3.4GHz CPU stuck at 1.6GHz

I am sort of beside myself at the moment. I have the sig rig with a Xeon E3-1240V2 3.4GHz processor. It is stuck at 1.6 and nothing I do in the BIOS ever fixes it. I even reflashed the BIOS and used all defaults same thing. Running the intel burn test and the CPU at 100% I don't even pass an average of 50C! Any ideas?

any power saver app?, or power saving mode?.. same issue happens with my cousin Laptop, they have a third party program that had the CPU max power saving mode on their i3.. so the CPU was always locked at 800mhz.. checked at bios fine, checked energy management fine, later i noticed the program "Super Posh" a third party app that monitors all in the whole system.... uncheck the CPU power saving. and done.. CPU unlocked to work as should work..

A long shot,

What's the maximum processor state set at in the Windows power options?

Control Panel --> Power Options --> Change plan settings (on currently selected option) --> Processor Power Management --> Maximum Processor State (in %)

Min and Max processor state are at 100%...wonder if it is the CPU or the board having an issue.

Try settings to high performance and rebooting, sounds like a windows issue.

Boot from an Ubuntu live CD and check it there... If it works properly, you got issues with Windows.

I tried a Windows 8 and a Windows 7 installation on separate drives - both have that problem. I'll do a live CD tonight and see.

ThrottleStop 5.00
http://www.techinferno.com/downloads/

The bios might not have any idea what a Xeon E3-1240 is so it will set the CPU to a safe multiplier which is the minimum 16 multiplier. Give ThrottleStop a try to see if it can help you out.

The other problem is that sometimes the board will send the CPU a BD PROCHOT signal which is basically something on the board telling the CPU that it is too hot so it forces the CPU into thermal throttle mode which locks the multi to 16. It's usually just a bad sensor. You can also use ThrottleStop to disable the BD PROCHOT signal if that is causing the problem.

Here's a guide if you need some help but it's a lot easier to just start the program and try clicking on a few options.

http://forum.notebookreview.com/har...arket-upgrades/531329-throttlestop-guide.html

Click on Set Multiplier and set that to Turbo and then click on the Turn On button. If that doesn't fix things, uncheck the BD PROCHOT option. If you are still having problems, post some screen shots of ThrottleStop so I can see how you have it setup.

Thanks for info on Throttlestop I will try it once I get home - for giggles I ran it on my work pc it did the TS Bench 1024M in 292.042 seconds. Lit that proc right up. Hopefully will work on the home PC.

I'm at the house now and here is what ThrottleStop says initially:


After turning the BD PROCHOT off everything goes back to normal.



Did a sensor go bad on my board ... wth?

Windows 7 same thing as Windows 8


Wonder if I will always have to keep that program running now...wth.

Is your Xeon processor on the motherboard's compatibility list? Sure sounds like the BIOS can't recognize it and makes a SWAG at what the multiplier should be.

It is on the compatible CPU list for that board. Plus it worked fine until the last week or so.

I started an RMA with Gigabyte yesterday but it is still in the waiting for approval stage.

BD PROCHOT stands for bi-directional processor hot.

This feature allows other sensors on the motherboard to send a throttling signal to the CPU. For example, if the voltage regulator was running too hot or drawing too many amps, it would send a signal to the CPU to tell the CPU to throttle back to its minimum speed. This in theory would allow the voltage regulator to cool down and then after a while, it would stop sending this signal to the CPU so it could get back up to full speed.

It is likely that this sensor is not working correctly so it is constantly sending this signal to the CPU. As soon as you uncheck BD PROCHOT in ThrottleStop, this tells the CPU to stop listening to this Processor Hot signal that is being sent to it. When you disable BD PROCHOT, your CPU will still slow down if it ever gets too hot and needs to. Disabling this signal path simply prevents anything else from throttling the CPU.

ThrottleStop has an INI option that is useful.

ExitTime=5

This will force ThrottleStop to exit 5 seconds after it starts. You can use ThrottleStop to disable BD PROCHOT and then automatically exit so you won't have to worry about this anymore. You can also add ThrottleStop to your Start Up sequence by using the Task Scheduler. Go to the second post in the Guide thread if you need an example of how to do that.

Glad to see your CPU running like it should.

With this problem, I do not think a future bios version will change anything unless this broken sensor is disabled.

Edit: Try doing some single threaded benchmarks like Super Pi Mod and keep a close eye on the multiplier that ThrottleStop reports while the bench is running. On some locked CPUs, you might need to have at least C3 enabled in the bios in order to get full Turbo Boost. A bench or two and watching the multi should be able to confirm if you need to do this or not.

Super Pi Mod 1.5
http://www.techpowerup.com/downloads/366/super-pi-mod-v1-5/

From the ubuntu live cd I am able to see the processor bump up to normal speeds. Cores idle at 1600MHz but when I launch somthing I will see some jump to 3800MHz. So wtf happened in windows lol. Going to try a totally fresh install and see what happens.

ubuntu@ubuntu:~$ cat /proc/cpuinfo | grep MHz
cpu MHz : 3801.000
cpu MHz : 1800.000
cpu MHz : 1600.000
cpu MHz : 3801.000
cpu MHz : 1600.000
cpu MHz : 1600.000
cpu MHz : 1600.000
cpu MHz : 1600.000

Depending on the motherboard, turning off the BD PROCHOT bit might last until you completely shut down your system. If you reboot to an Ubuntu CD, without shutting down first, BD PROCHOT might still be disabled. You can test for this by enabling or disabling BD PROCHOT in Windows before using Ubuntu.

Do you have SP1 installed in your Windows 7 install? Your GFlops look a little low, even for 1600 MHz so I was just curious. It might also be because you are using a very old version of IBT which probably has old Intel library versions.

IBT 2,54
http://www.xtremesystems.org/forums...urnTest-The-new-stress-testing-program/page43

scroll down to post 1074

Stasio does a good job of keeping the Intel libraries up to date. Not sure if his download link is still working in that post.
You need the new libraries to give modern Ivy based CPUs a good work out.

He's been keeping them updated on TweakTown: http://forums.tweaktown.com/gigabyt...system-info-benchmarking-stability-tools.html

Personally I don't use Linpack; instead, I favor Prime 95.

Well a absolutely fresh windows 7 install has the problem still. Going to check the power off ubuntu angle.

The problem you are having with BD PROCHOT is usually a hardware problem. It's possible that Ubuntu automatically turns BD PROCHOT off since this feature can cause more problems than it solves.

MSR 0x1FC - Bit[ 0 ] keeps track of whether BD PROCHOT is enabled or not. You can use my MSR Tool in Windows to see how this register is initially set up in Windows.

MSR Tool
http://www.mediafire.com/?myjkxzkzzmd

Enter 0x1FC in the MSR Number box at the bottom and then click on the Read MSR button. You can use ThrottleStop to toggle BD PROCHOT and then read this register again to see how it changes.

I don't know much about Linux but I think you can go to the command line and type something like this to accomplish the same thing.

rdmsr 0x1FC

If MSR 0x1FC - Bit[ 0 ] is cleared in Linux then it is being turned off by Linux.

Fresh boot of linux after power off on the power supply for 5 minutes shows
ubuntu@ubuntu:~$ cat /proc/cpuinfo | grep MHz
cpu MHz : 1600.000
cpu MHz : 1600.000
cpu MHz : 3500.000
cpu MHz : 1600.000
cpu MHz : 1600.000
cpu MHz : 3800.000
cpu MHz : 1600.000
cpu MHz : 1600.000


the register thing in Ubuntu is 2 - not sure what that means yet.

root@ubuntu:~# rdmsr 0x1FC
2
root@ubuntu:~#

Left has BD PROCHOT checked in ThrottleStop and the Right has it unchecked.


Results after:

In Linux MSR 0x1FC = 0x2

0x2 = 0010b in binary

bit[ 3 ] = 0
bit[ 2 ] = 0
bit[ 1 ] = 1 - C1E is enabled
bit[ 0 ] = 0 - BD PROCHOT is disabled

That explains why the multiplier works correctly when you are in Linux. This OS has disabled the BD PROCHOT feature.

The last digit in MSR 0x1FC when you are in Windows is 0xF. That means all 4 of the above bits are set so when you boot up in Windows, BD PROCHOT is set, a bad sensor tells your CPU to throttle so the CPU throttles.

When you use ThrottleStop to disable BD PROCHOT, the last digit in 0x1FC changes to 0xE.

0xE = 1110b in binary

bit[ 3 ] = 1
bit[ 2 ] = 1
bit[ 1 ] = 1 - C1E is enabled
bit[ 0 ] = 0 - BD PROCHOT is disabled

and now your CPU is allowed to use Turbo Boost in Windows too.

You should update CPU-Z to the latest version. Better yet, download RealTemp because it does a lot better job of tracking your multiplier, especially when your CPU is lightly loaded.

RealTemp T|I Edition
http://www.overclock.net/t/1330144/realtemp-t-i-edition

You are not the only person with this motherboard bug. The next version of RealTemp will let you disable BD PROCHOT. RealTemp and ThrottleStop share a lot of the same code.

I really appreciate the help everyone. This was quite the annoyance. I was about ready to toss my titan out the window when this first started and in world of tanks I was getting 20FPS. I suppose the sensor just goes bad, I run my board on a techbench. It isn't even in a case so it should never be overly hot.

I downloaded IBT 2.54 from TweakTown.

http://forums.tweaktown.com/gigabyt...system-info-benchmarking-stability-tools.html

Thanks SolidBladez for posting that link on the previous page and thanks to Stasio for updating those links on a regular basis.

An Intel® Xeon® Processor E3-1240 v2 CPU has a default multiplier of 34 and uses maximum Turbo Boost multipliers of 38, 38, 37, 36 depending on whether 1, 2, 3 or 4 cores are active. When running IBT, all 4 cores are active so the maximum CPU multiplier should be 36. If the Turbo Boost power limits are set high enough, you should see a constant 36.00 multiplier as reported by ThrottleStop for all 4 cores while IBT is running.

I don't have a Xeon handy so I locked my Core i5-3570K to the 36 multiplier so I could get some comparable results. I also bumped the multiplier up to 45 to show what the IBT benchmark numbers look like when overclocked.





The multiplier was steady at 45.00 when overclocked but I took the screen shot a few milliseconds after the benchmark had ended so you can see the multiplier and C0% load numbers starting to drop.

IBT / LinX / Linpack testing will show higher GFlops on CPUs that do not use hyper threading. When hyper threading is enabled, the GFlops drop about 10% so that explains some of the difference but there is still a significant difference between the Xeon and the 3570K. If you are bored, try running another test and snap a screen shot of IBT and ThrottleStop during the 9th test before it ends. If you are not getting the full 36.00 multiplier then you can try to increase the Turbo Power Limits in the bios. If this feature is not locked by the bios, you can also try increasing these power limits in ThrottleStop.

Always interesting to see this unique problem fully exposed. As I mentioned, you are not the first person with this issue so maybe someone else will end up at this thread and be able to solve their problem without having to return or without having to buy a new board.

Edit: My best guess is that your CPU, with hyper threading enabled, should be able to run at 80 GFlops in IBT 2.54.

Ok ran it and I get in the mid 70's.
Here is during test #10


Here is about 5 seconds after test 10 finished.

Check if you have a current limit in BIOS settings. It can also be caused by over heating, or a bad thermal detector. I once had a 3930K setup that is locked at base freq. The software says that the CPU is about 130C which is impossible coz the cooler is not even warm.

When BD PROCHOT is disabled, your CPU is running at its full Intel rated speed. With hyper threading enabled, I was expecting to see about 80 GFlops so 79.0957 GFlops looks normal.

If you look at my results, each of the individual GFlops for the 10 tests are very close to each other. That's because my system is very lean with not a lot of background tasks needing to be processed. Your C State percentages when idle also confirms that you have more sidebar gadgets and other junk running in the background on your system.

The Intel docs show that when 4 cores are active, the maximum turbo multiplier for your Xeon E3-1240 is 36. Your results across all 8 threads of 36.00 confirms that you are getting full Turbo Boost at full load so there is no need to adjust anything. You are at the maximum.

Some of the non K CPUs have a built in overclocking feature where they can be overclocked by a maximum of 4 bins of Turbo Boost. I am not sure if your Xeon supports overclocking or if your motherboard bios can access this feature if it does exist. Can you use the MSR Tool and post a screen shot of MSR 0x194 or just copy and paste the values for that MSR. This will show if your CPU supports limited overclocking. Also post the value in MSR 0xCE. This second register shows if this overclocking feature is locked or not.

Your turbo multis are 38, 38, 37, 36 so if your CPU supported overclocking, it might be able to go as high as 42, 42, 41, 40 which would be a nice free boost in performance if possible. MSR 0x194 will show how many bins of overclocking are possible. The Xeon is intended for business and 100% stability so overclocking might not be possible.

Without Hyperthreading I did it again..here is during test #9:


And a few seconds after:


The MSR Registers you were curious about:




So the 0x194 register has 1 not F so I would guess from your post here: Scroll to post 135 it does not support overclocking in that manner? Haven't been able to find a reference to show me what each register means yet.

