Ryzen 3950X Build Help

[erek]

i set my timings to 16-20-20-20-39-58
tRFC to 700
PROC ODT to 60 Ohms

SoC 1.1V, DRAM 1.360V

Still erroring at the same memory address on cpu 24

 

[A Computer Guy]

Look like progress. Step your DRAM voltage down to 1.35v and see if it is better or worse, then step 1.365v. sometimes more voltage isn't always better.

I don't see the nanosecond info for your prior screenshots.

I don't have my Corsair ram installed with XMP but here is an example:
(I would expect to see a section specifically for XMP.)

http://imgur.com/a/hV60RFv

You could try also settings BGS=diabled and BGS-ALT=enabled to see if that helps.

 

[erek]

So even non XMP 2133MHz getting errors on the same CPU and sane address

 

[A Computer Guy]

So even non XMP 2133MHz getting errors on the same CPU and sane address

View attachment 137998

Humm...3 passes ok 4th pass not good.

Dram Calculator doesn't have option for Hynix D-die option so this is a stab in the dark.

Just to try...

18-20-20-20-42-68 and match your tCWL to your tCL (a even number)

maybe bad sub-timing...some things to try...

tCKE=8
trdrdscl=6
twrwrscl=6
twrwrsd=8
twrwrdd=8
trdrdsd=6
trdrddd=6

 

[erek]

Welp got errors on different cpu and different memory address with totally different ram 4x8GB not even XMP

 

[Deleted member 185158]

What was the cpu temp again?

 

[A Computer Guy]

Here is where I am getting my current ideas from regarding the sub-timings...
( dram calculator doesn't have option for Hynix D-die option so this is a stab in the dark. )
( this calc screenshot is not accurate but maybe close enough to get an idea of something to try. )

http://imgur.com/a/Jcd1RXb

You could try SOC 1.125v (or 1.05v) but keep in mind in BIOS/UEFI the voltage you enter is not always the voltage you end up with so be very careful as you inch to 1.2v on SOC.

If someone has any Zen2 CPU and this ram kit we might get a good example of something to try if they are kind enough to post their Ryzen Master screenshot of their memory settings.

What was the motherboard again and what BIOS version are you on? This CPU is so new you might reach out to the motherboard vendor for help.

 

[erek]

What was the cpu temp again?

68c is the max I’ve seen in memtest but 53c running and up to 11 errors now on this other set of ram, far worse

Here is where I am getting my current ideas from regarding the sub-timings...
( dram calculator doesn't have option for Hynix D-die option so this is a stab in the dark. )
( this calc screenshot is not accurate but maybe close enough to get an idea of something to try. )

http://imgur.com/a/Jcd1RXb

You could try SOC 1.125v (or 1.05v) but keep in mind in BIOS/UEFI the voltage you enter is not always the voltage you end up with so be very careful as you inch to 1.2v on SOC.

If someone has any Zen2 CPU and this ram kit we might get a good example of something to try if they are kind enough to post their Ryzen Master screenshot of their memory settings.

What was the motherboard again and what BIOS version are you on? This CPU is so new you might reach out to the motherboard vendor for help.

Latest bios for the crosshair viii hero WiFi, I cant even remotely get the other set of ram to run without errors at non XMP 2133 stock : auto

 

[A Computer Guy]

68c is the max I’ve seen in memtest but 53c running and up to 11 errors now on this other set of ram, far worse



Latest bios for the crosshair viii hero WiFi, I cant even remotely get the other set of ram to run without errors at non XMP 2133 stock : auto

I feel like too many variables are changing at once. If swapping ram kits then be sure to reset all ram settings to auto as different kits may react differently to the stored settings we've been tinkering with.

I'm inclined to say we would need to start over with full AUTO and one ram stick in A2. Then at 2133 or 2400 re-validate setting one option at a time.

 

[erek]

I feel like too many variables are changing at once. If swapping ram kits then be sure to reset all ram settings to auto as different kits may react differently to the stored settings we've been tinkering with.

I'm inclined to say we would need to start over with full AUTO and one ram stick in B2. Then at 2133 or 2400 re-validate setting one option at a time.

It’s been reset full auto,this is the first attempt with no changes still

 

[A Computer Guy]

It’s been reset full auto,this is the first attempt with no changes still

Ok so which kit are you working with Crucial or GSkill? only 1 dimm in slot B2 correct?
Also are you still using that PSU from the screenshot with all the dust in it?

 

[erek]

Ok so which kit are you working with Crucial or GSkill? only 1 dimm in slot B2 correct?
Also are you still using that PSU from the screenshot with all the dust in it?

I’m stripped back one stick of the crucial in slot B2, yes still using that but it’s been thoroughly cleaned now

 

[Deleted member 185158]

That Crucial memory is no good for your configuration. Put the G.Skill back in.

Clear cmos.

Then post your XMP profile.

Take screen shots of all the memory timings including sub timings.

Then go to Memory SPD tab in bios (advanced?) and take a screen shot of that as well.

 

[erek]

That Crucial memory is no good for your configuration. Put the G.Skill back in.

Clear cmos.

Then post your XMP profile.

Take screen shots of all the memory timings including sub timings.

Then go to Memory SPD tab in bios (advanced?) and take a screen shot of that as well.

Bios

 

[A Computer Guy]

I’m stripped back one stick of the crucial in slot B2, yes still using that but it’s been thoroughly cleaned now

I have to sign-off for awhile. ShripBrime has good suggestion.

First we need all the info up front to be more methodical.
Second since you're having issues I would approach with validation at JEDEC timings first to see if we have any baseline of stability to work from. If it doesn't work properly at JEDEC speeds then it probably won't work until that issue is fixed. ( I can't tell if Asus recent bios is AGESA 1.0.0.4 with Patch B or something else, maybe that is needed. )
Third screen shot all settings and compare to see what we can get working properly vs. what is supposed to work from step 1.
Then plan strategy for changing and testing.

( I was also worried about the PSU, how old is it? )

 

[Deleted member 185158]

Yea, we can start at Jedec values. Need to see more than one in the bios, will need a screen shot of the SPD tab from cpu-z also.

This is D-die (Samsung?) memory correct? Let me see if I can find the white papers really quick.

Yea here's white papers for referencing.

https://www.samsung.com/semiconductor/global.semi/file/resource/2017/11/DS_64GB_TSV_DDR4_4Gb_D_die_RegisteredDIMM_Rev13.pdf

 

[erek]

I have to sign-off for awhile. ShripBrime has good suggestion.

First we need all the info up front to be more methodical.
Second since you're having issues I would approach with validation at JEDEC timings first to see if we have any baseline of stability to work from. If it doesn't work properly at JEDEC speeds then it probably won't work until that issue is fixed. ( I can't tell if Asus recent bios is AGESA 1.0.0.4 with Patch B or something else, maybe that is needed. )
Third screen shot all settings and compare to see what we can get working properly vs. what is supposed to work from step 1.
Then plan strategy for changing and testing.

( I was also worried about the PSU, how old is it? )

Corsair ax1200
Psu is from 2011

Yea, we can start at Jedec values. Need to see more than one in the bios, will need a screen shot of the SPD tab from cpu-z also.

This is D-die (Samsung?) memory correct? Let me see if I can find the white papers really quick.

Yea here's white papers for referencing.

https://www.samsung.com/semiconductor/global.semi/file/resource/2017/11/DS_64GB_TSV_DDR4_4Gb_D_die_RegisteredDIMM_Rev13.pdf

 

[Deleted member 185158]

Oh that's Hynix memory. NVM my above post. disregard. thanks.

This is Hynix Dual Rank PDF> https://www.skhynix.com/product/filedata/fileDownload.do?seq=8247

 

[erek]

Oh that's Hynix memory. NVM my above post. disregard. thanks.

This is Hynix Dual Rank PDF> https://www.skhynix.com/product/filedata/fileDownload.do?seq=8247

With the G.Skill in it consistently has errors only starting at 7641MB and usually CPU 24, but did see once 23.

 

[Deleted member 185158]

Corsair ax1200
Psu is from 2011




View attachment 138015

If JEDEC #6 isn't stable, you'll have to loosen cas to 16 from 15 and try again. All other timings should be similar if not identical.

You can test up to 1.26v at JEDEC which should keep you in spec for that speed.

 

[erek]

If JEDEC #6 isn't stable, you'll have to loosen cas to 16 from 15 and try again. All other timings should be similar if not identical.

You can test up to 1.26v at JEDEC which should keep you in spec for that speed.

Based on jedec 6, but 16-16-16 and 1.25v I’ve ran through multiple passes of the targeted CPU 24 and 7600-8000m address range for the typical test that were erroring, Got error

 

[A Computer Guy]

Based on jedec 6, but 16-16-16 and 1.25v I’ve ran through multiple passes of the targeted CPU 24 and 7600-8000m address range for the typical test that were erroring, Got errorView attachment 138033

Ok now swap the current stick in B2 with the other gskill stick and retest. Let's see if one stick of the pair may be defective or if we get the same result.

 

[Deleted member 185158]

Well what does the log say?

Go to MemTest86.log.

And also lemme quote from memtest.

3 Troubleshooting Memory Errors Elease be aware that not all errors reported by MemTest86 are due to bad memory. The test implicitly tests the CEUl 1 and 2 caches as well as the motherboard. It is impossible for the test to determine what causes the failure to occur. Howeverl most failures will be due to a problem with memory module. hen it is notl the only opton is to replace parts untl the failure is corrected. Sometmes memory errors show up due to component incompatbility. A memory module may work fne in one system and not in another. This is not uncommon and is a source of confusion. In these situatons the components are not necessarily bad but have mariinal conditons that when combined with other components will cause errors. Ofen the memory works in a diferent system or the vendor insists that it is iood. In these cases the memory is not necessarily bad but is not able to operate reliably at full speed. Sometmes more conservatve memory tminis on the motherboard will correct these errors. In other cases the only opton is to replace the memory with beter qualityl hiiher speed memory. Don't buy cheap memory and eupect it to work reliably. On occasion "block move" test errors will occur even with name brand memory and a quality motherboard. These errors are leiitmate and should be corrected. All valid memory errors should be corrected. It is possible that a partcular error will never show up in normal operaton. Howeverl operatni with mariinal memory is risky and can result in data loss and even disk corrupton. ven if there is no overt indicaton of problems you cannot assume that your system is unafected. Sometmes intermitent errors can cause problems that do not show up for a loni tme. You can be sure that Murphy will iet you if you know about a memory error and iinore it. e are ofen asked about the reliability of errors reported by MemTest86. In the vast majority of cases errors reported by the test are valid. There are some systems that cause MemTest86 to be confused about the size of memory and it will try to test non-euistent memory. This will cause a larie number of consecutve addresses to be reported as bad and ienerally there will be many bits in error. If you have a relatvely small number of failini addresses and only one or two bits in error you can be certain that the errors are valid. Also intermitent errors are without eucepton valid. Frequently memory vendors queston if MemTest86 supports their partcular memory type or a chipset. MemTest86 is desiined to work with all memory types and all chipsets. MemTest86 cannot diainose many types of EC failures. For euample a faulty CEU that causes indows to crash will most likely just cause MemTest86 to crash in the same way.

In short, I'm thinking you are making multiple passes with success and fail on a 4th. Could be coincidence it just happens to throw an error the same time over and over.

 

[erek]

Ok now swap the current stick in B2 with the other gskill stick and retest. Let's see if one stick of the pair may be defective or if we get the same result.

run just one stick only in B2? i did previously swap both sticks at the same time, it was the same results

same memory location, and same CPU reported, it was 7641MB again

 

[Deleted member 185158]

Try OCCT in windows.

OCBASE/OCCT : Free, all-in-one stability, stress test, benchmark and monitoring tool for your PC

Ocbase is the home of OCCT, the most popular all-in-one stability / stress testing / benchmarking / monitoring tool available for PC

www.ocbase.com