Automated RedPill Loader (ARPL)

[Rasmus86]

13 hours ago, idaanx said:

If you do this make sure you have an idea what you are doing, there is always a risk of losing data if something goes wrong.

 

modprobe acpi-cpufreq && echo "performance" | tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor

The modprobe loads the kernel module and on success it forces all the governor into performance mode.

 

Once loaded you can check if you get an output.

 

grep . /sys/devices/system/cpu/cpu0/cpufreq/*

 

Thank you very much for this - helped on my baremetal.

 

- DSM version DSM 7.1.1-42962 Update 1
- Loader version and model: Automated Redpill Loader 0.5 alpha 1, DS918+               
- Installation type: Baremetal (Gigabyte GA-H97N-WIFI – Intel i3-4170 - 8gb ram - 6 drives)

 

Before the above fix:

ARPL 918+

https://browser.geekbench.com/v5/cpu/19595117

Linux 4.4.180+ x86_64

198

Single-Core Score

419

Multi-Core Score

 

After the above fix:

ARPL 918+

https://browser.geekbench.com/v5/cpu/19595197

Linux 4.4.180+ x86_64

891

Single-Core Score

1851

Multi-Core Score

[Brunox]

For those who are interested, I would like to post my Geekbench test of my HP ProLiant MicroServer Gen10 Plus

Intel Xeon E-2236

https://browser.geekbench.com/v5/cpu/19595522

with Arpl 1.0-beta9  DS3622

Single-Core Score

1317

Multi-Core Score

6797

[Sebby75]

Testing this on QNAP TS253A in VirtualizationStation3...

Wanted to check out "Synology Photos" as I was never happy with QNAPs Photo Station and QuMagie at all.

 

Everything working spot on so far.

QNAP TS-253A VirtualizationStation3 (v3.6.39)
ARPL v1.0-beta9
running DSM 7.1.1-42962 Update 3 (DS920+)


BTW: When entering real Synology serial number, builder app seems to suggest that serial is incorrect... but it is correct and together with mac address of real Synology unit it allows me to enable QuickConnect.

 

Edited December 30, 2022 by Sebby75

[idaanx]

5 hours ago, apriliars3 said:

 

I can buy this nic ethernet. Is it compatible? I think so. 

 

Intel PRO1000 PT Dual Port Server Adapter Product Specifications

 

Is for a old pc socket 775 that freeze with actual nic, but if disabled internal nic and use only wifi, I test it that no freeze.

 

Looks like it uses the e1000e driver. Make sure you have a PCIe x4/x8/x16 slot on your motherboard, this dual port NIC needs at least x4 to fit. 

[idaanx]

12 hours ago, vbap said:

 

I am game, and went ahead with this on my Supermicro server, running ARPL 1.0b3 bare metal.

And, well, my geekbench multicore score has been tripled (https://browser.geekbench.com/v5/cpu/19587925) - still a bit below average, but already my VMM performance is much much better.

 

I executed the above after shutting down my VMs in VMM, but DSM was still running. Ran geekbench, then started up my VM again from DSM.

 

Thanks @idaanx @pocopico @fbelavenuto - hope you all enjoy the the festive season

 

 

The single score shows it working, multi is is lower which is probably just a throttling issue. Looking at an image of your motherboard it has a pretty small heatsink and fan for the amount of cores/threads.

 

Or it's still running some processes in the background, check the idle state in top before running the benchmark.

[gericb]

4 hours ago, Sebby75 said:

BTW: When entering real Synology serial number, builder app seems to suggest that serial is incorrect... but it is correct and together with mac address of real Synology unit it allows me to enable QuickConnect.

 

I've had this happen too, with 2 different REAL SN/MAC pairs.  Just thought it was me, but seems it really is something with the builder.  Hopefully we can get this quirk resolved, so I can actively use my other set. 🤠

[AuxXxilium]

On 11/3/2022 at 8:23 AM, wifi75 said:

 

hello  I'm trying your loader, could you also make a guide for proxmox?

what is Synocode Patch ?

and nvme (NVMe als Cache)?

