Load Line Calibration Yes or No?

OLDPHART

OLDPHART

OLDPHART

Soldato
Joined
19 Jun 2004
Posts
6,521
Location
n/east-the toon
A little while back l read up on LLC[Load Line Calibration] and basiclly running at a slighly higher Vcore to do away with using LLC. Does not necessarily mean higher temps them using LLC with a lower Vcore.

So this morning sitting at my PC, l decided to give it ago using Prime95 and Intel Burn Test, while l was waiting & checking if my sons flight had landed from the good old USA.

My PC spec > X58A UD5, CPU i7 920:4.2ghz, GPU 6970, 830 SSD, 12GB ram cooled by a Phobya 560 rad 900 rpm Fans on all tests below.
Test Results as follows-

First run of P95 AND IBT > Cpu Vcore set in Bios at 1.34375v / LLC>Standard below -
Prime95 - Blend > ran till Self Test 8k Passed 32 minutes with CPU Vcore at 1.34375v LLC>Standard, CPU-Z reports CPU Vcore 1.296v 100% Load, core temps 62>67'c room ambient 22.5'c.
IBT Standard 10 runs passed > CPU Vcore at 1.34375v / LLC>Standard, CPU-Z reports CPU Vcore 1.296v 100% Load, core temps 63>69'c room ambient 22.5'c / Note: core temps 2'c higher than P95 above.

Second run of P95 AND IBT > CPU Vcore set in Bios at 1.33750v / LLC>Level 1 -
Prime95 - Blend > ran till Self Test 8k Passed 32 minutes with CPU Vcore:1.33750v / LLC>Level 1, CPU-Z reports CPU Vcore 1.328v 100% Load, core temps 66>72'c room ambient 23'c
IBT Standard 10 runs passed > CPU Vcore at 1.33750v / LLC>Level 1, CPU-Z reports CPU Vcore 1.328v 100% Load, core temps 67>73'c room ambient 23'c / Note core temps only 1'c higher than P95 above.

So with the room ambient up 0.5'c to 23'c, l ran first IBT at 1.34375Vcore / LLC>Standard again as a quick check -
IBT Standard 10 runs passed > CPU Vcore at 1.34375v / LLC>Standard, CPU-Z reports CPU Vcore 1.296v 100% Load, core temps 64>69'c room ambient 23'c

Right if you compare below both the IBT test results with different CPU Vcore and LLC setting with the same room ambient 23'c, using LLC[Load Line Calibration] above standard gives higher temps -

IBT Standard 10 runs passed > CPU Vcore at 1.34375v / LLC>Standard, CPU-Z reports CPU Vcore 1.296v 100% Load, core temps 64>69'c room ambient 23'c
IBT Standard 10 runs passed > CPU Vcore at 1.33750v / LLC>Level 1, CPU-Z reports CPU Vcore 1.328v 100% Load, core temps 67>73'c room ambient 23'c

So using LLC at Standard Disables Load Line Calibration and sets VDroop following Intel specifications.(De- fault) CPU cores run cooler. I know this is just a couple of tests but the results speak for them selfs, has anyone else done this kind of test in the forums with LLC.

X58A UD5 Mobo Manual on using Load-Line Calibration -
Enables or disables Load-Line Calibration. This item allows you to adjust Vdroop at different levels.
Enabling Load-Line Calibration may keep the CPU voltage more constant under light and heavy CPU load.
Standard Disables Load-Line Calibration and sets VDroop following Intel specifications. (De- fault)
Level 1 Enables Load-Line Calibration and slightly adjusts VDroop.
Level 2 Enables Load-Line Calibration and moderately adjusts VDroop
Note: Enabling Load-Line Calibration may result in DAMAGE to your CPU or reduce the useful life of the CPU.

Well thats food for thought, l wonder if this basicly applies across the Intel range of CPU's with various mobo specifications, but does your PSU play apart in it as well?

Maybe the Guru's of this forum could shed some light on LLC Yes or No?

IBT Test Vcore at 1.34375v / LLC>Standard

P95 Blend Self Test 8k 32 minutes passed at Vcore 1.33750 / LLC>Level 1
 
I vote no. Dynamically changing the voltage based on load doesn't sit well with my limited understanding of electronics.

