Overclock.net › Forums › AMD › AMD Motherboards › ROG Crosshair VI overclocking thread
New Posts  All Forums:Forum Nav:

ROG Crosshair VI overclocking thread - Page 954

post #9531 of 16894
Quote:
Originally Posted by CeltPC View Post

UPDATE: I went into Task Manager, while leaving HWiNFO64 open, to look at what might be running in the background to push CPU usage and trip the threshold for the high P-state and volts. I saw that Asus Aura (which was not even "open" at the time, was a process eating 5% to 6% of CPU. I killed the process using end task, and bam, volts dropped to minimum with a browser, HWiNFO64, and Kodi all open.



To verify, I intended to open Aura again, but it crashed immediately. Despite uninstalling and cleaning the registry with CCleaner, then manually deleting anything left under aura folders and files, rebooting, re-installing, etc. - it still crashes right away!

So progress on one issue and now a new problem to tackle.. rolleyes.gif

I had exact same issue, same symptoms, same kill the process and Aura would crash upon trying to reopen. Removed from system did not try reinstalling. It didn't work anyways.
Destroyer
(13 items)
 
  
Reply
Destroyer
(13 items)
 
  
Reply
post #9532 of 16894
So i got my Corsair LPX 3200 ver 5.39 to boot at 3200mzh thumb.gif didn't get a straight 15 did a single boot loop then booted (a bit odd i know may go from 1.1soc to 1.15). I reflashed my bios, cleared cmos, and pulled the battery)

Ever since trying to OC my gpu and having it hard crash 8, when it crashes now (because msi ab still tends to crash when opening it up with some programs, no idea why either does it on both my systems) the PC just boot loops dram and cpu over and over and ends on code 50, clearing cmos and reloading my profile fixes the issue. So i reflashed to see if it helps any.

Started off changing soc voltage and ram at 2400, then loaded up changed it to 2933 then loaded up again and tried 3200. Booted and loaded into windows (removing the battery did delete my saved OC profile though so could thing i memorized it)

Going to try and shut off the pc now and do a cold/warm boot not sure which and see if it loads in without boot looping.

Update: After shut down and reboot, instant code 15 with 1.1v soc. Tried to load up the profile, instant 15 again. Loaded up into my other profile, with soc set to 1.0, changed it to 1.15 rebooted, reposted and changed it from 2933 and into 3200 and booted right up. Saved and reset and rebooted. I will try to shut down again and see if that soc 1.10 to 1.15 made a difference but it seems like the ram/mobo/cpu can do it just fine so far booting into windows (shows up correctly under task manager and hwinfo64 as well, timings at 16-18-18-18-36). So def has to do with soc voltage.

DRAM 1.36v
DRAM boot 1.45v
Soc 1.15v.
Edited by bluej511 - 4/15/17 at 8:49pm
    
CPUMotherboardGraphicsRAM
R7 1700X asus crosshair 6 Sapphire R9 390 Nitro CMK16GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Kingston HyperX 3K Seagate ST2000DM006 EKWB Supremacy Evo EK-XRES 100 Revo D5 - Acetal 
CoolingCoolingCoolingCooling
Alphacool GPX r9 390 m01 alphacool xt45 240 Alphacool xt45 360 Alphacool D5 PWM 
OSMonitorKeyboardPower
Windows 10 LG 29UM68 Corsair Strafe Corsair RM1000 
CaseMouseMouse Pad
Thermaltake X5 Corsair M65 Pro RGB Razer Goliathus Speed 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
R7 1700X asus crosshair 6 Sapphire R9 390 Nitro CMK16GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Kingston HyperX 3K Seagate ST2000DM006 EKWB Supremacy Evo EK-XRES 100 Revo D5 - Acetal 
CoolingCoolingCoolingCooling
Alphacool GPX r9 390 m01 alphacool xt45 240 Alphacool xt45 360 Alphacool D5 PWM 
OSMonitorKeyboardPower
Windows 10 LG 29UM68 Corsair Strafe Corsair RM1000 
CaseMouseMouse Pad
Thermaltake X5 Corsair M65 Pro RGB Razer Goliathus Speed 
  hide details  
Reply
post #9533 of 16894
Quote:
Originally Posted by noko59 View Post

I had exact same issue, same symptoms, same kill the process and Aura would crash upon trying to reopen. Removed from system did not try reinstalling. It didn't work anyways.