 

 

For Synocode patch please use Google. And "nvme" is Support for NVMe Cache with automatic find and patch your pcie paths.

[AuxXxilium]

6 hours ago, Rasmus86 said:

Thank you very much for this - helped on my baremetal.

 

- DSM version DSM 7.1.1-42962 Update 1
- Loader version and model: Automated Redpill Loader 0.5 alpha 1, DS918+               
- Installation type: Baremetal (Gigabyte GA-H97N-WIFI – Intel i3-4170 - 8gb ram - 6 drives)

 

Before the above fix:

ARPL 918+

https://browser.geekbench.com/v5/cpu/19595117

Linux 4.4.180+ x86_64

198

Single-Core Score

419

Multi-Core Score

 

After the above fix:

ARPL 918+

https://browser.geekbench.com/v5/cpu/19595197

Linux 4.4.180+ x86_64

891

Single-Core Score

1851

Multi-Core Score

You know that the cpufreq gov "performance" is setting your CPU to always operate at maximum frequency?

Edited December 30, 2022 by AuxXxilium

[idaanx]

2 hours ago, AuxXxilium said:

You know that the cpufreq gov "performance" is setting your CPU to always operate at maximum frequency?

 

This is false but I just noticed the 3615xs does indeed not scale down, this might be because of the mperf kernel module or old kernel, not sure... 

 

But I have previously successfully tested the freq. scaling on the 920+ (or 918+ can't remember) and just did it again on the 3622xs.

 

On DSM 7 and these models (and probably the others) there are only 3 governors to choose from, performance is the default set by Synology too.

 

Performance: scales up and down as needed

 

 

Powersave: stays looked in the lowest frequency

 

 

Userspace: can be set to a specific upper frequency and also scales down

 

 

As you can see /proc/cpuinfo doesn't show realtime results, which is also why a lot of people think frequency scaling and turbo boosting is not working even on a real Synology.

[idaanx]

8 hours ago, Sebby75 said:

BTW: When entering real Synology serial number, builder app seems to suggest that serial is incorrect... but it is correct and together with mac address of real Synology unit it allows me to enable QuickConnect.

 

3 hours ago, gericb said:

I've had this happen too, with 2 different REAL SN/MAC pairs.  Just thought it was me, but seems it really is something with the builder.  Hopefully we can get this quirk resolved, so I can actively use my other set. 🤠

 

The serial validation is done in combination with a model. Your real serial is either from a different model or the template for generating and validating is incorrect. Check the script from TCRP to validate the template, this one is easier to read.

 

Maybe not a good idea to use QuickConnect with a fake Synology...

 

[AuxXxilium]

23 minutes ago, idaanx said:

 

This is false but I just noticed the 3615xs does indeed not scale down, this might be because of the mperf kernel module or old kernel, not sure... 

 

But I have previously successfully tested the freq. scaling on the 920+ (or 918+ can't remember) and just did it again on the 3622xs.

 

On DSM 7 and these models (and probably the others) there are only 3 governors to choose from, performance is the default set by Synology too.

 

Performance: scales up and down as needed

 

 

Powersave: stays looked in the lowest frequency

 

 

Userspace: can be set to a specific upper frequency and also scales down

 

 

As you can see /proc/cpuinfo doesn't show realtime results, which is also why a lot of people think frequency scaling and turbo boosting is not working even on a real Synology.

simply not possible with "performance", i wrote linux/android kernel over years. so you only see the requested freq, not the real. i can see it in my ilo, if i change i have around 280 watts of power with "performance" and with the stock i only have about 70 watts. you there must be a bit more than this.

 

"

2. Governors In the Linux Kernel ================================ 2.1 Performance --------------- The CPUfreq governor "performance" sets the CPU statically to the highest frequency." <- from kernel.org

Edited December 30, 2022 by AuxXxilium

[cpresco]

3 hours ago, AuxXxilium said:

You know that the cpufreq gov "performance" is setting your CPU to always operate at maximum frequency?

