Scheduled Maintenance: We are aware of an issue with Google, AOL, and Yahoo services as email providers which are blocking new registrations. We are trying to fix the issue and we have several internal and external support tickets in process to resolve the issue. Please see: viewtopic.php?t=158230
Debian Jessie wifi 3x faster than Stretch
Debian Jessie wifi 3x faster than Stretch
I have two identical hardware ThinkPads. I have Jessie installed on one, and I recently installed Stretch on the other.
Jessie's wifi is coming in at 150Mbps on its N speed wireless card. Using the same access point, the Stretch install is coming in at 50Mbps.
Anybody know what could be causing this? Is there anything I can do to Stretch to improve its wifi speed? I assume the drivers are different between Jessie & Stretch, might this explain it? 150 vs 50 is a huge difference.
Jessie's wifi is coming in at 150Mbps on its N speed wireless card. Using the same access point, the Stretch install is coming in at 50Mbps.
Anybody know what could be causing this? Is there anything I can do to Stretch to improve its wifi speed? I assume the drivers are different between Jessie & Stretch, might this explain it? 150 vs 50 is a huge difference.
Re: Debian Jessie wifi 3x faster than Stretch
Since the two ThinkPads are identical could you swap the N cards to see if the situation changes?
Can we assume that you have the appropriate non-free firmware in place?
Can we assume that you have the appropriate non-free firmware in place?
Re: Debian Jessie wifi 3x faster than Stretch
The N cards are the same in each, I confirmed with lshw.
And in both I installed the non-free firmware (which was required for wifi to even work).
And in both I installed the non-free firmware (which was required for wifi to even work).
Re: Debian Jessie wifi 3x faster than Stretch
Many years ago I was a tech in a large company. One of our clients was using two identical machines in the same room. One of them was a chronic source of service calls and several techs had tried to fix it. One day, out of shear frustration, I took all of the circuit boards out of both machines and swapped them over. The boards from machine A went into machine B and those from machine B went into machine A. We never again had problems from either. Techs see this stuff and it doesn't make sense. We had a term for it using the acronym FM. The M stands to magic; you can figure out the F.
Have you done any searches for your problem? Either here of through Google? Yours isn't the only complaint about Stretch being slower than Jessie. I don't think that Stretch itself is the problem.
Try disabling ipv6.
Have you done any searches for your problem? Either here of through Google? Yours isn't the only complaint about Stretch being slower than Jessie. I don't think that Stretch itself is the problem.
Try disabling ipv6.
Re: Debian Jessie wifi 3x faster than Stretch
You don't give much data to base a speculation on. Are the two units sitting side-by-side, so that in theory they could have the same reception? Even something as simple as antenna orientation can make a difference. If they are in different parts of a room, it is possible for reception to vary, radio waves can be affected by many things in a building. Since you confirmed the cards were the same, why didn't you mention what cards those are, in case there are reports of flaky cards of that type? That might be something to check.kcbagr wrote:The N cards are the same in each, I confirmed with lshw.
I too have seen the "FM" that Bulkley mentions and even cards that worked better when they were reseated in their slot. I think I would have tried what was suggested. Might have made a difference, or maybe not but you can't be sure if you don't try and it's very difficult to speculate without some kind of comparison data, at least, try moving the slower system nearer to the Access Point.
It could still be difference between versions and the advice Bulkley gave, to search about that, would be a good thing to look into and see if others report something that sounds like what you are seeing.
Re: Debian Jessie wifi 3x faster than Stretch
Thanks for all the posts, guys. I appreciate it.
I post on this because it's such a huge discrepancy. 3x difference is not minor. I was hoping someone would have seen this before or something. I guess it's just ghost in the machine.
This is something I'm interested in. Was there a slowdown with Stretch? Is this at least partly to blame because Stretch is slower? My next test is going to be to install Stretch on the "fast" box and see what happens. But thought I'd post here first just to see what comes up.
I tested side-by-side. I also put 1st one on desk, ran test. Then I swapped its position with the 2nd one and ran test again. I've run several tests trying to account for things like this.Thorny wrote: Are the two units sitting side-by-side, so that in theory they could have the same reception? Even something as simple as antenna orientation can make a difference. If they are in different parts of a room, it is possible for reception to vary, radio waves can be affected by many things in a building.
I post on this because it's such a huge discrepancy. 3x difference is not minor. I was hoping someone would have seen this before or something. I guess it's just ghost in the machine.
Since the cards are the same, I didn't think that mattered. They are Centrino Ultimate-N 6300 cards.Thorny wrote: Since you confirmed the cards were the same, why didn't you mention what cards those are, in case there are reports of flaky cards of that type? That might be something to check.
Did you mistype this? If my "complaint" is not the only one saying Stretch is slower, doesn't that support the possibility that Stretch IS slower than Jessie?Thorny wrote: Yours isn't the only complaint about Stretch being slower than Jessie. I don't think that Stretch itself is the problem.
