Print Page | Close Window

Having trouble getting my RAM above 2133Mhz.

Printed From: ASRock.com
Category: Technical Support
Forum Name: AMD Motherboards
Forum Description: Question about ASRock AMD motherboards
URL: https://forum.asrock.com/forum_posts.asp?TID=4864
Printed Date: 12 Dec 2024 at 7:17am
Software Version: Web Wiz Forums 12.04 - http://www.webwizforums.com


Topic: Having trouble getting my RAM above 2133Mhz.
Posted By: trainergames
Subject: Having trouble getting my RAM above 2133Mhz.
Date Posted: 16 Apr 2017 at 2:58pm
" rel="nofollow - I am using a R7 1700 on an ASRock X370 Taichi. 

I am using a GSkill Ripjawz V 3200mhz kit, Model:F4-3200C16D-16GVKB

I tried XMP,i tried setting the speed and timings myself,i even tried setting the RAM voltage.

But it will only post at 2133Mhz.

I noticed that is it not on the QVL list,but i was hoping to at least get 2933mhz if the the 3200.


Can you guys please help me find out what is wrong?



Replies:
Posted By: Sensai
Date Posted: 17 Apr 2017 at 12:13am
" rel="nofollow - I am getting the same thing with my gskill ram f4-3200c14d-16gtz, using the same processor. Are you getting a wierd reboot dance? that is shut off turn on shut off turn on.. etc half a dozen times before it goes to default?  what is your latency setting to? mine is going 2133 with 15 latency.


Posted By: Xaltar
Date Posted: 17 Apr 2017 at 12:28am
That "boot dance" is by design, it is something new that came with Ryzen. When you set new RAM settings the system will train the memory up to a preset number of times trying to make the settings work before resetting to defaults. The setting can be changed in the UEFI, I believe the default value is 5.

From what I understand the board will try and train the RAM a few different ways until it reaches the limit set in the UEFI. If you are trying to get a higher frequency to post and having trouble doing so then increasing this limit (resulting in more post looping) may give you a better chance of it working. If you are just looking for the quickest stable values then you can lower the setting and be subjected to less looping.

I still don't have a Ryzen system to test with but Parsec and Wardog have been testing extensively and the above is my understanding of what they have found so far. I am sure when one of them see this they will clarify further.


-------------


Posted By: Brandon
Date Posted: 17 Apr 2017 at 1:56am
Originally posted by trainergames trainergames wrote:

" rel="nofollow - I am using a R7 1700 on an ASRock X370 Taichi. 

I am using a GSkill Ripjawz V 3200mhz kit, Model:F4-3200C16D-16GVKB

I tried XMP,i tried setting the speed and timings myself,i even tried setting the RAM voltage.

But it will only post at 2133Mhz.

I noticed that is it not on the QVL list,but i was hoping to at least get 2933mhz if the the 3200.


Can you guys please help me find out what is wrong?  Clear CMOS, this is important. Disable cool and quiet.
leave dram and soc voltage default. Set timings to 2667 16-15-15-15-36 1T. Try that out.


Posted By: Nerio_Lab
Date Posted: 17 Apr 2017 at 2:12am
Use A2 slot (the second from CPU)

-------------
The One


Posted By: trainergames
Date Posted: 17 Apr 2017 at 11:59am
Originally posted by Brandon Brandon wrote:

Originally posted by trainergames trainergames wrote:

http://forum.asrock.com/RTE_textarea.asp%3cmode=quote&ID=27403&CACHE=1526" rel="nofollow - I am using a R7 1700 on an ASRock X370 Taichi. 

I am using a GSkill Ripjawz V 3200mhz kit, Model:F4-3200C16D-16GVKB

I tried XMP,i tried setting the speed and timings myself,i even tried setting the RAM voltage.

But it will only post at 2133Mhz.

I noticed that is it not on the QVL list,but i was hoping to at least get 2933mhz if the the 3200.


Can you guys please help me find out what is wrong?  Clear CMOS, this is important. Disable cool and quiet.
leave dram and soc voltage default. Set timings to 2667 16-15-15-15-36 1T. Try that out.


