But why forking immortalwrt? Why not openwrt? @VIKINGYFY
Because heās in China, and ImmortalWrt tends to support more China-market devices and packages out of the box, so it made more sense for his use case than OpenWrt.
1 Like
Hello,
Iād appreciate any information if anyone knows the answer.
Are there any OpenWrt-compatible APs that support 802.11be (Wi-Fi 7) and also have hardware offloading similar to NSS? From what Iāve researched so far, it doesnāt seem like there are any.
Iām still using an 802.11ax AP (5950AX12), and Iām very satisfied with it since I can get around 2 Gbps in real-world use. However, more of my devices now support 802.11be, so Iām starting to consider upgrading my AP as well.
That said, from my limited testing of 802.11be at friendsā houses, even with MLO enabled, lower- to mid-range APs didnāt seem dramatically faster than my AX12 setup. The best result I saw was with an EAP773, using MLO with 6 GHz 320 MHz + 5 GHz 160 MHz at close range, reaching around 2.4ā3 Gbps. Considering the price of those APs, NSS-enabled APs still seem very compelling.
Iād prefer to stay with OpenWrt because of its flexibility, feature set, and the responsiveness of the web UI. If anyone has already tried an 802.11be OpenWrt-compatible AP, Iād really appreciate hearing about your experience.
Thanks!
First look here but you have to read very carefully that really long thread and then decide for yourself.
Additionally they are hardly available on the market currently. Personally I've bought four, eight months ago when they were just 30$ a piece.
1 Like
I have WRX36 running AgustinLorenzo's build for over two years, a few weeks ago I tested using W1700k as main router, and was hoping it will be able to replace WRX36+BE805 AP.
It's mostly working fine as router but I found that WiFi BE at 6GHz is not as fast as BE805 on factory firmware, so I went back to WRX36 as main with BE805 as AP, it will take some time with enough feedback from users to get to decent output, but your experience might be different. I didn't test MLO because my surrounding neighbors are using all the 2.4/5 frequency including the DFS channels, 2.4GHz is so crowded that latency is up to 100ms, a lot of neighbors simply set transmission power max hoping to get better throughput, without realizing it's detrimental for everyone.
1 Like
Thank you very much for the highly useful information.
The thread you pointed me to was quite long, but it contained a lot of interesting information. When I looked into the BPi-R4 around 1ā2 years ago, I remember the situation being that the mt79 NPU/PPE offloading was not working properly yet, so Iām glad to learn that hardware offloading has gradually started to become functional recently.
It still seems like playing around with the W1700K is somewhat like early access, but Iām a bit jealous that you were able to join in for only $30 
I also considered jumping into this game, but with things like HW Rev2 starting to appear, it seems wiser to wait a little longer for more information.
In any case, itās very useful to have found a candidate AP for my next upgrade, so Iāll keep an eye on it.
Thank you again!
1 Like
What do you mean by "HW Rev2"?
It's the hardware version of W1700k, it has 1.0 1.1 under quantum fiber, and 2.1 which is under Lumen, it's discussed in the long thread of W1700k.
1 Like
Sorry, I meant Rev 2.1. Please see around this post.
worked, very fast, thanks!
So is Qosmio out of the picture forever? Because I tried to build a new fw and I saw the last time he did any work was like 3 months ago?
How can we build our own now? I don't like using prebuilts, I also saw VIKINGYFY and Augustin builds? Which are are regularly updated?
Go with Augustin builds.
2 Likes
Nope, he isn't out, quite the contrary. But he is probably really busy, seems to be working on a lot of other projects. He is active on his github account and you can just check his other activities there.
With regard to the NSS development, here is his latest activity just some hours ago.
And yes, not committing anything for over two months gave me a chance to test the NSS build for stability and as of today my last build (with a ton of services on it) based on his repo is running for over 55 days without a reboot.
Really impressive work. Well Done, @qosmio!
Let's not forget anyone else who contributed for NSS to be working right now.
Cheers everyone!
6 Likes
Because these are Chinese developers, IMM is more user-friendly for Chinese users.
I've tried to dry-run your patch to see what I will get. Using latest qosmio rebase.
File to patch: build_dir/target-aarch64_cortex-a53_musl/linux-qualcommax_ipq807x/linux-6.12.92/drivers/net/wireless/ath/ath11k/core.c
checking file build_dir/target-aarch64_cortex-a53_musl/linux-qualcommax_ipq807x/linux-6.12.92/drivers/net/wireless/ath/ath11k/core.c
Hunk #1 succeeded at 18 with fuzz 2 (offset -5 lines).
Hunk #2 succeeded at 1540 with fuzz 1 (offset 149 lines).
Hunk #3 FAILED at 2096.
Hunk #4 FAILED at 2116.
Hunk #5 FAILED at 2137.
Hunk #6 FAILED at 2156.
Hunk #7 FAILED at 2185.
Hunk #8 FAILED at 2200.
