unlocked ramcodes #562
Comments
|
What device are you using? I got something like 650mbs or a bit more on the MT7615DN wireless chip. You also can decrypt the stock firmware if you want And you might be able to access the console interface (depending on the device and firwmare). |
|
it's my own spin, but the ceiling seems to be the same whether or not i use this driver or the one from openwrt: http://github.com/openwrt/mt76 i am pretty sure hardware offloading is enabled by default? there is no configuration flag for it, from what i can see. i will try to decrypt the firmware and see if maybe their eeproms are better, but i won't be able to get the ram codes. i had another user of my firmware confirm that their ceiling was also around 350-400MBps. for upload i can, and have, hit 500-550MBps but nothing higher. this is for the MT7615E. i think the 7615D is 2x2 or 3x3? it's very possible they didn't limit the lower-end chip (no offence intended). the 7615E is the flagship with either 3x3 but more commonly 4x4. |
|
That is weird then. offloading is compiled by default but not enabled. It is for NAT though so if your wireless network is on the same subnet as your gateway it should not make a difference. It is enabled in the wan section I have tested just then with iperf3 on a different subnet with hw offloading and here are the results. I did both download and upload tests. Again my device uses MT7615DN (DBDC). |
|
while i don't have iperf running (i don't know how to open the ports locally lulz), i wouldn't doubt i get similar uploading performance. are you using this build? i am quite sure it's the rom patch binary. the one from openwrt and this build are identical. it seems almost all post 5.x mt_wifi builds all have teh same rom patch. |
|
AX200+7615E+padavan，it's very easy to hit 800 Mbps. |
|
iperf3 -R means downloading test and we both get good download speed (7615dn 580Mbps and 7615e 800Mbps). |
this is what it says now. it could be my wireless card (BCM943602CDP WiFi Bluetooth Airport Card), but i thought this was rated for 1733Mbps. also, when this test is performed, it is using only the wifi interface, right? i want to eliminate the ethernet driver as a potential problem. i get high speed via ethernet no problem. while the speed above is lower than usual, it is reflective of the issue. i don't even know if i have a card capable of 1733Mbps if this card is not able to. other device is WUSB600N ver 2 which can get similar speed. what do you think the problem could be? |
|
another thing i don't udnerstand: how can iperf3 performance be lower than speedtest? |
|
here is my configuration file. the bigger issue is (i think) the performance is the same on mt76 driver and this driver. i don't know why this is the case. and i don't know why upload is faster than download. it is possible an antenna is loose in the actual router, and i will check that, but i think there must be some other issue related to software.. |
You are using DD-WRT and it doesn't support wifi hardware nat as openwrt. |
|
are you talking about the raeth driver? i am using the mtk_eth_soc driver and it does have hardware nat... i tried using raeth driver two years ago and it did not work with linux 4.14. are you saying the raeth driver has been fixed for newer kernels? i wanted to use this driver for some time. and the hardware nat (in net/nat/hw_nat) needs raeth to work |
openwrt only support ethernet hardware nat,no wifi hardware nat support. |
|
but wifi hardware nat need raeth wire driver, which does not work for linux 4.x last time i have checked. do you know if anyone has updated raeth driver for linux 4.x? wifi hardware nat needs this driver: https://github.com/hanwckf/rt-n56u/tree/master/trunk/linux-3.4.x/drivers/net/raeth |
No one has done that.So in all,the firmware that doesn't support wifi hardware nat cuased ur low wifi speed.Padavan has no problem. |
|
someone needs to do that man! linux 3.18 is no longer supported lol. i wanted to use the raether driver for a long time!! and hardware nat needs this driveR! :( |
it's about hardware nat and has nothing to do with wifi driver. |
Here is DLINK XDR1860 (7621+7915+linux 4.4_MTK HNAT) GPL source code.Maybe u could try to port it. |
|
