Real life utility of SoC speed and RAM size in routers

[tiddlywink]

..

 

[sfx2000]

The CCP controls the default uboot and I care about security. Once replaced, I have no issue with the flint 2.

Whatever... you have your reasons, I'm sure, and like they say for infosec, only the paranoid survive

I'll reach out to the GL-Inet folks to see if they can post their uboot source over on their github - they've done it for most of the other platforms they support - e.g. RAMIPS (MediaTek MIPS based), IPQ40xx, IPQ60xx (QualcommAX), MVEBU, Atheros MIPS, so it should be easy for them to import the source for their newer Mediatek uboot

 

[RMerlin]

it should be easy for them to import the source for their newer Mediatek uboot

As long it's fully open sourced. Broadcom's uboot code for instance contains proprietary code, so parts of it are released as binary blobs by manufacturers.

But that's Broadcom... The company that has put "PROPRIETARY" headers on various Makefiles of their SDK...

 

[sfx2000]

As long it's fully open sourced. Broadcom's uboot code for instance contains proprietary code, so parts of it are released as binary blobs by manufacturers.

But that's Broadcom... The company that has put "PROPRIETARY" headers on various Makefiles of their SDK...

true, even if there are binary blobs, they're likely safe - the Qualcomm IPQ60xx has blobs, but the source there is known... MediaTek should be similar here with their ARM uboot - the MIPS uboot is well known...

At least with MediaTek - home ROC vs the PRC...

I've been noted in the past - politics doesn't really have a place with tech, might be unpopular for some, but we must appreciate how folks can be very inventive when blocks are presented...

where is most of the interesting work in RISC-V happening these days - yes, in china due to sanctions - that will at some point bite everyone in the *ss

same goes with chip fabs - a lot of folks were surprised that Huawei did a 7nm mobile phone chip from SMIC with machines that were spec'ed out for 14nm - they found a different path, and it worked.

Not everything is stolen technology, and not all software is "backdoor" spyware...

I realize sometimes MediaTek gets a bum rap as they are one of the more common vendors over in the low-end mobile space from some questionable PRC based mobile phone vendors - but note that most of the issues there are not in the chipset or bootloader - it's in the applications and operating system as most are not stock Android but forks of AOSP with questionable local applications.

 

[Metroguy]

I am not a very technical person . I used resource monitor in the Asus app to see CPU and ram usage spikes . This is what I found
My router is the Asus RT-88u( dual core 1.4ghz, 512 mb ram) which is a 4x4 for both bands . This router model was released 9 years ago and was top end.

I have

9 ac devices connected currently.
My isp plan is 500mbps fibre.

This is what i found while doing some tests that would stress the router a bit more than usual

1) when using media server , from SSD attached to the router to stream movies to 2 smart TVs at the same time , the CPU utilisation stays in the lower 20s and ram usage also stays mostly the same ( 43% used) as with idle or web surfing . So this shows that this activities are not resource consuming.

2) the CPU usage increases and especially ram spikes to 92% when 2 torrents are downloading in download master and the router is streaming to 2 tvs at the same time .

3) on normal webpage surfing the CPU says below 20% and ram stays at 43% same as 1)

Based on My observation, using the download master to download torrents actually uses up lot of ram since the data maybe has to be held in the ram before its to be going to the SSD. At this time sometimes the web UI hangs when downloads in progress probably due to a lack of ram

So this leads me to believe that ram is more important than dual core or quad core CPU. This is why routers with usb ports and designed to run things with the usb port usually come with 512mb ram to cope with

I have read comments from others that getting a WiFi 6 router will help improve WiFi 5 device performance by miles due to some optimisations. I watched a YouTube video that shows a increase of almost always 40 % for older WiFi 5 devices when used with a newer WiFi 6 router .

This is mostly due to the newer hardware architecture and CPU efficiency and other components in WiFi 6 routers ( just like every generation of intel CPU or graphics csrds or others ) that is designed to cope with massively faster WiFi 6 speeds.

So my take on this issue is that one should get a router based on antenna and range design and ram. Do not go below 512 mb ram if U are going to use the router for other functions frequently.

 

[Metroguy]

This is mostly true for Broadcom based hardware, whereas MTK and QCA have at least traditionally, done the WiFi offloading on the main SoC. This might well have changed with more recent WiFi standards, but Broadcom appars to have gone for the beefiest hardware in the WiFi chips compared to its competitors. It also seem to depend if the SoC has integrated WiFi or not.

Don't forget that the OS is also running in RAM on most routers. This is why often half of the RAM or more isn't available to the user.

On my 512mb half is not available

 

[sfx2000]

I'll reach out to the GL-Inet folks to see if they can post their uboot source over on their github - they've done it for most of the other platforms they support - e.g. RAMIPS (MediaTek MIPS based), IPQ40xx, IPQ60xx (QualcommAX), MVEBU, Atheros MIPS, so it should be easy for them to import the source for their newer Mediatek uboot

GitHub - gl-inet/mt798x-boot: U-BOOT for MT2500/MT3000/X3000/XE3000/MT6000

U-BOOT for MT2500/MT3000/X3000/XE3000/MT6000. Contribute to gl-inet/mt798x-boot development by creating an account on GitHub.

github.com

Done