seems that the autoswitching has gotten good enough to trust that it's making better decisions than you are about whether the 5GHz network is superior.

I don't get this behaviour with my (latest-gen) Airport Extreme. I have a separate 5GHz network and get consistently 360Mbps to 450Mbps connection to it. Are you sure your Airport isn't faulty?

Changing it to a single name for both networks leaves my iMac always connected via 2.4GHz and no obvious way to switch to 5GHz.

He wasn't using the default settings. When he left everything on fully-automatic, it was all fine.

This is a classic case of someone geeking out over settings and over-optimizing themselves into a corner. Hence the last line of the article, and an excellent moral.

My point was that I'm running a customised set like the one he describes and my speeds are fine, I sometimes even get a 450Mbps link speed.

Saying that, I see no downsides to leaving it fully auto and I think I'll leave it at these settings now.

Are you sure? I don't have an iMac, but I've tested with two MBPs. They will always use the 5GHz network if it's available. This may have something to do with how good your connection to the airport is - does your signal quality say Excellent?

I can't say that this configuration change will help for everybody - I'm just reporting what I'm seeing.

My MBP and MBA fall back to 2.4GHz sometimes. This is annoying because microwave interference is common in my building.

So now I keep separately named networks and can usually connect at 450 mbps (mbp) or 150 (mba). At least I believe that is the case, I can double check at home.

I found that my aTV, MBP (Lion) and iPad 2 were always connecting to the 2.4 Ghz network with my dual band Airport extreme as well. Setting them separately, I have forced it to 5Ghz. I live in a condo building so 5Ghz has far less interference, so I don't understand the reason for the behaviour.

Try this on the MBP:

* Connect to the 2.4Ghz network

* Hold "Option" and click the wifi icon in the top menu

* Note the "Transmit Rate"

* Repeat this for the 5Ghz network

When I first setup my Time Capsule I noticed that my Mac often times was attaching to the 2.4Ghz network. At that point I separated the networks and did the above.

Low and behold, each time the Mac picked the 2.4Ghz network it did so because it had a higher transmit rate than the 5Ghz one in a given location.

Turns out I had to delete the old network from Network preferences and add it again, it was probably set to 2.4GHz only like it was originally.

Now I'm on the 5GHz again I seem to be getting a more consistent 450Mbps link speed although actual throughput seems the same.

I think I might just leave it on fully auto now, it seems to work pretty well.

I'd like to know a little more about the network setup. Are the Expresses bridging via Ethernet, or acting as repeaters? I don't know if that could be part of what he's seeing, but I'm still interested.

This airport extreme is connected directly to outside (cable modem). There are several airport expresses extending the network (I have no idea how to tell if the client devices are connected directly to them, and I'm not sure it matters). There is a second airport extreme bridged to that one, which is running a dedicated 802.11g network to keep those devices off the same network as the faster devices.

I'd really like to know if the situation changes if you:

1. Disable your other airport expresses and the 2nd airport extreme

2. Place your iPad as close to the airport extreme as is reasonably possible (the one connected to the cable modem... just to try to get the max speed – it would be interesting to see what the speed dropoff with distance is in a second, separate test)

3. Try the different options again (especially connecting on 5ghz using a different SSID with the airport extreme's 5ghz network set to n-only)

It's not just the wifi interference (which shouldn't be a problem, but could you also post how many other wifi SSIDs are visible from your iPad?) – but multiple airport expresses mixed with an airport extreme falls outside the realm of what 99% of users do (I think it could be a bug when the airports interact).

As other posters have said, this kind of performance drop is abnormal.

I will run some other tests when I get a chance. I've done #2, with no effect.

I don't think the Airport Expresses have anything to do with it - I've had some reports of other people being able to replicate my results without that.

Have you tried using specific channels (choosing channels with less interference from other access points in your area -- i use wifi-explorer) for 5ghz and 2.4ghz, instead of leaving them at automatic?

I have not in this case, but I've found in the past that the appropriate channels tend to shift with other usage in the area, and leaving it on automatic allows it to compensate for that, while hardcoding a channel can work for a while and then stop being the best.

When it comes to any mass-market consumer product, 99% of the users (I'm exaggerating, no, probably not) will leave all settings at the default. Therefore the default settings have orders of magnitude more real-world testing than all non-default setting combinations COMBINED. Nearly any mature product is therefore bound to behave dramatically better on its default settings than in any other configuration in sheer defiance of what "should" happen.

That really depends. It's completely untrue for, say, TVs:

http://prolost.com/blog/2011/3/28/your-new-tv-ruins-movies.h...

100% true. For this device, just roll with the defaults.

Reply on news.ycombinator.com