Flint 2 (GL-MT6000 ) - bug reports - collective thread

Might be an issue with the Pixel then.

I have the same mobile.
With 40Mhz forced:

With 20/40Mhz forced:

Not big difference.
The only strange thing is your mobile isn’t connecting to the 40Mhz

Wow! What a great result. I’m curios too on the reason of that.

Pixel 7 Pro here, Flint 2 firmware 4.5.8 R2, 2.4ghz channel 13 (clearest in my area), 40mhz forced, upload speed was very flaky, up and down, approx 3 meters away.

This is forced 20/40mhz, approx 3 meters away, all channels remain the same

And forced 20mhz, all at approx 3 meters away, download started off well with this one but slowly dropped, but best result yet.

Where is the comparison between both devices (TUF AX6000 vs Flint2)?

On TUF AX6000 I noticed you found 207 Mbps (2.4 GHz @ 20 MHz) and 369 Mbps (2.4GHz @ 40 MHz).
On Flint2, what speed are you reaching, on the same client?

Should I compare it with 4.5.7/4.5.8 (OpenWRT 21 + OEM driver) or 4.5.6 (OpenWRT 23 + OpenSource driver)?

If you have a chance, it would be nice to see:

  • ASUS TUF AX6000 with the latest native firmware (2.4 GHz);
  • ASUS TUF AX6000 with OpenWRT 23.05.3 (2.4 GHz);
  • Flint 2 with 4.5.6 Firmware (2.4 GHz);
  • Flint 2 with latest 4.5.8 beta Firmware (2.4 GHz);
  1. Do not keep both routers powered on the same time to avoid interference between them;
  2. Do not connect anything on the router’s USB port;
  3. For consistent results, use the same client to perform the test at the same distance
2 Likes

You could create a custom snapshot for both the GL-MT6000 and ASUS TUF AX6000. That way they’re using the exact same firmware version, packages and settings on both routers. And the OpenWrt snapshots contain mt76 fixes that GL.iNets lacks.

I’d avoid using 23.05.3, since it’s months behind in terms of fixes and I’ve heard that 160 MHz doesn’t work correctly.

Both have the same chipset.

Is fair enough to just install stable versions and test it as they are.

Flint 2 was released more than 5 months ago. Doesn’t makes sense using a fresh snapshot to use it.

By the way, in this comparative we are giving a chance to Flint2, testing their snapshot (firmware 4.5.8)

I’m aware. And that’s why I’m suggesting this.

Both of these routers use the same SoC, so the idea is to run identical firmware on both of them as that’ll make it a fair 1:1 comparison.

Sure, you can also compare against GL.iNet’s 4.5.8 firmware, but that’s not a 1:1 comparison since that’s an old version of OpenWrt with a proprietary WiFi driver. And GL.iNets 4.5.6 firmware would be a poor comparison too, since their fork with the mt76 driver is missing fixes.

I personally use OpenWrt snapshots so that the mt76 driver isn’t 7 months behind and 160 MHz works correctly.

Manufacturer “A” and Manufacturer “B” are using the same SoC.

I’m suggesting to test the stable firmware from both manufacturers.

If GL-iNet stable firmware 4.5.6 is not good and the 4.5.8 (snapshot) is also not good, we have a huge problem…

And I’m saying he should create identical snapshots for both of them because…

OpenWrt snapshots are more stable than both 4.5.6 and 4.5.8 and it’s been that way for months.

If you compare near identical firmwares on routers with the same SoC and one performs much better than the other then it should be clear that it’s a hardware issue rather than a software issue. But if you compare OpenWrt (snapshot or 23.05.3) against GL.iNets fork then you can’t rule out software as the issue, since GL.iNets fork is older, it lacks dozens of fixes and it has many different patches applied to it.

There is no stable firmware for Gl.inet’s Flint 2. The company has never developed any bug-free firmware. Since the first version there have been problems with the 2.4 GHz wifi, among others.

Tests with OpenWRT would be better, as they use the same version of the MT76 drive.

3 Likes

Though between the flint 2 and the AX6000 tuf indeed it is the same soc.

But the wireless hardware is a little different its also discussed here:

Flint2:
2.4GHz: MT7976GN 4T4R, 5GHz: MT7976AN 4T4R

AX6000 Tuf:
MT7986AV, MT7976DA

Normally there is not super much difference between the two routers, but the small detail in wireless chips are when it is about performance, mediatek is known to retrofit sometimes combinations in their chips.

Also it appears to me, if you look into the supported drivers for mediatek inside the linux kernel its not really there so far only the mt7986 (tufs one) not the mt7976, but this does not mean its in OpenWrt.

So I’m at the conclusion there are very small differences on the chip this may explain the performance difference, but it doesn’t mean its interesting to test the difference nevertheless though :+1:

Is it a hardware defect?, due to the difference i think not, the least of a hw issue i can think of is the interference problem with usb sticks.

GL-iNet offers stable firmware.
Whether it’s good or bad is a different story.

Their firmware is built based on OpenWrt.

For the most reliable performance, it’s best to compare products using the manufacturer’s recommended firmware.

If an alternative firmwares is giving a better result, this demonstrate that the manufacturer is missing something.

If similar products from different brands show significantly different results with their manufacturer-recommended firmware, it could indicate an optimization issue with one of them.

If GL-iNet most recent snapshot is still resulting bad, the test will demonstrate that they are doing something wrong. We need to point this NOW and not when they move the firmware to the stable version.

So, the conclusion is: GL-iNet is missing something in their patches.

The beta firmware 4.5.8 was released less than 1 week ago.

If OpenWRT snapshots is more stable than the manufacturer firmware, so the manufacturer should stop what they are doing and start from the scratch using the OpenWRT snapshot as initial reference, otherwise we will see another useless firmware like 4.5.7

2 Likes

Not sure whether you are saying that the stable firmware is good or bad ! You really need to make your mind up here. For me all versions have been terrible (not tried vanilla OpenWRT obviously and talking about official firmware) due mostly to frequent VPN disconnects (not to mention WiFi issues as I have given up on these being fixed and have been using a different wifi access point instead).

All these previous answers were about the suggestion I wrote here:

If the comparative show that ASUS is giving a better result, using a similar hardware, why GL-iNet cannot do the same? Are they missing any patches?

If OpenWRT firmware is resulting in a better performance on ASUS, why are we finding a poor performance on GL-iNet on similar hardware?

MT7986AV is the CPU, not the wireless chip, both the asus and the GL.inet have that CPU.
The WiFi chips are identical for both.
GL.iNet GL-MT6000
ASUS TUF-AX6000

2 Likes

Some of the information is wrong on that TUF-AX6000 wiki page.

MT7976GN (2.4 GHz) and MT7976AN (5 GHz) is correct.

Everything is within both the mt76 driver and the linux kernel.

Both the GL-MT6000 and TUF-AX6000 will use the MT7986 firmware.

Yes. As I mentioned in this post, GL.iNet forked a pre-release version of OpenWrt 23.05.0-rc1, applied their changes and then failed to merge in many of the newer OpenWrt fixes. And I gave a brief example of that.

As you can see in the post that I linked to, I said the same thing in January :sweat_smile:

The OpenWrt snapshots have been flawless compared to any of the GL-MT6000 firmware updates from GL.iNet. And myself and many others from the OpenWrt forums can attest to that.

You’ll still experience mediocre speeds when using 2.4 GHz from non AX devices, but there’s no crashes and you can apply any setting without any weirdness. Plus none of the supplied packages are years out of date.

1 Like