Overclock.net › Forums › AMD › AMD Motherboards › My experience with the Asus PRIME X370-Pro
New Posts  All Forums:Forum Nav:

My experience with the Asus PRIME X370-Pro - Page 548

post #5471 of 5920
Quote:
Originally Posted by Lermite View Post

I tried 14-13-13-13-28-41 because the minimum value of tRC is tRP+tRAS
But it's far from stable.

Then I tested the timings I just posted.
I got only one error with 16 instances of HCI Memtest with 1370% as completion.
Perhaps I'll manage to get rid of this flaw by playing again with CLDO_VDDP.

My IMC are lazy as **** sozo.gif

Interesting :-)

1370% isn't bad though (?), has to be really, really close to stability? I honestly never run my stability tests that long even if I should. ;-)

EDIT: Leaving hci memtest running while sleeping is smart.
Edited by makatech - 10/20/17 at 11:58pm
post #5472 of 5920
Quote:
Originally Posted by makatech View Post

1370% isn't bad though (?), has to be really, really close to
:thumb:stability? I honestly never run my stability tests that long even if I should. ;-)

From my experience, such a result means the RAM will causes one or two errors per weeks during a normal use.
The stability is almost reached, but it's not really because a single error can make a program to crash, which is always bothering.
So while my RAM is not completely stable, I'll keep on playing with its settings.

A completion of 400% may be enough for a quick test, for a partial evaluation of new settings, but it far from enough to be sure the RAM is perfectly stable, without any error ever.
The easiest way to reach a 1000+% completion is to let Memtest run for the whole night, and even two or three if the first one is successful.
Multipurpose PC
(12 items)
 
  
MotherboardGraphicsRAMHard Drive
Asus Prime X370 Pro EVGA 1050 Ti HOF 3600 C17 Intel 600P 512MB 
Hard DriveCoolingOSMonitor
WD Red 3TB Alphacool Eisbaer 280 Windows 10 Enterprise BCC Acer G246HL 
KeyboardPowerCaseMouse
Microsoft Natural Ergonomic Keyboard 4000 v1.0 Seasonic Prime 650W Titanium SSR-650TD Antec P280 Logitech G502 
  hide details  
Reply
Multipurpose PC
(12 items)
 
  
MotherboardGraphicsRAMHard Drive
Asus Prime X370 Pro EVGA 1050 Ti HOF 3600 C17 Intel 600P 512MB 
Hard DriveCoolingOSMonitor
WD Red 3TB Alphacool Eisbaer 280 Windows 10 Enterprise BCC Acer G246HL 
KeyboardPowerCaseMouse
Microsoft Natural Ergonomic Keyboard 4000 v1.0 Seasonic Prime 650W Titanium SSR-650TD Antec P280 Logitech G502 
  hide details  
Reply
post #5473 of 5920
1 week ago i low my voltage soc to 0.95v (llc3) stable. before i use 1.18v.

History:
i use 0.95v in SoC with 3200mhz ram
change motherboard in warranty
i need 1.18v in SoC (I blamed the mother even though I knew that SoC is CPU-specific.)
few weeks ago i change my cpu-cooler and need to change my rams to other slots...
again i need only 0.95v in SoC.

i dont know if the problem is the slots, the bus to IMC, bios version, etc. but in my opinion, if you have high voltage needed in soc try to use other slots for ram and try with the new bios at the moment.

you not lose nothing to try.
(1.18v to 0.95v is much difference)
InsanePowah
(16 items)
 
  
CPUMotherboardGraphicsRAM
Ryzen7 1600 Asus Prime X370 Pro Zotac 1060Gtx 3Gb G.Skill TridentZ F4-3200C16D-16GTZB 
Hard DriveHard DriveCoolingOS
Samsung 850 Pro Western Digital Black Deepcool lucifer v2 (With koolance coolers) Windos 10 Pro x64 
MonitorKeyboardPowerCase
Samsung s23c350h Razer BlackWidow Ultimate 2014 Stealth Thermaltake Smart 1200w Thermaltake Overseer rx-1 
MouseMouse PadAudioAudio
Thermaltake Theron Thermaltake Ladon Old AIWA Minicomponent (NSX-F7) Sony XB950 
  hide details  
Reply
InsanePowah
(16 items)
 
  
CPUMotherboardGraphicsRAM
Ryzen7 1600 Asus Prime X370 Pro Zotac 1060Gtx 3Gb G.Skill TridentZ F4-3200C16D-16GTZB 
Hard DriveHard DriveCoolingOS
Samsung 850 Pro Western Digital Black Deepcool lucifer v2 (With koolance coolers) Windos 10 Pro x64 
MonitorKeyboardPowerCase
Samsung s23c350h Razer BlackWidow Ultimate 2014 Stealth Thermaltake Smart 1200w Thermaltake Overseer rx-1 
MouseMouse PadAudioAudio
Thermaltake Theron Thermaltake Ladon Old AIWA Minicomponent (NSX-F7) Sony XB950 
  hide details  
Reply
post #5474 of 5920
Nooo, just over 13 hours, error detected.... uumm... balls..... ok, better than before, not sure what to alter now, seems so close.

