Question Regarding Turbo Boost Voltage

Toggle sidebar Toggle sidebar
Y

Yeoman1000

Honorable
Aug 7, 2012
67
0
10,630
First off, if this in the wrong place, my apologies, but Turbo boost is a form of overclocking so I thought this would do.

For as long as I've had my 2500k, I've been running it with turbo boost disabled (long story), and until only recently have I enabled Turbo boost. I gave it a whiz in Intel burn test just to look at temperatures before/after.

Disabled they're around 57C, and with turbo boost enabled they're at 68C. I was using hwinfo to monitor. I was quite surprised at the large temperature increase for a 0.4GHz increase, but then I realised it must be down to voltage.



When I'm in the Bios, if I recall correctly, the CPU voltage is at 1.200 at 3.3GHz, . Hwinfo records the VID (I'm not sure if this is the same thing) at 1.251v and 1.351v disabled/enabled respectively, whilst doing a Standard 10 runs of IBT.

Now, finally to my question, and I'll just mention that overclocking is something I have never tried and have little knowledge of.

The voltage with turbo boost enabled seems quite high, is it normal for it to be so high? I've seen before people mentioning that they're overclocked to 4.3 on >1.300 volts. I know that turbo boost might be different when the voltage is left on auto, as opposed to a manual overclock, but should I be worried at how high the voltage is?

Edit: I have an Asus P8P67
 
The turbo boost is designed to provide extra performance while keeping the processor within its TDP, not to use the lowest voltage possible. Sure it's possible to get 4.2+ Ghz on 1.3 volts but it may not be completely stable and over time it will become unstable. Intel warranties their processors as being stable so the stock settings for normal and turbo operation leaves quite a bit of voltage headroom at all times.
 
Okidoke, thanks for that.

I'll probably leave it disabled until I would want to overclock it manually myself (if i ever choose to).

Why would such an overclock become unstable overtime? Would it just be general 'old age', or does the overclock (even a modest one) quickly eat away at the processors 'endurance'?

 


Processors degrade over time due to electromigration, a process that degrades the conductive electrical paths. It's very similar to the way engines in vehicles wear out due to friction. Electromigration occurs in all electrical devices but is particularly strong in semiconductors due to their drastically lower conductivity. It also scales exponentially with voltage and temperature. This is why overclocked devices will fail sooner than devices that are run within the manufacturer's specifications.

Intel warranties their processors for 3 years so long as the device is kept within the electrical and thermal specifications set out in the manufacturer's datasheet. Despite being warrantied for 3 years, most should have at least a 10 year lifespan which means that they will be replaced long before they fail. Mild overclocking such as that performed by turbo boost might reduce this to 8 years of moderate usage, still long enough to last through the warranty and be replaced by the overwhelming majority of users. Moderate overclocking such as the common 4Ghz+ on current generation Sandybridge chips should still see between 3 and 5 years of lifespan before failure. This kind of overclocking is not covered under normal factory warranty but can be covered with the Intel performance tuning plan.

Excessive overclocking beyond 4.5Ghz will often reduce the lifespan to less than 3 years with many failures happening within 2 years or less on lower end chips. Most people who run their components in this range replace their computer every couple of years anyway.

Extreme overclocking in the 5Ghz+ range will result in failure within months on most chips.

The lifespan of each chip depends on the quality of the chip itself and the condition it is run in. Temperatures below 70 degrees centigrade are fine at all times but temperatures beyond 85 should not be sustained for long. Similarly, voltages above the absolute maximum specified by Intel (absolute maximum voltage is different from the maximum standard voltage) should not be used without extreme caution.

Overclocking can be fun and it's a great way to get to know how PC internals work together. However, it has its risks and the risks are all too often ignored on these forums. If reasonable care is taken then the risks can usually be mitigated.

You can leave Turbo Boost on if you'd like. It's pretty harmless and its operation is warrantied by Intel.
 
That is high voltage for turbo, mines only 1.21 enabled. Maybe you're mistaking the auto ocer. Use cpuz to see your actual speed. Also turbo is only .1 ghz on all cores.

If you got those lifespan expectations from release datasheets, current reports are showing they're lasting much longer. The average oc is ~4.5ghz and is expected to last 4-5 years. There have been a couple of posts of 4.7ghz+ becoming unstable and needing a voltage raise but there are still people running 5ghz since day 1 and it's been over a year.
 
My data is several years old but it really hasn't changed that much. Intel's EIST greatly reduced the degradation of chips by running them at full speed only when the load requires it. 5Ghz 24/7 is different than 1600Mhz 20/7 and 5Ghz 4/7. The multiple multipliers change things a bit.

When I say "most chips" I mean "most chips" and not "top binned performance chips". I also don't discriminate between AMD and Intel and prefer to err on the side of caution. Naturally, anyone who has the time and money to invest in a proper overclock will be able to squeeze significantly more performance and lifespan out of the product.

Sandybridge has only been out for a year and a half but the failure distribution isn't flat, it's going to be heavily clustered around a timeframe which is still in the future. Given that Intel's IMC appears to be the weakpoint in many chips (not surprising considering that the memory controller was also the weakpoint on most LGA 775 platforms) I wouldn't be surprised if there's a huge influx of failures 12-18 months from now when Haswell hits the market.
 
Thanks for the informative post Pinhedd. I don't think I want to overclock any more then haha.

Hmm, I did a quick run of intel burn-test, using CPU-Z, this is with TB disabled. The core voltage reached a max of 1.192, and this VID was at 1.251 again. I'm guessing this VID thing isn't the same as core voltage? Or if so, why is it showing such large differences?
 
VID is vcore, my guess is hwinfo is getting incorrect readings. You could use other monitoring software like hwmonitor or core temp to make sure what is getting a correct reading. Did you see what speeds you were running at? And just making sure, but you didn't change any other settings did you?

I don't want to hijack the thread with a lifespan discussion, but don't be afraid to oc, even if you only do it in the later part of it's lifespan.
 


VID is the automatic vcore. The VID is an 8 bit value which is presented by the CPU to the VRM and is programmed into the CPU's microcode during binning. The VRM uses this value to determine what voltage to provide to the CPU.

The VCore will fluctuate a little bit and will drop under load to prevent damage. Thus, VCore should be slightly below VID.

Separate VID values can be used for idle / standard / turbo
 
The speeds were 3300mhz stock and 3700mhz when TB is enabled. If I use burn test at 3300mhz, CPU-Z shows a max vcore of 1.192, whilst Hwinfo (I put it up for temps) shows the VID as 1.251.

At 3700mhz (TB enabled), CPU-Z is showing a max Vcore of 1.280, whilst Hwinfo records this VID at 1.351.

I saw it described somewhere that VID was some sort of reading for the 'requested voltage', and the Vcore was the actual voltage being received by each core. But, this is something I know anything about.

I was thinking I might have a go at overclocking at some point. I was thinking that I might set a manual overclock for say, 3.7ghz, and then at least I've learnt the process; I can undo or increase it if I choose. Although, when I looked in the bios I couldn't see how to alter my cpu multiplyer.
 
Right but it's simpler to just look at vcore since that is what will be reported/set in bios. Are you only testing 1 core in ibt because it will only go to 3.7 if only 1 core is used. Otherwise you are overclocked and would explain why auto is setting the vcore higher.
 
Oh. In Hwinfo, all four cores are at 3700 mhz. So it must be auto overclocked?

That actually makes a little sense. I think when I first put my pc together, I was confused by the motherboard manual, and I had my TPU switch on I think(P8P67), because I recall in the first week of building it; I had trouble with high temps. I was looking in the bios, and saw that it was aiming for a speed of 4300mhz, I remember being surprised by this, but I just switched it down to a x37 multiplyer, as I knew that was what turbo boost went up to. Shortly after that, I turned turbo mode off anyway I think.

But it is likely my cpu has been overclocked using the tpu switch for a long while then?

Yeah, I just got out my manual, and I can see through my case window that the TPU switch is on. In the manual it says "For ensuring the system performance, turn the switch setting to enable when the system is powered off". I remember being really worried that the whole thing wouldn't work if I didn't listen to the manual and put my TPU & EPU switches on. Silly.

How to I easily get my bios back to a default non clocked state? Just, turn off, disable the TPU, and then set to restore factory settings in the BIOS?

 
My EPU switch is also set to enable, would you recommend disabling this? Or is this a relatively harmless feature implemented by Asus?

Also, since I have had turbo mode disabled for pretty much 99% of the time I've had my PC up and running, would having the TPU enabled still have been 'affecting' my cpu/motherboard even with the Turbo Boost feature disabled? As in, some of the settings/voltages that might have been changed for the purpose of OC'ing Turbo boost would still be in their changed state if TB was disabled?

The reason I'm asking, is that I've been plagued with infrequent BSOD's since completing my build, which until yesterday, I thought had been fixed. I'm just wondering if it could be down to some unstable overclocking thanks to enabling the TPU switch.

Thanks for your help so far btw guys, and sorry for the wordy questions.




 
Okay, I disabled the TPU switch on my P8P67, and loaded optimised defaults in the BIOS. After saving and restarting, the CPU voltage in the Bios has dropped from 1.200 to 1.126. Interesting.
When stress testing in IBT, CPU-Z records the voltage at 1.144v Max, whilst Hwinfo is still seeing 1.271v for the VID. I guess that's normal.

I then went back in and set my X.M.P profile in the ai-tweaker thing, I noticed this automatically overclocks turbo mode, so I set the option back to Auto regarding Turbo Boost core multipliers.
CPU-Z is still seeing 1.144, although I caught very rare climbs up to 1.152. I noticed also, that the down-clocked voltage was a little higher (0.896 vs 0.928) after enabling my X.M.P


However, I have noticed an odd quirk in my pc now. When restarting either from bios, or windows, or just before logging into windows, it has a tendency to shut-down, before booting up again. Any ideas on how to fix this?
 


The double shutdown is a common effect on Asus motherboards. I'm not sure what causes it but it seems to be related to the EPU and TPU DSPs. It is expected behaviour, nothing to be worried about.
 
Okay, I see. Thanks for that.

I might try disabling the EPU switch, I'm assuming it doesn't really do anything useful. A long time ago I had a similar problem, but it was when booting the PC up. disabling PLL Overvoltage fixed it that time. I guess i'll just have to experiement.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom