Overclock.net banner
Cancel
Create thread New Forums

AMD DDR5 OC And 24/7 Daily Memory Stability Thread

93 reading
Jump to Latest
8,241 - 8,260 of 31,499 Posts
#8,241 ·
discomute said:
I'm the person Dasa is helping. VDDP at 1.10 and 5400 means it won't post and I need to pop the battery.

Interesting comment re: CPU. I'm new at this and had thought it was probably my motherboard (seems to me the AX version gets all of gigabyte's love) as it has a few quirks like crashing if a specific usb port is used (as did my previous mb even before the LTE firmware upgrade/replacement). Didn't realize it could be the CPU. 5200 isn't the worst thing in the world, maybe in a year or two bios updates will sort it?
Click to expand...
maybe is the ram kit, is not on GB QVL, so just enabling EXPO may not just work.
Try to set both VDD and VDDQ to 1.35v which is rated for this kit.
If still no boot, set CPU_VDDIO_MEM to 1.35v too (probably not needed this much for 6000 loose timings)
 
#8,242 ·
Dasa said:
I am trying to help somebody stabilize there RAM which wont run EXPO 6000, hell it wouldn't even post at 5200 without setting these V (CPU_VDDIO_MEM 1.25v, DDR_VDD 1.375v, DDR_VDDQ 1.25v, VSOC 1.2v )
5400 results in needing to remove the battery to reset BIOS and make it post again.

Unfortunately I am more familiar with DDR4 than DDR5 so perhaps you can spot something that looks wrong here.

Image


MB has been replaced under warranty after a bad firmware update but no change and he has tried each stick individually at EXPO but it wouldn't post.

Thanks.
Click to expand...
Same board that I got but my XMP profile ram boots just fine. Think you got Hynix M die there, so there could be something going on with that, my sticks have Samsung B die. Tried raising the VSOC a little? but it really look like manually setting the other voltages as well, like VDDIO_MEM, etc..
 
#8,243 ·
dansi said:
Try to set both VDD and VDDQ to 1.35v which is rated for this kit.
Click to expand...
VDD was already at 1.375, okay to leave as is?

VDDQ being put to 1.35 and then run at 5400... Booted after a very long time. Seemed to run okay.

Upon a reset it kicked me to bios and said it has boot failure.

dansi said:
If still no boot, set CPU_VDDIO_MEM to 1.35v too
Click to expand...
Upon doing this
It wouldn't post / pop battery
 
#8,245 ·
discomute said:
VDD was already at 1.375, okay to leave as is?

VDDQ being put to 1.35 and then run at 5400... Booted after a very long time. Seemed to run okay.

Upon a reset it kicked me to bios and said it has boot failure.



Upon doing this
It wouldn't post / pop battery
Click to expand...
try VDD at 1.35 too. sometimes too much is bad.

CPU_VDDIO_MEM set to auto or normal.

find memory context restore options in GB bios and disable it. disable fast boot option too.
 
#8,246 ·
dansi said:
try VDD at 1.35 too. sometimes too much is bad.

CPU_VDDIO_MEM set to auto or normal.

find memory context restore options in GB bios and disable it. disable fast boot option too.
Click to expand...
At 5200 it booted with VDD 1.35

Following everything you say here with VDDQ at 1.35 (fast boot is always diabaled) and setting it to 5400 it gave me a boot failure and kicked me to bios
 
#8,247 ·
discomute said:
At 5200 it booted with VDD 1.35

Following everything you say here with VDDQ at 1.35 (fast boot is always diabaled) and setting it to 5400 it gave me a boot failure and kicked me to bios
Click to expand...
Did you try each ram stick separately ?
 
Save
Reply Quote Rep+ Rep+
#8,249 ·
Save
Reply Quote Rep+ Rep+
#8,251 ·
Yuke said:
Increasing VDDIO helped me getting to tRCD 43 it seems...

View attachment 2627139

stopped at 2 hours for now because im dying with 30°C in my room already....

next target is tCL 34....but that might be too much to ask.
Click to expand...
Well, looking good I guess

Rectangle Font Screenshot Parallel Number


1.49 VDDIO
 
  • Rep+
Reactions: DedEmbryonicCe1 and Kaliz
Save
Reply Quote Rep+ Rep+
#8,252 · (Edited)
Dasa said:
I am trying to help somebody stabilize there RAM which wont run EXPO 6000, hell it wouldn't even post at 5200 without setting these V (CPU_VDDIO_MEM 1.25v, DDR_VDD 1.375v, DDR_VDDQ 1.25v, VSOC 1.2v )
5400 results in needing to remove the battery to reset BIOS and make it post again.

Unfortunately I am more familiar with DDR4 than DDR5 so perhaps you can spot something that looks wrong here.

Image


MB has been replaced under warranty after a bad firmware update but no change and he has tried each stick individually at EXPO but it wouldn't post.

Thanks.
Click to expand...
Hello,

I have the same memory, I'm not sure that you can have better performances in 6000

Font Electronic device Multimedia Screenshot Software


Look at this
Intel Alder Lake DDR5 Memory Scaling Analysis With G.Skill Trident Z5
 
#8,253 · (Edited)
tCL 34 looking good so far

Product Azure Rectangle Font Line


but feeling uncomfortable with VDDG 1.05V ... 950mV i had before was crashing out fast so not really confident in reducing it by much

:edit: NOPPEEEEE, not a chance. ****ing hate this Kit/Board
 
Save
Reply Quote Rep+ Rep+
#8,254 ·
Curious behaviour w/ a 7800x3d and 7900 xtx.
3dMark benches score lower in all tests when comparing 6400 CL28 vs EXPO 6000 CL30. Same goes by 6000 CL26 vs EXPO 6000 CL30. It is stable, 40K in karhu, ram temp below 40C always.
However, games that have built-in benchmark score higher, but not that much.
CPU Profile have a difference of max 2% in between same speed combos above.
I'm starting to believe is not worth it to tight timings with the 7800x3d w/ 6000Mhz ram, maybe running 7800mhz+ would be a different story.
 
#8,255 · (Edited)
thomasck said:
Curious behaviour w/ a 7800x3d and 7900 xtx.
3dMark benches score lower in all tests when comparing 6400 CL28 vs EXPO 6000 CL30. Same goes by 6000 CL26 vs EXPO 6000 CL30. It is stable, 40K in karhu, ram temp below 40C always.
However, games that have built-in benchmark score higher, but not that much.
CPU Profile have a difference of max 2% in between same speed combos above.
I'm starting to believe is not worth it to tight timings with the 7800x3d w/ 6000Mhz ram, maybe running 7800mhz+ would be a different story.
Click to expand...
What FCLK? Am hoping you're comparing 6000MT/s with a 2000MHz FCLK vs 6400MT/s with a 2133MHz FCLK for apples to apples. Also the IF starts error correcting if it can't go high enough/lacks voltage. Some people get best results at 6200MT/s with 2066MHz, some with 6600MT/s 2200MHz, then some even higher (but moving into 1:2 territory).
 
#8,257 ·
metalshark said:
What FCLK? Am hoping you're comparing 6000MT/s with a 2000MHz FCLK vs 6400MT/s with a 2133MHz FCLK for apples to apples. Also the IF starts error correcting if it can't go high enough/lacks voltage. Some people get best results at 6200MT/s with 2066MHz, some with 6600MT/s 2200MHz, then some even higher (but moving into 1:2 territory).
Click to expand...
6000mhz fclk 2000 and 6000mhz fclk 2133 (less aida latency) and 6400mhz fclk 2133 (even lower aida latency). I'm gonna try 6400mhz with fclk 2000. I should not change anything but lemme check. I don't believe fclk 2133 is unstable, 2166 clearly is, but 2133 is solid in any test do I don't think there's that regression of performance. I believe it's just the L3 cache invalidating (being much faster) the timings tightening.. when with 5800x3d + 3800 cl14 flat I could see much more score improvement compared with 3600 cl16, there was no performance regression.
 
#8,258 ·
thomasck said:
6000mhz fclk 2000 and 6000mhz fclk 2133 (less aida latency) and 6400mhz fclk 2133 (even lower aida latency). I'm gonna try 6400mhz with fclk 2000. I should not change anything but lemme check. I don't believe fclk 2133 is unstable, 2166 clearly is, but 2133 is solid in any test do I don't think there's that regression of performance. I believe it's just the L3 cache invalidating (being much faster) the timings tightening.. when with 5800x3d + 3800 cl14 flat I could see much more score improvement compared with 3600 cl16, there was no performance regression.
Click to expand...
About the only way I've found to spot FCLK instability at the edge is set your sound card sample rate as high as possible with the buffer size as small as possible, belt a tune out with your mouse on the highest polling rate doing circles and kick off some y-Cruncher FFT. If you notice any mouse jerkiness or stutters in the audio the FCLK needs to come down or the voltages need to go up.
 
#8,259 ·
metalshark said:
About the only way I've found to spot FCLK instability at the edge is set your sound card sample rate as high as possible with the buffer size as small as possible, belt a tune out with your mouse on the highest polling rate doing circles and kick off some y-Cruncher FFT. If you notice any mouse jerkiness or stutters in the audio the FCLK needs to come down or the voltages need to go up.
Click to expand...
BuildZoid was also getting some feedback on FCLK instabilities by using Linpack Xtreme here
 
8,241 - 8,260 of 31,499 Posts
About this Discussion
31K Replies
1.1K Participants
Last post:
gupsterg
Overclock.net
posts
28M
members
562K
Since
2004
The premier forum for overclocking experts and enthusiasts. Discuss hardware optimization, custom builds, benchmarking, cooling solutions, and pushing the boundaries of computing performance. From beginner guides to extreme overclocking, join our technical community to master system tuning.
Full Forum Listing
Explore Our Forums
Intel CPUs Hardware News NVIDIA PC Gaming Water Cooling