-
-
Notifications
You must be signed in to change notification settings - Fork 142
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Test EDUP PCIe Intel AX200 WiFi 6 Card #22
Comments
|
Rebuilding the kernel, here's where the driver is according to menuconfig:
|
From
So it seems to init correctly, but then when the driver loads:
Looks like I need to install some firmware... |
Looking at Intel's page https://wireless.wiki.kernel.org/en/users/Drivers/iwlwifi — it looks like it needs MVM (which I have installed, as well as the firmware linked in the table for
|
Yikes, tons of error output, going to have to blacklist the module and load it manually to see what's going on...
|
Then concurrently,
|
First see:
And just noting:
lol, there's an open bug from a few years back for this thermal issue: https://bugzilla.kernel.org/show_bug.cgi?id=201761 But moving on, I also see:
|
And every time the driver resets itself, it increments (now it's up to 464... waiting a few seconds, now it's 482...) |
Related? QubesOS/qubes-issues#5615 |
Maybe this patch will help—that's where the error is coming from, anyways... https://lore.kernel.org/linux-wireless/3cab5072-17a2-4d9a-2077-93788971c6c4@invisiblethingslab.com/T/#u diff --git a/drivers/net/wireless/intel/iwlwifi/mvm/fw.c
b/drivers/net/wireless/intel/iwlwifi/mvm/fw.c
index 98263cd37944..943831c9b685 100644
--- a/drivers/net/wireless/intel/iwlwifi/mvm/fw.c
+++ b/drivers/net/wireless/intel/iwlwifi/mvm/fw.c
@@ -1111,15 +1111,6 @@ int iwl_mvm_up(struct iwl_mvm *mvm)
goto error;
/* Init RSS configuration */
- if (mvm->trans->trans_cfg->device_family >= IWL_DEVICE_FAMILY_22000) {
- ret = iwl_configure_rxq(mvm);
- if (ret) {
- IWL_ERR(mvm, "Failed to configure RX queues: %d\n",
- ret);
- goto error;
- }
- }
-
if (iwl_mvm_has_new_rx_api(mvm)) {
ret = iwl_send_rss_cfg_cmd(mvm);
if (ret) { |
In my cross-compile environment:
Note: that patch causes the following warning to emit during compilation:
Recompiling the kernel the 18th time this week :D |
Well heck. What do you know?
And:
The Pi seems much happier. Now, to connect to my WiFi... |
Hmm... this isn't super encouraging though:
|
But... after setting up config in
So... 🤷♂️ Let's get some performance testing! |
On the Pi:
On my Mac:
As I used to say in college: "Noice!" Compared the the maximum 80 Mbits/sec I can get via the CM4's internal WiFi, even with an external antenna... that's rather speedy! Edit: A few tests:
(For comparison, wired interface gives ~940 Mbits/sec.) Edit 2: OOH... if I reverse the roles (Pi is client, Mac is server), I can pump through 928 Mbits/sec over WiFi! Nice—that's faster than I can get through my old Mac mini to my new MacBook pro...
|
Dangit, now that I finally had some success I should probably make a video about it. But who has the time!? I think the other thing I want to test and see how it works is using the PCIe splitter/switch with this WiFi card and my Samsung 970 EVO Plus, and see if I can max out the WiFi bandwidth and write a full ~650 Mbps (80 MB/sec) so we could have a worthy little WiFi-only NAS that still competes with a wired NAS (for general usage... not gonna be pumping ProRes RAW through one of these little dudes!). |
So... I was thinking, "hey, it would be a great idea to test this WiFi card on a real WiFi 6 network, using a router that works with a 10G network interface, allowing me to test the full throughput against my Mac over 10G wired network..." Then I was glancing around at the (very few, so far) good WiFi 6 ax routers with 10G network ports, and they all cost $450+ and are out of stock pretty much everywhere. I was looking mostly at the ASUS RT-AX89X, mostly because I've stuck with ASUS wireless gear for like 5 years now and never had an issue... but it's out of stock everywhere. So I'll stick with testing over my ASUS AC2900-based 802.11ac network for now. I mean, getting 1 gig over my current network is nothing shabby, as that's also probably hitting the max throughput I've seen through two of my Macs (a 2016 MBP and a 2019 MBP) over the same network... Heck, maybe my Mac is the slower of the two devices in terms of raw potential throughput! All that to say, I think the 'year of WiFi 6' in the Geerling household is not in 2020. Unless maybe I can get ASUS to sponsor a video (a video for a router, eh?). |
How much would it cost to just have a direct ad-hoc connection between two Wi-Fi 6 clients, instead of buying a router? |
@PixlRainbow - I've been looking around unsuccessfully—does anyone make a WiFi 6 / ax adapter for ThunderBolt 3, USB-C/3.2, or any other connection besides internal/PCIe? I don't have any other devices currently that support PCI express :( |
Your Mac is a laptop? Or a desktop? I'm guessing it's an older model that doesn't have WiFi 6 built in. Unfortunately, it appears the speed of WiFi 6 doesn't really allow regular USB to be used, and there aren't direct thunderbolt WiFi adapters either, not even for older WiFi standards. I presume the manufacturers are assuming that the general thunderbolt user would already have wifi integrated in their device. I did find thunderbolt to PCIE adapters but they are still quite pricy. |
Both my Mac and Dell are laptops, and I just realized the Dell has a replaceable WiFi module—so I just ordered a AX200NGW Wireless Card in M.2/NGFF, and it's coming tomorrow. Getting ready to do a little surgery on my XPS 13 again (I upgraded the SSD a couple years ago), and I'll let you know how it goes. It looks like iperf is also available on Windows, so I'll have to see what happens! (Totally unfamiliar with Windows' networking stack since the last time I did anything networking-related was probably around the time Windows NT was the new hotness). |
The card works great! Working on a short video now just to show the process of upgrading the Dell card. |
Upgraded the Dell, made a video: How to Upgrade a Dell XPS 13 to WiFi 6 (802.11ax). |
Heh. This is probably the most expensive of all the tests so far. I figured...
