Realistic speed between wifi devices over ac1750?

[mzaur8]

Hey guys,

I have an Apple Airport Extreme AC1750 router connecting a few devices in a small apt. I have a PC running Windows 10 with a wifi N pcie card that occasionally gets lag spikes while gaming so I figured I'd update the wifi card. I know wifi isn't the best for gaming but my router is 10ft away and I'm on 5ghz with minimal interference so it shouldn't be happening. Anyway, I got a TPlink Archer T8E AC1750 PCIe card. With my old card I can transfer at about 10MB/s to my macbook (connected at a rate of 1gbit over wifi). The new card at first was only getting 7MB/s. I'm not sure why/how but after uninstalling/reinstalling drivers and rebooting several times it seems to be stable now at 27-30MB/s between the two wifi devices. If I connect the desktop to gigabit ethernet and transfer to the macbook, it goes at 70MB/s

is 240mbit good or bad between two AC devices through an AC1750 router? I cannot find the max speed of the Macbook pro but found in system info that it's Broadcom BCM43xx. The link rate is 878mbps in OSX and in Windows for each device.

I don't really care so much about speed but just want to make sure the new wifi card is working properly. The fact that it was transferring at 7MB/s for the first hour is kind of worrying, and I don't particularly know how I got it to go faster. I am thinking of returning this card and getting the more expensive Asus PCE-AC68 (cannot find the PCE-AC66 anywhere) and just want to know what sort of transfer speeds I should be aiming for with this router and two AC devices to know that it's working properly

Thanks!

EDIT: nevermind not stable at all. I reboot and now it's going at 8-10MB/s. Going to try the Asus wifi card. Would appreciate if someone could let me know what transfer speeds I should be getting though

 

[L&LD]

Try changing the channels, changing the location and / or making sure the way too hot running AE is running as cool as possible.

I am not familiar with the TP Link products, but try another brand after trying the suggestions above first and see if it might be your specific environment that is causing these fluctuating throughputs.

With an indicated and strong connection rate of 867Mbps you should be getting close to half that in throughput. So about 50MB/s. This is assuming line of sight (less than 20') and with no obstacles or other environmental interferences.

If trying different channels or router location and antennae positioning (with the AE, simply rotate the router) doesn't help, I would be inclined to either try another router (same model, but new) or try a different brand altogether, like the Asus PCE-AC68 you're considering.

As a last resort, you can try testing a better router or commit to wiring your gaming computer instead.

 

[mzaur8]

Try changing the channels, changing the location and / or making sure the way too hot running AE is running as cool as possible.

I am not familiar with the TP Link products, but try another brand after trying the suggestions above first and see if it might be your specific environment that is causing these fluctuating throughputs.

With an indicated and strong connection rate of 867Mbps you should be getting close to half that in throughput. So about 50MB/s. This is assuming line of sight (less than 20') and with no obstacles or other environmental interferences.

If trying different channels or router location and antennae positioning (with the AE, simply rotate the router) doesn't help, I would be inclined to either try another router (same model, but new) or try a different brand altogether, like the Asus PCE-AC68 you're considering.

As a last resort, you can try testing a better router or commit to wiring your gaming computer instead.

Hey, I tried changing channels. Right now it's 161 but I tried 40 and a few others. Didn't make a difference. With line of sight I am getting 30MB/s but this is from one AC device to the other. I am sure that it would be faster from wifi card to ethernet but unfortunately my macbook has no ethernet port. That's why I was asking what sort of throughput I can expect from one AC device to another on this router.

Is this the correct benchmark? 5ghz updown throughput? http://www.smallnetbuilder.com/tools/charts/router/bar/116-5-ghz-updn-c

 

[L&LD]

Did you reboot both the router and the computers you're testing after changing the channel?

From a client to the router to the router to another client, 30MB/s may be the limit of your equipment. That is 60MB/s or 480Mbps.

I think the link you have shows multiple connections from a single client, not multiple clients.

http://www.smallnetbuilder.com/wire...32478-how-we-test-wireless-products-revison-8

Hey, I tried changing channels. Right now it's 161 but I tried 40 and a few others. Didn't make a difference. With line of sight I am getting 30MB/s but this is from one AC device to the other. I am sure that it would be faster from wifi card to ethernet but unfortunately my macbook has no ethernet port. That's why I was asking what sort of throughput I can expect from one AC device to another on this router.

Is this the correct benchmark? 5ghz updown throughput? http://www.smallnetbuilder.com/tools/charts/router/bar/116-5-ghz-updn-c

 

[mzaur8]

Did you reboot both the router and the computers you're testing after changing the channel?

From a client to the router to the router to another client, 30MB/s may be the limit of your equipment. That is 60MB/s or 480Mbps.