i will look into this. thanks for this information. leave the ticket open though. i think net/nat/hw_nat may not need raeth driver, but i thought it did. i will also poke at this source. thanks very much mt_wifi master. |
|
source doesn't have new raeth driver but it suggests it has been updated grep: DIRX1860_A1_GPL102B06DLink/alpha/pkg.priv/mtk/linux-4.4: No such file or directory :( |
It used new mtk driver instead of reath.u could also found hnat information through .dts files. |
|
are you saying "new mtk driver" as in the one from openwrt? i see in wifi driver there is WHNAT config option, but i thought this needed net/nat/hw_nat? |
openwrt rewrote mtk driver so there is much similarity between sdk and openwrt driver code. |
|
nevermind, it needs FAST_HW_SUPPORT which needs HW_NAT argh :P |
|
i am using the openwrt driver but from what i can gather, wifi hardware nat won't work without ethernet hardware nat in the DIR x1860 source, they reference RAETH for 4.4 but it's not included. i don't think openwrt driver can work with WIFI HW_NAT unless i can somehow port HW_NAT from net/nat. my openwrt driver has HNAT for mt7621 but it does not have 2 gmac or anything like that. i am not sure what to do. i want rated speed :P |
|
fast_nat_support is designed for raeth driver and new mtk driver can support wifi hwnat without fast_nat_support. |
|
so if i remove dependency in embedded/Kconfig: config WHNAT_SUPPORT and also in mt7615_ap/Makefile: ifeq ($(CONFIG_FAST_NAT_SUPPORT),y) then i should get rated speed? |
|
No.WIFI-HWNAT has dependency on ethernet driver. |
|
but i can't find the mtk ethernet driver in the GPL source. you're saying this source is different from the openwrt, has hardware nat, and will allow WHNAT to work, right? |
|
target/linux/ramips/files/drivers/net/ethernet/mediatek |
|
OH that's in 101 GPL source not 102! GagansMacPro:tools Gagan$ ls DIRX1860_A1_GPL102B06DLink/target/linux/ramips/files/drivers/net/ethernet/mediatek SWEET SHIT dude i'll work on this ina few hours after i get proper practice done. this was distracting me. so this driver is the one that will work OMFG |
|
i checked. it's the same as the one from DIRX1860_A1_GPL102B06DLink source. it's possible you made another change to the wifi driver but i don't know. what i do know is this is frustrating! :( the module is hooked in properly but for some reason, it's not accelerating the way it should be. i am hoping someone can provide a suggestion as to what else could be the problem. |
|
i'm off to play civilization revolution for the xbox for now. this sucks. at least everyone in the "chinese router scene" is now aware i can produce bacon :) |
|
i can confirm that it must be the ramcodes. stop the bullshit brother, no offense. just tried GPL MTK driver with hnat, here is bootlog. and i get SAME iperf3 performance via 2.4G (no reason to think 5.0G is any different). i spent one week RIGOROUSLY evaluating your claim that it was the lack of hardware nat, but as i've shown: this is simply not true. the performance is definitely being limited by the ramcodes. |
|
again i want to thank you for trying to help, i really do appreciate it. but my time isn't cheap. i have spent a reasonable amount of time making this platform good, and i feel the lack of performance for some reason is upsetting. i am aware of the high suicides at FOXCONN and associated entities (REALTEK included). i am sorry for that and i don't intend to sound like one of the supervisors at those corporations who drive the hard working employees to such tragic ends. this is frustrating. there is no reason performance should be low. there should be a real desire from the chinese router enthusiasts to help me out here (i appreciate your efforts). the only thing i can point to is the ramcodes, which are able to drastically affect performance. |
|
This is my wireless speedtest which runs in openwrt master（linux 5.4).It test the speed between wired lan and wifi.From my test,the wireless driver has bo problem. |
|
The tested router is xiaomi redmi ac2100(7621+7615e).I used the MTK proprietary driver. |
|
i like the proprietary driver. i don't know what the problem is! this is a frustrating situation. is your ac2100 7615x2? sounds like similar SOC to the dir 882. what kind of SPI/nand? |
7615+7603,nand. |
|
The wan<->wifi speed is about 400Mbps in default situation and I can hit 800Mbps If I use nat_flow since it support wireless hw_nat. |
