alirz wrote:

Chadster766 wrote:

Honestly I kind of wish OpenWRT continued development with the completely opensource wlan7 driver. I know this isn't helpful but I couldn't help but mention it and wonder how far it could have been customized by now for the WRT1900AC V1. It's a pet peeve I have

I agree. I used your AA build for a LONG time. It gave me the fastest wireless speeds. Even the latest CC is still about 30% slower in throughput.

You can still use AA. It's your router.

gufus wrote:

alirz wrote:

I agree. I used your AA build for a LONG time. It gave me the fastest wireless speeds. Even the latest CC is still about 30% slower in throughput.

You can still use AA. It's your router.

I can't remember if I tried the "disassoc_low_ack=0" fix for Apple devices on AA. AFAIK it's supported. I also don't know if AA has the same critical issue #20 CC has.

Also, my town house is wired with cat3 cable so I don't use wifi much but like the AC speeds when I do.

OK, I can now trigger a wan/lan freeze for the first time.

Have laptop connected to 5GHz (N).

Have android connected to 5GHz (N).

Start an on-going ping on the laptop, everything OK.

Switch android to 2.4GHz - pings time out, no router login access, no Internet access.

Switch android back to 5GHz and everything resumes normally.

Trunk r45950, uptime 9d 14h 2m 13s

DavidMcWRT wrote:

OK, I can now trigger a wan/lan freeze for the first time.

Have laptop connected to 5GHz (N).

Have android connected to 5GHz (N).

Start an on-going ping on the laptop, everything OK.

Switch android to 2.4GHz - pings time out, no router login access, no Internet access.

Switch android back to 5GHz and everything resumes normally.

Trunk r45950, uptime 9d 14h 2m 13s

Good troubleshooting but this is related but in addition to issue #20

https://github.com/kaloz/mwlwifi/issues/20

gufus wrote:

Also, my town house is wired with cat3 cable so I don't use wifi much but like the AC speeds when I do.

Was this a typo?

I'm sure you'd see higher speeds over wifi than over any cat3 cabling.

Did you mean Cat5e, Cat6, Cat6a or Cat7?

Cat3 is what was used for 10Base-T (10Mbit/s Ethernet) and sometimes, 100Base-T4 (Special case of 100Mbit/s Ethernet).

// Stefan

Chadster766 wrote:

Honestly I kind of wish OpenWRT continued development with the completely opensource wlan7 driver. I know this isn't helpful but I couldn't help but mention it and wonder how far it could have been customized by now for the WRT1900AC V1. It's a pet peeve I have

I agree. I used your AA build for a LONG time. It gave me the fastest wireless speeds. Even the latest CC is still about 30% slower in throughput.

cptn_brittish wrote:

I've been rewriting my fan control program to be threaded and it requires libpthreads.
I can compile the program but it will not run on the router as it cannot find libpthread..so.0.
Libpthreads wasn't initially installed but once I installed it through 'opkg install libpthread' I still cannot use the library and it does not seem to have been installed on the router. Does anyone know how to fix this?
I'm currently on RC2.
Thanks

Even though this might not be helpful - why use "proper threading" and not just fork()?

EDIT: Depending on libc, it might still require libpthread, so it may not help.

Also, was there supposed to be two dots in a row there?

libpthread..so.0

// Stefan

alirz wrote:

Chadster766 wrote:

Honestly I kind of wish OpenWRT continued development with the completely opensource wlan7 driver. I know this isn't helpful but I couldn't help but mention it and wonder how far it could have been customized by now for the WRT1900AC V1. It's a pet peeve I have

I agree. I used your AA build for a LONG time. It gave me the fastest wireless speeds. Even the latest CC is still about 30% slower in throughput.

What are your current throughput speeds and what were they before?  Also, what channel are you using for both spectrums?

stesmi wrote:

gufus wrote:

Also, my town house is wired with cat3 cable so I don't use wifi much but like the AC speeds when I do.

Was this a typo?

