Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Nov 18, 2020.
@kapu You win a silicon lottery probably
Even my sample is better ,it seems too big difference on average it is 150-200Mhz any given scenario.
Would think a early 20.11.2 review driver was used for the initial GPU's then either 20.11.2 public or 20.11.3
20.12.1 has the above fix and 20.12.2 is mainly the AMD RGB Coolermaster support thing but also contains a newer Radeon Settings Software application though quirkiness with Wattman and OC results shouldn't be affected by that.
Push the power limiter up, test the clock scaling, dial in a nice min/max 3D clock speed setting or just push them up and then see if a tad lower voltage can be stable while also sustaining the boost speeds.
Complain about the locked max speeds for GPU core and memory though I suppose for regular usage none of the cards can hit the max speed while being stable anyway though they can get close.
Touch more voltage perhaps but that too is a bit locked unless you go via More Power Tool, power limit and such might have some restrictions as well though changing the GPU's allowed power total can work around that leaving the 15% max but now for a higher set base value.
(2400 - 2500 for the 6800, 2700 for the 6800XT and 6900XT -50Mhz or so from the Wattman specified values due to some thing keeping it like that.)
EDIT: Hmm actually if all the testing and benchmarks consistently show worse figures I wonder what AMD sent out, maybe there's just a higher variance for the GPU's although many reviews do show overclock results fairly consistent closing in on the max allowed before stability issues become a concern and either getting close to or completely maxing out the VRAM clock speeds before error correction starts reducing performance instead of improving it.
Maybe it's not a given anymore that hardware is tested and the best available seeded out for review purposes, I like that though in turn the sites getting the less performant hardware will be criticized for doing it wrong or hindering things for favor of the competing businesses here (All like two of them in total since it's AMD/Intel and AMD/NVIDIA pretty much.) and you get that little mess instead, no way around it I suppose, sigh.
When i compared benchmarks it seems my card is around 5-9% faster at stock VS Hilbert's card. Also overclocking gain is another 5-9% depending scenario . It's quite alot.
Typical STOCK scenario *( here cyberpunk).
Max OC :
Is there anyway to push OC beyond radeon software ? seems that there is more potential in the card ....
Seems like your card is using limits of XT version. Weird is temperature. Do you have model with better than reference cooling? Because reference non-XT cards use weaker cooling than reference XT cards.
I have reference Sapphire card. Max temp i seen is 72C ( long gameplay) OC adds around 1-2C. Card is very quiet . My case is CLOSED Fractal R4. Not really good ventilated ( 3 silent 600 rmn fans)
Unigine Heaven Extreme benchmark OC : 232 FPS STOCK : 216 FPS.
It is solid gain tbh, for free.
Anyone has score of 6800 in Horizon Dawn 1080p/ Ultimate Quality preset ? I failed to find one to compare...