I don't care much about idle voltage since it seems unlikely that the cpu accumulates damage rapidly while it's idle, even at high voltage. For one thing the temperature is low when it's idle. Load voltage worries me, and LLC doesn't decrease that - but it does jerk the voltage up and down a bit while under load - which doesn't sound great.

Any electrical guys here?

Common consensus will vote yes on the basis that lower voltage must be better.
 
Tests done on LLC showed there was no significant spikes or ripple with LLC enabled. Generally you do want at least some level of it to counter-act VDroop, but high levels will tend to overvolt the chip slightly, which increases heat.

Regardless when they say high voltages may shorten the lifespan of the CPU, they mean instead of it lasting until the death of the universe it will live for maybe 15 years. I mean most of us replace our CPU after 2 or 3 years and at most 5 anyway.
 
Ive used it for years on various rigs, in daily use it hasnt caused any issues. Had an i7 920 that ran at either 4ghz or 4.2ghz for 90% of the time i owned it, (3 years iirc), an i7 930 based setup still running at 4.0ghz for the best part of 2 years. Llc enabled on both. Heck im not fussed about cpu lifespan lol, ive had 3 cpu's in the last year. An i5 3570k, an i7 3770k and the current i7 4770k.
 
From my perspective if set up properly using LLC should mean a longer life for your CPU. It should allow the default 24/7 voltage to be reduced to whatever is required to maintain stability at idle whilst allowing the CPU to draw higher volts only when circumstances require it. The disclaimer about reducing CPU life is most likely there because if people enable LLC but don't reduce VCore (or leave it set to auto) then the system will over-volt.

In the above example with LLC set to level 1 it should be possible to further reduce Vcore by a fair margin whilst still allowing a peak of 1.296 if that is what is required for stability under load.
 
Last edited:
Just tried it to see the results, l have used level 1 on my X58A UD5 for nearly 4 years at the same CPU Vcore for 4.2ghz 24/7 stable. The only complaint l have about the mobo is the stepping of the voltage.

Left it on the higher CPU Vcore in the Bios at 1.34375v / LLC>Standard, played WAW core temps at room ambient of 23'c > CPU 49 to 54'c / GPU 46.5'C. This was in a private server.

Also played a few other games no problems with stability, temps where slightly higher room ambient 24.5'c.
 
If it were me I'd be looking at the reason behind why your overclock can be stable at a max of 1.296v when not using LLC, but when you use LLC you need 1.328v.

What happens if you have LLC at level 1 and drop VCore below 1.3375? The difference between peak voltages is 0.032v so theoretically you should be able to subtract this from VCore (giving 1.3055v) whilst still having the ability to peak at 1.296v.
 
PieEater said:
If it were me I'd be looking at the reason behind why your overclock can be stable at a max of 1.296v when not using LLC, but when you use LLC you need 1.328v.

What happens if you have LLC at level 1 and drop VCore below 1.3375? The difference between peak voltages is 0.032v so theoretically you should be able to subtract this from VCore (giving 1.3055v) whilst still having the ability to peak at 1.296v.
Click to expand...


Hi Pie, when l first built this PC it only 6GB ram with a 5870 at LEVEL 1 = 4.2GHz:Vcore-1.3312v.

When l upgraded the GPU to a 470 them 6970 also now have 12GB ram Had to up the Vcore to 1.3375v. If l drop the Vcore to 1.3312v l will get just the odd BSOD so 1.3375v. l'v tried adjusting the Bios QVP/VTT, etc, its unstable so 1.3375v is needed.

My mobo is a X58A UD5, l can take the LLC to level 2 but read in a couple of threads not to use it others its ok, but prefere not to take the chance so l might well be able to lower the Vcore using level 2.

Don't forget its a middle of the range mobo and maybe not being able to fine tune the Bios settings compared to the X58A UD7.
 
Last edited:
Well im undecided,although I use the offset method my self , when using a fixed voltage im only using about 10watts with no loading on the cpu and temps are good, not sure it makes much difference as my 4.3ghz o/c uses about 1.25v,although it might make more of a difference if vcore was 1.4v
To run offset I use 1.272/260v
 
I remember an article many years ago about LLC, and it was suggesting you don't use it, for one big reason. As the chip comes off load, it can cause a voltage spike to the CPU. Things may have improved since the Core2/775 days, but I doubt it.
 
You must log in or register to reply here.
Back
Top Bottom