This is something I'm interested in. Was there a slowdown with Stretch? Is this at least partly to blame because Stretch is slower? My next test is going to be to install Stretch on the "fast" box and see what happens. But thought I'd post here first just to see what comes up.
Re: Debian Jessie wifi 3x faster than Stretch
Ideally the test machines should be running same OS version to determine whether it's hardware related or not. So I'd encourage you to do it like you already plan. I don't think it's impossible though, that there's e.g. regression in wireless drivers for Stretch. Those drivers are proprietary and as such, are not "Stretch issue" in a sense. Not that it's any help when it comes to user experience.
Before upgrading OS you could try to find out what driver versions you have on both machines and check whether there are reported bugs or regression. To find used versions, I'd go about it with something like this. Someone wiser could provide better solution, drivers are not my strong area.
Before upgrading OS you could try to find out what driver versions you have on both machines and check whether there are reported bugs or regression. To find used versions, I'd go about it with something like this. Someone wiser could provide better solution, drivers are not my strong area.
Code: Select all
Get kernel module name for wireless:
$ lsmod
As root:
$ modinfo $(find /lib/modules/$(uname -r) -type -f -name <kernel_module_from_lsmod>.ko) | grep -i version
Debian 10.2 Stable with i3
Secure your stuff: Securing Debian Manual
Don't break your stuff: Source List Management DontBreakDebian
Secure your stuff: Securing Debian Manual
Don't break your stuff: Source List Management DontBreakDebian
Re: Debian Jessie wifi 3x faster than Stretch
No, I didn't mistype that, I didn't type that at all, that was a comment Bulkley made. I think it was supposed to be a suggestion to use a search engine to look for other reports that might be similar. I think that was good advice.kcbagr wrote:Did you mistype this? If my "complaint" is not the only one saying Stretch is slower, doesn't that support the possibility that Stretch IS slower than Jessie?Thorny wrote: Yours isn't the only complaint about Stretch being slower than Jessie. I don't think that Stretch itself is the problem.
I also think Bulkey's advice to swap the network cards and see if the problem followed one of the cards might have been a quick way to test hardware since your original report had very little data to troubleshoot with. (That would also have reseated both cards and tested if a bad connection might have been involved.)
The Wizard made a sane suggestion after you identified what cards you were working with, illustrating why posters are urged to give information about their system with their questions. Clearly, Wizard1000 knew something about those cards and was able to offer a suggestion.
Re: Debian Jessie wifi 3x faster than Stretch
It is always possible the antenna is crippled. Of course, you do not need to install a new OS just to test, this is why Linux liveCD's exist. I personally keep SystemRescueCD always in my pocket in a USB stick.
- alan stone
- Posts: 269
- Joined: 2011-10-22 14:08
- Location: In my body.
Re: Debian Jessie wifi 3x faster than Stretch
In addition to what's already been suggested it could be a combination of causes, for ex. both laptops are from a different production site assembled with bezels made of different composition/material.kcbagr wrote:I have two identical hardware ThinkPads...
Anybody know what could be causing this?
Re: Debian Jessie wifi 3x faster than Stretch
This helped a lot, thanks!wizard10000 wrote:Yup - I would have guessed they were Intel cards.
As root, create /etc/modprobe.d/iwlwifi.conf and add this single line to the file.
Reboot. Enjoy. Headed for work but I can explain more later if you're interested.Code: Select all
options iwlwifi led_mode=1 power_save=0 11n_disable=8 swcrypto=1
Good luck -
edit: Check out this thread for a more detailed explanation - http://forums.debian.net/viewtopic.php?t=121696
My tests are now scoring between 89Mbps and 116Mbps with a better bufferbloat score. Still not as good as Jessie but surely a significant improvement. Doesn't this lend evidence toward the idea that Stretch has slower wifi than Jessie?
Although wizard10000 has improved things, Jessie still is getting 50Mbps faster and even better bufferbloat scores. I'm seriously thinking about going back to Jessie.
I assume Debian can't fix this, that the problem is in the iwlwifi driver (which is proprietary preventing Debian from fixing)? This stinks that we're getting a huge reduction in performance from update. Updates shouldn't land when they severely reduce performance.
Re: Debian Jessie wifi 3x faster than Stretch
Just wanted to leave a big Thanks to everyone for your posts. Thank you so much for your efforts guys. And Wizard10000, thank you for chiming in with what resulted in a big improvement
Re: Debian Jessie wifi 3x faster than Stretch
Ugh. Making the same iwlwifi parameter changes that Wizard10000 posted about to Jessie brought Jessie's internet speed up to 200Mbps.
Wifi performance really took a dive in Stretch. Should a bug report be filed about this? Maybe Debian developers can do something to improve the situation for Buster's release?
Wifi performance really took a dive in Stretch. Should a bug report be filed about this? Maybe Debian developers can do something to improve the situation for Buster's release?