• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Poll: Ryzen 7950X3D, 7900X3D, 7800X3D

Will you be purchasing the 7800X3D on the 6th?


  • Total voters
    191
  • Poll closed .
why with GDM disabled i get worse copy and write aida score by 0.5%
Is it stable? Often if you push things too far performance decreases, ECC is a thing even if set to disabled. Also, having the FCLK too high can nerf performance, example below and verified with Karhu

2167Mhz FCLK
xeInpdV.jpg

If I run 2200 FCLK, those values drop to 100MB.

I get best overall results, including lowest latency, with GDM off, 6400MhzC28, 1.51v. However being able to run GDM off at higher memory frequencies on AM5 usually depends on having the right RTT/impedance settings in the BIOS. You can try copying mine to see if this helps.
2UcQyRp.jpg

I found that X3D adds about 4ns+ of latency in Aida64 vs non X3D AM5 parts.

Memory timings, FCLK and curve optimiser values tested and verified stable overnight Karhu, TM5 and Y Cruncher runs.

YmYBuxs.png

Tz6KqWa.jpg
TkPF0r5.jpg
 
Last edited:
Is it stable? Often if you push things too far performance decreases, ECC is a thing even if set to disabled. Also, having the FCLK too high can nerf performance, example below and verified with Karhu

2167Mhz FCLK
xeInpdV.jpg

If I run 2200 FCLK, those values drop to 100MB.

I get best overall results, including lowest latency, with GDM off, 6400MhzC28, 1.51v. However being able to run GDM off at higher memory frequencies on AM5 usually depends on having the right RTT/impedance settings in the BIOS. You can try copying mine to see if this helps.
2UcQyRp.jpg

I found that X3D adds about 4ns+ of latency in Aida64 vs non X3D AM5 parts.

Memory timings, FCLK and curve optimiser values tested and verified stable overnight Karhu, TM5 and Y Cruncher runs.

YmYBuxs.png

Tz6KqWa.jpg
TkPF0r5.jpg
Thanks for the tips Matt.
am not sure if its 100% stable yet but I played some warzone 2 games and all was fine but i would need to run many hours of tests like TM5 OCCT Y-CRUNCHER to call it stable.

Watchdog legion benchmark was very similar but the low 1% and 0.1 seemed a tad better.
Just weird that aida64 showing worse copy and write
 
Last edited:
Thanks for the tips Matt.
am not sure if its 100% stable yet but I played some warzone 2 games and all was fine but i would need to run many hours of tests like TM5 OCCT Y-CRUNCHER to call it stable.

Watchdog legion benchmark was very similar but the low 1% and 0.1 seemed a tad better.
Just weird that aida64 showing worse copy and write
Aida is not too important in the grand scheme of things tbf other than it's nice to see as low a latency as possible and consistent results each time you run it.
 
What ram are you using there matt, is it A or M die?

Also is 6400 faster then, I thought the sweet spot for AMD was 6000 and anything higher was worse overall?
 
Last edited:
Just got an email saying my 7950x3d wont be here till end of april now... ordered on 5th mar.... are AMD shipping a single item per day....? :mad:
Yikes, what was the original estimation?

News gets worse by the day, lol. It'll be interesting to see the 7800x3d launch.
 
What ram are you using there matt, is it A or M die?

Also is 6400 faster then, I thought the sweet spot for AMD was 6000 and anything higher was worse overall?
The actual kit is shown in the zen timings screenshot. Enter that kit number into Google and you'll find places to buy it.

As fast as you can run is always best, as long as its stable. Generally 6600Mhz is the fastest you can run on AM5, but most people settle on 6200/6400Mhz as the max attainable speeds.

6000Mhz offers easy plug and play with good performance with very little effort required from the user. With higher speeds, particularly 6400Mhz+ that's not the case.
 
Last edited:
when I first ordered it was saying end of march, I wasn't expecting one to turn up on my door step the next day, but to be looking at a 6-8 week wait.... absolute **** take
Yeah that's kind of ridiculous. I expected these CPU's to go out of stock for a while, but didn't really expect supply to be this poor. I don't usually agree with the term paper launch, but man, it almost feels like it here.
 
LTmatt I looked up the number and I notice you have Gskill there, can you do me a favour, does the RGB turn off in hibernate? AFAIK it wont turn off in sleep, but Gskill said in an email that it wont turn off in hibernate either, is this true? Only way to turn off RGB is to completely switch off PC with the Gskill?
 
Last edited:
LTmatt I looked up the number and I notice you have Gskill there, can you do me a favour, does the RGB turn off in hibernate? AFAIK it wont turn off in sleep, but Gskill said in an email that it wont turn off in hibernate either, is this true? Only way to turn off RGB is to completely switch off PC with the Gskill?
GSkill are confused. If you power down the system or use hibernate, the DRAM RGB turns off. If you put the system to sleep, it stays on.
 
The actual kit is shown in the zen timings screenshot. Enter that kit number into Google and you'll find places to buy it.

As fast as you can run is always best, as long as its stable. Generally 6600Mhz is the fastest you can run on AM5, but most people settle on 6200/6400Mhz as the max attainable speeds.

6000Mhz offers easy plug and play with good performance with very little effort required from the user. With higher speeds, particularly 6400Mhz+ that's not the case.
I believe the 7000 series runs better with tighter timings rather than higher Mhz, at least from HUB

 
I believe the 7000 series runs better with tighter timings rather than higher Mhz, at least from HUB

Since HUB use the Watchdogs legion benchmark sequence and I have a 4090, I can probably copy that to compare to HUBs numbers.

I'm expecting 6400Mhz C28 to be a bit faster than 6000Mhz C30.

EDIT, they also test Horizon Zero Dawn so could test that too.
 
Last edited:
Since HUB use the Watchdogs legion benchmark sequence and I have a 4090, I can probably copy that to compare to HUBs numbers.

I'm expecting 6400Mhz C28 to be a bit faster than 6000Mhz C30.

EDIT, they also test Horizon Zero Dawn so could test that too.
That would be great to see some comparison numbers
 
Back
Top Bottom