AC68U DDR Configuration Upgrade Tools

[jayboyyyy]

Whenever you are asked to reset to factory defaults, you should never restore a saved backup afterward, as it would totally defeat the purpose of resetting to default values - you would simply reload the same old values on top of them.

after you run the tool and everything is running, can you not restore your settings then? What would cause problems?

 

[RMerlin]

after you run the tool and everything is running, can you not restore your settings then? What would cause problems?

Your router has setting1=badvalue. You make a backup of it.
You reset to factory default. Router then sets setting1=goodvalue
You reload your saved setting. What happens? You will reset setting1 to badvalue. Therefore you are back to square one.

 

[dmorales023]

Is the Reset necesary?

I think it worked without reset:

RMerlin, is it possible to do what Kris404 did? I mean, if all the tool is doing is updating a value in the bootloader, I dont see the need to reset it.

 

[jayboyyyy]

Your router has setting1=badvalue. You make a backup of it.
You reset to factory default. Router then sets setting1=goodvalue
You reload your saved setting. What happens? You will reset setting1 to badvalue. Therefore you are back to square one.

i was under the impression the tool changed BL settings/values. Didn't think restoring the back up of the setting would also change the BL settings. Just figured the back up was of web gui settings but I'm not familiar with what the back up file actually backs up.

 

[RMerlin]

RMerlin, is it possible to do what Kris404 did? I mean, if all the tool is doing is updating a value in the bootloader, I dont see the need to reset it.

I haven't personally tested the update yet (I have some tests to do before that), but I suspect that the bootloader update changes the clkfreq default value as well. That means if you were to upgrade the bootloader without also using the new default clkfreq value, then you would end up still running your RAM at the old 533 MHz clock.

When you do a factory default reset, what happens is that the default values that are device-specific (such as clock, MACs, etc...) stored in the bootloader gets written to nvram. On the first firmware boot, then the firmware-specific default values are also added to nvram.

That's most likely why a factory default reset is required after the upgrade. The new bootloader might also have brought other changes, but I don't know, as I haven't checked yet.

 

[chunlianghere]

i wonder if i do this DDR upgrade, updated to 306. after than IF hav the same problem when running utorrent, will it affect the updated ram speed if revert back to 205?

 

[bigmag]

Hi

how can I check the flash was successful and DDR 666 Ram works

 

[nurv2600]

Just ran the tool, and it backs up, upgrades, and restores your settings all in one go. No problems, no restoring, but I did backup just in case. I recommend everybody do a backup anyway, since if it's updating the bootloader, I'd imagine a settings restore wouldn't overwrite those settings. I think (hope?) that the backup feature only backups the features you see in the web GUI.

 

[nurv2600]

Hi

how can I check the flash was successful and DDR 666 Ram works

SSH or telnet into the router, then run "nvram get clkfreq" without the quotes.

 

[Smedley]

I think I am missing the point of this "upgrade". Is this fixing an error in the bootloader that will likely be remedied in newly manufactured units? Is this offered as a way of boosting performance?

I am using my AC68u strictly as an Access Point. I have had no issues with any of the firmware revisions. I have only rebooted to upgrade firmware. I am getting my full wired ISP throughput (about 70 Mbps) on my 2.4GHz, 802.11n connections.

Is this memory clock speed change recommended or required? Is there any downside to applying it (e.g. would the ac68u run hotter with a faster clock speed on its memory)?

 

[TonyH]

Hi,
Increasing clock rate will increase heat as well.

 

[RMerlin]

I have run some benchmarking tests before and after upgrading the bootloader. The process upgrade the bootloader from 1.0.1.1 to 1.0.1.6.

Some interesting results, but at a cost for overclockers.

The test setup was a laptop connected through Ethernet on the LAN side. The WAN side was connected to my regular LAN, where I ran the "iperf" daemon on an HTPC.

HW acceleration was forced disabled, so I could truly evaluate the raw CPU muscle in terms of routing.

I did two tests:

1) I ran iperf between the laptop and the HTPC to measure routing performance (keep in mind this was done with HW acceleration disabled, so it was raw CPU muscle). The client was run like this:

iperf -c 192.168.10.130 -M 1400 -N -l 64K -t 30

2) I ran a simple openssl benchmark on the router:

openssl speed aes-128-cbc

This was done with a rough build of my FW with 374_306 code, so it has my OpenSSL my optimizations in it.

OpenSSL 1.0.0j 10 May 2012
built on: Tue Oct 29 19:45:57 EDT 2013
options:bn(64,32) rc4(ptr,char) des(idx,cisc,16,long) aes(partial) idea(int) blowfish(ptr) 
compiler: arm-brcm-linux-uclibcgnueabi-gcc -fPIC -DOPENSSL_PIC -DOPENSSL_THREADS -D_REENTRANT -DDSO_DLFCN -DHAVE_DLFCN_H -ffunction-sections -fdata-sections -DTERMIO -O3 -Wall -DOPENSSL_BN_ASM_MONT -DSHA1_ASM -DSHA256_ASM -DSHA512_ASM -DAES_ASM -DGHASH_ASM

I ran the two tests in four scenarios:

1) Pre-upgrade, with CPU @ 800 MHz and DDR @ 533 MHz
2) Pre-upgrade, with Turbo mode enabled (which bumps the CPU at 1 GHz)
3) Post-upgrade, with CPU @ 800 MHz and DDR @ 666 MHz
4) Post-upgrade, with Turbo mode enabled

CPU: 800 MHz
DDR: 533 MHz

iperf:
[152] local 192.168.1.100 port 1841 connected with 192.168.10.130 port 5001
[ ID] Interval Transfer Bandwidth
[152] 0.0-30.0 sec 1.16 GBytes 333 Mbits/sec

openssl:
The 'numbers' are in 1000s of bytes per second processed.
type 16 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
aes-128 cbc 25299.61k 27983.66k 28966.83k 29250.56k 29392.90k


CPU: 1 GHz (through Turbo mode)
DDR: 533 MHz

iperf:
[152] local 192.168.1.100 port 2086 connected with 192.168.10.130 port 5001
[ ID] Interval Transfer Bandwidth
[152] 0.0-30.0 sec 1.30 GBytes 371 Mbits/sec

openssl:
The 'numbers' are in 1000s of bytes per second processed.
type 16 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
aes-128 cbc 31755.66k 35012.59k 36352.85k 36820.51k 36711.59k


CPU: 800 MHz
DDR: 666 MHz

iperf:
[152] local 192.168.1.100 port 2301 connected with 192.168.10.130 port 5001
[ ID] Interval Transfer Bandwidth
[152] 0.0-30.0 sec 1.23 GBytes 351 Mbits/sec

openssl:
The 'numbers' are in 1000s of bytes per second processed.
type 16 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
aes-128 cbc 25468.94k 28007.64k 29075.80k 29366.61k 29455.70k


CPU: 1 GHz (through Turbo mode)
DDR: 666 MHz

<Router failed to boot>


Conclusion:
The good news is, this does improve raw routing performance by around 6%. The bad news is, Turbo mode no longer works with the upgraded bootloader (either due to the higher DDR clock, or because the feature is simply broken in that new BL).

While increasing the CPU clock yields a higher performance boost than increasing the DDR clock, the DDR clock is less likely to generate additional heat, and it's also giving you that increase within the HW specifications (while overclocking is pushing it beyond the official, tested specs, so stability can vary).

 

[ronan_zj]

CPU: 1 GHz (through Turbo mode)
DDR: 666 MHz

<Router failed to boot>

I think I have this problem. My DDR has overclocked to 666 MHz, so when I push the cpu overclock button, the router fail to boot and work.

 

[RMerlin]

I think I have this problem. My DDR has overclocked to 666 MHz, so when I push the cpu overclock button, the router fail to boot and work.

That confirms what's happening then. No idea however if it's because Turbo Mode is broken in the new bootloader, or if it's because of stability issues when you increase both the CPU and RAM clocks.

 

[hackintoshsr]

Have you tried manually setting the clock to see if its turbo mode specific?

I have a stable 900,666 clock going on.

Maybe your turbo mode could have a setting in 50mhz increments inside the gui?

Example

dropdownlist:
850
900
950
1000

Also first post. Just want to thank you for your work Merlin

 

[geops]

I can't use this tool for some reason. During the "backing up" portion I get "Retrieve 2.4G MAC Fail!!!". I logged into the router and I saw that the LAN MAC and the Wireless 2.4Ghz MAC are identical. Is this normal?

 

[RMerlin]

Have you tried manually setting the clock to see if its turbo mode specific?

I have a stable 900,666 clock going on.

Maybe your turbo mode could have a setting in 50mhz increments inside the gui?

Example

dropdownlist:
850
900
950
1000

Turbo Mode doesn't work that way. Turbo Mode is handled by the bootloader. At boot time it will check if the button is pressed - if it is, it will set the CPU clock to 1 GHz. This is hardcoded in the bootloader, this is why even with Turbo Mode enabled, clkfreq still says 800.

I don't want to implement manual overclocking, for safety reasons. Turbo Mode was actually a feature Asus was working on, and it's perfectly safe (unlike changing clkfreq) since even in case of failure to boot, you can easily disable it with the press of a button. I have seen situations where the bootloader would no longer acknowledge the reset or the WPS button - the only way to recover from this was by using a serial cable. So if you were to end up in that situation with invalid clkfreq values, you would be stuck with a brick.

 

[ronan_zj]

I can't use this tool for some reason. During the "backing up" portion I get "Retrieve 2.4G MAC Fail!!!". I logged into the router and I saw that the LAN MAC and the Wireless 2.4Ghz MAC are identical. Is this normal?

My lan and 2.4G wifi also have same mac address, so it should be normal.

 

[geops]

My lan and 2.4G wifi also have same mac address, so it should be normal.

Thanks. Now, why isn't the tool working on my ac68u

 

[ronan_zj]

Turbo Mode doesn't work that way. Turbo Mode is handled by the bootloader. At boot time it will check if the button is pressed - if it is, it will set the CPU clock to 1 GHz. This is hardcoded in the bootloader, this is why even with Turbo Mode enabled, clkfreq still says 800.

I don't want to implement manual overclocking, for safety reasons. Turbo Mode was actually a feature Asus was working on, and it's perfectly safe (unlike changing clkfreq) since even in case of failure to boot, you can easily disable it with the press of a button. I have seen situations where the bootloader would no longer acknowledge the reset or the WPS button - the only way to recover from this was by using a serial cable. So if you were to end up in that situation with invalid clkfreq values, you would be stuck with a brick.

is it possible to let users to choose which clock frequency they want to run before they push the button?
for example, under performance turning tab, when we choose LED button behavior to Overclock, and it will also ask us which clock frequency we would love to run, such as: 900, 950, 1000.( maybe 900 or 950 may work for DDR 666 ?)