I think the link you have shows multiple connections from a single client, not multiple clients.

http://www.smallnetbuilder.com/wire...32478-how-we-test-wireless-products-revison-8

Yeah I did reboot router and both clients. I took a look at that link and honestly it doesn't do a good job at explaining what each of the benchmarks actually mean.Looking at 5ghz downlink profile, the fastest average is around 300mbps. Is that from a wired source? With my router I can get a consistent speed of 560mbps from a wired source using my macbook on 5ghz wifi so I'm confused about that benchmark. There's no way my Airport Extreme is that much faster than all of the routers tested. Anyway just trying to figure out what would be a good speed for this router, and specifically transfer speed between two AC devices, both of which are connected at ~1gbit

 

[azazel1024]

What source to what destination? Is this between two computers BOTH connected via wifi? If so, 30 odd MB/sec isn't too bad. If this is from Wifi to wired, that is pretty bad. My laptop with just a 867Mbps connection manages up to about 60MB/sec of real through put.

A 1300Mbps 3:3 11ac link at fairly close ranges SHOULD be able to manage closer to 80-100MB/sec of real through put. But only if it is wired to wireless or wireless to wired (NOT though the internet).

70MB/sec is also frankly pretty bad for a wired connection...unless what you are meaning is now you are testing one of the clients are WIRED and the other is still wireless. If that is the case, it isn't too bad. It should still be somewhat faster, but if one of the clients is 867Mbps, then yeah, pretty good.

Wait, scratch all of that, I read closer. No, those numbers are pretty decent. If you have two Wifi clients connected at once, you are going to halve your maximum transfer rate. So, yes, 30MB/sec between the clients with BOTH in wifi is decent, but not great and 70MB/sec with one client on wireless is decent, but not great.

Example, with my laptop with a 2:2 11ac adapter (867Mbps link speed) to my AC1200 and AC1750 routers (same link speed though) I can hit about 62MB/sec with a tail wind. If I go laptop, AC1750 and then wireless bridge, so it is two WiFi hops, I get about 30MB/sec. If I connect wired to my AC1750 router and then do a wireless bridge to my AC1200, I can get about 74MB/sec (more powerful radio, bigger antennas, probably better wireless chipset).

Every time you have to hop wireless, you have to divide the through put. No hops, full through put. 1 hop, halve it. 2 hops, you get 1/3rd. 3 hops, 1/4. Etc. This is why wireless repeaters suck, unless there are zero choices in having to use one.

Also Wifi tends to be about 65-75% net yield for 11n on the latest routers (IE 2.4GHz), but 11ac is still in its relative infancy and you should only expect about 50-60%. Maybe in a couple of years you might expect closer to a 70% net yield with improving chipsets, radios, drivers, etc.

Not that net yield is for a very GOOD connection, with no interference.

 

[mzaur8]

What source to what destination? Is this between two computers BOTH connected via wifi? If so, 30 odd MB/sec isn't too bad. If this is from Wifi to wired, that is pretty bad. My laptop with just a 867Mbps connection manages up to about 60MB/sec of real through put.

A 1300Mbps 3:3 11ac link at fairly close ranges SHOULD be able to manage closer to 80-100MB/sec of real through put. But only if it is wired to wireless or wireless to wired (NOT though the internet).

70MB/sec is also frankly pretty bad for a wired connection...unless what you are meaning is now you are testing one of the clients are WIRED and the other is still wireless. If that is the case, it isn't too bad. It should still be somewhat faster, but if one of the clients is 867Mbps, then yeah, pretty good.

Wait, scratch all of that, I read closer. No, those numbers are pretty decent. If you have two Wifi clients connected at once, you are going to halve your maximum transfer rate. So, yes, 30MB/sec between the clients with BOTH in wifi is decent, but not great and 70MB/sec with one client on wireless is decent, but not great.

Example, with my laptop with a 2:2 11ac adapter (867Mbps link speed) to my AC1200 and AC1750 routers (same link speed though) I can hit about 62MB/sec with a tail wind. If I go laptop, AC1750 and then wireless bridge, so it is two WiFi hops, I get about 30MB/sec. If I connect wired to my AC1750 router and then do a wireless bridge to my AC1200, I can get about 74MB/sec (more powerful radio, bigger antennas, probably better wireless chipset).

Every time you have to hop wireless, you have to divide the through put. No hops, full through put. 1 hop, halve it. 2 hops, you get 1/3rd. 3 hops, 1/4. Etc. This is why wireless repeaters suck, unless there are zero choices in having to use one.

Also Wifi tends to be about 65-75% net yield for 11n on the latest routers (IE 2.4GHz), but 11ac is still in its relative infancy and you should only expect about 50-60%. Maybe in a couple of years you might expect closer to a 70% net yield with improving chipsets, radios, drivers, etc.

Not that net yield is for a very GOOD connection, with no interference.

Thanks for your detailed response! Glad to hear that the numbers are decent. I have no idea what sort of adapter is inside my 2014 13'' Macbook Pro. It says Broadcom BCM43xx in system info but online cannot find info. I'm guessing it's 2:2 ac adapter like yours since I"m getting the same speed. Thanks for explaining about hops and halving expected throughput. Very informative.

BTW, sort of unrelated question but I couldn't find much about this online. Do you know if latency is better with 11ac than with 11n?

 

[L&LD]

Just had to say... I said the same thing in my post above.

