OK, so just tried unRAID on a few machines here, results as follows:

 

AMD FX8320E (Vishera) on 760G chipset board (Asus M5A78L-M/USB3) - CPU throttles down to 1400MHz OK, doesn't turbo to 4GHz.

 

Intel Xeon E3-1225v3 (Haswell) on C226 chipset board (Lenovo TS140) - CPU constantly pegged at 3600MHz, never throttles, stuck in turbo

Intel i3-4350 (Haswell) on B85 chipset board (MSI B85M Eco) - CPU constantly pegged at 3600MHz, never throttles

Intel [email protected] (Devil's Canyon) on Z97 chipset (Asus Maximus VII Gene), CPU sits at 3.5GHz, never throttles or turbos

 

Intel i3-2120 (Sandy Bridge) on B75 board (Asus B75M-Plus) - CPU throttles down to 800MHz OK

Intel Celeron G1620 (Sandy Bridge I think?) on B75 board (same B75M-Plus) - CPU throttles down to 800MHz OK

 

 

So, there we go, seems unRAID can't cope with Haswell chips.

I guess you didn't read the link a gave you in the other CPU throttle thread.... 

This is an issue with the Intel-Pstate scaling driver, which is caused by the CPU state being to sensitive and causes the CPU and always run at max.

 

If you disable the Pstate driver, the apci-cpufreq scaling driver will take over, and handle the scaling correctly.

 

More details can be found in this thread.

 

This is an issue with the Intel-Pstate scaling driver, which is caused by the CPU state being to sensitive and causes the CPU and always run at max.

 

If you disable the Pstate driver, the apci-cpufreq scaling driver will take over, and handle the scaling correctly.

 

More details can be found in this thread.

 

The pstate driver has been fixed in beta13.  We manually patched it.

This is an issue with the Intel-Pstate scaling driver, which is caused by the CPU state being to sensitive and causes the CPU and always run at max.

 

If you disable the Pstate driver, the apci-cpufreq scaling driver will take over, and handle the scaling correctly.

 

More details can be found in this thread.

 

The pstate driver has been fixed in beta13.  We manually patched it.

 

That's really good news! Thanks!

Cool, will try upgrading to b13 and see how it goes. 

As a Haswell owner, I am still seeing high cpu frequencies at idle with b13. I have reported it in the other thread.

Yes, we thought we had it resolved, but apparently only on certain systems.  We are still tracking this one.

So you've replicated the issue OK? 

 

I'm planning on moving my unRAID to my Haswell machine, but I'd rather not until his issue is resolved.  The AMD sucks power, but actually sucks less than the Haswell running it's cores at full speed. 

 

 

So you've replicated the issue OK? 

 

I'm planning on moving my unRAID to my Haswell machine, but I'd rather not until his issue is resolved.  The AMD sucks power, but actually sucks less than the Haswell running it's cores at full speed. 

 

The oddest thing is that even though the cores are reporting faster speeds, they are still entering into a C7 state.  Yes we can replicate the issue on haswell systems.

Over a month since the last reply on this thread, I was wondering how you're getting on with this issue?

 

I'm still keenly waiting for 6rc1 or 6.0 to be released.  Currently running OMV in the meantime on an older machine, because I don't want to fully deploy unRAID until this issue is fixed.

 

Going by the power consumption, I don't think my CPU is going into C7 at all, because it's still using close to 40W with the drives spun down, I expect more in the high 20W area (22W is typical, though I have seen 19W in Windows 2012R2 on the odd occasion).

No update just yet. Intel hasn't responded to us on the issue. Since it doesn't affect system stability, we aren't going to delay rc / final to wait for this to be resolved.

That might explain why my celeron G3258 is saying 3200mhz all the time, yet sucking 15-20 watts less than my AMD A6-6400K Richland build that did throttle down.

No update just yet. Intel hasn't responded to us on the issue. Since it doesn't affect system stability, we aren't going to delay rc / final to wait for this to be resolved.

 

Fair plan.  It's not a massive issue, really, it's more my OCD than anything.

 

Any idea when RC is due?

Any idea when RC is due?

 

Soon. Stay tuned for more info later.

Do not mean to rehash an old topic, but I am experiencing this issue as well on 6.1.3. Any news/anything for me to do/try?

Interesting that it's still showing that.  I believe this is a reporting issue -- i.e. that the CPU is actually throttling down okay, but it's not being reported correctly ... but I can't find the thread where this was discussed in more detail and that was stated.    If you check your power consumption with a Kill-a-Watt, I believe you'll see that it's clearly drawing far less than max power when the system is idling ... a good indication that it's indeed just a reporting problem and not a real issue to be concerned about.

 

This is absolutely still an issue, and still needs more investigation to fix.

I have done tests with my kill-a-watt and recorded ~15 watts higher after an extended period of time with the server doing nothing.

I had also noticed higher CPU temps.

 

If you poke around through the threads you will see that many people will recommend the intel_pstate=disable to resolve this issue.

I have had to use this in my syslinux.cfg for a long time now, and every time I remove it and think it may be resolved, it is certainly not.

 

I have done tests with the GUI loaded, not loaded, server idled for a long time, etc... Checking status with htop and whatever other commands as suggested through SSH always shows my cores maxed out, all 8.

 

I also notice that in reporting with the intel_pstate=disable, that my CPU never reports its turbo frequency, only it's base max core speed.

Without the intel_pstate=disable, my cores all show up as the turbo speed.

 

I have been wanting to post a new defect thread in order to bring more attention to this issue, however have not prepared the information as much as I'd like to open that thread at this point.

 

In summary, no command through SSH or GUI ever shows anything other than max without the intel_pstate=disable for my configuration.

~15 watts extra at idle, consistently, and additional CPU temps reported.

 

Main thread is here http://lime-technology.com/forum/index.php?topic=36575.0

More topics here http://lime-technology.com/forum/index.php?topic=41805.msg397144#msg397144

and here http://lime-technology.com/forum/index.php?topic=42063.msg400364#msg400364

I can report using the pstate=disable shows the CPU throttling via the unRaid GUI and the terminal command line. I drop from my stock to around 900-1000mhz, on an e3-1245v3. My desktop which uses same chip gets as low as 800, but obviously has far less going on.

 

Power consumption is less with pstate=false. I was averaging 150 watts when sitting with low CPU utilization before, I now sit around 100 watts. (I still wonder where this additional power is coming from. My desktop idles around 40). I suppose ~10-20 per video card(have 2), then 10-20 for the raid controller. A couple additional fans...