idaanx

The easy suggestion is, stick with what works. ARPL and TCRP have some road ahead before they're as stable and usable as DSM 6 and Jun's loader was. Another solution I can give you, try out a Hypervisor like Proxmox or similar and run your Windows and DSM in separate VMs.

I don't know what tool you're using here, but it shows options that aren't available in DSM, before you break your system don't use this tool in DSM. The governor issues are fixed since the updates from this year, mainly the misc add-on. When updating from within ARPL make sure to also update the add-ons.

When running DSM inside a VM the CPU management is handled by the Host (Proxmox). Since you've disabled Speedstep you might not get you're boost speeds and the rest of its benefits. The processor info is baked into the system and is the CPU info from the model you've chosen.

The governor is now loaded by the system again as intended, as it is on real Synology hardware. No need to load it yourself any longer, ie. the fix I've provided. Run the command below and you'll see it still loads the performance governor by default for all your cores. cat /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor

Just tested it on my main system, on Skylake and VM and now the module is loaded on my main system and not loaded on the other two as aspected. Thanks for the fix 👍 One small issue remains though. The ARPL loader/configurator still has the governor set to userspace with the lowest speed. This is only on the pre-Skylake system, Skylake and up use the intel_pstate driver with different settings, VM doesn't use this at all. The issue with this is that the configurator, the installation itself and the boot is unnecessarily slow, until the acpi-freq module is loaded in DSM. Setting it to 1 of the other 3 available governors will help, if possible. Another possibility is to set the scaling_setspeed inputs to the output of scalling_max_freq. This also has the result of showing this lowest frequency in the Info Center, looks like system is checking the speed before the module is loaded. This is just cosmetic but it is an indication it is still off at boot. I have also set a custom frequency in the configurator and booted from the menu with that frequency being displayed in Info Center and having a faster boot-time.

Same as everyone else. Load the the configurator, and in the ARPL menu go the update option... At the moment beta 9 is the latest, so no need to update.

Could you check if you have Intel Hardware P-state enabled in DSM by running grep flags /proc/cpuinfo | grep -m1 hwp No output means it disabled, with output it should be enabled. If disabled you should check your BIOS and enable Intel Speed Shift, or just might be called Hardware P-state. This allows the CPU to manage the Power states without the need of the OS to do so. My BIOS doesn't have an option to disable it so can't test what happens when it is off. Let me know it goes. Edit: Could you also run a Geekbench test, if you haven't already, maybe it's not related to this issue at all.

There is no output apart from the loader itself, a real Synology doesn't even have a video out.

Do what you want with that answer, I'm out...

No issues with the install no. New idea. echo nameserver 1.1.1.1 >> /etc/resolv.conf And run the nslookup command again.

First part is the file which loads the necessary modules, acpi-cpufreq still has a hash in front of it which normally disables the loading of it. This is done by the misc add-on. The second part, starting with xt_tcpudp are the loaded modules with a filter for the text cpu. As you can see it shows acpi-cpufreq in the second row from the bottom. No clue why it is loaded anyway if you didn't do anything.

Looks like the DNS settings on your ARPL is pointing to itself, don't know why that is. Just wipe the flash drive and flash it again with BalenaEtcher and try again, easier than trying to fix this mess. And if you haven't done so already turn off and on again your router and/or modem.

Where did you check it is 1500 and not 1501? You're still using model 920+ right, I'm wondering how your governor was set to performance? Can you post the result for cat /lib/modules-load.d/70-cpufreq-kernel.conf && lsmod | grep cpu

Just measuring the idle state is a bit pointless. Under some load the 6700 will easily exceed 50+ W, where as the J4105 will most likely stay under 20 W. For my N3150 the difference between idle and full load is only around 3 W, but a 65 W TDP chip like the 6700 will never have that. With the scores you previously posted for the J4105, it is very unlikely you only getting 1.5 GHz. If you're getting reports of the frequency being 1501 in DSM, in reality it is 2500 / 2.5 GHz. Can you post the results for the J4150 with grep . /sys/devices/system/cpu/cpu0/cpufreq/* And can you please answer the question if you used the CPU scaler script on the J4105 or set it manually?

