[Solved] Archer C7 V2.0 Overclock

Hi guys,

Has anyone successfully tried to overclock the C7 V2.0 or C5 V1.2 ?
If so, how to do it?

Why would you do something like that?

why people are doing it in PC world ??

A PC has active cooling, but the Archer don't.

So I think the risk of burning the cpu is to high.

@Pedro: Have you used the search function?

I did use the search function. Nothing related came by. I also did the same on the old OpenWRT forums and on google without success.

Regarding your "why would you do something like that" question:
Because I want to extract the maximum amount of performance from the device, and overclocking is one of the ways that you can use to do that.
Regarding the risk of burning the CPU, yes, the risk is there, but maybe it is low instead of high. Do you know how much hot it would run on an overclocked state? I don't, but I would like to find out, or maybe someone has tried it already.

and

:smiley:

1 Like

That thread is not C5/C7 specific. I'm looking for something more specific to my device so that the risk of bricking it is reduced.

I've downloaded the uboot partition and looked at it with an hex editor but I see no correlation with the generic thread for arxxx overclocking.

It's possible with Breed uboot. You must know that it's in Chinese only and not open source. There are lede builds here that has writable mtd partition. I think eko openwrt does too, I'm not sure. Otherwise you'll have to use a usb-ttl and putty.
English screenshot - no idea how they did that.
MIne is chinese. I flashed it after I bricked it; now it's unbrickable with uboot recovery. I could post the translated menu items I saved as text.

But can I use breed on the C7 V2? it's not listed on the compatibility page.
My personal lede build has a writable mtd.

ok, so the dump of my uboot partition has 128k and the breed uboot that suposedly is the one for the C7 has 91k!

https://breed.hackpascal.net/breed-qca9558-ar8327n.bin

Despite the size, it's working for me. On power up, the first time system light blinks off, hold the reset, then the lights will blink 3 times, leave reset button, and the webui is at 192.168.1.1. Also set the mac address.
This may be a useful read - pepe's uboot mods although C7 isn't there.

English menu
  1. (index.html)System information

  1. (upgrade)Firmware update
    Conventional Firmware
    Bootloader [...] <----update breed
    Firmware [...] <---- normally
    ART [...]
    Flash layout
    [] Automatic restart
    Programmer firmware
    []"
    Flash layout [1]Automatic identification [2]TPLINK [3]Atheros SDK
    [] Automatic restart
    [] Retain Bootloader
    [] Retain ART

  1. (backup)Firmware backup
    ART
    Firmware
    Full

  1. (clock)Freq settings
    Tip: if overclocking after cannot boot, press and hold reset button to power, system will be default frequency start. warning: please careful consideration to memory for overclocking, if memory overclocking ??, after need please to default frequency start, then modified.

  1. (reset)Restore factory settings
    (Please choose the right firmware type, error select can damage firmware. click [implementation] to restore factory settings.)
    Firmware types [1]TP-LINK original factory [2] TP-Link Openwrt [3] TP-Link ddwrt [4] Atheros SDK Openwrt

  1. (tplink)TPLINK settings- (MAC/WPS pin)

  1. (envconf)Environment variable settings
    Enabling environment variables allows some of Breed's settings to be modified and saved.
    Depending on the type of firmware you are using, select the appropriate location, which may damage the firmware or even Breed.
    • For the TP-LINK firmware, select [Breed Internal], environment variables will be stored at the end of the Breed partition, and updating Breed will cause the environment variable data to be lost.
    • For LSDK and QSDK firmware, select [LSDK / QSDK], updating Breed causes the environment variable to be disabled, but does not clear the environment variable data.
    • For the UBNT firmware of the AR724X, select [UBNT.XM] and have the same characteristics as [LSDK / QSDK].
    • For other types of firmware, select [Breed Internal].
      Use a custom location only if you are fully familiar with the firmware structure and are confident.
    Control (radio)[.] Disable [] Enable
    Position: Custom/Breed internal/(LSDK/QSDK)/UBNT.XM
    Start address _________
    Size (bytes) ________

  1. (Reboot))Restart
    click [reboot] button to restart routing

  1. About

how did you flash breed?
from lede mtd write to mdt0 or from the tplink bootloader?
if from lede, did you erase the partition first with mtd?

I flashed the c7v2 from usb-ttl. The flashing memory address as per your flash rom size, chipset, and partition(art/uboot) from Breed forum. I've flashed breed on a tp-link 1043nd from openwrt. Sorry, I don't remember if I erased.

Out of curiosity, what exactly are you intending to utilize your router for?

  • A computer a router may be, but it is not a PC and does not function like a PC.
    • Unless you're going to be utilizing CPU intensive tasks, such as compiling, heavy data transfers via USB, etc., your CPU may be lucky if it ever sees utilization above 25 - 33%.

Normal home consumer usage does not create enough traffic to even minutely stress the CPU.

maybe I'm not a normal "home consumer". If I was I wasn't running LEDE!

You can stress the router in many forms, one being using SQM under large bandwidth pipes. Or for instance running transmissionBT from the router, or even running some VPN connections to/from it.

A different question is: will a 10% overclock result in a 10% performance increase and is this noticiable? let's see..

Two points:

  • Next, @rj-45 covers the second

Yes, I have experience with conditions of CPUs degrading and failing due to many conditions, including thermal degradation.

It seems you're not necessarily interested...though, I am interested to see how long your router's CPU runs at these specifications.

  • If you find some way to exert the CPU...without proper cooling the "highways" and transistors begin to fail...
  • The CPU was not really designed to need cooling...
  • This cooling will cause stress in other places on the board
  • The board is likely thin - like a consumer motherboard
  • Other [micro] cracks and breaks will ensue

It's magical how we make metal, glass and sand...with a sprinkle of code and electricity...do wonders...but it is sill just sand...plastic...and breakable...

Let me know your results...as for luck...?

Edit: And to add,...

EXAMPLE, your CPU is 500 MHz...you clock it to 550...

10% increase of what? running at 100% of the overclock? 10% over the normal usage?

Yes, any increase can be measured, and is therefore noticeable.

Easy...because the processors of those machines do 2 things routers don't usually do:

  • Get configured to run to "multitask" in user environments
  • Need CPU in terms of completing tasks "faster" - in being ran though the queue faster due to software processing from the point above

Process on routers generally use a small piece of their CPU space, and don't take much more or less.

this is not an answer to a "why..."
People are doing it in PC world to increase the performance, that is the reason why

Sorry but I didn't get the point of your reply.

This thread's intention was to gather knowledge about how to overclock the C5 v1.2 and C7 v2.0. Neko gave us some great insights, resumed they boil down to:

  • it is possible to overclock them using a special bootloader ( breed )
  • he has personally flashed this bootloader on his C7
  • he flashed the bootloader via ttl-usb

I plan to do the same, albeit trying to flash the bootloader from LEDE, and if successful, write a step-by-step guide on this thread on how to do it. I also plan to publish the maximum overclock my C7 is able to sustain without errors. As flashing the bootloader is quite risky, I am waiting for the arrival of a new router I bought ( a dual core, 1.7GHZ Zyxel NBG6817 that is LEDE compatible ). This way, if I kill the C7 I won't be without internet for too long.

Having said this, I am more than happy that this thread is also used to discuss the pros and cons of overclocking, so please fell free to keep contributing with your opinion.

Let me state mine, starting with some FACTS:

  • router performance has increased from old models to new models. New routers of the same price range usually have more ram, more flash and faster CPUS be it because they are clocked higher or have more cores or have a better design or some or all of those.
  • The number of transistors on CPU's and RAM dictate the price of the chips. More transistors equal more expensive chips
  • The design of the chip also dictates the price of the chip. New, faster designs are more expensive.
  • the consumer market is very thin on revenue margins.

Having stated all the above facts, the only reason I see for manufacturers to go for faster devices even when the cost increases is that there is a need for faster devices. If it wasn't so, we would still be using 2005 router hardware.

Some more FACTS:

  • It is possible to use mainstream router hardware to the maximum. There are several threads in this forum
    of users that are unable to max their internet connections on mainstream hardware and LEDE because the router's CPU is maxed out long before full use of the internet can be reached.
  • My use case is a clear example of this. I have a 100/100 internet connection that I can max out if downloading torrents from my PC but that will throttle to 50 or 60mbits if I use transmissionBT on the router instead of on my PC. The load average during this time hovers around 1.8 to 2.2, meaning I would need a 2.2x faster hardware to cope with the load.

So, the fact that manufacturers keep increasing device performance even at a cost of price with the fact that some users report lack of being able to fully utilize their internet service due to hardware being constrained leads me to conclude that, yes, at least for some users, me included, faster devices are needed.

Regarding the danger of overclocking, it depends on alot of factors. Some FACTS:

  • CHIPS are degraded by a process known as electromigration
  • electromigration process is afected by these factors: temperature, voltage, current, and process node of the chip.

My very personal opinion, as someone that has been overclocking his CPU's since the pentium mmx, having killed a few cpu's along the road is that changing the clock of the CPU from 720MHZ to 800MHZ or even 900MHZ ( a 10 to 20% increase in clock ) without touching on the voltage is unlikely to cause any harm that can be noticed on the device's service life ( I'm assuming around 6 years ). I would be very surprised to be able to reach 900MHZ without touching the voltage but I think 800MHZ is doable. Having said that, running the chip on his thermal/power limit, eg, increasing the clock until it fails without touching the voltage will not cause a degradation of the IC's due to electromigration that will be noticed during its service life.

So for me the question is: is it worth the effort to overclock the hardware by 10 or 20%? will it solve my problem. The answer I believe is no, because I need hardware that is at least 2.2x faster and so I would need to clock the CPU to 1600MHZ, something it is not possible. Even 900MHZ is too little, so I guess what remains here is that it is also fun. Maybe there are other use cases that can benefit from an extra bit of juice.

@Pedro You're trying to utilize your router for something it was never intended for, such as a torrent server or client. This doesn't mean it can't be used for the aforementioned, but that you're expecting quite a lot out of a 500MHz SoC.

There's a reason processor manufacturers, even on processors advertised as overclockable, state clearly that should one overclock their processor, it voids the warranty. This is because of the "silicon lottery" and the fact no two processors are identical due to chemistry and the extremely tight tolerances.

  • What this means specifically for you is you're wanting to overclock a SoC not intended for overclocking, nor have the PCB or PCB components been stress tested and/or designed in such a way the excess heat will not damage them.

  • While you could rig up a fan, there's still no way to know if any microscopic damage is being done to the board and it's components. This means you may get 5 years out of the router or 6 months.

.
None of this means you shouldn't, but that if you do, the likely course of events is the router will fail or exhibit instability due to physical damage to the PCB components at some point in the near future. I personally would recommend simply spending the $150 - $300 on a router or SBC that can do what you want without issue of throttling.

  • The Linksys WRT AC Series is an excellent choice in the $150 - $200 range

  • For SBCs, I'd recommend SolidRun's ClearFog Pro which is extremely versatile and I believe it's $269 shipped.