Don't know if you saw the followup Post I made, but in case you do want a fix:

"I have some news on the Aura issue. I had tried uninstalling with control panel, and with CCleaner, and then cleaning the registry with the same program, but no success. In googling around on uninstalling Aura, I found a program called "Advanced Uninstaller PRO 12". I installed it, ran an uninstall of Aura 1.04.26 and a registry clean with the program. It did seem to find more files and references to fix. I then ran an install of Aura 1.04.29. It went a bit wonky and brought up the damn Doctor Dump thing, but then did install. I was not optimistic, but tried a startup of Aura and what do you know, it opened, showed the G.Skill RGB! I ran the calibration thing, and tried out some effects and now it all seems to work. I rebooted and opened it again, and still works. Some may be solved!"

Since that post have not had any further problems, so I declare it working for me.
post #9534 of 16894
I have been able to run my 1800x at 4.0 stable and G.Skill 14cas memory at 3200 stable both at 1.355v and SOC 1.2v. I know that is not extreme, but my problem is I cant boot it up all the time. Sometimes my C6H gives 3 beeps and so I have to reboot into safe bios mode for no reason then save and restart, sometimes to or three times, then everything is fine.

How do I get my system to boot up consistently without having to run shutdowns and reboots so often? Thanks!
R51600
(11 items)
 
R71800X
(11 items)
 
 
CPUMotherboardGraphicsRAM
R5 1600  Gigabyte AB350M Gaming 3 ASUS RX 480 Strix 8gb OC G.Skill Trident Z 3200 
Hard DriveHard DriveCoolingOS
SSD Seagate Noctua D15 Windows 10 Creators Mess 
MonitorKeyboardPower
Acer 1080p Velocifire EVGA 650 
CPUMotherboardGraphicsRAM
R7 1800X ASUS Crosshair VI Hero ASUS RX 480 Strix 8gb OC G.Skill Ripjaws 3200 
Hard DriveHard DriveCoolingOS
Plextor M.2 Seagate Custom Water Cooled Hard Tube Loop  Windows 10 Creators Spyware 
MonitorPowerCase
Acer 1080p EVGA Supernova 750 G3 Thermaltake Core P5 
  hide details  
Reply
R51600
(11 items)
 
R71800X
(11 items)
 
 
CPUMotherboardGraphicsRAM
R5 1600  Gigabyte AB350M Gaming 3 ASUS RX 480 Strix 8gb OC G.Skill Trident Z 3200 
Hard DriveHard DriveCoolingOS
SSD Seagate Noctua D15 Windows 10 Creators Mess 
MonitorKeyboardPower
Acer 1080p Velocifire EVGA 650 
CPUMotherboardGraphicsRAM
R7 1800X ASUS Crosshair VI Hero ASUS RX 480 Strix 8gb OC G.Skill Ripjaws 3200 
Hard DriveHard DriveCoolingOS
Plextor M.2 Seagate Custom Water Cooled Hard Tube Loop  Windows 10 Creators Spyware 
MonitorPowerCase
Acer 1080p EVGA Supernova 750 G3 Thermaltake Core P5 
  hide details  
Reply
post #9535 of 16894
i solved the problem by using 1.38v for ram and 1.45 vboot, had exactly the same problems.
i'm using the 1002 bios.
post #9536 of 16894
Quote:
Originally Posted by gupsterg View Post

@Gadfly

I hope this does not come across the wrong way. I do not believe RealBench is as stressful as it may seem.

Firstly on my i5 4690K (2x which I owned) the custom x264 on OCN (which is highlighted as the "cool AVX stress test") made both of them fall over quicker than RB. And temps were ~5°C lower on x264 vs RB. Secondly both the R7 1700 I have owned again failed x264 very quickly vs RB, some data posted here.

I have never increased LLC on any of the chips mentioned for 24/7 use. Besides what can be found on the net about LLC The Stilt, Elmor and Raja@Asus have shared here some valuable info IMO.

All I can say is without the level of stability testing I did on the i5 4690K it would not have held up for f@h for the length of time I can be doing that, so the R7 had to pass that. Yeah my OC is not high, 3.8GHz/3200MHz.

Anyhow as always stability testing is a subjective matter, which I reckon there will always be disagreement as much as agreement between members wink.gif .

Not at all, and I agree with you. It is a benchmark, not a stress test.

For stress testing I use Prime95 & OCCT. Folding @ home, which I assume is what F@H stands for, is also a very good stress test, but as you mentioned, It will beat down your GPU as well as your CPU so If I am purely looking to test the CPU, It is not my first choice. Stress testing is one of those things. It is absolutely possible to push the stress levels higher with a synthetic stress test than your CPU will ever go under any circumstances during it's life time of normal use. IBT is especially bad in that regard, it is possible to generate so much heat that a good number of CPU's will not pass with a factory cooler at stock speeds. My 4770k was a good example of this, it would fail before the 10th run at stock speeds, stock voltage, with the stock factory air cooler. The CPU was perfectly stable at stock speeds, and in fact with nothing more than a 280mm AIO it was more than capable of a 4.6ghz OC with some decent timings on DDR3 2400 memory. It was just that the IBT stress test put a very high artificial load on the CPU. Which makes perfect sense when you look at what IBT was made for. It was designed to be used as a bench marking utility on super computers, solving Linear equations and stressing the floating point calculations to the absolutely max On commercial CPU's it is an interesting tool, but not one that should be taken very seriously for a home / Gaming PC. Even F@H will not generate the CPU cache stress and heat that IBT creates. Which is also why I feel it is not the best stress testing tool for Ryzen CPU's. It was designed for CPU's with external cache, not on die cache like we find in the Zeppelin die.

After the posted on here from Raja, I don't mind LLC 2 (or even 3 to a limited degree) It helps with voltage drop under load without having to set the base voltage too high. The overshoots on LLC2 are very quick and low, and not really outside of what I expect during normal operation. As I am not setting my vcore to high levels, It really is not much to worry about. Speaking to the low base voltage and drop under load, It appears that others are having real good results using a P State OC, thus far I have not messed with that yet. I am doing pure manual OC's at the moment while I play with memory speeds and thermal solutions.

The more I test Ryzen CPU's the more I am convinced that the cooling requirements of this CPU when overclocked are grossly underestimated. Even a run of the mill custom loop with a 360mm radiator really does not do a good enough job over @4 ghz. The need to remove heat from the die quickly under load is much greater than any recent Intel CPU. It reminds me a lot of my old Opteron 150; which was a great OC'ing CPU in it's day, but only when actively cooled (Phase, chiller, TEC's etc.) I feel Ryzen is very similar. They both have similar voltage scaling, "walls", and that immediate cooling requirement, even more so when the cache is heavily loaded.

I am attempting to convince myself NOT to go fire up one of my old Vaporchill phase changers... but it is getting harder and harder to make that internal argument.
post #9537 of 16894
Quote:
Originally Posted by Gadfly View Post

Not at all, and I agree with you. It is a benchmark, not a stress test.

For stress testing I use Prime95 & OCCT. Folding @ home, which I assume is what F@H stands for, is also a very good stress test, but as you mentioned, It will beat down your GPU as well as your CPU so If I am purely looking to test the CPU, It is not my first choice. Stress testing is one of those things. It is absolutely possible to push the stress levels higher with a synthetic stress test than your CPU will ever go under any circumstances during it's life time of normal use. IBT is especially bad in that regard, it is possible to generate so much heat that a good number of CPU's will not pass with a factory cooler at stock speeds. My 4770k was a good example of this, it would fail before the 10th run at stock speeds, stock voltage, with the stock factory air cooler. The CPU was perfectly stable at stock speeds, and in fact with nothing more than a 280mm AIO it was more than capable of a 4.6ghz OC with some decent timings on DDR3 2400 memory. It was just that the IBT stress test put a very high artificial load on the CPU. Which makes perfect sense when you look at what IBT was made for. It was designed to be used as a bench marking utility on super computers, solving Linear equations and stressing the floating point calculations to the absolutely max On commercial CPU's it is an interesting tool, but not one that should be taken very seriously for a home / Gaming PC. Even F@H will not generate the CPU cache stress and heat that IBT creates. Which is also why I feel it is not the best stress testing tool for Ryzen CPU's. It was designed for CPU's with external cache, not on die cache like we find in the Zeppelin die.

After the posted on here from Raja, I don't mind LLC 2 (or even 3 to a limited degree) It helps with voltage drop under load without having to set the base voltage too high. The overshoots on LLC2 are very quick and low, and not really outside of what I expect during normal operation. As I am not setting my vcore to high levels, It really is not much to worry about. Speaking to the low base voltage and drop under load, It appears that others are having real good results using a P State OC, thus far I have not messed with that yet. I am doing pure manual OC's at the moment while I play with memory speeds and thermal solutions.

The more I test Ryzen CPU's the more I am convinced that the cooling requirements of this CPU when overclocked are grossly underestimated. Even a run of the mill custom loop with a 360mm radiator really does not do a good enough job over @4 ghz. The need to remove heat from the die quickly under load is much greater than any recent Intel CPU. It reminds me a lot of my old Opteron 150; which was a great OC'ing CPU in it's day, but only when actively cooled (Phase, chiller, TEC's etc.) I feel Ryzen is very similar. They both have similar voltage scaling, "walls", and that immediate cooling requirement, even more so when the cache is heavily loaded.

I am attempting to convince myself NOT to go fire up one of my old Vaporchill phase changers... but it is getting harder and harder to make that internal argument.

A 360mm has a dissipation of around 300w or so to get a delta t of 10°C over ambient, if youre keep your room at 23°C lets say, your water temp will already be 33°C (which is around average) then add the fact that your cpu at 1.3+v will reach around 150w or so if not more then you can see why temps aren't as they're cracked up to be. Most people really REALLY overestimate watercooling when it comes to a cpu.

Example, my 1700x at 1.199v under load reaches around 50°C tctl and my gpu at 40°C, running a 360 (push/pull with 6 fans) and a 240mm in push. Adding more fans will increase the dissipation but not by a crazy amount, running fans at a faster speed will, and so will running the pump a bit faster (depending on what cpu block you're using)

Compare your temps to people hitting 70-80°C on air with the same voltages and water does FAR better then air if you have the loop for it. It is after all an 8c/16t cpu.
    
CPUMotherboardGraphicsRAM
R7 1700X asus crosshair 6 Sapphire R9 390 Nitro CMK16GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Kingston HyperX 3K Seagate ST2000DM006 EKWB Supremacy Evo EK-XRES 100 Revo D5 - Acetal 
CoolingCoolingCoolingCooling
Alphacool GPX r9 390 m01 alphacool xt45 240 Alphacool xt45 360 Alphacool D5 PWM 
OSMonitorKeyboardPower
Windows 10 LG 29UM68 Corsair Strafe Corsair RM1000 
CaseMouseMouse Pad
Thermaltake X5 Corsair M65 Pro RGB Razer Goliathus Speed 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
R7 1700X asus crosshair 6 Sapphire R9 390 Nitro CMK16GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Kingston HyperX 3K Seagate ST2000DM006 EKWB Supremacy Evo EK-XRES 100 Revo D5 - Acetal 
CoolingCoolingCoolingCooling
Alphacool GPX r9 390 m01 alphacool xt45 240 Alphacool xt45 360 Alphacool D5 PWM 
OSMonitorKeyboardPower
Windows 10 LG 29UM68 Corsair Strafe Corsair RM1000 
CaseMouseMouse Pad
Thermaltake X5 Corsair M65 Pro RGB Razer Goliathus Speed 
  hide details  
Reply
post #9538 of 16894
Quote:
Originally Posted by infraredbg View Post

External Digi+ Power Control -> CPU Current Capability. I use 140%, but you can try 120 - 130.
Not sure what to recommend though. I avoid recommending anything, just sharing what I use.