|
it's possible something is wrong with kernel but i don't see how kernel could interfere with speed like this.... the drivers are from GPL and this github, so i want to believe kernel isn't the issue here. seems like only big difference. |
|
padavan and my openwrt are using the same the proprietary driver.All the the proprietary driver code is shown on the github. |
|
i know. so i don't understand how this is possible. sigh |
|
You can plug your network cable into the lan instead of wan so you can test wireless speed in the situation that exclude the impact of nat performance |
|
i just run iperf -s off my router, i know you said i shouldn't, but i don't think this is the problem. it doesn't matter whether i use wan or wan, the cap is about the same. i'm wondering what would cause it. also, idisabled "DRACTRL_LIMIT_MAX_PHY_RATE=1300" in the makefile, but i don't know if this would cause any problem since it just seems to set fgphymode check rt-n56u/trunk/proprietary/rt_wifi/rtpci/4.4.2.1/mt7615/embedded/ap/ap.c Lines 2234 to 2279 in 88a26f9 |
|
it has to be something with the remove_vlan_tag function. i don't think it's being untagged properly or something. i'm going to try the one on this git since padavan seems to have rewritten it. edit: actually nevermind, there are too many changes in the macros for me to wade through it all. just have to focus on this function i guess edit2: in older kernel KConfig it explicitly states hw_vlan_tx is not supported for dual gmac.i thought this would be a problem, but commenting out CONFIG_RAETH_HW_VLAN_TX in raeth_config didn't help |
|
btw @MeIsReallyBa i noticed that dirx1860 102 gpl has slightly newer RAETH with 1200 mode enabled whereas 101 does not. i'm wondering if that is partially the reason you don't get your switch picked up. |
|
What I mean is that the wan will work even if switch is not loaded properly.In fact after modifying the dts, the switch was successfully loaded |
|
i know that. what i am saying is, it is possible that newer raeth from 102 may allow the switch to work. i finally got the "Reentry frame" error to go away. i had to add non-"ethX" interface to br0. i create two vlans on eth2 and added vlan1. it's still not going over 400Mbps though. i suspect this issue has to do with the vlans and the bridge. but i'm not exactly sure what to do aside from havign vlan1 on the bridge. |
|
@MeIsReallyBa show me your bootlog if you can, please. also, any relevant vlan information etc. i hate to be honest but i don't think you're telling the full truth. the linux dir you've provided doesn't even have a properly-hooked wifi driver into the hw_nat. so either you're using some stock router fw and testing on that (and not telling me), or you're not telling the truth. it's not possible for you to get that performance with that linux directory in the build tree, if we are to believe the lack of hW_NAT was the reason for bad performance. since you are the one who provides the newer drivers, the only thing i can think of is you have some sort of incentive to not be forthcoming about the lack of full performance. in other words: your evidence is extremely lacking in comparison to mine |
|
it sort of works now! i suspect the reason i'm not maxing out is because APPUL (@apple) wifi licks balls. i'm pretty sure if i was in windows i'd get more. either way. hitting 500/500 consistently now with the HARDWARE NAT baby the broken english log output is so amazing rn my users are probably so happy hahahahah. REALTEK FOR LIFE
@MeIsReallyBa i know what you mean now when you're talking about CPU. one core will just max out when it starts pushing. it starts sort of slow then all of a sudden it keeps going up and up. i'm often running out of packets by the time this shit is warmed up. THIS THING IS SO SICK. THE HARDWARE NAT HAHHAHA |
|
[ 0.000000] Linux version 4.4.177 (bibi@ubuntu) (gcc version 7.3.0 (OpenWrt GCC 7.3.0 r4046-beaa2ecc10) ) #0 SMP Tue Nov 19 05:51:15 2019 |
|
Why do you think the wifi driver has not been hooked into hw_nat? |
|
#562 (comment) |
|
I was assuming you used the driver as it is on the source. But when I looked at your source in 7615+7615 it seems you've uncommented the appropriate portions. My apologies on that.