Sorry for the late reply i was at work,i will try that now and post back how it goes.


EDIT:Ok i tried that and it didn't work.


Posted By: Onslaught2k3
Date Posted: 21 Apr 2017 at 11:30am
There are other causes for the dreaded F9 boot loop aside from bad RAM overclocking settings.  While Ultra Fast Boot is enabled, it WILL give you F9 boot loop from a cold boot.  Restarting or diasabling ultra fast boot doesn't seem to cause this issue.  Try increasing the VTT_DDR voltage to 0.7V and the DRAM voltage to 1.4V and see how that works with XMP enabled.  The Ripjaws V kit is not single-ranked Samsung B-die ICs.  There's a good chance you won't get far with them.

-------------
https://valid.x86.fr/lcd1c8" rel="nofollow">


Posted By: parsec
Date Posted: 21 Apr 2017 at 1:14pm
Originally posted by Onslaught2k3 Onslaught2k3 wrote:

There are other causes for the dreaded F9 boot loop aside from bad RAM overclocking settings.  While Ultra Fast Boot is enabled, it WILL give you F9 boot loop from a cold boot.  Restarting or diasabling ultra fast boot doesn't seem to cause this issue.  Try increasing the VTT_DDR voltage to 0.7V and the DRAM voltage to 1.4V and see how that works with XMP enabled.  The Ripjaws V kit is not single-ranked Samsung B-die ICs.  There's a good chance you won't get far with them.


Since Ryzen memory compatibility is virtually random at this point, even with so called Ryzen specific memory like the FlareX I use, would you please give us the full G.SKILL model name of your TridentZ memory? We can add it to our unofficial list of truly Ryzen compatible memory models. Wink  Thanks!

I'm giving up on compatibility lists provided by the memory manufactures. Models they say will work may not, and some models they don't list as compatible work better then the models they say are compatible. Angry

I agree about not using the Ultra Fast setting of the Fast Boot option. That setting really causes a reduced POST time, skipping or reducing some POST procedures. POST memory initialization and training is critical for Ryzen systems with the not ready for prime time AGESA versions from AMD yet.

The F9 boot loop you mentioned, with its starts and stops is caused by the AMD Advanced Boot Training option in the OC Tweaker screen. I'm not saying this option is bad, just that it is the cause of the start-stop behavior when memory settings fail POST. The latest versions of at least some ASRock Ryzen boards now have an option to set the number of repetitions of this option, in the Boot screen. The default is five reps. Have I ever had it succeed in training the memory to work at the applied settings? You (anyone) tell me. Confused

If you have a POST beep speaker connected to the board, and the POST beep option enabled in the UEFI, you'll get three POST beeps if the memory settings fail, before the board shuts off and restarts again. If it continues to fail, three more beeps. The last rep of the cycle will beep once, shut off, and then reset the memory to its default settings, and attempt to start the PC. More about this below.

If you want to interrupt that cycle, shut off the PSU with its power switch when the PC turns off. You also MUST clear the UEFI/BIOS with the board's jumper or Clr CMOS button (if your board has one) before you try starting the PC again. Otherwise it will start the on-off cycle over again.

If you endure all the stop start reps and don't clear the UEFI/BIOS, the board will set the memory to the default speed and timings, and boot into Windows, if it can work at default settings. But there's more to it than that. The default memory settings are only used for one boot of the PC. The memory settings you tried to use and failed are STILL SET in the UEFI! Restart the PC, or shut it down and start it later, and guess what happens? That's right, the start-stop cycle of the AMD Advanced Boot Training begins again.

I've had no faith in XMP profiles designed for Intel memory controllers, but are you saying you were able to use the XMP profile in your memory?

For the OP
, you are the second user of Ripjaws 5 memory that has not had success running it above its default speed in this forum with a Ryzen system. You may never get it much higher than that. Apparently, a user of Ripjaws 4 memory was successful running it at higher speeds.


-------------
http://valid.x86.fr/48rujh" rel="nofollow">


Posted By: Onslaught2k3
Date Posted: 21 Apr 2017 at 2:12pm
Hey, Parsec, 

They are F4-3200C14D-16GTZR, the kit I'm currently using.

The XMP profile loaded works 100%, with 0.7V on VTT_DDR.  0.68 seemed fine.  

I gathered the f9 boot loop was a memory initialization failure.  I'm taking it that the DDR4-3200 bootstrap may not be fully stable on my end.  The odd thing is that when doing memtest86, prime95 blended, aida64 there are no issues or crashes.  I only seem to have issues from a cold boot.  

On another note, I ran in to another oddity currently with v2.0 on the x370's BIOS.  I am able to (theoretically) bring my CAS timings down to 13 from the 14 default, but the Tcas timing stays stuck at 14 ONLY when trying to make it 13.  TRCD, TRP can go down to 13 with no issue.  Tras goes to 26, and Trc/Trfc are 75 and 312 respectively.  Command rate stays at 1T always.  This may work in A1 and B1 slots on the Taichi, but the Trfc skyrockets to ~562T from 312T.  When I try to lower the Tcas timing to 12, it won't boot so I assume it's a stability limit.  May need to test further but don't want to go beyond 1.55V on the DRAM.


-------------
https://valid.x86.fr/lcd1c8" rel="nofollow">


Posted By: datonyb
Date Posted: 21 Apr 2017 at 11:04pm
Originally posted by Onslaught2k3 Onslaught2k3 wrote:

Hey, Parsec, 

They are F4-3200C14D-16GTZR, the kit I'm currently using.

The XMP profile loaded works 100%, with 0.7V on VTT_DDR.  0.68 seemed fine.  

I gathered the f9 boot loop was a memory initialization failure.  I'm taking it that the DDR4-3200 bootstrap may not be fully stable on my end.  The odd thing is that when doing memtest86, prime95 blended, aida64 there are no issues or crashes.  I only seem to have issues from a cold boot.  

On another note, I ran in to another oddity currently with v2.0 on the x370's BIOS.  I am able to (theoretically) bring my CAS timings down to 13 from the 14 default, but the Tcas timing stays stuck at 14 ONLY when trying to make it 13.  TRCD, TRP can go down to 13 with no issue.  Tras goes to 26, and Trc/Trfc are 75 and 312 respectively.  Command rate stays at 1T always.  This may work in A1 and B1 slots on the Taichi, but the Trfc skyrockets to ~562T from 312T.  When I try to lower the Tcas timing to 12, it won't boot so I assume it's a stability limit.  May need to test further but don't want to go beyond 1.55V on the DRAM.


my f4 3200c15d-16gtz kit is doing similar
its supposed to be 15/15/15/35 and is set to that in bios both manually and in xmp

but it will set itself to 16/15/15/35 in windows despite going back into bios and keep setting to 15 it wont run it in windows even though its in the bios and boots cleanly cold or restart
(on a side not i can get 14/13/13/35 to restart and be stable but similar to you it will not cold boot at that)
i am waiting until agesa code 1005 is released before i waste any more time playing with the ram settings myself
current setting is
16/15/15/35 @3200 and 1.35volts  SOC @ 1.000v LLC level 2


-------------
[url=https://valid.x86.fr/jpg250][/url]

3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold


Posted By: Onslaught2k3
Date Posted: 21 Apr 2017 at 11:21pm
Thanks for your response, datonyb.

What's interesting is that I believe the BIOS is reserving timings for certain frequencies (ie. ddr4-2133 taking cas15, ddr4-3200 taking cas14, etc.).  I too hope that the next BIOS update will provide more fine-tuning for the memory timings/frequencies.


-------------
https://valid.x86.fr/lcd1c8" rel="nofollow">



Print Page | Close Window

Forum Software by Web Wiz Forums® version 12.04 - http://www.webwizforums.com
Copyright ©2001-2021 Web Wiz Ltd. - https://www.webwiz.net