No need to adjust this setting unless the rig is shutting down when running heavy workloads. Current capability just sets the trip point for the max power the VRM will supply. Auto should be fine for all 24/7 setups.
post #9539 of 16894
Quote:
Originally Posted by Praz View Post

G.Skill Samsung B-Die







Nice result.
post #9540 of 16894
so on the advice i gave my buddy, he picked up the gigabyte gaming 5 over the crosshair and i do truly believe a lot of issues plaguing ryzen is stemming from the use of the crosshair in regards to high temperatures, ram compatibility, and voltages. i helped him set everything up and before we built his computer together he let me use his gaming 5 for a shortwhile to further test somethings out. honestly, i just wanted to see if my 1800x and gskill 16gb flare x kit played well again like it did prior when i had the gaming 5... and lo and behold, they worked great at 3200 like before. no cold boot issues.

but i did some more testing regarding the complaints i made eailer about my 1800x running hotter and at higher voltage at stock settings with xfr enabled on the crosshair compared to my previous gaming 5. from digging into i noticed the difference. the gaming 5 is simply not as aggressive as the crosshair in regards to xfr boosting. i noticed on the gaming 5 my 1800x RARELY boosted to 4.1ghz during an xfr boost. it only boosted to 4ghz the MAJORITY of the time when it boosted high with xfr. that little difference makes a WHOLE A LOT of a difference. with the gaming 5 only pegging 4ghz most of the time for the top end, the volts stayed below 1.4 volts range with an average of 1.34v's after monitoring for 40 minutes. this also caused the fans not to ramp up and down like crazy since any load, even the slightest didn't cause xfr to jump straight to 4.1ghz and 1.5v's like the crosshair does. i also noticed how my 1800x on the gaming 5 didn't stay at 4ghz for long and it didn't jump up to 4ghz frequently. it would mostly boost up to around 3.8 - 3.9ghz. only a few times would it boost to 4ghz, let alone 4.1. unlike the crosshair where its always pegging 4.1ghz anytime anything touches a core, regardless how heavy it is. idle, then open firefox? boom 4.1. load foobar? boom 4.1 with 1.5v's. alt tab to swtich a lot, boom 4.1 with 1.5v's. why i noticed a high 1.4v average with the crosshair.

i also noticed another odd ball. the gaming 5 when under abusive load, like intel burn test, both regular and avx, along with prime, my 1800x would operate at 3.5ghz rather than 3.7ghz it does on the crosshair. so this caused my 1800x to run cooler since it used far less volts at 3.5ghz.

after doing the testing and putting my buddies computer together i ran the same tests on his 1800x. what i observed with mine i observed with his. what i find odd the most was the heavy load when all 16 threads are used at 100% with how it runs at 3.5ghz instead of the stock 3.6ghz. let alone 3.7ghz because iirc, with xfr enabled, all cores operate at 3.7ghz instead of 3.6ghz stock. so i don't know why its dropping to 3.5ghz on the gaming 5. the cooler used for mine was a nh-d15s while my buddy is using a corsair h110i. so cooling shouldn't be an issue.

so for those having ram issues, temps, and voltage it could just be the motherboard. even regarding overal ram stabilitly for the perfect case scenerio, samsung single rank 16gb kits. that's what puzzles me the absolute most is how i have cold boot issues with the SAME flare x kit on my crosshair but NOT on the gaming 5. for xfr with voltages and temperatures, for whatever reason i don't know. could be asus themselves tweaking bios settings to make xfr more aggressive since they figure those wanting the crosshair to run at stock settings can get the most out of it, or maybe since the crosshair has a better power delivery system it allows ryzen to boost to those high frequencies, and voltages at stock. i'm sorta leaning towards both with how the gaming 5 drops to 3.5ghz at full, 16 thread load since it does have a weaker vrm setup compared to the crosshair.

either way i would rather have my 1800x boosting to 3.8 - 3.9ghz frequently, with an occasion 4ghz, and rare 4.1ghz if it meant it kept voltages down allowing for lower temps and fan speeds.... but i also like the crosshair overall features wise. i just reconmended the gaming 5 to my friend because its an easier set it and forget it experience.
Edited by muffins - 4/15/17 at 11:13pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD Motherboards
Overclock.net › Forums › AMD › AMD Motherboards › ROG Crosshair VI overclocking thread