Overclock.net banner
2401 - 2420 of 3625 Posts

·
Registered
Joined
·
693 Posts
New MPT Beta: "RDNA2: The setting options for the frequencies have been expanded considerably and pretty much everything that could be disclosed has been disclosed."

Does this mean non-XTH GPU's can increase their memory past 2150MHz ??? It still didn't work for me
 

·
Registered
Joined
·
731 Posts
Is there a bug currently?

How is this possible...


23.5K Graphic Score at 2500 Mhz average???
That seems very strange. To break 23K i need 2800MHz set clk.
I suspect it's just a reporting bug.

I've found that the clock speeds and averages are not always 100% accurate on some runs.

I was reading on hwbot that raising the min SOC voltage to 950mv can help squeeze a bit more out of mem oc.
Good find Jon. I will have to try bumping the min SOC voltage to 0.950mv next time I'm doing some benching to see if more can be gained.
 

·
Registered
Joined
·
458 Posts
New MPT Beta: "RDNA2: The setting options for the frequencies have been expanded considerably and pretty much everything that could be disclosed has been disclosed."

Does this mean non-XTH GPU's can increase their memory past 2150MHz ??? It still didn't work for me
No, there's nothing new on MPT that makes any difference unfortunately.
I've tested it all.

Maybe for gaming that support REBAR since you can now mess with infinity fabric clockspeed but from my testing it does absolutely nothing.
 

·
Premium Member
Joined
·
241 Posts
Anyone with a 6900XT check on HWINFO what Voltage is your VRAM at.

View attachment 2517907

Pretty sure it's supposed to say 1.356V or something
According to my HWINFO the memory (VDDIO) is steady at 1.356V, no load or loaded.
 

·
Registered
Joined
·
731 Posts

·
Invalid Media
Joined
·
4,786 Posts
It made no difference for me. After extensive testing in Timespy 2150 MHz gives the best results.
2150 MHz VRAM w/ fast timings seem to work w/o issue (ie loss of efficiency) for me as well, in fact I wish I could adjust the VRAM to a bit higher speed...

...or better yet, AMD adding GDDR6X for the 6900 XT(/H) 😎
 

·
Professional Lurker
Joined
·
259 Posts
Unigine Superposition "4k Optimized" is MUCH more sensetive to memory speed changes than anything I've found in the 3dMark suite.

I do 3 runs at each memory speed to rule out background tasks, it's worth a shot since yall are getting nowhere with Time Spy.
 
  • Rep+
Reactions: J7SC

·
Registered
Joined
·
458 Posts
Messing with any of the settings on MPT does nothing on my card, except of course, increasing the Power Limit and Current limits.

These new drivers however, despite being more efficient, made my previously super stable and performance increase 2150 Mhz lose performance and I have to run around 2050-2070 Mhz to get the best performance.

This is specially noticeable on Superposition as someone already mentioned. Time Spy is too inconsistent to give proper results in terms of scaling.

Unless there are combination of settings or hidden stuff we don't know about, I call placebo.

Another thing I would like to know is how the hell the top 2 Graphic Score cards on TS leaderboard are reference AMD cards, one of them at 2800 and the other at 2660 Mhz. One of them even has the core at 61C which would suggest at best, water cooling so nothing exotic.

Are people using EVC to post 24K+ Graphic Score on non XTXH cards?
 

·
PC Evangelist
Joined
·
47,675 Posts
Messing with any of the settings on MPT does nothing on my card, except of course, increasing the Power Limit and Current limits.

These new drivers however, despite being more efficient, made my previously super stable and performance increase 2150 Mhz lose performance and I have to run around 2050-2070 Mhz to get the best performance.

This is specially noticeable on Superposition as someone already mentioned. Time Spy is too inconsistent to give proper results in terms of scaling.

Unless there are combination of settings or hidden stuff we don't know about, I call placebo.

Another thing I would like to know is how the hell the top 2 Graphic Score cards on TS leaderboard are reference AMD cards, one of them at 2800 and the other at 2660 Mhz. One of them even has the core at 61C which would suggest at best, water cooling so nothing exotic.

Are people using EVC to post 24K+ Graphic Score on non XTXH cards?
Maybe the early XTX could have been XTXH.
 

·
Registered
Joined
·
190 Posts
It made no difference for me. After extensive testing in Timespy 2150 MHz gives the best results.
I think it will only help if you're just almost stable at 2150Mhz and need something little to get you there. Me and the guy that originally posted it were both only stable at 2130MHz before the tweak.

Also, enable SAM. It really helps boost Timespy scores.

Nah, not that.

Look at this score...


How is this even possible?
Nothing is strange about that score, unless that's a reference card on air lol. My best GPU score was higher than that, but I have XTXH under water. I had to run 2770/2150 with SAM and +35% PL/TDC/SOC to get that high of a score.
 

·
Registered
Joined
·
731 Posts
I think it will only help if you're just almost stable at 2150Mhz and need something little to get you there. Me and the guy that originally posted it were both only stable at 2130MHz before the tweak.

Also, enable SAM. It really helps boost Timespy scores.



Nothing is strange about that score, unless that's a reference card on air lol. My best GPU score was higher than that, but I have XTXH under water. I had to run 2770/2150 with SAM and +35% PL/TDC/SOC to get that high of a score.
What's your best Timespy score Jon on 21.7.1?
 

·
Premium Member
Joined
·
241 Posts
Is your memory frequency running at maximum when idle?
My memory frequency, according to HWINFO, is running at the set frequency when at idle.
 
2401 - 2420 of 3625 Posts
Top