How did you get rid of the "no prim_ch value for adjust?"
From: MeIsReallyBa
Sent: Monday, 18 January 2021 8:49 PM
To: hanwckf/rt-n56u
Reply To: hanwckf/rt-n56u
Cc: gagan sidhu; Mention
Subject: Re: [hanwckf/rt-n56u] unlocked ramcodes (#562)
Why do you think the wifi driver has not been hooked into hw_nat?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub<#562 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AITLEGYTU6WNAZNAANUQVJ3S2T6LPANCNFSM4V35NZEA>.
|
The only 7615+7615 TP router in my hand only supports one GMAC, so I have not tested the 7615+7615 driver. The 7615+7615 may even be the version under linux 5.4, maybe I didn't pay attention to it when I packaged it.I have never met "no prim_ch value for adjust", so I can't help. |
What firmware are u using?padavan? |
|
No problem.
I'm getting better speeds now. I don't have a windows device with good wifi. Just an xbox series x. I'll try speedtest on it tomorrow.
The fact I can hit 500Mbit via speedtest is encouraging.
Mac is very weird. First few speedtest runs it climbs past 500 then successive runs slow down to about half that speed. Then I have to disable and re-enable the interface to get better speeds.
May very well be a mac issue heh
From: MeIsReallyBa
Sent: Monday, 18 January 2021 9:42 PM
To: hanwckf/rt-n56u
Reply To: hanwckf/rt-n56u
Cc: gagan sidhu; Mention
Subject: Re: [hanwckf/rt-n56u] unlocked ramcodes (#562)
I was assuming you used the driver as it is on the source. But when I looked at your source in 7615+7615 it seems you've uncommented the appropriate portions. My apologies on that. How did you get rid of the "no prim_ch value for adjust?" From: MeIsReallyBa Sent: Monday, 18 January 2021 8:49 PM To: hanwckf/rt-n56u Reply To: hanwckf/rt-n56u Cc: gagan sidhu; Mention Subject: Re: [hanwckf/rt-n56u] unlocked ramcodes (#562<#562>) Why do you think the wifi driver has not been hooked into hw_nat? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#562 (comment)<#562 (comment)>>, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AITLEGYTU6WNAZNAANUQVJ3S2T6LPANCNFSM4V35NZEA.
The only 7615+7615 TP router in my hand only supports one GMAC, so I have not tested the 7615+7615 driver. I have never met "no prim_ch value for adjust", so I can't help.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub<#562 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AITLEG3VJBZDL4MNF36T4W3S2UETNANCNFSM4V35NZEA>.
|
|
@MeIsReallyBa when do you think we'll see a newer mt_wifi driver with full PMF functionality? ithink this is where most errors are coming from. |
|
@MeIsReallyBa can you show me your dat files for the radios? i want to make sure i have the optimal settings. my airport is reporting MCS index of 9 when i've set it to 33 in the dat file. wondering if that is part of the reason for suboptimal performance. |
|
@MeIsReallyBa are you alive? why is 2.4ghz performance so bad? is it bad for you? i never get over 100mbit. one guy is complaining about signal dropping. anything i can do to possibly fix it? |
ok @hanwckf
i know you guys are holding out on us ok. there is no way you aren't.
both the mt76 driver (when i did get vht160 stable enough to run speedtest a few times) AND the proprietary driver seem to never pass 400 Mbps down, or 500 Mbps upload.
literally right when it hits about ~400 Mbps, it starts to throttle down and seems to top out around 371Mbps.
it is highly unusual both drivers yield very similar performance that is far below the specification and capabilities of the driver. this would explain why the factory firmware from dlink is encrypted (if i recall correctly) and also you cannot access terminal via serial.
i understand why these unlocked ramcodes aren't public, but please be considerate. i have been supporting this platform for a few years and now all that's missing is the full performance. there is no way that two drivers can yield the same performance cap if it is not being done artificially at the driver (or ram code, is my guess) level.
The text was updated successfully, but these errors were encountered: