Go Back   Overclock3D Forums > [OC3D] Processors & Platforms > Overclocking
Reply
 
Thread Tools Display Modes
 
  #1  
Old 29-09-12, 01:21 AM
Master&Puppet's Avatar
Master&Puppet Master&Puppet is offline
OC3D Elite
 
Join Date: Sep 2011
Location: London
Posts: 1,923
LLC and Vcore

This is some research I’ve been doing on LLC and overclocking, I don’t think it is a particularly well explained or well understood topic so I hope a few people find it useful. I certainly have been learning whilst experimenting (and my pc still works too – bonus)!

At this particular moment I am running my 3570K at 4.6GHz. If I were to say that I had the choice between running the Vcore at 1.32v or at 1.40v (and both being stable) I’m sure that most people would know which one is the obvious choice and would therefore be surprised to find that I chose 1.4v. Why? Well, believe it or not, manipulating the LLC gets me lower temperatures at 1.4v than at 1.32v.

Setting a voltage in BIOS is only a fraction of the story. If you set 1.32v into BIOS then that doesn’t mean that the CPU simply runs at 1.32v. In fact 99.9% of the time the CPU will be running on voltages which are completely different yet still very much related. What you’ve set is more like a target which the motherboard doesn’t want to exceed.

When the CPU is running it demands power from the motherboard. The motherboard carefully supplies power as a reaction to this demand but, because it is a reaction, the actual change in power delivery is always a bit late. This means that the CPU is very briefly supplied with more power than it needs - a power spike. You won’t see these on any hardware monitor because they happen so fast but they do appear on an oscilloscope.

Power spikes are bad. In order to protect itself from any potential damage the system will run the CPU underneath this limit. Therefore when these spikes happen they also remain underneath the limit that has been set. This is called the Voffset. When the CPU is working hard and demanding more power the size of these power spikes will increase so Intel added a second function called Vdroop which simply increases the offset, by another name, when the CPU is under load so that these larger spikes still remain under the limit. Here’s a graph:



Not too long ago enthusiasts noticed that they weren’t getting the volts that they asked for and the market response was to create Load-Line Calibration AKA Vdroop Control. This does exactly what it says on the tin – the greater the Vdroop control, the less able the motherboard is to reduce volts under load. On the face of it overclockers were happy because their hardware monitors were now reporting higher voltage which meant less Volts set in BIOS for the same overclock. What wasn’t realised (because you can’t see them) is that these voltage spikes still happen, only now the large spikes are more likely to break the limit because the overall Voffset is much less without Vdroop:



There is also another drawback. On some small level LLC reduces the responsiveness of the power supply and that means instability. As I understand it, by limiting voltage control you limit one of only two methods (volts and amps) which the motherboard can use to control power.

Here’s a practical example. I setup my rig to run a 4.6GHz clock on fixed volts. I then selected a Vcore in BIOS which gave me a stable system within two conditions – LLC off (level 7) and LLC on fully (level 0). To get some reliability I stipulated the Vcore had to be the minimum required (to the nearest 0.005v) to run 10 passes of IBT on Very High stress. This gave me 1.4v/LLC7 and 1.32v/LLC0. I then logged the sensors whilst running IBT for a second time:



As you can see the LLC0 condition supplied a much more varied amount of volts due to the free Vdroop. For those who are concerned about volts it’s interesting to note that the 1.4/LLC7 condition supplied noticeably few volts under load but that really isn’t the total picture:



You can see that, even though the voltage supplied was vastly different between the two conditions, the actual Watts was very similar. However there is a noticeable increase in watts supplied to the CPU under the LLC0 condition in the later runs. The control that the LLC is having over the voltage is affecting the control that the motherboard has over the power supplied to CPU resulting in a tiny increase in power which is reflected in the temperature graph.



The resulting cost in temperature was 2C. The important point is that the temperature mirrors the wattage and not the volts.

The moral of the story?
High volts aren’t bad, high watts are – and watts make heat.
Vdroop is a good thing – at least you know what the upper voltage limit is.
If your default is to run with LLC on then try it the other way around, or somewhere in the middle – you might get better temps or even a better clock.

That's as far as my research has got me at the moment. If anyone has anything to add it would be great to hear it!

M&P

Credit due here.

__________________
Ivy Bridge Overclocking
Overclocking pt2-LLC & Vdroop

OC3D Overclockers Club Member #006 (Alec Trevelyan was always harder than Bond anyway)
Reply With Quote
  #2  
Old 29-09-12, 02:03 AM
Josh Weston Josh Weston is offline
OC3D Elite
 
Join Date: Feb 2012
Location: Essex, UK
Posts: 3,512
Another great, and informative, article - well done, M&P
+1 rep for you (imagine it's still here, just for a second )
Reply With Quote
  #3  
Old 29-09-12, 04:13 AM
airdeano airdeano is offline
OC3D Elite
 
Join Date: Feb 2012
Posts: 2,895
in automotive tuning we called it tip in stumble. on high out
forced induction engines, it was critical to maintain a proper
air-fuel ratio. at wide-open throttle that wasnt as a problem..
but drivability was the pain in the arse to tune. so 4-6hrs of
load dyno and road testing to fix the same aspects you are
chasing. slight opening of the throttle meant the engine was
going to go lean or fat rich.

in PC, the jargon is different, but the results usually are the
same with nothing etched in stone (liability). usually it took
someone 90+hrs to write another code section to suppliment
the "stock or factory" settings. 16MB+ BIOS files and limited
"option" lists means there is 15MB+ for unknown function
code.
ive opened a BIOS in an editor and holy gobstoppers..
the definition files are so blank and FFFFF strickened...
if A and X while B is increasing delta Y pass-through C
only if Z.. and that is when you start POST....

like in automotive.. unless you know what you are doing
its either gunna scream speed or an expensive puff of smoke
will occur. and its usually the latter.

expecting the "general public" to understand the ramifications
of changing the number or unchecking the box and using the RMA
process as a fall-back is poop...

i thank you for the enthusiasts review of more over-clocking
information... hopefully most will understand the exercise
without letting the magic smoke out...

airdeano
Reply With Quote
  #4  
Old 29-09-12, 04:24 PM
Master&Puppet's Avatar
Master&Puppet Master&Puppet is offline
OC3D Elite
 
Join Date: Sep 2011
Location: London
Posts: 1,923
Cheers guys

I did a follow up test this morning, mostly for personal interest but some interesting things turned up. My 24/7 overclock is a x44 offset overclock which I described in that thread I wrote a little while ago. I wanted to compare it to setting a fixed voltage and judge which is better. It was the same test as last night but with new conditions at x44:

auto(offset)/LLC0
fixed 1.325/LLC7

Obviously it would have been better to test this with a true offset overclock but I don't have that option on the GD65. It might have been better to do a test with auto/LLC7 too but that would have meant using a x42 overclock which doesn't seem worth the effort! As I mentioned in the overclocking thread I used LLC to effectively raise the Vcore to get to x44.

Anyway back on topic and here are the results.



Sadly I don't know what the auto voltage profile is but it's still clear that the fixed profile is taking big lumps out of the volts for Voffset and Vdroop which is as expected. The auto/LLC0 profile is doing the opposite and increasing the voltage with load which, again, is exactly what I would expect. Because I've got the profile on auto/LLC0 it is impossible to tell how much of the volts under load is due to the LLC. Judging by the first experiment I'd say that the difference is probably quite a lot to do with the LLC and that in a true offset profile with LLC off I think the load voltages would be very similar.

The entire point of a Turbo/offset overclock is to supply the power when needed but otherwise keep the power as low as possible. A fixed voltage is perhaps less able to do this and on the right-hand side of the graph you can see where the idle voltages lay - and it is a huge difference. Is this important? Well it really depends upon the other half of the power equation:

Volts (v) x Current (Amps, a) = Power (W)

If the motherboard can reduce the amps low enough then having a high voltage will make no difference to power supplied.



Sadly this wasn't the case. I can't speak for every motherboard but at idle the fixed overclock was drawing ~25.5W@~1.284v and the offset just ~9W@~0.875v. If I put those figures into the equation:

x44 fixed/LLC7 1.284v x 19.86a = 25.5W
x44 auto/LLC0 0.875v x 10.29a = 9W
x46 fixed/LLC7 1.352v x 19.56a = 26.4W

I'm not an electrical engineer so this is where I keep things as simple as possible but the higher voltage is encouraging a higher current. This might be for many reasons I'd probably expect that the motherboard profile limits the minimum current it can push through at a certain voltages. Interestingly the x46 current didn't increase with the extra idle volts.

At the load end of the graph

x44 fixed/LLC7 1.212v x 65.6a = 79.5W
x44 auto/LLC0 1.230v x 65.2a = 80.2W
x46 fixed/LLC7 1.272v x 66.8a = 85W

The motherboard was supplying around 66 amps. It would be interesting to see whether this changed at different voltages but I think that is for another day!



The temperatures again responded to the wattage supplied and not the directly to the volts. Ohms law states that Current = Volts/Resistance. It is resistance which generates the heat which we should be primarily afraid of. Provding 80W of power is better done on a higher voltage as this requires less current which means less resistance and therefore heat. The caveat to that is of course that we have no control over the current and an increase in volts will probably correspond to some increase in current.

As a summary in terms of what I've seen over the last 24hrs I'd have to say that the less we try and control the Volts, the better. I haven't seen any benefit at all to the LLC so far, with a tight LLC I'm having to provide a tiny bit more power to overcome the instability that arises with the mechanism at full load. Situationally, as in my auto offset profile, it has some use giving a vastly reduced idle power supply with only a small increase at full load. Just applying an LLC to lower what you type into BIOS is a mistake when setting a higher figure will probably get you lower volts, watts and temps overall when combined with a free Vdroop.

If your overclock is too large for an offset then you don't really have a choice but to accept higher idle wattage on a fixed voltage. Is that detrimental to the life of the CPU? Well on the face of it you might be sending 2 or 3 times the wattage through the system at idle but I've only seen that make about a 3C rise in idle temps. Volts, amps, watts, resistance and heat all affect the life of a chip and it maybe in your interest to keep them as low as possible depending on your circumstances. The biggest difference between an offset and fixed voltage is when idling and it makes sense to have an offset overclock for everyday use to maximise CPU life. If you know that you are in for a session using your CPU at high loads and you want more CPU power then a carefully calibrated fixed voltage overclock with a free Vdroop looks ideal.

In any case it is certainly worth having a play with your settings to find the best setup for you.

M&P
__________________
Ivy Bridge Overclocking
Overclocking pt2-LLC & Vdroop

OC3D Overclockers Club Member #006 (Alec Trevelyan was always harder than Bond anyway)
Reply With Quote
  #5  
Old 29-09-12, 07:48 PM
yassarikhan786's Avatar
yassarikhan786 yassarikhan786 is offline
OC3D Elite
 
Join Date: Aug 2011
Location: United Kingdom
Posts: 8,363
Really awesome work M&P, I genuinely learnt something reading this

As Josh mentioned it's for threads like this that we need the +1 option back.
__________________
Motherboard: Asus P8P67 Pro (B3) || CPU: i7 2600k || Graphics Card: EVGA GTX 780 Ti ACX || Memory: 16GB - Mushkin Blackline 8GB + Corsair Vengeance LP 8GB || SSD's + HDD's: Crucial M4 128GB (OS + Programs) + Caviar Black 2TB (Games + Storage) || CPU Cooling: Noctua NH-D14 || Power Supply: BeQuiet Dark Power Pro P9 750W || Case: Corsair Obsidian 650D || Monitor(s): Dell U2410 + HP w1907v

