Overclock.net banner

3321 - 3340 of 10217 Posts

·
Registered
Joined
·
1 Posts
Hey everyone,

System config:

CPU - AMD Ryzen 1600X under NH D-14

M/B - Asus Crosshair VI Extreme

RAM - Crucial Ballistix Elite DDR4 3466 [email protected]



According to Thaiphoon Burner this kit uses samsung b-die.

However, I can't get it to run anything faster than 2800 CL 15 for the life of me, although I've been trying for a week now. I've read lots of stuff, including reddit, guides here on the forum and just googled a lot.

XMP profile just won't POST no matter what i try, as well as Stilt's presets included in C6E bios.

I did find some correlation between TRFC/2/4 and lowest CL this kit can run. Using DRAM Calculator 3200 safe settings (16-16-16-32-307-228-140) the kit doesn't POST. Nor does it post @16-18-18-42 at anything above 2666. If I use TRFC settings which are in SPD profile (560/418/256) it will occasionally boot at something like 3000 15-15-15-34 but far from being stable. Not even once have I managed to boot it at 3200 14-14-14-34.

I tried bumping SOC voltage up to 1.17v (with LLC on SOC up to level 5) and DRAM voltage up to 1.5V. Also, messed with ProcODT (40-80 ohm). Btw, if I set everything in BIOS to auto and just set the DRAM clock to 3200, it boots with timings 20-22-22-53 trfc 560 with 1.34V on DRAM and 1.14V on SOC, passes some tests but I didn't even bother to test stability on these settings properly. It seems to me that the only things I haven't touched are dram vref voltage and cldp_vddp, will those improve the situation drastically or are they used to fine tune stability?

I've tried BIOS 3508, 6201, and the latest 6301 none of them making any noticeable difference.

I would very much appreciate any help I can get, i'm getting desperate at this point.
 

·
Registered
Joined
·
1,424 Posts
Today I did a new try to get 3733Mhz working on my Micron E-die kit.

Took 6 hours and now I found my "first" stable settings that pass cycles in TestMem5. I was about trying all and everything and was narrowing down things that didn't work and other things that didn't work but were just a nudge away from working. A lot of back and forth testing various settings to find what was giving less errors than others.

It finally clicked together... then... cycle 2 fail... NO!... HEAT! for sure the memory was giving way to much heat at the voltage I was using, 1.490V.
Added a fan on my memory and now it runs several cycles more.. Closed the side panel and it seems it failed after a while again.

It for sure is heat that is my issue here.

Almost a success image attached. failed a second after the image was taken.

Doing a retry with open chassis.

I found out voltage works a little finicky. Every 2-3 straps is more stable than the rest giving less issues. Good voltages are 1.380 , 1.410 , 1.430 , 1.460 , 1.490. Values in between are utter failure in stability. 1.490 was the "click" (going from less to more) but at the same moment that voltage causes too much heat running this speed. More voltage is fine @ 3600Mhz with other settings.
 

Attachments

·
Registered
Joined
·
32 Posts
Hey @1usmus and everyone.
Just reporting i could successfully oc my RAM F4-3200c14d-16gtzr using 3200 fast preset and its been stable for several days. (memtest 120%+ run and tm5 no errors plus everyday running without issues)

***EDIT: How can i change BGSalt to disable on my motherboard? I cant seem to find it (since calculator says to change to off for gaming)***

Anything else i can squeeze from my ryzen (running stock cooler atm) for gaming purposes?
Perhaps tuning llc for lower voltages and lower temps for better auto oc?
Any guide for cpu core setup you recommend like this one for ram? Thanks and great work!
 

Attachments

·
RadeonMOD Author
Joined
·
741 Posts
Today I did a new try to get 3733Mhz working on my Micron E-die kit.

Took 6 hours and now I found my "first" stable settings that pass cycles in TestMem5. I was about trying all and everything and was narrowing down things that didn't work and other things that didn't work but were just a nudge away from working. A lot of back and forth testing various settings to find what was giving less errors than others.

It finally clicked together... then... cycle 2 fail... NO!... HEAT! for sure the memory was giving way to much heat at the voltage I was using, 1.490V.
Added a fan on my memory and now it runs several cycles more.. Closed the side panel and it seems it failed after a while again.

It for sure is heat that is my issue here.

Almost a success image attached. failed a second after the image was taken.

Doing a retry with open chassis.

