What's new
Carbonite

South Africa's Top Online Tech Classifieds!
Register a free account today to become a member! (No Under 18's)
Home of C.U.D.

Time Spy Results Thread

Well, I did a quick tinkering with help from @VPII and here is the result on Time Spy for now. Will play around a bit again next weekend, to see what can be done haha.

 
Completely stock run. The card boost 2500Mhz out of the box and getting this performance for at less than 300W

 
Completely stock run. The card boost 2500Mhz out of the box and getting this performance for at less than 300W

Wow!

I think you have definitely won the silicon lottery!!!

I have had 2 RX6900 XT MERC 319's and I struggled to get a Time Spy Graphics Score of over 20 300 on either of them by overclocking with 115% power & stock air cooler or water block.
 
Last edited:
Wow!

I think you have definitely won the silicon lottery!!!

I have had 2 RX6900 XT MERC 319's and I struggled to get a Time Spy Graphics Score of over 20 300 on either of them by overclocking with 115% power & stock air cooler or water block.
Interesting is that I tried overclocking the card now, but at 2.6Ghz or 2.7Ghz I stock lower (19000 flat) vs the stock. I am not sure if I am running out of juice. Will try to lower the voltage and see if it make a difference. I also noticed that if I touch the memory it crashes.
 
Dropped the voltage to 1100mV and ran again with 2.6Ghz. Some improvement, but really not worth it. Just going to run it stock

 
Dropped the voltage to 1100mV and ran again with 2.6Ghz. Some improvement, but really not worth it. Just going to run it stock

Wow, that is a great score!

I also found the pushing frequency at 1175mV would generally run but drop the benchmark score on both of my cards.

On the card I kept, as far as I can see the sweet spot for the highest score was at
Freqency Min 2550MHz
Frequency Max 2650MHz
Voltage: 1080mV
Power: +15% (115%)
Memory Timing: "Normal" ("Fast" timings was less stable would crash much earlier & more often
Memory: 2100MHz (2150MHz would run but give lower score)

Since my card stays very cool with the water block, I plan on using More Power Tool (MPT) to push the max power limit to 375W - 400W to see what scores I can reach.

Enjoy your card!
 
Wow, that is a great score!

I also found the pushing frequency at 1175mV would generally run but drop the benchmark score on both of my cards.

On the card I kept, as far as I can see the sweet spot for the highest score was at
Freqency Min 2550MHz
Frequency Max 2650MHz
Voltage: 1080mV
Power: +15% (115%)
Memory Timing: "Normal" ("Fast" timings was less stable would crash much earlier & more often
Memory: 2100MHz (2150MHz would run but give lower score)

Since my card stays very cool with the water block, I plan on using More Power Tool (MPT) to push the max power limit to 375W - 400W to see what scores I can reach.

Enjoy your card!
Thanks Manticor3d.

I did what you say. For now just running 2550mhz @ 1100mV and the results seem much better. I am going to use this for gaming and see how it works.

 
@baasgene slitgh update here, I actually forgot about this run and right now with temps I can forget about trying to replicate. Again, with the Galax 1Kw bios left at 100% card draws around 600watt power. Oh and I was happy about the 22K graphics score. CPU score still a little iffy, difficult to get constant cpu score results with AMD Ryzen cpu it seems.

 
@baasgene slitgh update here, I actually forgot about this run and right now with temps I can forget about trying to replicate. Again, with the Galax 1Kw bios left at 100% card draws around 600watt power. Oh and I was happy about the 22K graphics score. CPU score still a little iffy, difficult to get constant cpu score results with AMD Ryzen cpu it seems.


This is the limit of my card with the original bios, but overall it only pushes 450w at most. It would probably fall in that same margin as yours in the end if I use the 1000w bios judging by the max clock.


Overall I'm extremely happy with this card, it was a absolutely worthy purchase over my Tuf 3080 :)
 
Last edited:


Pretty meh Time Spy results, but how much can one expect from a 5900X and a non-XTXH 6900 XT. Did these while my town was freezing around 3AM-ish. These are pretty much topped out short of hard modding the card (I don't have enough money to risk doing that).