Reply With Quote
  #6  
Old 30-09-12, 01:22 AM
Master&Puppet's Avatar
Master&Puppet Master&Puppet is offline
OC3D Elite
 
Join Date: Sep 2011
Location: London
Posts: 1,923
Glad I'm not the only one who found it useful!
__________________
Ivy Bridge Overclocking
Overclocking pt2-LLC & Vdroop

OC3D Overclockers Club Member #006 (Alec Trevelyan was always harder than Bond anyway)
Reply With Quote
  #7  
Old 30-09-12, 04:17 PM
Henrik's Avatar
Henrik Henrik is offline
Member
 
Join Date: Apr 2011
Location: Sweden
Posts: 137
Awesome work dude! I have read the AnandTech article my self a while back and thought i might try this for my self. This was on a LGA1156 i5-660, that was running 4.5Ghz and like you mentioned I was able to run prime stable on lower volts than i could with LLC on, that's after Vdroop ofc..

This makes me curios to try it out on my 3930k and see if or how much lower Vcore I can run.

Edit: Oh I forgot, it ran significantly cooler as well!
__________________
My Rig
CPU: Intel 3930k @ 4,6Ghz
Motherboard: Rampage 4 Extreme
Ram: Crucial Ballistix 1600Mhz 4x4GB
GPU: EVGA GTX580
OS Drive: Corsair force GT 120GB
3930k @ 5.2Ghz
Reply With Quote
  #8  
Old 01-10-12, 10:08 PM
Master&Puppet's Avatar
Master&Puppet Master&Puppet is offline
OC3D Elite
 
Join Date: Sep 2011
Location: London
Posts: 1,923
Quote:
Originally Posted by Henrik View Post
Awesome work dude! I have read the AnandTech article my self a while back and thought i might try this for my self. This was on a LGA1156 i5-660, that was running 4.5Ghz and like you mentioned I was able to run prime stable on lower volts than i could with LLC on, that's after Vdroop ofc..

This makes me curios to try it out on my 3930k and see if or how much lower Vcore I can run.

Edit: Oh I forgot, it ran significantly cooler as well!
If you do every try it, lets us know! The more info the better!
__________________
Ivy Bridge Overclocking
Overclocking pt2-LLC & Vdroop

OC3D Overclockers Club Member #006 (Alec Trevelyan was always harder than Bond anyway)
Reply With Quote
  #9  
Old 09-10-12, 11:51 PM
Perturabo Perturabo is offline
Advanced Member
 
Join Date: Nov 2011
Location: Guildford
Posts: 391
Really interesting read, nice work. I shall have a play with my OC also and see what happens.
Reply With Quote
  #10  
Old 10-10-12, 03:40 AM
Tindo's Avatar
Tindo Tindo is offline
Member
 
Join Date: Feb 2012
Location: Birmingham, UK
Posts: 240
thanks for the post
__________________
Cpu - AMD Phenom II 955 @ 3.8 Ghz, Cooler - Arctic Freezer Pro Rev 2.0, Ram - 8 Gb Kingston HyperX @ 1600, MBoard - Asus M4A87TD Evo, Gpu - Zotac GTX 570 Synergy Edition, HDD - Seagate 1TB, Psu - Cooler Master GX 650W, Case - Cooler Master Storm Enforcer, Sound Card - Asus Xonar DG, - Mouse, - Mad Catz Cyborg R.A.t. 3, - G510 - Rig Name - "SkyNet", SSD - Samsung 840

Reply With Quote
Reply

Tags
llc, overclocking, vcore, vdroop, volts

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump






All times are GMT. The time now is 04:27 AM.
Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2014, vBulletin Solutions, Inc.