| Forum Home | ||||
| Press F1 | ||||
| Thread ID: 112237 | 2010-08-29 00:26:00 | Internet Speed | PPp (9511) | Press F1 |
| Post ID | Timestamp | Content | User | ||
| 1132810 | 2010-08-31 00:19:00 | True, but going from Auckland to Auckland should still give you a relatively fair indication. If your exchange is overloaded, then you're going to be getting crap results, which are a fair indication of your crap connection. If things are good, you're on ADSL2+, and nobody else at your home is downloading, then you should again get a pretty decent / accurate indication. Internationally though you're right Yep sure ish , but keep in mind you may be connecting to a competing ISP that is hosting the Speed test server..... so there may be additional hops than some else testing in Auckland, not to mention the possibility of load balancing in favour of the Host ISP (that part speculation). Bottom line, nothing beats real live multi source downloading to test speed and eliminate any "one" hop being slow. Test sites are truly worthless. |
Battleneter2 (9361) | ||
| 1132811 | 2010-08-31 01:02:00 | Hmm, I just checked my D-Link, which is one of these (www.dlink.com), turns out DD-WRT doesn't work on it. bugger. Might have to get a WRT54GL after all... kind of defeats the purpose of getting the wireless N Netcomm a bit now! :p Anyhow, it will die a week out of warranty, with bad caps, and I'll have to replace it. :lol: :badpc: Actually Chill, while we're on the subject, do you know of a reliable wireless N router that tomato will run on? I've not done much with N-Wireless to be honest, as all the routers that I've tested with the exception of the Apple Airport-range and the NetComm NB6Plus4Wn, they've all been total crap and highly unreliable. I know that recently there's been support added for a few N-Wireless cards into DD-WRT (Potentially from OpenWRT upstream?) however I'm not sure how much of those changes have filtered down through to the like of Gargoyle or Tomato. I'm still very much a fan of G-Wireless. If you don't need ethernet ports, then there's some nice open-mesh OM1P devices on trademe from a guy who's heavily involved in the ROBIN firmware. Good bloke. If you *do* need the ethernet ports, then go with the WRT54GL. Since the WL-520GU has been discontinued, my vote goes there, not to mention it's *incredibly* easy to do the firmware upgrade: c2s.co.nz Yep sure ish , but keep in mind you may be connecting to a competing ISP that is hosting the Speed test server..... so there may be additional hops than some else testing in Auckland, not to mention the possibility of load balancing in favour of the Host ISP (that part speculation). Yes but you're failing to see that that is real-world results. It's not a "This is the max attainable line speed of your router" test, that's just stupid, you can look in your router for that. It's a REAL "this is what you COULD expect to attain in the real world" test. I know that all the NZ ones are hosted in data centers, and from what I understand they're all gigabit connections to the internet also. I know, when I was recently at Auckmageddon, on a well-loaded gigabit connection to the internet, I still managed to attain over 180mbps to my virtual server in a competing ISP's datacenter. Is that the max that either of the connections could do? Nope. Is it a real-life expectation of what you can expect to attain? You bet it is! Bottom line, nothing beats real live multi source downloading to test speed and eliminate any "one" hop being slow. Test sites are truly worthless. Why? Do you do all your file downloads that way, from multiple sources? When browsing PressF1, do you expect to be able to use multiple mirrors? When gaming, do you connect to multiple hosts to distribute your bandwidth between all of them? For a friend of mine in Henderson Valley, their local exchange is highly congested. They can do QoS, but they've got to limit to around 90kbps (bits, not bytes), but if they do that then it's reliable QoS. Occasionally without QoS they can peak at around 300kbps. What does speedtest.net show them? Exactly what they're going to get in real-world results: 300kbps What if they multi-thread downoads from different sources? Same deal ... What's their line sync speed? 7200kbps! Do you have a false expectation of what speedtest.net should be showing you? I think so ... |
Chilling_Silence (9) | ||
| 1132812 | 2010-08-31 01:11:00 | I've not done much with N-Wireless to be honest, as all the routers that I've tested with the exception of the Apple Airport-range and the NetComm NB6Plus4Wn, they've all been total crap and highly unreliable. I know that recently there's been support added for a few N-Wireless cards into DD-WRT (Potentially from OpenWRT upstream?) however I'm not sure how much of those changes have filtered down through to the like of Gargoyle or Tomato. I'm still very much a fan of G-Wireless. If you don't need ethernet ports, then there's some nice open-mesh OM1P devices on trademe from a guy who's heavily involved in the ROBIN firmware. Good bloke. If you *do* need the ethernet ports, then go with the WRT54GL. Since the WL-520GU has been discontinued, my vote goes there, not to mention it's *incredibly* easy to do the firmware upgrade: c2s.co.nz Cool, few things to think about there, cheers. |
wratterus (105) | ||
| 1132813 | 2010-08-31 03:04:00 | Yes but you're failing to see that that is real-world results. It's not a "This is the max attainable line speed of your router" test, that's just stupid, you can look in your router for that. It's a REAL "this is what you COULD expect to attain in the real world" test. I know that all the NZ ones are hosted in data centers, and from what I understand they're all gigabit connections to the internet also. I know, when I was recently at Auckmageddon, on a well-loaded gigabit connection to the internet, I still managed to attain over 180mbps to my virtual server in a competing ISP's datacenter. Is that the max that either of the connections could do? Nope. Is it a real-life expectation of what you can expect to attain? You bet it is! Why? Do you do all your file downloads that way, from multiple sources? When browsing PressF1, do you expect to be able to use multiple mirrors? When gaming, do you connect to multiple hosts to distribute your bandwidth between all of them? For a friend of mine in Henderson Valley, their local exchange is highly congested. They can do QoS, but they've got to limit to around 90kbps (bits, not bytes), but if they do that then it's reliable QoS. Occasionally without QoS they can peak at around 300kbps. What does speedtest.net show them? Exactly what they're going to get in real-world results: 300kbps What if they multi-thread downoads from different sources? Same deal ... What's their line sync speed? 7200kbps! Do you have a false expectation of what speedtest.net should be showing you? I think so ... KA??? I am saying the REAL WORLD DOWNLOAD in caps is often faster than Speedtest.net result due to single source hops, didn't even mention max attainable line speed. You do realise ISP, often take different hops to the same destination don't you? So to dumb this down...... PCworld "may be" faster for you and "Cnn.com" may be faster for me if you are talking single source web links. Without writing 300 line response you wont read, Speedtest.net is "often" not accurate testing faster connections (a hop cant handle the speed), however can be accurate testing slower connections, which makes a lot of sense when you stop to think why. |
Battleneter2 (9361) | ||
| 1132814 | 2010-08-31 04:43:00 | ...yeah because 9/10 people here are on 100m/bit connections or greater, of course, makes sense, tailor your advice for them ... | Chilling_Silence (9) | ||
| 1132815 | 2010-08-31 04:57:00 | Cool, few things to think about there, cheers. Yeah so depending on what features you want the most will probably depend on which firmware appeals to you more, Tomato or Gargoyle. Same for routers ... The $40 difference between the OM1P and the WRT54GL is 2/3 of the way towards buying you a second OM1P you could use as a repeater / WDS node :D |
Chilling_Silence (9) | ||
| 1132816 | 2010-08-31 05:04:00 | ...yeah because 9/10 people here are on 100m/bit connections or greater, of course, makes sense, tailor your advice for them ... Even when I had just a 2Mbit connection I would often have to multisource my downloads with GetRight to ensure I got the maximum download speed. |
Agent_24 (57) | ||
| 1132817 | 2010-08-31 05:16:00 | Yup, especially with ADSL1, and especially where ISP's traffic-shape. That same shaping is mostly applied to speedtests as well, which means it's an accurate example of what you can expect to attain. Sure, you may be able to get more if you multi-thread downloads, but that's not what speedtest tries to show you. In NZ, if you're with a semi-decent ISP that's not grossly oversubscribing, on ADSL2+, you should be able to attain very close to your max line speeds. If there are issues with you not attaining it, and you know that from other places you *can* (with a single thread), then you know that it's an issue with the remote host. I still stand by my statement that speedtests aren't a waste of time. |
Chilling_Silence (9) | ||
| 1132818 | 2010-08-31 05:22:00 | I have always found Speedtest.net to accurately reflect the maximum line speed (if you use a local speedtest server) If your real download speed is slower, it's almost always because of the server you are downloading from. If you can't download at maximum line capacity from the NZ Ihug Linux mirror then something IS wrong. |
Agent_24 (57) | ||
| 1132819 | 2010-08-31 05:26:00 | Agreed :) | Chilling_Silence (9) | ||
| 1 2 3 4 | |||||