I found out voltage works a little finicky. Every 2-3 straps is more stable than the rest giving less issues. Good voltages are 1.380 , 1.410 , 1.430 , 1.460 , 1.490. Values in between are utter failure in stability. 1.490 was the "click" (going from less to more) but at the same moment that voltage causes too much heat running this speed. More voltage is fine @ 3600Mhz with other settings.
Grats, good effort.
Im trying to get >3500MHz now but with CPU @ 3.96GHz ;)
3639MHz RAM is doable on my RIG with 16-17-17-17 1T GD
 

·
Premium Member
Joined
·
20,154 Posts
Can anyone beat this Mem Latency?
 

Attachments

·
AMD overclocker/developer
Joined
·
1,651 Posts
Discussion Starter #3,329 (Edited)
Thank you for the advice, senpai.
I'm reworking my entire approach, I was coming from an old school methodology: raise it all to the limits, work your primaries down till they don't post, work the voltages, tweak the secondaries; finalize with the stable and the reasonable according to your own premisses.

I already foun better results: verbatim from the calc's recommendations; pic below. 100% is already a win for me after seeing it fail so many times before I could even open the 4th instance - actually that's why I grew fond of it, it'll tell you almost immediatly when you're being plain stupid (and that's a good thing).

I also have two extra questions:
first, you still didn't invent the import profile button, neither the r-xmp (up to the the 66th page, december 2017). So i've been importing values from an html export from thaiphoon, should have I been using the r-xmp. I noticed that values it generates are quite tighter than the exported html and i'm not quite sure what they're showing me.
second, you had this (2nd pic) posted way back, which implies that undershooting vtt_ddr by a small margin is better for stability. I had a suspicion derived from my previous lame attempts that overshooting it (on the napkin formula ddr voltage divided by 2) gave slightly better stability - especially since the granularity for the crosshair vii on those settings is quite bad; 250mv steps, so you either lock yourself on multiples of 4 or you always end up over or undershooting it.

edit: bah it failed at 343% in one instance. gonna try it at the next voltage bump.
1) I think you should try to import an XMP profile at the very least, you shouldn't focus on it, it's not the best solution for Ryzen. Perhaps in the future I will add automatic imports (I have little time for such things, so if anyone wants to help with the implementation, I will be happy to talk).
2) This picture was relevant at the time of publication, with each microcode PMU / SMU has changes, maybe now everything is different. Your find is permissible and natural. Margin is in both directions :)

5h of testing, Tested BIOS 6301 & 6101 -> Close but no cigar
Can't stabilize at 3500MHz CL15-15-15-15 1T GD (shows CL16) nowhere near the 1403 results !

My RAM 'likes' Low CL and +2 on other
4133 CL19-23-23-23 & 4000MHz CL19-21-21-21 (1.35v Kits so there is room for improvement)
For my setup Best is BIOS 1403 (It's magical :D)

Yo, Now im stable with this one (fresh pic ;))
@1usmus I know, now im trying to get Stable LLT at 3400 or 3500MHz (Hoping for some tighter secondary timings)

==
I disassembled the test bench with CH6, to be honest, I was very tired of the fact that the company Asus decided to deprive this product of support. The latest bios was the worst in history for this board (IMHO). But I like your diligence, maybe you can discover something new :)

It is usually between 1, 2 and 6 test.
These tests are basic for a stable system. It is very strange that GSAT does not find errors.
Are there any errors in games and programs?

you know when you're trying to get to the next milestone in your stable testing and then you start bumping this, that and at some point it starts spiraling down again and giving out errors at 10-50% again?
(or worse, immediatly)
yeah. to hell with that 3266 multiplier freq.
starting anew with 3200, presumably conservative settings. I just want a stable platform to start working on upwards, I'm tired of always going downwards.
got new cad_bus settings too, and bumped up the procODT to 60ohms.
I always have the twr to fall back to, raising it to 12. Also that sexy tRRDS of 4 (tFAW x6) could get bumped if I encounter another error.

edit: bye bye twr of 10, you were loved very dearly during all of your 247% :sadsmiley (please don't take my tRRDS and scl's)
the problem is in the signals, here are the optimal settings for your system

procODT 48 / 53
RTT PARK 40 or 48

Soc 0.98 - 1.03 v , DIMM 1.36 v

sometimes you need to swap modules А2<->B2

Try SOC at 1.187v or 1.193v (when observed it shows 1.155v actual)
your proposal is relevant only for very high frequencies + the higher the voltage for the SOC, the lower the probability of a stable system

Trying to go below 60 on this G.Skill 3200 FlareX. Any tips? Currently using 3466 Fast Preset. Tfaw 67 won't stick. It goes down to 54.
You entered incorrect values, on the first page there is a video instruction ;)

First time noticing V1 / V2 option. I knew I don't have high quality RAM and I didn't change it, I thought it has to do with XMP version. I knew that 3200c14/3600c14 is HQ and 3200c15/c16 is medium quality RAM, I have 3200c15 b-die, so I choose 3333 fast V2 and I never been so stable. I even got 25mhz more on CPU. The gaming feels so smooth in every second, before I noticed some hiccups every 10-15 seconds.
3400 fast v2 is somewhat less stable, will experiment more later with 3400 and 3466.
I have 3000CL14 modules, they work fine with the V1 profile, I also have 3400CL16 and they also work perfectly with V1. V2 is in most cases great for the Patriot Corsair and so on.

+

https://www.addictivetips.com/windows-tips/empty-standby-memory-fix-game-stuttering-windows-10/

Hey guys,
thanks @1usmus for the great calculator!

I'm using Crucial Ballistix Sport Memory Kit 2x8gb DR 2666Mhz 16-18-18-38-61 which uses Micron D-Die according to Thaiphoon. But experiencing problems using the Micron D-Die selection in your dram calculator. Program keeps giving me error message "input string was not in a correct form". I know the support for Micron D-Die is a just recently added feature, just so you're aware of that. Calculation of other Die-types works just fine.

So far I got my memory running stable @ 3200Mhz 16-19-19-36-61, 1,35V DRAM and 1,1V SoC-Voltage on a ASRock AB350 pro4 with Ryzen 1700X @3700Mhz 1,35625V Vcore via pStates, everything else left untouched. Wanted to start going in with secondary and tertiary timings but don't really know where to start since I'm lacking experience in that and was hoping your tool would give me a hint or first-start on that.

Thanks again for your great work!
Try to enter data manually. You have a good result of overclocking this memory. Show RTC and error please
 

·
Registered
Joined
·
1 Posts
No matter how low the frequency I set is, it just won't stabilize enough to run a stress test for 10 seconds. I'm new at this, can someone give me a light of what is going on? I'm following the steps included in the program but nothing has helped at all. I have no problems with temperatures, CPU is always between 30~60c and I can keep my CPU stable at 3.9GHz 1.37v and RAM 3000MHz 1.34MHz but with timings on auto.

Ryzen 1700X btw.
 

·
AMD overclocker/developer
Joined
·
1,651 Posts
Discussion Starter #3,331
Hey everyone,

System config:

CPU - AMD Ryzen 1600X under NH D-14

M/B - Asus Crosshair VI Extreme

RAM - Crucial Ballistix Elite DDR4 3466 [email protected]



According to Thaiphoon Burner this kit uses samsung b-die.

However, I can't get it to run anything faster than 2800 CL 15 for the life of me, although I've been trying for a week now. I've read lots of stuff, including reddit, guides here on the forum and just googled a lot.

XMP profile just won't POST no matter what i try, as well as Stilt's presets included in C6E bios.

I did find some correlation between TRFC/2/4 and lowest CL this kit can run. Using DRAM Calculator 3200 safe settings (16-16-16-32-307-228-140) the kit doesn't POST. Nor does it post @16-18-18-42 at anything above 2666. If I use TRFC settings which are in SPD profile (560/418/256) it will occasionally boot at something like 3000 15-15-15-34 but far from being stable. Not even once have I managed to boot it at 3200 14-14-14-34.

I tried bumping SOC voltage up to 1.17v (with LLC on SOC up to level 5) and DRAM voltage up to 1.5V. Also, messed with ProcODT (40-80 ohm). Btw, if I set everything in BIOS to auto and just set the DRAM clock to 3200, it boots with timings 20-22-22-53 trfc 560 with 1.34V on DRAM and 1.14V on SOC, passes some tests but I didn't even bother to test stability on these settings properly. It seems to me that the only things I haven't touched are dram vref voltage and cldp_vddp, will those improve the situation drastically or are they used to fine tune stability?

I've tried BIOS 3508, 6201, and the latest 6301 none of them making any noticeable difference.

I would very much appreciate any help I can get, i'm getting desperate at this point.
You have problems with procODT + RTT, I am sure you need to try all the options offered by the calculator.

It is also possible your memory loves the CL16 and CWL16 :)



Today I did a new try to get 3733Mhz working on my Micron E-die kit.

Took 6 hours and now I found my "first" stable settings that pass cycles in TestMem5. I was about trying all and everything and was narrowing down things that didn't work and other things that didn't work but were just a nudge away from working. A lot of back and forth testing various settings to find what was giving less errors than others.

It finally clicked together... then... cycle 2 fail... NO!... HEAT! for sure the memory was giving way to much heat at the voltage I was using, 1.490V.
Added a fan on my memory and now it runs several cycles more.. Closed the side panel and it seems it failed after a while again.

It for sure is heat that is my issue here.

Almost a success image attached. failed a second after the image was taken.

Doing a retry with open chassis.

I found out voltage works a little finicky. Every 2-3 straps is more stable than the rest giving less issues. Good voltages are 1.380 , 1.410 , 1.430 , 1.460 , 1.490. Values in between are utter failure in stability. 1.490 was the "click" (going from less to more) but at the same moment that voltage causes too much heat running this speed. More voltage is fine @ 3600Mhz with other settings.
Thanks for the information, at high frequencies I also noticed that there is a dead voltage gap in the RAM. And maybe it's the same for everyone.
Your voltages are currently the same as mine for high frequencies, 1.43, 1.46 and 1.49 :thinking:

LLC do not touch, automatic mode works well in most cases, and manual mode may require to achieve extreme results



Hey @1usmus and everyone.
Just reporting i could successfully oc my RAM F4-3200c14d-16gtzr using 3200 fast preset and its been stable for several days. (memtest 120%+ run and tm5 no errors plus everyday running without issues)

***EDIT: How can i change BGSalt to disable on my motherboard? I cant seem to find it (since calculator says to change to off for gaming)***

Anything else i can squeeze from my ryzen (running stock cooler atm) for gaming purposes?
Perhaps tuning llc for lower voltages and lower temps for better auto oc?
Any guide for cpu core setup you recommend like this one for ram? Thanks and great work!

try tWTRL 8 and tFAW 16, you may need to increase the RAM voltage by one step
BGS is off on screenshots

in your case it is better not to touch LLC, MSI auto mode works well
, all fine :)
 

·
Premium Member
Joined
·
20,154 Posts
No matter how low the frequency I set is, it just won't stabilize enough to run a stress test for 10 seconds. I'm new at this, can someone give me a light of what is going on? I'm following the steps included in the program but nothing has helped at all. I have no problems with temperatures, CPU is always between 30~60c and I can keep my CPU stable at 3.9GHz 1.37v and RAM 3000MHz 1.34MHz but with timings on auto.

Ryzen 1700X btw.

You might want to try working on the RAM while CPU is at Stock. That and follow the video in the op, which what im about to do. You need to manually set all the data in the Calc. Avoiding as much Auto settings as posible. Make sure you save a profile of your 3.9GHz OC in BIOS before setting Optimized Default, then play with RAM.

@1usmus, thanks. I'll fix it.
 

·
DRAM Obsessor
Joined
·
1,033 Posts

·
Overclock.net user
Joined
·
447 Posts
Stable ram settings

Hi 1usmus, really happy to see you here.
Thank you for all your support. Your calculator is really an help tool for Ryzen.
Much appreciated.

For who is interested, I'm stable from the day one of the CH6 bios release 6301 (and no problems or fan issues).
With this last releases finally I've reach a stable (near Rock solid) configuration with a little overclock of my usual (24/7) ram settings of 3466 Mhz with tight timings (sammy b-die RGB kit 3600c16). 101 Bck give me the "old" 3500 MHz for ram with the timings of 3466 MHz. Also with the Asus CB11.5 trick activated! ;)
This is a screen shot with some test done to compare.
 

Attachments

·
Registered
Joined
·
11 Posts
Can not understand Step 4

Cannot understand step 4 , what does it mean ?

do i have match the readings from thaiphoon burner with Dram calculator ?
do i have to check if values on both software match ?


 

·
Registered
Joined
·
1,424 Posts
@Nighthog
Very nice result :). Also nice IMC and nice Ram. Going for DDR4-3800 now :D?
It can boot but it's so damn unstable better left to not try again.

Grats, good effort.
Im trying to get >3500MHz now but with CPU @ 3.96GHz ;)
3639MHz RAM is doable on my RIG with 16-17-17-17 1T GD
thanks. though I was celebrating a little premature.

Thanks for the information, at high frequencies I also noticed that there is a dead voltage gap in the RAM. And maybe it's the same for everyone.
Your voltages are currently the same as mine for high frequencies, 1.43, 1.46 and 1.49 :thinking:
I have to add that I didn't get better than to cycle 4 at best before I tried to tighten my primary timings. I still would get fails around cycle 3-4 at best in random parts and had to look elsewhere as something else was not 100%.

I went back to my secondary timings and adjusted tWTRS, tWTRL and it increased success rate to pass all 5 cycles!
Will need to try the primary timings again for tightening to see how far I can go.
tCL 14 needs 1.500++ volts so unlikely my cooling sufficient at the moment, will stick to tCL 16 for now.
 

·
Premium Member
Joined
·
20,154 Posts
Sure, just let me start up Photoshop...
Haha. Its a fluke. Did twice. Can't get any lower than 61. Prolly use 3533. But, even at stock 3200, the RAM temps gets close to 45 during gaming. Some parts of Africa can get really warm during summer. A/C struggles.
 

Attachments

·
AMD overclocker/developer
Joined
·
1,651 Posts
Discussion Starter #3,338
Cannot understand step 4 , what does it mean ?

do i have match the readings from thaiphoon burner with Dram calculator ?
do i have to check if values on both software match ?


You must use the profile V1 or V2.

0) chose the type of memory, the frequency that we want to get
1) chose V1 or V2 profile
2) pressed the R-XMP
3) pressed "Safe" or "Fast"
4) done

In the debug profile, values are entered that are in Thaiphoon.
There is no need to compare anything, you just have to enter these numbers or make an import, as in the video instructions.
 

·
Registered
Joined
·
10 Posts
Hey everyone,

System config:

CPU - AMD Ryzen 1600X under NH D-14

M/B - Asus Crosshair VI Extreme

RAM - Crucial Ballistix Elite DDR4 3466 [email protected]



According to Thaiphoon Burner this kit uses samsung b-die.

However, I can't get it to run anything faster than 2800 CL 15 for the life of me, although I've been trying for a week now. I've read lots of stuff, including reddit, guides here on the forum and just googled a lot.

XMP profile just won't POST no matter what i try, as well as Stilt's presets included in C6E bios.

I did find some correlation between TRFC/2/4 and lowest CL this kit can run. Using DRAM Calculator 3200 safe settings (16-16-16-32-307-228-140) the kit doesn't POST. Nor does it post @16-18-18-42 at anything above 2666. If I use TRFC settings which are in SPD profile (560/418/256) it will occasionally boot at something like 3000 15-15-15-34 but far from being stable. Not even once have I managed to boot it at 3200 14-14-14-34.

I tried bumping SOC voltage up to 1.17v (with LLC on SOC up to level 5) and DRAM voltage up to 1.5V. Also, messed with ProcODT (40-80 ohm). Btw, if I set everything in BIOS to auto and just set the DRAM clock to 3200, it boots with timings 20-22-22-53 trfc 560 with 1.34V on DRAM and 1.14V on SOC, passes some tests but I didn't even bother to test stability on these settings properly. It seems to me that the only things I haven't touched are dram vref voltage and cldp_vddp, will those improve the situation drastically or are they used to fine tune stability?

I've tried BIOS 3508, 6201, and the latest 6301 none of them making any noticeable difference.

I would very much appreciate any help I can get, i'm getting desperate at this point.
I have the same memory, try these settings:
soc votlage 1.03125 LLC2
ddr voltage 1.365

I did a lot of testing with this memory and is a bit tricky to get stable,
but these settings run 100 stable for me (tested with tm5 10 cycles, ram test 2400%, mem test pro 2400%, prime 2h blend)

important for me with these dimms:
- cl 16-16-16-16 crashes instantly, cl 16-17-16-16 works fine, but trCDWR has to be 17 even at 3200 mhz)
- cl 14 cant get stable at all
- cl15 2T geardown disabled also works, eg: cl 15-17-16-16
- TRDRDSCL 4 (cant do lower)
- TWRWRSCL 4 (cant do lower)
- Ram stability best at 1.365 -1.370, after that they get less stable.
- TRFC can go as low as ~ 255ns, im using (default for ram is 350ns)
 

Attachments

·
Registered
Joined
·
3,326 Posts
I want to thank 1usmus for his work. I got two sets of Hynix 16 Gigabyte kits stable using the DRAM Calculator. I tried everything on my own and basically got nowhere and gave up.
 
3321 - 3340 of 10217 Posts
Top