So I ran and grabbed an ASUS RT-AX86U AX5700 WiFi 6 router, which is not the top of the heap, and is only dual band... but is better than my older ac router, and has a 2.5 Gbps LAN uplink port (which can be swapped for the WAN port—though here in St. Louis it'll probably be 50 years before we end up getting gigabit ethernet to the home!). I installed it, got it configured, and now I'm planning on doing some more testing from 10G wired to AX wireless on both my Dell laptop and the CM4. But that has to wait a bit as I'm trying to finish up my NAS explorations with the SATA card first! |
This is exceedingly interesting. So - similar to Danish Guy's project, I've also been thinking about how the Pi CM4 and recent kernel patches now make a Pi-based Laptop feasible. This suggests that the best way to implement Wi-Fi on such a device would actually not be through the CM4's SDIO chipset, but through a standard m.2 card. I guess the ideal approach would be to design the laptop to allow an antenna cable to be routed to either a PCIe card or a Wireless CM4; and offer the former through one of those fancy m.2 slots with three screws that could take anything from a wifi card to an NVMe drive? |
@mo-g yes, I think that would be ideal. Basically give the user some flexibility. |
@geerlingguy We need to stop running into each other like this. People will talk. Yeah, I'm actually having the same problem as you had here but on a Razer Blade Stealth that I'm using while waiting for my System76 laptop be be delivered. I really don't want to have to recompile the Kernel as you had done here. Surely that shouldn't be the solution given the fact that I'm ...
Will update once I've got this working. |
Hey, just stumbled across this after watching your latest video! Regarding the testing of raw 802.11ax performance with
If you are using Asus network gear, you might be able to make use of some other community-developed tools running directly on the router (adblocking, improved firewalls, ...). |
Interesting... after upgrading my router, I retested everything today, and I can get 875-900 Mbps both ways, regardless of which host is the iperf3 server now. Further testing is warranted—I now have a 10 GbE interface hooked up to my Mac, a 2.5 Gbps connection to my WiFi router, and a 10 GbE switch (though I'm not sure if it can drop to 2.5G on the router connection, so that'll have to be something I test later). I'm guessing I could push more bandwidth through this card with my new router :-O |
Now that |
Testing with the Merlin firmware:
Format a USB flash drive to set up Entware:
(After formatting, the router reboots.) Now install Entware so you can install
Thanks again to @st0nedB for the suggestion! First quick run from my Mac over ac:
|
Hmm... the version installed via
It looks like that's a fairly outdated version though, and doesn't support On my Mac:
|
Regular old
(Just noting these results and the ones above are from my Mac in a non-optimal location; can do better if I'm closer to the router. This is just me debugging the process.) |
On the Pi:
Monitoring with So the CPU isn't causing any bottleneck—it seems like this card can pump quite a few wireless bits through, beating even my MacBook Pro from a similar range! (The MBP gets ~900 Mbps in the same spot as the Pi). I'm guessing I could tweak some things, move some things around, and get more, but 1.2+ Gbps on WiFi on the Pi is pretty darn toasty! |
And a blog post: WiFi 6 gets 1.34 Gbps on the Raspberry Pi CM4. |
Hi! I just stumbled over your excellent youtube channel. So here are my question...
If so then one could build a wifi system that automatically passes a connected user around between different access points. Cisco sells these kind of systems so I thought that it would be fun to build an open source version. I was thinking that you could build a rest like interface with fastapi to run on the rasberry and the control all ap:s from a webserver. |
@christofer-f - It seems the AX200 might not be the best chip for the purpose (though it probably can be made to work). It would be cool to be able to do an open source setup like that, though! |
All right, after going through another round of testing for the Intel AX200 Desktop Kit adapted by the MZHOU WiFi/BT adapter in #38, I'm going to re-run the benchmarks today, this time both with the wired network enabled and disabled, and see what kind of results I get: First testing with
Then with
So... this card is not breaking the gigabit barrier (except for RX traffic from the router to the Pi), and I think it would be safe to say I can get better speeds than the same card in my Dell XPS 13 (likely due to the larger external antennae), but even my i9 MacBook Pro can get about the same speeds over 802.11ac (again, likely due to it's well-designed antennae). Oopsie! I'll have to work on a correction video. Don't want to mislead people! See #38 for more speculation and links to reasons why this happened. |
Hmm...
It's using 80 MHz, why not 160 MHz? Source: Dmitri Arekhta's comment indicated I might not be seeing the full potential of the AX200... |
Closing issues where testing is at least mostly complete, to keep the issue queue tidy. I'm going to move the rest of the testing of the AX200 with 80 vs 160 MHz over to #56. |
The URL for amazon is not working. |
@lealog - That's weird. Even if I go into my order history on Amazon, and click through the order link, it gives me a Page Not Found... I guess EDUP dropped the card or something :( |
What a fun repo to have found! The Amazon link does work again now. |
I bought an EDUP PCIe Intel AX200 WiFi 6 Card, which uses the Intel AX200 chip, which is supposedly friendlier with Linux than the Realtek chip I tried integrating in the ASUS PCE-AC51 over in #20.
Note: It should work with the
iwlwifi
driver—supposedly.The text was updated successfully, but these errors were encountered: