Overclock.net banner

4741 - 4760 of 4797 Posts

·
Registered
Joined
·
291 Posts
Hello. New 'here. Board: X570 Aorus Elite Wifi (F30a). Memory: Ballistix Sport LT (E-Die) 2x16GB 3200 stock (Now running @3733 1.38v)
So..... on my first run I was able to do 3733 16-20-16-38-60 at 1.35V. I ran memtest up to 400% and reported no errors, but GSAT was spitting errors like no tomorrow (miscompare on CPU). I raised the voltage to 1.38 and ran it again for 8 hours without issues at all.

Does anyone know what this miscompare error is? Also, is it too bad to be golden on Memtest but have errors on GSAT?
BTW, is ProcODT at 53.3 too high? Some people ar running it at 40 something or 34, but i really don't know.

I attach screenshots of zen timings, memtest and a photo of GSAT (I can't record a log since I run it from a flashdrive).

IMG_20200912_104523.jpg ZenTimings_Screenshot.png memtest pass.png
 

·
BOINC Cruncher
Joined
·
902 Posts
Hello. New 'here. Board: X570 Aorus Elite Wifi (F30a). Memory: Ballistix Sport LT (E-Die) 2x16GB 3200 stock (Now running @3733 1.38v)
So..... on my first run I was able to do 3733 16-20-16-38-60 at 1.35V. I ran memtest up to 400% and reported no errors, but GSAT was spitting errors like no tomorrow (miscompare on CPU). I raised the voltage to 1.38 and ran it again for 8 hours without issues at all.

Does anyone know what this miscompare error is? Also, is it too bad to be golden on Memtest but have errors on GSAT?
BTW, is ProcODT at 53.3 too high? Some people ar running it at 40 something or 34, but i really don't know.

I attach screenshots of zen timings, memtest and a photo of GSAT (I can't record a log since I run it from a flashdrive).

View attachment 2458749 View attachment 2458750 View attachment 2458754
Read the author's explanation. Miscompare reason
 
  • Rep+
Reactions: Leito360

·
Premium Member
Joined
·
5,017 Posts
More voltage often is NOT better. If I have my RAM at 1.47v, VOC at .900v and VDDGs at .975v and 1.025v GSAT fails immediately.

If I reduce the VDDGs to .950v and 1.000v GSAT passes just fine. It also happens when I raise my RAM voltage on my Dual Rank b-dies.
 

·
Registered
Joined
·
291 Posts
More voltage often is NOT better. If I have my RAM at 1.47v, VOC at .900v and VDDGs at .975v and 1.025v GSAT fails immediately.

If I reduce the VDDGs to .950v and 1.000v GSAT passes just fine. It also happens when I raise my RAM voltage on my Dual Rank b-dies.
Of course, the less the voltage the better, unfortunately, it was the only way to pass the GSAT. I will try with less voltage and some other values later and see how it goes.
What is a good value for ProcODT?
 

·
Registered
Joined
·
5 Posts
@Veii
tWR <16 nicht stabil bekommen, bevor ich CsOdtDrvStr von 24 auf 20 gesenkt habe.
Tatsächlich war zuvor mit höherem procODT mit CsOdtDrvStr 20 auch instabil. ( nur in HCIM )
Es muss also eine Beziehung zwischen tWR und CsOdtDrvStr geben
 

·
Registered
Joined
·
882 Posts
Please, this is an English speaking forum
 

·
Registered
Joined
·
20 Posts
It shouldn't do much to the boosting result, but you can work with per CCX or later a fine adjusted PBO
PBO without voltage limits only does harm and nothing good
Your voltages are low, it has no scaling inside
VDDP -> VDDG is 100mV.
VDDG -> SOC is 150mV
unsure if this will preserve stability, but if you go with 75mV, it should work
Example:
cLDO_VDDP 825mV
VDDG IOD (is your board daisy chain?) 875
VDDG CCD 925
VSOC 1050

VDDG without splitting it would be 900mV
With a split you can push CCD higher where avg VDDG is 900mV between both
Yes i ment to disable PBO unless you know your exact workload voltages, know which CCX are good and know how to PBO (TDC,EDC) limit them
Continue from the quote of this discussion, to the next page
Should forward you to the CH7 page and that to the X370 Taichi one

But PBO will mess up the memory OC, like every core OC normaly would
For fixed allcore, focus on staying under 1.2875v vCore ~ till you know your actual safe upper limits
4.1ghz AC should work under 1.1, 4ghz under 1v and 4.2 near 1.2175v/1.2v

Without PBO, it's suggested to use 1usmus powerprofile or the latest ryzen balanced with enabled CPPC and CPPC preferred cores functionality from the bios
Hey, it's me again! Thanks for helping me back then.
Do you happen to have a working link to that discussion? Seems like it's broken. Wanna try your suggestions on a per-CCX overclock (just discovered my motherboard added it with a BIOS update).
 

·
Registered
Joined
·
88 Posts
Hi
For my configuration / timings I think the latency (64.0ns) is a very high value.
What do you think?
2458803
 

·
Edgy & on the edge
Joined
·
1,529 Posts
Hi
For my configuration / timings I think the latency (64.0ns) is a very high value.
What do you think?
View attachment 2458803
Looks normal to me. You'll have to OC your CPU and go way tighter on the memory to get 63ns and under.
 
  • Rep+
Reactions: jcpq

·
Registered
Joined
·
882 Posts
I've seen systems here with the same CPU and the same timings where there are less than 62ns.
I do not understand.
Your AIDA screenshot shows the frequency of your CPU when you ran the benchmark.

The 64ns is because your CPU is only running at 4250mhz.

If you ran an all core overclock of 4.3Ghz and run the benchmark you score would be less than 63ns and if you ran a core core overclock of 4.4Ghz it would be less than 62ns ...
 

·
Registered
Joined
·
882 Posts
So finally tuned my second set of Vipers 4400mhz, not as bad as i intially thought but not comparative to the set in my sig

2458812


Perhaps on a better board they could do SCL lower than 4 and tRCDRD @15 ......
 

·
Edgy & on the edge
Joined
·
1,529 Posts
Your AIDA screenshot shows the frequency of your CPU when you ran the benchmark.

The 64ns is because your CPU is only running at 4250mhz.

If you ran an all core overclock of 4.3Ghz and run the benchmark you score would be less than 63ns and if you ran a core core overclock of 4.4Ghz it would be less than 62ns ...
It's not that easy to get under 62. :)
 

·
Registered
Joined
·
882 Posts
It's not that easy to get under 62. :)
Yeah maybe over stated as his tCL/tRCDRD is @15

But its also not that hard

:p

** EDIT **
With the right hardware and knowledge

:)
 

·
Registered
Joined
·
211 Posts
How important is TRDWR for performance, is the difference between 9, and 10 negligible?

I managed to do 20 rounds TM5 v3 preset with 32gb with TRDWR at both 10, and 9 but the next reboot TRDWR at 9 failed after 10 runs, should I just keep it at 10, and 3800CL15-15-15-15-30-46 1T 276TRFC?
 

·
Registered
Joined
·
25 Posts
@Veii This Mobo won't let me set 7 as CDWR timing :mad:
8 is the lowest manual value -auto doesn't post.
Maybe sometime in the future I'll swap the Asus board in here and see what it can do with these sticks.

I've just gotten the R3 back together, and soon I'm gonna start working the Rev. E properly.

I've seen systems here with the same CPU and the same timings where there are less than 62ns.
I do not understand.
This is what it's taken for me to get sub 62 at 4.32 all core.
I'm sure this is a redundant post as these timings are already posted by me not long ago, but this is to hopefully highlight the not-insignificant difference in timings.

You definitely need more core clock though.
Also, optimize your os as best you can for running the bench
 

Attachments

  • Rep+
Reactions: jcpq

·
....
Joined
·
225 Posts
Here is an update for people having Zen+ CPUs, 3600 CL16 with safe settings was fine.
I managed to pass a couple of hours of TM5 (anta777) with no errors, I will try to test more as soon as I have more time.

2458874
 

·
Registered
Joined
·
80 Posts
hi guys.
i am mostly at stock now (PBO auto, XMP loaded, and some settings for 1usumus power plan), just doing some testing, before trying some dram calc.
i tried TM5 with [email protected], after 10mins temps were 48/47 on ram (B-die G.Skill F4-3600C18-8GTZRX) are those safe temps for ram on stock settings 1.350V??
Also my VSOC temps seem a bit higher than what jcpq posted above. is that ok?
thanks for help.

2458878
 
4741 - 4760 of 4797 Posts
Top