GL.iNET Flint 2 (GL-MT6000) discussions

I didn’t observe a noticeable difference of CPU load with packet steering enabled / disabled. So I left it disabled and didn’t care about irqbalance. I did these tests while having torrent upload activity around 800 Mbps through a WireGuard VPN.

1 Like

Hi,
Is there anyone running OpenWrt 25.12.2 on GL-MT6000 without any issue?

I’m always running custom builds of 25.12-SNAPSHOT, so not exactly tagged version, but it has been fine (apparently I completely missed the bad commit in 25.12.1 with my builds).

1 Like

I am running OpenWrt 25.12.2 on GL-MT6000 (26 hour) without any issue!

2 Likes

Does anyone see the new mq-cake option? My mt6000 is running 25.12.2 with sqm-scripts 1.7.2-r1 but I don’t see the option anywhere in luci.

The LuCI update didn’t make it into 25.12 yet, but it’s probably for the best since cake-mq is still not fully sorted, at least for this device.

4 Likes

Yes no issue for my MT6000 25.12.2 but it runs only for a day and is lightly used

3 Likes

For sysupgrade of both 23.05 > 24.10 and 24.10 > 25.12 , can option Do not keep configuration be unchecked, so we actually can keep configuration? Or is it always recommended to first check it for flashing firmware, and then to manual restore configuration via System > Backup / Flash Firmware > Restore backup after successful flash?

https://openwrt.org/toh/gl.inet/gl-mt6000 > “Installation” states to not check it, but it is not clear, if this is just for initial installation (upgrade from GL-MT6000 fork of OpenWrt) or for each OpenWrt upgrade.

I got some older routers left, that need to upgrade from 23.05, and am assuming one shouldn’t jump over multiple major versions.

Thanks much for clarifications.

  • Never keep settings, if you go from vendor firmware to OpenWRT.
  • you can safely keep settings, if you sysupgrade +1 version update step at a time
  • it is not just about not keepin settings, it is also: avoid sysupgrading more than +1 version steps at a time, if you don’t know what to expect, as the outcome is a surprise.
  • if you do each step consecutively, you can keep settings in both steps (23.05 > 24.10 and 24.10 > 25.12)
  • it does not matter technically, whether you keep settings or do a second step config restore. Keeping settings during sysupgrade is just way more convenient. Both allow later resetting config to restart with default settings, in case a config drama goes south.

I am having these issues: OpenWrt 25.12.2 - Service Release - #232 by Azorean

Yea we didn't mention keeping settings on sysupgrade between major yearly releases because although it's reliable I just don't trust it. There are many people that post to the release threads with little minor issues that are sometimes resolved by not keeping settings. I never keep settings between major OS updates. It's up to you, it'll very likely work perfect, but there are edge cases.

Edit: after giving it some thought I added a sentence in there mentioning keeping settings because I think most people want to do it if they can.

2 Likes

Wow, I never believed WiFi could be this fast. I'm testing with the Galaxy S25 FE and the Flint 2 using OpenWrt 25.12.2.

PS: the client matters a lot, iPhones can’t seem to support more than 1.5gbps

3 Likes

And are there any real-life benefits from those speeds? Usually the paths through the external internet or the servers in the other end will be bottlenecks, if you try to maintain a sustained data load with those speeds.

These top speeds are a nice marketing gimmick for Wifi vendors, but in general, the speeds are already higher than most of the real-life use cases.

2 Likes

Yeah, more speed doesn’t necessarily means better experience for end users but I was playing with my new Android phone and I needed to check how was the WiFi performance :sweat_smile:

Yes, you can use your phone as a webcam with good image quality.

Why so critical of someone sharing the joy of fast WiFi?
Yes, there are real world benefits, obviously. Does nobody do wireless local file transfers or backups? Just my phone has 100s of GiB of media transfered from my local file server, where ~4x faster (~200mbps 802.11n vs ~800mbps 802.11ac, in my case) speeds make an obvious real-world impact.

Your comment feels like the Bill Gates missquote meme of "640K (of system memory) ought to be enough for anybody".

I apologize for being critical of your post, it just seems very strange on a WiFi-centric forum, of all places.

6 Likes

The screenshot showing 2000 Mbps WiFi speed on a Samsung Galaxy S25 FE is almost certainly done with 160 MHz channel width. Depending on the housing density, I might stick to 40 MHz just because there are only a handful of channels and using 8 of them at once could mean being rude to neighbors.

100 GB of data at 2000 Mbps speed means roughly 8 minutes. Unless you’re doing video editing on Adobe Premiere, which you shouldn’t be doing on a Galaxy S25 FE smartphone anyway, having that kind of speed is simply not worth the potential annoyance to neighbors.

But this obiously has nothing to do with the GL-MT6000 in question.

I’ve tested this but with OpenSpeedTest-Server_2.1.10.exe on a LAN PC connected with Ethernet at 2.5 Gbps. The Wireless client is a Xiaomi Pad 7 Pro.

1 Like

What is the CPU usage on Flint at those speeds?

I’m curious if the SOC can handle it or it needs WED.

I don’t know what’s the best tool to measure that but here’s a BTOP screenshot. I took the screenshot just after starting the 5th test.

3 Likes