5 minutes ago, idaanx said:

 

This is false but I just noticed the 3615xs does indeed not scale down, this might be because of the mperf kernel module or old kernel, not sure... 

 

But I have previously successfully tested the freq. scaling on the 920+ (or 918+ can't remember) and just did it again on the 3622xs.

 

On DSM 7 and these models (and probably the others) there are only 3 governors to choose from, performance is the default set by Synology too.

 

Performance: scales up and down as needed

 

Powersave: stays looked in the lowest frequency

 

Userspace: can be set to a specific upper frequency and also scales down

 

As you can see /proc/cpuinfo doesn't show realtime results, which is also why a lot of people think frequency scaling and turbo boosting is not working even on a real Synology.

 

Per https://www.kernel.org/doc/Documentation/cpu-freq/governors.txt

 

2.1 Performance ----- The CPUfreq governor "performance" sets the CPU statically to the highest frequency within the borders of scaling_min_freq and scaling_max_freq.

 

Per https://wiki.archlinux.org/title/CPU_frequency_scaling#Scaling_governors

 

Governor         Description

performance   Run the CPU at the maximum frequency, obtained from /sys/devices/system/cpu/cpuX/cpufreq/scaling_max_freq.

 

Per https://documentation.suse.com/sled/15-SP3/html/SLED-all/cha-tuning-power.html

 

Performance governor

The CPU frequency is statically set to the highest possible for maximum performance. Consequently, saving power is not the focus of this governor.

 

Per https://wiki.gentoo.org/wiki/Power_management/Processor

 

Option                             Module                         Supported Processors                                                                               Note

'performance' governor cpufreq_performance Sets the frequency statically to the highest available CPU frequency. For Intel core processors, this should be selected as default.

 

 

[nemesis122]

The Problem is when i set this command after a reboot or when the server was switched off you have to set again ...

modprobe acpi-cpufreq && echo "performance" | tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor

 

[gericb]

14 minutes ago, idaanx said:

 

 

The serial validation is done in combination with a model. Your real serial is either from a different model or the template for generating and validating is incorrect. Check the script from TCRP to validate the template, this one is easier to read.

 

Maybe not a good idea to use QuickConnect with a fake Synology...

 

Yeah, I am not using a fake generated data, and not actually actively using QuickConnect....that was the previous guy. But I am 100% sure that the serial # that was flagged as being incorrect or invalid for the DS920+ is in fact correct, for that model.  I checked several times.  I just moved on, but wanted to mention the quirk.  I'll check it again for kicks and giggles.  Thanks for the feedback though! 🫡

[idaanx]

Just now, AuxXxilium said:

simply not possible with "performance", i wrote linux/android kernel over years. so you only see the requested freq, not the real. if see it in my ilo, if i change i have around 280 watts of power with "performance" and with the stock i only have about 70 watts. you there must be a bit more than this.

 

Could you be a bit more specific, what is the stock governor in your case?

[cpresco]

6 minutes ago, nemesis122 said:

The Problem is when i set this command after a reboot or when the server was switched off you have to set again ...

modprobe acpi-cpufreq && echo "performance" | tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor

 

 

You could try setting that with a Boot-up User-defined script in Control Panel/Task Scheduler.

[idaanx]

7 minutes ago, nemesis122 said:

The Problem is when i set this command after a reboot or when the server was switched off you have to set again ...

modprobe acpi-cpufreq && echo "performance" | tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor

 

 

That's by design as the fix is only for testing, if you want it permanently be creative with the task scheduler.

[AuxXxilium]

38 minutes ago, cpresco said:

 

Per https://www.kernel.org/doc/Documentation/cpu-freq/governors.txt

 

2.1 Performance ----- The CPUfreq governor "performance" sets the CPU statically to the highest frequency within the borders of scaling_min_freq and scaling_max_freq.

 

Per https://wiki.archlinux.org/title/CPU_frequency_scaling#Scaling_governors

 

Governor         Description

performance   Run the CPU at the maximum frequency, obtained from /sys/devices/system/cpu/cpuX/cpufreq/scaling_max_freq.

 

Per https://documentation.suse.com/sled/15-SP3/html/SLED-all/cha-tuning-power.html

 

Performance governor

The CPU frequency is statically set to the highest possible for maximum performance. Consequently, saving power is not the focus of this governor.

 

Per https://wiki.gentoo.org/wiki/Power_management/Processor

 

Option                             Module                         Supported Processors                                                                               Note

'performance' governor cpufreq_performance Sets the frequency statically to the highest available CPU frequency. For Intel core processors, this should be selected as default.

 

 

correct. static highest. i see it in power meter in ilo.

[nemesis122]

like this ?

 

 

[cpresco]

1 minute ago, nemesis122 said:

like this ?

 

 

 

Yes. I have the User field set as root on the General tab with a similar scheduled task.

[cpresco]

38 minutes ago, idaanx said:

 

Could you be a bit more specific, what is the stock governor in your case?

 

This thread could be relevant to this discussion about governors in Xpenology/TCRP/ARPL.

 

[idaanx]

52 minutes ago, AuxXxilium said:

correct. static highest. i see it in power meter in ilo.

 

Maybe you should read it again, and this time the important part.

 

2.1 Performance ----- The CPUfreq governor "performance" sets the CPU statically to the highest frequency within the borders of scaling_min_freq and scaling_max_freq.

 

The second link is about the intel_pstate scaling driver, so not relevant in this case.

 

See the default governor for cpufreq on the last link in @cpresco post.

 

'performance' → For Intel core processors, this should be selected as default.

 

Here is a dump of a real 220+ with the acpi-cpufreq governor set to performance.

 

 

And again, what is the stock governor and driver on your server? Please post your results of "grep . /sys/devices/system/cpu/cpu0/cpufreq/*".

 

Newer kernels have better scaling drivers but we are stuck with the acpi-cpufreq driver.

 

For my research I've also done some power draw tests with ARPL and the 920+ model, I've verified the top result with Ubuntu.

 

 

The burst frequency is reported as 1601 in the available frequencies but scales up perfectly to 2080. Again tested in Ubuntu with power draw and php test. The tiny differences in the idle state are from rounding errors.

 

Edited December 30, 2022 by idaanx

[Sebby75]

1 hour ago, idaanx said:

 

 

The serial validation is done in combination with a model. Your real serial is either from a different model or the template for generating and validating is incorrect. Check the script from TCRP to validate the template, this one is easier to read.

 

Maybe not a good idea to use QuickConnect with a fake Synology...

 

Well... serial and mac number I use is from real DS920+ unit that was damaged beyond repair a while ago...
When I enter that obviously correct serial in arpl builder app I get message that it is incorrect... Still allows me to accept/save it.

Eventually I will disable QuickConnect and forward ports that I need on the router.
https://kb.synology.com/en-me/DSM/tutorial/What_network_ports_are_used_by_Synology_services

 

[idaanx]

3 minutes ago, Sebby75 said:

Well... serial and mac number I use is from real DS920+ unit that was damaged beyond repair a while ago...
When I enter that obviously correct serial in arpl builder app I get message that it is incorrect... Still allows me to accept/save it.

Eventually I will disable QuickConnect and forward ports that I need on the router.
https://kb.synology.com/en-me/DSM/tutorial/What_network_ports_are_used_by_Synology_services

 

According to the validation test the serials for 920+ should start with 2030SBR, 2040SBR, 20C0SBR or 2150SBR. Is that the case for you?

 

I've tested it with a random key-mashing serial and still worked

 

You might want to take a look at Cloudflare tunnels if you don't want to open up ports, it's also free.

[Sebby75]

26 minutes ago, idaanx said:

 

According to the validation test the serials for 920+ should start with 2030SBR, 2040SBR, 20C0SBR or 2150SBR. Is that the case for you?

 

In my case it starts with 21C0TERxxxxxx