I'm sure you'd see higher speeds over wifi than over any cat3 cabling.

Did you mean Cat5e, Cat6, Cat6a or Cat7?

Cat3 is what was used for 10Base-T (10Mbit/s Ethernet) and sometimes, 100Base-T4 (Special case of 100Mbit/s Ethernet).

// Stefan

I should have said cat5e, I get 1.0 “gigabit”
http://www.gypsy-designs.com/1.jpg

stesmi wrote:

cptn_brittish wrote:

I've been rewriting my fan control program to be threaded and it requires libpthreads.
I can compile the program but it will not run on the router as it cannot find libpthread..so.0.
Libpthreads wasn't initially installed but once I installed it through 'opkg install libpthread' I still cannot use the library and it does not seem to have been installed on the router. Does anyone know how to fix this?
I'm currently on RC2.
Thanks

Even though this might not be helpful - why use "proper threading" and not just fork()?

EDIT: Depending on libc, it might still require libpthread, so it may not help.

Also, was there supposed to be two dots in a row there?

libpthread..so.0

// Stefan

It was meant to have one dot not two.
I'm using proper threading mostly as a learning experience as it is my first attempt at threading and because I am sharing a variable between both threads so I am trying to avoid contention on the variable. I'm not sure if that is possible with fork().

I worked around the problem by updating the router to the snapshot builds which allowed libpthread to install.

JW0914 wrote:

alirz wrote:

I agree. I used your AA build for a LONG time. It gave me the fastest wireless speeds. Even the latest CC is still about 30% slower in throughput.

What are your current throughput speeds and what were they before?  Also, what channel are you using for both spectrums?

The only AC wifi device is my nexus 5 phone. I had spend days on testing wifi speeds using all sorts of methods. Once I found the fast file transfer method for me, which turned out to be WebDAV . I used that for benchmarking file read speeds from a networked SSD drive.

My phone being a limited AC Chipset can only connect to the router at 433mbps. With AA I was able to get 22MB (megabytes) per second pretty much all the time. With CC I get 14-15MB/sec.

alirz wrote:

JW0914 wrote:

What are your current throughput speeds and what were they before?  Also, what channel are you using for both spectrums?

The only AC wifi device is my nexus 5 phone. I had spend days on testing wifi speeds using all sorts of methods. Once I found the fast file transfer method for me, which turned out to be WebDAV . I used that for benchmarking file read speeds from a networked SSD drive.

My phone being a limited AC Chipset can only connect to the router at 433mbps. With AA I was able to get 22MB (megabytes) per second pretty much all the time. With CC I get 14-15MB/sec.

I've never heard of a limited AC chipset, and I must not be using the right search parameters on Google; could you elaborate on what that means (I think it's referring to the spatial streams, with the NX5 only being 1x1)?

With a 1x1 spatial stream, the NX5's max speed at the 20mHz width is 75mb/s (9.16MB/s) and 150mb/s (18.31MB/s) at the 40mHz width.  If your Nexus 5 has a 1x1 spatial stream, it's physically impossible for it to garnish a speed above 18.31MB/s.  The only way it could possibly do so is if it utilizes a similar technology to PC wifi cards that utilize both wifi and bluetooth together to garnish faster speeds. 22MB/s = 176mb/s and 15MB/s = 120mb/s

You should be able to hit 150mb/s however and would recommend trying channels in the 150 range, along with making sure the band width is set to 40mHz.  When it comes to the 5gHz spectrum, you'll almost always experience faster speeds at higher channels.  I'd also recommend boosting your dB level to 20 - 25 to see if that helps and setting the distance to 50M.

This offers a more in depth response: XDA

cptn_brittish wrote:

stesmi wrote:

Even though this might not be helpful - why use "proper threading" and not just fork()?

EDIT: Depending on libc, it might still require libpthread, so it may not help.

Also, was there supposed to be two dots in a row there?

libpthread..so.0

// Stefan

It was meant to have one dot not two.
I'm using proper threading mostly as a learning experience as it is my first attempt at threading and because I am sharing a variable between both threads so I am trying to avoid contention on the variable. I'm not sure if that is possible with fork().

I worked around the problem by updating the router to the snapshot builds which allowed libpthread to install.

Great. I was just making sure you weren't getting that from ldd which meant something was wrong during the build.

stesmi@Defiant:~$ ldd /bin/bash
        libncurses.so.5 => /usr/lib/libncurses.so.5 (0xb6f14000)
        libdl.so.0 => /lib/libdl.so.0 (0xb6f00000)
        libgcc_s.so.1 => /lib/libgcc_s.so.1 (0xb6ee4000)
        libc.so.0 => /lib/libc.so.0 (0xb6e88000)
        ld-uClibc.so.0 => /lib/ld-uClibc.so.0 (0xb6f50000)

For instance.

You can allocate a memory area as shared if you wish and use some kind of locking (or make it atomic).

mmap() with MAP_SHARED is one solution. For instance make a struct of a set size and you're golden.

mmap for sizeof(struct your_struct) and assign the pointer you get from mmap() to the pointer of your struct and you can use that between threads (or processes really since that's what fork() gives you in some implementations).

Now we've gotten Off Topic though (sorry all!).

I'm glad it worked out for you in the end!

// Stefan

Hi guys.
I'm in the stage of purchasing the WRT1900AC and looking for the best reseller.
I just have one question:

Can somebody please tell me if the v1 and the v2 have the same amount of RAM?
In the Hardware table, the v2 info says "512 RAM" but if you enter the details page it says 256 RAM for both versions.

Thank's. :-)

Just for comparison, maximum results of my synthetic wireless bandwidth tests for OpenWRT CC: 540 Mbps. Connection: WLAN-LAN, 5 GHz, 802.11ac 80 MHz, 36 ch., 866.5 Mbps.

WLAN-WAN (internet) will be a bit less, because of NAT it becomes CPU-bound: 500 Mbps (CPU usage 100%).

JW0914, what is theoretical limit for out spatial stream system (4x4:3, 802.11ac, 80 MHz)?

JW0914 wrote:

alirz wrote:

The only AC wifi device is my nexus 5 phone. I had spend days on testing wifi speeds using all sorts of methods. Once I found the fast file transfer method for me, which turned out to be WebDAV . I used that for benchmarking file read speeds from a networked SSD drive.

My phone being a limited AC Chipset can only connect to the router at 433mbps. With AA I was able to get 22MB (megabytes) per second pretty much all the time. With CC I get 14-15MB/sec.

I've never heard of a limited AC chipset, and I must not be using the right search parameters on Google; could you elaborate on what that means (I think it's referring to the spatial streams, with the NX5 only being 1x1)?

With a 1x1 spatial stream, the NX5's max speed at the 20mHz width is 75mb/s (9.16MB/s) and 150mb/s (18.31MB/s) at the 40mHz width.  If your Nexus 5 has a 1x1 spatial stream, it's physically impossible for it to garnish a speed above 18.31MB/s.  The only way it could possibly do so is if it utilizes a similar technology to PC wifi cards that utilize both wifi and bluetooth together to garnish faster speeds. 22MB/s = 176mb/s and 15MB/s = 120mb/s

You should be able to hit 150mb/s however and would recommend trying channels in the 150 range, along with making sure the band width is set to 40mHz.  When it comes to the 5gHz spectrum, you'll almost always experience faster speeds at higher channels.  I'd also recommend boosting your dB level to 20 - 25 to see if that helps and setting the distance to 50M.

This offers a more in depth response: XDA

Correct. i had used the term "limited" to avoid going into details about spatial stream explanation. Yes it is 1x1. Well i did used to be able to get sustained speeds of over 20MB for large size file transfer.. Perhaps the app used was generous in telling me the speed!.
I am on channel 149 80MHZ. Its the cleanest spectrum around me.