CPU/RAM supporting the GPU:
5900X 4.85GHz/4.75GHZ (CCD0/CCD1) @ 1.425V MSI LLC Mode 4
DDR4 2X8 (SR) 5000 2T-17-25-25-23-50 (tRC-78, tRFC-780, tWR-12, tWTRS-4. tWTRL-8, tRRDS-4, tRRDS-4, tFAW-16, tRTP-10, tCWL-16, tCKE-1) @ 1.65V VDIMM (0.825V Termination), 1.2V SoC, 1.1V CLDO VDDP, 1.0V VDDG CCD, 1.15V VDDG IOD
 

this was with my old mobo and my ram timings much tighter than they are now. Looks completely mid tier against you guys xD
 


For what it's worth, here is a score I got in June when I had 2x 6900XT's (One OC'd and the other one stock).

This was after a few hours of AMD's mGPU mode not activating, Timespy failing to complete and (my favourite) 3D Mark error: "Benchmark Tesselation Load Modified". Now I definitely understand why SLI/Crossfire isn't popular any more.
 


Pretty meh Time Spy results, but how much can one expect from a 5900X and a non-XTXH 6900 XT. Did these while my town was freezing around 3AM-ish. These are pretty much topped out short of hard modding the card (I don't have enough money to risk doing that).

CPU/RAM supporting the GPU:
5900X 4.85GHz/4.75GHZ (CCD0/CCD1) @ 1.425V MSI LLC Mode 4
DDR4 2X8 (SR) 5000 2T-17-25-25-23-50 (tRC-78, tRFC-780, tWR-12, tWTRS-4. tWTRL-8, tRRDS-4, tRRDS-4, tFAW-16, tRTP-10, tCWL-16, tCKE-1) @ 1.65V VDIMM (0.825V Termination), 1.2V SoC, 1.1V CLDO VDDP, 1.0V VDDG CCD, 1.15V VDDG IOD
I find these results really subjective. I take your GPU score in Time Spy and Time Spy Extreme and usually the GPU score is the main factor in the result. Your cpu score with you 5900X is actually pretty good from what I have seen. Now in Time Spy you kill my score but in Time Spy extreme my score is higher. Looking at the gpu score, ity is strange to see that when your result handedly beat mine in Time Spy, it is lacking in Time Spy Extreme.

Look the CPU score counts for something I will not deny that. What I will suggest, not in TS Extreme but in normal Time Spy is to disable SMT or Simultaneous multithreading as the benchmark is not reaaly coping or using the additinal threads properly. When running with SMT my cpu score would be around 16800 maybe 1700 but without SMT it would be 18000.

My scores are as follow:


 
I find these results really subjective. I take your GPU score in Time Spy and Time Spy Extreme and usually the GPU score is the main factor in the result. Your cpu score with you 5900X is actually pretty good from what I have seen. Now in Time Spy you kill my score but in Time Spy extreme my score is higher. Looking at the gpu score, ity is strange to see that when your result handedly beat mine in Time Spy, it is lacking in Time Spy Extreme.

Look the CPU score counts for something I will not deny that. What I will suggest, not in TS Extreme but in normal Time Spy is to disable SMT or Simultaneous multithreading as the benchmark is not reaaly coping or using the additinal threads properly. When running with SMT my cpu score would be around 16800 maybe 1700 but without SMT it would be 18000.

My scores are as follow:


I was lucky enough to secure some interesting kit and a better bin 69XT, I think. I need my Comet Lake combo to sell first, but I’m going to be re-doing the entire suite with a 12900K. A bit worried my 750W PSU won’t be sufficient but we’ll have to see.
 
I was lucky enough to secure some interesting kit and a better bin 69XT, I think. I need my Comet Lake combo to sell first, but I’m going to be re-doing the entire suite with a 12900K. A bit worried my 750W PSU won’t be sufficient but we’ll have to see.
I do not think it will be in my honest suggestion. Just saying.
 
I do not think it will be in my honest suggestion. Just saying.
The 6900 XT with a 450A current limit might hit 350W. The i9 during the graphics portion might sit at 150. Combined might be a bit of a challenge, but as long as the switching frequency of my GPU is sufficient, it should cut the load quickly enough before physics loads although physics doesn’t draw that much current anyway. I think I might just be fine 🤣.

Don’t forget about how power efficient DDR5 is 🤣🤷‍♂️.
 
The original timespy should be disregarded imo, it only really scales well up to 10 cores. The extreme test scales properly. And the reason why the original TS scores better on the amd 6000 series gpu is because of core clock speed, it helps for higher fps which is why amd excels at lower resolution.
 
Last edited:
I got this score a while ago when I was seeing what my new ram kit could do
 

Users who are viewing this thread

Back
Top Bottom