Overclock.net › Forums › Intel › Intel CPUs › == Haswell-E Overclock Leaderboard & Owners Club ==
New Posts  All Forums:Forum Nav:

== Haswell-E Overclock Leaderboard & Owners Club == - Page 1361

post #13601 of 22164
Quote:
Originally Posted by Blameless View Post

Cache ECC errors usually present as performance loss before more obvious problems, but there are so many factors that can cause small variances in the x265 bench times that it's going to take a lot to prove that vccsa was the key factor.

I agree, was using it as a first step, then onto more (like what you suggested),
Actually running 0.94v for a few days, gaming, encoding etc will test that out too right?
7820x TUF
(8 items)
 
  
CPUMotherboardGraphicsRAM
7820x Asus x299 TUF Mark 1 Asus GTX1080ti Strix Corsair Vengeance LPX CMK32GX4M4B3200C16 32GB (... 
CoolingOSMonitorPower
Corsair H1150i Windows 10 Acer Predator XB271HU Corsair HX850i 
  hide details  
Reply
7820x TUF
(8 items)
 
  
CPUMotherboardGraphicsRAM
7820x Asus x299 TUF Mark 1 Asus GTX1080ti Strix Corsair Vengeance LPX CMK32GX4M4B3200C16 32GB (... 
CoolingOSMonitorPower
Corsair H1150i Windows 10 Acer Predator XB271HU Corsair HX850i 
  hide details  
Reply
post #13602 of 22164
Quote:
Originally Posted by cookiesowns View Post

Try even lower wink.gif You will have dead spots. I generally lower 0.01 to 0.05 at a time.

These dead spots are actually more of an alignment issue between the VCCSA and memory related circuits. Praz explained this better a few months ago. Of course the voltage needs to be ample enough for the given frequency and timings too.


EDIT: Generally speaking, because Google Stress App is much quicker at finding errors than HCI (especially with 32GB and upward), you're more likely to worm out training related issues and VCCSA issues. Basically, it's the best test that I know of at least that helps with memory related instability in general.
Edited by Silent Scone - 10/18/15 at 1:14am
post #13603 of 22164
Quote:
Originally Posted by Silent Scone View Post


EDIT: Generally speaking, because Google Stress App is much quicker at finding errors than HCI (especially with 32GB and upward), you're more likely to worm out training related issues and VCCSA issues. Basically, it's the best test that I know of at least that helps with memory related instability in general.

I still need to get around to installing Linux..lol.
Though it is interesting playing with the VCCSA voltages changes the x265 outcome.

Just changed mine from 0.94v to 1.04v and this is the result I got now:

7820x TUF
(8 items)
 
  
CPUMotherboardGraphicsRAM
7820x Asus x299 TUF Mark 1 Asus GTX1080ti Strix Corsair Vengeance LPX CMK32GX4M4B3200C16 32GB (... 
CoolingOSMonitorPower
Corsair H1150i Windows 10 Acer Predator XB271HU Corsair HX850i 
  hide details  
Reply
7820x TUF
(8 items)
 
  
CPUMotherboardGraphicsRAM
7820x Asus x299 TUF Mark 1 Asus GTX1080ti Strix Corsair Vengeance LPX CMK32GX4M4B3200C16 32GB (... 
CoolingOSMonitorPower
Corsair H1150i Windows 10 Acer Predator XB271HU Corsair HX850i 
  hide details  
Reply
post #13604 of 22164
Not sure if they really means anything but looks good all the same smile.gif


Just a quick snap of temperature with the sig configuration on my loop when running 4x overkill. Case ambient 21c, water 22c



Edited by Silent Scone - 10/18/15 at 3:19am
post #13605 of 22164
Quote:
Originally Posted by Silent Scone View Post

Not sure if they really means anything but looks good all the same smile.gif


Just a quick snap of temperature with the sig configuration on my loop when running 4x overkill. Case ambient 21c, water 22c

Ok I'm sitting in Linux Mint 17.2 now, stressapptest is install, ran the default and passed, but whats the best way to run it, I have 16GB ram...
7820x TUF
(8 items)
 
  
CPUMotherboardGraphicsRAM
7820x Asus x299 TUF Mark 1 Asus GTX1080ti Strix Corsair Vengeance LPX CMK32GX4M4B3200C16 32GB (... 
CoolingOSMonitorPower
Corsair H1150i Windows 10 Acer Predator XB271HU Corsair HX850i 
  hide details  
Reply
7820x TUF
(8 items)
 
  
CPUMotherboardGraphicsRAM
7820x Asus x299 TUF Mark 1 Asus GTX1080ti Strix Corsair Vengeance LPX CMK32GX4M4B3200C16 32GB (... 
CoolingOSMonitorPower
Corsair H1150i Windows 10 Acer Predator XB271HU Corsair HX850i 
  hide details  
Reply
post #13606 of 22164
type stressapptest -W -s 3600 into terminal


This will run the test for one hour, the test should allocate 94% of memory by default
post #13607 of 22164
Quote:
Originally Posted by schoolofmonkey View Post

Good point, will give that a go today, want to get back into MGS V, haven't had the time, though it barely stresses the CPU.
I finally matched your overclock, just not the cache (no OC socket), I'm guessing you're running 1.24v because you're cache is a lot higher (Mines only 3.3Ghz).