6 out of 8 hunks FAILED
I don't understand the reason for the failures. Outdated patch or another patch (by @qosmio) has been applied to /ath/ath11k/core.c.
is it the same patch? https://github.com/qosmio/openwrt-ipq/issues/104
Nope. My build (main-nss) with qosmio's yesterday commits compiled OK. I just wanted to use @BrainSlayer BDF regdomain patch in order to test some wifi settings that I cannot change otherwise.
With that patch the following should be possible to be changed from OpenWrt.
Linux/OpenWrt CAN override:
* VHT160 flags
* regdomain fields
* bandwidth flags
* target power tables
* antenna gain tables
* board capability flags
Because the patch modifies the BDF in RAM before Ath11k firmware consumes it.
1 Like
I finally bought a W1700K as well. (Including shipping, it ended up costing about $100 per unit. Oh no!)
However, after spending some time testing it, I have to say that it still feels very much like early access. For my main AP, it looks like I'll be sticking with qosmio's NSS build for at least another year. (MLO doesn't always behave as intended, some hardware offloading features are still incomplete, roaming isn't working reliably, etc.)
This experience has really reminded me how polished and refined that build has become.
1 Like
the patch was made a long time ago. of course its outdated. but i dont think its a hard thing to adjust it a little to make it fit again.
1 Like
@qosmio I'm attempting to build 24.10-nss commit 2a9a392 for Dynalink DL-WRX36 and am getting these errors. I did give 25.12-nss commit 72c0e5b a shot as well, it's fails as well I did not grab the output from that build though.
Any ideas what's going on?
Edit: I was using the config file supplied and only edit was to enable the Dynalink DL-WRX36.
pplying /home/fred/openwrt-ipq/target/linux/qualcommax/patches-6.6/0600-4-qca-nss-ecm-support-net-bonding-over-LAG-interface.patch using plaintext:
patching file drivers/net/bonding/bond_3ad.c
Hunk #1 succeeded at 115 (offset -4 lines).
Hunk #2 succeeded at 1196 (offset -1 lines).
Hunk #3 FAILED at 2124.
Hunk #4 succeeded at 2134 (offset 3 lines).
Hunk #5 succeeded at 2907 (offset 7 lines).
1 out of 5 hunks FAILED -- saving rejects to file drivers/net/bonding/bond_3ad.c.rej
patching file drivers/net/bonding/bond_main.c
Hunk #1 succeeded at 290 (offset 2 lines).
Hunk #2 succeeded at 1257 (offset 10 lines).
Hunk #3 succeeded at 1919 (offset 10 lines).
Hunk #4 succeeded at 2387 (offset 16 lines).
Hunk #5 succeeded at 2461 (offset 16 lines).
Hunk #6 succeeded at 2491 (offset 16 lines).
Hunk #7 succeeded at 2514 (offset 16 lines).
Hunk #8 succeeded at 2851 (offset 16 lines).
Hunk #9 succeeded at 2892 (offset 16 lines).
Hunk #10 succeeded at 2952 (offset 22 lines).
Hunk #11 succeeded at 4211 (offset 24 lines).
Hunk #12 succeeded at 5597 with fuzz 1 (offset 69 lines).
Hunk #13 FAILED at 5789.
1 out of 13 hunks FAILED -- saving rejects to file drivers/net/bonding/bond_main.c.rej
patching file include/net/bond_3ad.h
Hunk #1 succeeded at 309 (offset 1 line).
patching file include/net/bonding.h
Hunk #2 succeeded at 682 (offset 4 lines).
Hunk #3 succeeded at 716 (offset 5 lines).
Hunk #4 succeeded at 827 (offset 5 lines).
Patch failed! Please fix /home/fred/openwrt-ipq/target/linux/qualcommax/patches-6.6/0600-4-qca-nss-ecm-support-net-bonding-over-LAG-interface.patch!
make[4]: *** [Makefile:33: /home/fred/openwrt-ipq/build_dir/target-aarch64_cortex-a53_musl/linux-qualcommax_ipq807x/linux-6.6.141/.prepared_dba0667ad2e78b144239817826bc9338] Error 1
make[4]: Leaving directory '/home/fred/openwrt-ipq/target/linux/qualcommax'
make[3]: *** [Makefile:12: compile] Error 2
make[3]: Leaving directory '/home/fred/openwrt-ipq/target/linux'
time: target/linux/compile#6.03#2.79#6.51
ERROR: target/linux failed to build.
make[2]: *** [target/Makefile:32: target/linux/compile] Error 1
make[2]: Leaving directory '/home/fred/openwrt-ipq'
make[1]: *** [target/Makefile:25: /home/fred/openwrt-ipq/staging_dir/target-aarch64_cortex-a53_musl/stamp/.target_compile] Error 2
make[1]: Leaving directory '/home/fred/openwrt-ipq'
make: *** [/home/fred/openwrt-ipq/include/toplevel.mk:233: world] Error 2