Trying to find info in this but its a lot to read lol. http://www.intel.com/content/dam/www/public/us/en/documents/datasheets/xeon-e3-1200v2-vol-1-datasheet.pdf

The 1 in MSR 0x194 means the maximum non-Turbo ratio is adjustable but in my opinion, this is a useless feature. This lets you lower the max non-Turbo ratio but it does not let you increase the non-Turbo ratio above the CPU's default value. This feature doesn't give you any control over the turbo ratios.

Bits 19, 18 and 17 in MSR 0x194 will show you how much turbo overclocking a CPU supports. If these bits were all enabled, combined with bit 16 enabled, that's when you would see an F in the middle of this register. That indicates the CPU supports unlimited turbo overclocking. You see an F in this register with the desktop K or XM mobile CPUs. Unfortunately in your Xeon, none of those bits are set so that means your Xeon does not support turbo ratio overclocking.

Bit 28 in 0xCE is not set and that also confirms that the turbo ratios are not programmable. Bit 29 shows your turbo TDP power limits are programmable but you are already getting full turbo boost when fully loaded so there is no need to increase your turbo power limits. If these values are locked in the ThrottleStop - TPL window then the only way to adjust them is in the bios. You could go lower to reduce the amount of turbo boost but going higher will not gain you anything since you are already at the max for your CPU.

Your 94 GFlops in IBT with hyper threading disabled is right in line with what I am getting on my 3570K when run at the same 3.6 GHz. Perhaps your extra 2 MB of L3 cache is giving you a slight advantage.

I am starting to get a little OCD about useless background junk running on my system. The RealTemp - C States feature is a great tool when trying to hunt down useless Windows background junk as well as poorly written drivers.



The individual cores are averaging 99.5% of the time in C6 and the entire CPU package is spending 97.5% of the time in C6. In the C6 low power state, cores are taken off the power grid so the core voltage is reduced to basically zero which is far lower than any offset voltage that enthusiasts like to talk about. As soon as any load needs to be processed, the CPU is back up to 4500 MHz for a nice combination of good performance and low power consumption. On my system, a high C2 package percentage when idle was fixed by switching my keyboard and mouse from a USB2 port to a USB3 port. I guess the software engineers are putting a little more time into driver development these days.

Unclewebb - thanks for all the information this has been quite interesting. I recieved an RMA number from Gigabyte but I went ahead and picked up an Asus Maximus V Formula/ThunderFX from Amazon Warehouse so I wont be without a system. I am considering picking up a good OC chip and see how it does vs the Xeon.

It will be interesting to see your Xeon running on a different board. Hopefully that cures the BD PROCHOT issue. My Asus P8Z77-V Pro board boots up into Windows with BD PROCHOT enabled by default but it is not causing any problems. I am very happy with this board. A few minor adjustments in the bios and it had my 3570K stable at 4500 MHz. All I had to do was set the Turbo multipliers to 45 and increase the voltage and that was about it. I also adjusted my memory from 1333 MHz to 1600 MHz. I remember the Core 2 days when I was spending months trying to get the most out of a CPU. Now it is almost too simple to build a fast and reliable system.

If you need hyper threading but do not need the Xeon's ECC memory feature then for the money, you might as well get a 3770K. They sell for a similar price and even a really bad 3770K will outperform a Xeon E3-1240 V2. Probably an even better idea to wait another month and get a Haswell based system.

Here's an Intel video about the benefits of ECC.

http://www.youtube.com/watch?v=UEGJCAnDr5E

Well I do a fair bit of VM work so I do have need for the VT-d.

Too bad Intel left VT out of their K CPUs. The best non-Xeon choice is the regular Core i7-3770 which supports VT.

http://ark.intel.com/products/65719/Intel-Core-i7-3770-Processor-8M-Cache-up-to-3_90-GHz?q=3770

I am pretty sure these CPUs also support overclocking with +4 bins of additional Turbo Boost. If you are interested, find someone with a regular 3770 and get them to run the MSR Tool and have a look at MSR 0x194. You should see the digit 9 in that register where you are currently seeing 1.

1 0 0 1 binary = 9 hexadecimal

Bits 19, 18 and 17 should show 1 0 0 which corresponds to 4 bins of turbo boost.

This Microsoft tool should show if VT is enabled in a CPU.

http://www.microsoft.com/en-us/download/details.aspx?id=2896

Might have to do a search at the Microsoft site for VT.

Well here it is on the new motherboard:
Before starting


And after


Temps were pretty good with the True on there too.

good to see that you are back with us! I just had a similat issue but mine was a multiplier issue. Good Luck!