I'm running adaptive voltage on my core with a -ve offset on non-turbo bins. That basically means, any frequency below 3.3GHz, the CPU will be running at a lower voltage than stock. My final Vcore is 1.24V, so that's the Vcore when my chip is pushed to the max frequency of 4.3GHz. When you start off with adaptive, never try to use -ve offset until you're absolutely stable with the turbo voltage. Once that's out of the way, you can start messing about with -ve offset for an extra voltage reduction on idle and low power states. With my cache, I have the O.C. socket, so I am running 4.0GHz on that. I wouldn't necessarily say that Vcore is dependent on cache voltage at my medium OC settings, as I've tried stock cache with 1.23V and have failed x.264 before. At higher (and less efficient) overclocks, it may well be wise to bump up Vcore a tiny bit when messing around with cache.

The one thing that Vcore and Vcache depend on are the VCCIN, which is the input voltage. For testing purposes, always set your VCCIN up to 1.95~2.00V (or higher, but I don't personally) with a low LLC, and only reduce VCCIN after you are sure everything else (including VCCSA) is stable. As for VCCSA, that's a weird voltage which is dependent on RAM frequency, cache, RAM timings - it changes vastly with things like adding 8 instead of 4 RAM sticks, or bumping up the frequency. Usually setting "Auto" gets VCCSA correct, but if it doesn't you can always change it in increments of 0.03V and see what happens.
post #13608 of 22164
Quote:
Originally Posted by Desolutional View Post

h.265 encode with Handbrake, and make the encode last 4 hours (with a 2 pass). If it passes, you're good to go for all practical intents and purposes. Give it something "natural" like a video encoding task instead of "NUMBERS MASON, WHAT DO THE NUMBERS MEAN, CRUNCH THE NUMBERS MASON". Prime95 is yuck - it's just a test designed to draw maximum current through the CPU, not really a daily test of real life stability. The best "natural" test is a video game.
try ASUS realbenchv2.4 stress trest, doe sa pretty good job as a system-wide stressor. (if you have air cooled GPUs, enable your gaming fan curve. wink.gif )
Quote:
Originally Posted by schoolofmonkey View Post

Good point, will give that a go today, want to get back into MGS V, haven't had the time, though it barely stresses the CPU.
I finally matched your overclock, just not the cache (no OC socket), I'm guessing you're running 1.24v because you're cache is a lot higher (Mines only 3.3Ghz).
Cache is a separate voltage rail on ASUS but some vcore will be needed at higher cache.

Quote:
Originally Posted by Blameless View Post

Cache ECC errors usually present as performance loss before more obvious problems, but there are so many factors that can cause small variances in the x265 bench times that it's going to take a lot to prove that vccsa was the key factor.
^^ This!
with x265, W7, W8.1 and W10 really behave differently.
Quote:
Originally Posted by Silent Scone View Post

These dead spots are actually more of an alignment issue between the VCCSA and memory related circuits. Praz explained this better a few months ago. Of course the voltage needs to be ample enough for the given frequency and timings too.
EDIT: Generally speaking, because Google Stress App is much quicker at finding errors than HCI (especially with 32GB and upward), you're more likely to worm out training related issues and VCCSA issues. Basically, it's the best test that I know of at least that helps with memory related instability in general.
Lol - with those new 32GB sticks.. .set up HMT and go on Holiday. tongue.gif
Quote:
Originally Posted by schoolofmonkey View Post

I still need to get around to installing Linux..lol.
Though it is interesting playing with the VCCSA voltages changes the x265 outcome.
Just changed mine from 0.94v to 1.04v and this is the result I got now:
Sweet! At some poinmt, just use the rig as you normally would (eg.. "step away from the stress tests" wink.gif ). If the ram acts up, tweak VSA again.
x99
(22 items)
 
Z370
(8 items)
 
x299
(18 items)
 
CPUCPUMotherboardGraphics
i7 5960X i7 6950X Asus Rampage V Extreme 10 2 GTX Titan X Pascal 
RAMHard DriveHard DriveHard Drive
GSkill 3200c14 TZs 8x8GB @ 3400c13 Intel 750 NVMe 400GB 2x Plextor SSD 256 Raid 0 (Win 7) Samsung NVMe 950 Pro M.2 
Optical DriveCoolingCoolingCooling
Plextor 810 2x XSPC RX360 Koolance 380i D5 
CoolingOSOSOS
Aquaero 6 Windows 10 x64  Windows 7 Pro Linux Mint 
MonitorKeyboardPowerCase
Seiki 50" 4K monitor (720P-2160i) 240-30Hz Das Keyboard Model S Pro Corsair AX1200 Case Labs SM8 
MouseAudio
Steelseries Rival Ultimate Ears TF10 
CPUMotherboardGraphicsRAM
8700K ASUS Maximus X Apex GTX 1080 2x8 GB G.Skill 4400c19 
Hard DriveCoolingOSCase
Samsung 960 Pro 360Rad + Chiller Win 10 HWBOT OBT 
CPUCPUMotherboardGraphics
7740X 7980XE ASUS Rampage Vi Apex 2x Nvidia Titan Xp SLi 
RAMHard DriveHard DriveHard Drive
G.Skill 3600c15 2x8GB kits Samsung 960 Pro, NVMe WD Raptor 1T Plextor SSDs Raid 0 
Optical DriveCoolingOSOS
Plextor DVD/BR R/W (very) Custom Water Windows 10 Pro Windows 7 Pro 
MonitorKeyboardPowerCase
ASUS 144Hz Ducky Corsair 1500i Microcool Benchetto 101 
MouseAudio
Steelseries Rival Ultimate ears TF10s (IEMs) 
  hide details  
Reply
x99
(22 items)
 
Z370
(8 items)
 
x299
(18 items)
 
CPUCPUMotherboardGraphics
i7 5960X i7 6950X Asus Rampage V Extreme 10 2 GTX Titan X Pascal 
RAMHard DriveHard DriveHard Drive
GSkill 3200c14 TZs 8x8GB @ 3400c13 Intel 750 NVMe 400GB 2x Plextor SSD 256 Raid 0 (Win 7) Samsung NVMe 950 Pro M.2 
Optical DriveCoolingCoolingCooling
Plextor 810 2x XSPC RX360 Koolance 380i D5 
CoolingOSOSOS
Aquaero 6 Windows 10 x64  Windows 7 Pro Linux Mint 
MonitorKeyboardPowerCase
Seiki 50" 4K monitor (720P-2160i) 240-30Hz Das Keyboard Model S Pro Corsair AX1200 Case Labs SM8 
MouseAudio
Steelseries Rival Ultimate Ears TF10 
CPUMotherboardGraphicsRAM
8700K ASUS Maximus X Apex GTX 1080 2x8 GB G.Skill 4400c19 
Hard DriveCoolingOSCase
Samsung 960 Pro 360Rad + Chiller Win 10 HWBOT OBT 
CPUCPUMotherboardGraphics
7740X 7980XE ASUS Rampage Vi Apex 2x Nvidia Titan Xp SLi 
RAMHard DriveHard DriveHard Drive
G.Skill 3600c15 2x8GB kits Samsung 960 Pro, NVMe WD Raptor 1T Plextor SSDs Raid 0 
Optical DriveCoolingOSOS
Plextor DVD/BR R/W (very) Custom Water Windows 10 Pro Windows 7 Pro 
MonitorKeyboardPowerCase
ASUS 144Hz Ducky Corsair 1500i Microcool Benchetto 101 
MouseAudio
Steelseries Rival Ultimate ears TF10s (IEMs) 
  hide details  
Reply
post #13609 of 22164
I'm just running stressapptest again.
Made the mistake of running it from a USB instead, 5 minutes in the machine rebooted itself lol.
My bad, Mint is installed properly now 45 minutes into the test.
Freaked a bit when I saw the whole thread power spike thing but it's normal and didn't hang.
I'm guessing that happens a few time, just saw it again "pausing worker threads in preparation for a power spike".

Quote:
Originally Posted by Silent Scone View Post

type stressapptest -W -s 3600 into terminal
This will run the test for one hour, the test should allocate 94% of memory by default

Ok it passed, but a strange thing happened once it was finished, I was there the last 2 minutes, once it finished and brought up the results, 0 errors, the machine rebooted a few seconds later.
Now I don't know why that happened or if it was Mint trying to reallocate ram, I'm not sure, but as far as the test was concerned it passed. redface.gif
Edited by schoolofmonkey - 10/18/15 at 8:13am
7820x TUF
(8 items)
 
  
CPUMotherboardGraphicsRAM
7820x Asus x299 TUF Mark 1 Asus GTX1080ti Strix Corsair Vengeance LPX CMK32GX4M4B3200C16 32GB (... 
CoolingOSMonitorPower
Corsair H1150i Windows 10 Acer Predator XB271HU Corsair HX850i 
  hide details  
Reply
7820x TUF
(8 items)
 
  
CPUMotherboardGraphicsRAM
7820x Asus x299 TUF Mark 1 Asus GTX1080ti Strix Corsair Vengeance LPX CMK32GX4M4B3200C16 32GB (... 
CoolingOSMonitorPower
Corsair H1150i Windows 10 Acer Predator XB271HU Corsair HX850i 
  hide details  
Reply
post #13610 of 22164
Quote:
Originally Posted by scoobied77 View Post

Can't say I would be happy running them volts, don't see the point in pushing it that much. I'm more interested in 24/7 stable clocks.

My old 5930K died on me (memory controller fried) made me realise these chips are more fragile than you think.

Fortunately my new replacement 5960X runs on incredibly low volts, it's a amazing chip and I want to look after this one.

thumb.gif

I hear what you are saying bit I have the Performance Tuning Protection Plan so i may even try 1.52v on a cool day
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › == Haswell-E Overclock Leaderboard & Owners Club ==