The 3615xs loads the kernel module and sets the governor to performance like it is intended by Synology. Unlike the other models when the misc add-on is enabled, which disables this. There is no difference of setting it yourself with my previously provided command. Like I said Skylake and up don't need a fix and from my testing don't even work with the module. It might even brake your setup when used improperly. I've also told you before the scores for your 6700 look good and it works as intended. If you want to be sure Intel Hardware P-state is enable run the command below, when you get output it is. grep flags /proc/cpuinfo | grep -m1 hwp Any idea how your J4105 got good scores all of a sudden? Do you use the CPU scaler script on it?

The NIC works otherwise you wouldn't have gotten an ip from the router, 192.168.x.x are local handed out by the router. The ping in ARPL is a bit different, try "ping -w5 1.1.1.1" again. If you get a positive response it has internet, if not it doesn't. And restart your router and/or modem, and try to ping github.com or something else again. If you are able to ping 1.1.1.1 but a url doesn't work it's a DNS issue. Run "nslookup github.com" to see what DNS server is used.

Local network is not the same as the internet. If you don't know how to use SSH use a monitor and keyboard on the computer you run ARPL on and wait until it is done loading. On the console type "ping -t3 github.com" without the quotes and if it works you'll get a connection, if it doesn't work try "ping -t3 1.1.1.1". What do you mean with driver?

It fixes the part of the CPU freq being in the lowest state, in your case the 800 MHz you saw in the Info Center. Very likely being in the base clock, this is the advertised freq without the turbo speeds. On your E3-1246v3 this could be around a 10% lose, for me it's 20%. If you want all performance you still need to load the acpi-cpufreq and performance command.

Like the name suggests it boots directly into DSM without first loading the ARPL kernel . ARPL gets loaded, reboots for a clean slate and loads DSM, so no conflicts between the 2.

You could also try BalenaEtcher to flash the drive, works on Windows, Mac and Linux. It will also verify the data written to the flash drive. Are you sure the computer you are trying to install on, is connected to the web? Could you try to ping github.com on it, either directly or using SSH? Maybe restart your router, just to be sure...

Looks like you're in the wrong section, here you can find discussions and help for TCRP.

@fbelavenuto @pocopico Since posting this I've also discovered a probable cause for this. I've compared the results of the cpufreq directories in the ARPL configurator and noticed the driver is different. My main board with Intel uses the intel_cpufreq driver and the newer Skylake uses intel_pstate driver to manage the states. This uses something called Intel Hardware P-State (HWP) which was introduced with Skylake and is easy to test for and exclude the acpi-cpufreq module. The real 3622xs still uses pre-Skylake Xeons so makes sense that model loads the module. To test older Intel chips for frequency scaling there is Intel Dynamic Acceleration (IDA), which is also easy to test for. @Dreadnought just confirmed this in a way, since this even older Atom doesn't have the issue and also no IDA feature. Verified the CPU flags online. This just leaves AMD. I've extracted the the 1621+ pat file and checked the 70-cpufreq-kernel file, for this Ryzen model it still loads the acpi-cpufreq module, so should be good...

Just tested this, had to do a clean install again (was testing something with Jun and DSM 6.2 before). It is faster than before and a lot of people will think this fixes things. But I can tell you with certainty, it is still not fixed, it's basically the same as TCRP. The /sys/devices/system/cpu/cpufreq/ directory is still empty so the acpi-cpufreq is still not loaded. Resulting in the base frequency for my CPU instead of the full performance, just like TCRP. Like I said before in my detailed issue analysis, it is a flawed test in the misc add-on, this needs to be fixed.

This CPU doesn't support the needed scaling, just checked its features. So you're good.

Please don't! Like I said it just messes up your settings, and at the time I didn't know how else to fix the performance yet. Just run the performance command on your NAS, if you have low Geekbench scores on it. It should work on this CPU, if it doesn't a reboot will put it back to normal.