30MB/s between two wireless clients is about max.

 

[stevech]

If not already said, if both clients are on the same SSID/RF channel, or nearly so, then they time-share that access. It's not like switched ethernet.

 

[sfx2000]

Thanks for your detailed response! Glad to hear that the numbers are decent. I have no idea what sort of adapter is inside my 2014 13'' Macbook Pro. It says Broadcom BCM43xx in system info but online cannot find info. I'm guessing it's 2:2 ac adapter like yours since I"m getting the same speed. Thanks for explaining about hops and halving expected throughput. Very informative.

BTW, sort of unrelated question but I couldn't find much about this online. Do you know if latency is better with 11ac than with 11n?

Your MacBook Pro is a Broadcom BCM4360 configured as a 3*3:3 config - MacBook Airs are the same chip, but as a 2*2:2 config.

11ac vs 11n latency - not much different - depends more on the AP's implementation when using the same client.

11ac based routers do tend to have better performance, even with 11n clients mostly due to better chipsets, better drivers, and improvements in the radios.

 

[azazel1024]

Technically 11ac would be lower latency because it is transmitting it's packets faster. However, on a GOOD link, the difference is going to be measured in microseconds. Too many variable to say on a bad link, but since generally 11ac has better performance at similar ranges, sometimes significantly better performance at medium and long range, then yes, it would likely be a fair amount better latency for 11ac because you have fewer dropped packets, or those packets that are dropped get retransmitted faster/sooner.

 

[mzaur8]

Thanks guys. Really appreciate your replies!
I got the Asus PCE-AC68 wifi adapter and the transfer speed is about the same, 30MB/s. I'll take it
Strangely i'm having the same sort of issues where the transfer speed is inconsistent. I rebooted all devices and router several times and now it seems to be at 30mb/s but before was at 10mb/s. Seems to be an issue either with Win 10 or the Apple router (bootcamp win10 on macbook didn't help). I'm too busy to troubleshoot. Oh well, not a big deal. -- Actually might be related to Win 10 drivers not being available for this chipset (both TP link and Asus drivers are from Win 8)

 

[sfx2000]

Technically 11ac would be lower latency because it is transmitting it's packets faster. However, on a GOOD link, the difference is going to be measured in microseconds.

I was more speaking towards an 802.11n client working with an 11ac vs. 11n router, so for the most part it would be the same - 11ac AP's can, and often do, a better job of 11n...

Better Software, Better Hardware, Better RF == Better Pizza!

 

[mzaur8]

Hey guys,

These sort of lag spikes pinging the router over wifi aren't normal right? I am having this happen with both the Asus and the TP link wifi cards. http://i.imgur.com/7zJIk7P.png Same happens with pinging google.com. I have inSSIDer open and the signal strength is fine. Just the ping is jumping around. I also ran ping test on my MBP and it's fine, averaging 2.5ms so clearly not an issue with router. Maybe related to Broadcom 4360 win8 drivers not being compatible with win10? Really strange. A reboot seems to have temporarily fixed the issue, now getting 2ms average. Must be something related to windows?

 

[azazel1024]

It could be something related to windows. It could be coincidental to the reboot. A lot of times lag spikes (not consistent latency) is because of periodic interference. Or it could be because of other clients sharing the channel.

Packets aren't necessarily perfectly interleaved. So if you have two clients talking it isn't necessarily going to be client A gets to send a pack, client B gets to send a packet, back to A and so on. It can easily be that Client A sends 20 packets, then client B sends 10, then Client A 12, Client B 18 and so on. If one of the clients drops a packet or two, retransmits will greatly increase the time involved too.

This is partly why latency at range on Wifi can be so high. Wifi doesn't operate where each packet is verified and if needed retransmitted. It is similar to TCP/IP (but half duplex) where several packets are sent before an acknowledgement is received. In some cases this can lead to all packets having to be resent, or just one, or several. Just depends on the nature of how they were dropped.

 

[mzaur8]

It could be something related to windows. It could be coincidental to the reboot. A lot of times lag spikes (not consistent latency) is because of periodic interference. Or it could be because of other clients sharing the channel.

Packets aren't necessarily perfectly interleaved. So if you have two clients talking it isn't necessarily going to be client A gets to send a pack, client B gets to send a packet, back to A and so on. It can easily be that Client A sends 20 packets, then client B sends 10, then Client A 12, Client B 18 and so on. If one of the clients drops a packet or two, retransmits will greatly increase the time involved too.

This is partly why latency at range on Wifi can be so high. Wifi doesn't operate where each packet is verified and if needed retransmitted. It is similar to TCP/IP (but half duplex) where several packets are sent before an acknowledgement is received. In some cases this can lead to all packets having to be resent, or just one, or several. Just depends on the nature of how they were dropped.

Thanks. So this could be the case with only 2 devices (both 11ac) connected to an 11ac router less than 10ft away and only one device seeing the latency spikes?

 

[azazel1024]

It is certainly possible. It is also certainly possible it is an issue with the netstack and/or wireless drivers on the one device as well.