You could be right with 3333, 3466 with CL 14 should be a obv step up, but id be even more surprised if that would stick.

Having said that, 3066 (14 16 16 16 32 54) to 3333 (14 14 14 14 32 50) hasn't suddenly made my machine 10% quicker, might not be worth it other than keeping my mind entertained biggrin.gif

Yea, 14,14,14,14, the attachment was what got me to the 13hrs
post #5475 of 5920
Quote:
Originally Posted by inserf1 View Post

Nooo, just over 13 hours, error detected.... uumm... balls..... ok, better than before, not sure what to alter now, seems so close.

You could be right with 3333, 3466 with CL 14 should be a obv step up, but id be even more surprised if that would stick.

Having said that, 3066 (14 16 16 16 32 54) to 3333 (14 14 14 14 32 50) hasn't suddenly made my machine 10% quicker, might not be worth it other than keeping my mind entertained biggrin.gif

Yea, 14,14,14,14, the attachment was what got me to the 13hrs

There are obvious performance gains between different memory speeds and timings.

If having any doubts I recommend looking at these post from the Swedish forum (all from the same user, not using the Asus Prime board though). If you don't have time using google translate then the graphs and timings are pretty obvious too. He is using the same memory but with different speeds and timings comparing performance gains. Performance gains may be dependent on type of cpu and gpu as well but it will for sure give you a hint.

You will find the timings he used if you click on the "Visa innehåll" button.
https://www.sweclockers.com/forum/post/17022467

He is doing some more testing here
https://www.sweclockers.com/forum/post/17024530

Slightly better timings for 3333
https://www.sweclockers.com/forum/post/17031006

and here
https://www.sweclockers.com/forum/post/17039604

Running 3466
https://www.sweclockers.com/forum/post/17039921

New performance graphs including 3466
https://www.sweclockers.com/forum/post/17041245

and finally slightly better timings for 3466
https://www.sweclockers.com/forum/post/17044833

I believe he also is showing that running 1800X cpu stock (not overclocking cpu) with fast memory speeds and good timings may give faster system performance comparing to an overclocked 1800X but with slower memory speed which I find really cool.
Edited by makatech - 10/21/17 at 8:08am
post #5476 of 5920
Quote:
Originally Posted by makatech View Post

There are obvious performance gains between different speeds and timings.

If have any doubts then I recommend looking at these post from the Swedish forum (all from the same user, not using the Asus Prime board though). If you don't have time using google translate then the graphs and timings are pretty obvious too. He is using the same memory but with different speeds and timings comparing performance gains. Performance gains may be dependent on type of cpu and gpu as well but it will for sure give you a hint.
https://www.sweclockers.com/forum/post/17022467

You will find the timings he used if you click on the "Visa innehåll" button.

He is doing some more testing here
https://www.sweclockers.com/forum/post/17024530

I believe he also is showing that running 1800X cpu stock (not overclocking cpu) with fast memory speeds and good timings may give faster system performance comparing to an overclocked 1800X but with slower memory speed which I find really cool.

Thanks, will have a look, interesting to see he uses lower tRC than the RAS Pre + tRAS norm

Have noticed tRAS and tRC, seem to have a larger say on stability, not so large with AIDA's bench, but higher didnt necessary mean more passes, maybe there's a window or a relation between them or another setting...
post #5477 of 5920
Quote:
Originally Posted by inserf1 View Post

Thanks, will have a look, interesting to see he uses lower tRC than the RAS Pre + tRAS norm

Don't know if this specific tRC timing value is optimal (?) my point was showing the performance gains using higher memory speeds together with good manual timings on a Ryzen system (if somebody still had doubts).
Edited by makatech - 10/21/17 at 8:05am
post #5478 of 5920
Quote:
Originally Posted by makatech View Post

Don't know if this specific tRC timing value is optimal (?) my point was showing the performance gains using higher memory speeds together with good manual timings on a Ryzen system (if somebody still had doubts).

I saw, tRC just jumped out being different to others

His (i think) 3333 'tuned' vs 3333 with only primaries were noticible
post #5479 of 5920
I reverted to 0810 and now i'm testing some tightened timings, calculated with the DRAM Calculator by 1usmus

This are the calculations results:
th_2RyzenDRAMCalculatorFast.png

And this are the timings i used:
th_3200_Ryzen_DRAM_Calculator_FastTimings.jpg

System is booting with 38x oc and i passed a 400% quick memtest, but OCCT fails after ~25/30 minutes.
I'm playing with CLDO_VDDP, moving up by 10 or 20 at a time, starting from 425, i'm actually testing 470.

Do you guys think i can correct something?
Ty and have a nice w-e
Edited by komodikkio - 10/21/17 at 2:14pm
post #5480 of 5920
How do you calculate tRFC (ns)? here it shows "----"

I assume these are tRFC1 and tRFC2 right? where is tRFC3? or do i need to calculate those manually? (or set only tRFC1 and leave the other two in Auto in the BIOS?)


Edited by SaccoSVD - 10/21/17 at 2:28pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD Motherboards
Overclock.net › Forums › AMD › AMD Motherboards › My experience with the Asus PRIME X370-Pro