Current Generation Games Analysis Technical Discussion [2020-2021] [XBSX|S, PS5, PC]
- Thread starter Shortbread
- Start date
- Status
Silent_Buddha
Legend
D
Deleted member 2197
Guest
Oh I saw that - I mean the part where it is shown visually that they are indeed the same. Instead of just saying it and assuming as much. Basically - actually demonstrating what the settings are by going through them.
snc
Veteran
you mean around 5:02
you mean around 5:02
Which shows that settings aren't equivalent. Aside from the lack of moving shadows, you can also see that the LOD is reduced on the consoles in that centre tree which looks noticeably more dense on the PC version.
Note the scenes where the drops in PC performance were shown were foliage heavy.
Also as a bit of a nitpick, his overclocked 2070 isn't really a 2070 Super level GPU. He's running it at around 1700Mhz core while the Super has a 1770Mhz boost which it should be able to comfortably exceed. He's also reporting memory speeds 1700Mhz whereas the 2070s runs at 1750Mhz.
snc
Veteran
Sure its hard to match exact same settings in most games, differences in clocks is 4% on core and 2.9% on memory :d
Clock speed isn't the only difference between the 2070 and 2070 Super. 2070 has 2304 CUDA cores and 144 TMUs while the Super has 2560 CUDA and 184 TMUs. 2070 Super uses a cut down version of the 2080 die (TU104?) and I think it retains the extra cache it had over the stock 2070 also.
snc
Veteran
correct tough 160 tmu on super, still according to techpowrup 2070 super in 1440p is 13.6% faster than 2070 so oc 2070 will be very close
It's certainly hard to match settings exactly. But it's quite easy to not claim the settings are identical when they're not, as a basis for a performance comparison. He has a whole section on the lack of moving foliage and foliage shadows on the next gen console versions (which is a big deal!) then chalks it up to a bug and proceeds to ignore it in the performance comparison and ultimate ranking of the versions.
Also it's more like 8% on the core (see the link I posted) as the 2070S will comfortably exceed it's boost clock during heavy gaming loads. I'm taking the 1840Mhz figure as a point of comparison there but it clearly can sustain even higher. Then there are the core differences as see colon mentioned which make the difference even greater.
I have one more minor nitpick about the conclusion regarding the Steam hardware numbers as well. He says "unless you're in the 4% of PC gamers that are running better hardware than me" however it's more like 6% at the 2080 or above level or 8.3% at the real 2070S and above level. In real terms that's something like 7-10m systems (based on the count of 120m active Steam users in the last month) or quite comparable in overall numbers to next gen console sales. Stating it as 4% makes it seem like a insignificant number barely worthy of mention, which I believe is somewhat misrepresenting the situation.
Not like you'd want to run those benches again, but it would have been interesting to see the average minimum framerate as well as the CPU/GPU split for those cases. I'd imagine it would be pretty similar, but there might be some interesting things there potentially. The savings in rendering time from VRS could be getting used in other parts of the rendering pipeline. Unfortunately, while the Gears 5 benchmark gives us significantly more information that other build in benchmarks in games, the information isn't fine grained enough to determine that.
As well, it should be noted that some amount of rendering time will be used for scaling the out back down to 1080p from 4k. Overall, however, it looks to be about what would be expected.
Thanks for going through all that effort.
Regards,
SB
Side note - I said everything was at max, but I noticed after the SSGI Ray count is at 8. That goes up much higher. 32 I think is the max. But I left it the same so my results were comparable. Also, I skipped testing balance since it's performance is almost the same as quality. And I only tested at 1440p.
My 2070 Super spikes up to 1950mhz. Not sure if that's the fastest it's ever been but it's the fastest I've noticed.
Oh yeah, I forgot nVidia early review docs had the wrong value for TUMs listed and many reviews never corrected it. 160 TMUs is right.
It's not that it won't be close. Even at stock speeds I wouldn't put the 2070 in a different performance bracket than Super. It's just that the Super is faster clock for clock, and he isn't matching the clocks either.
I don't mean to pile on NXG. I enjoy his videos. And, he has a really enjoyable speaking voice.
Yeah - the consoles are definitely not across the board Ultra settings. Next Gen consoles even have lower grass draw distance and LOD than Xbox One X, for example.
I was mistaken on the core clocks earlier. He seems to be running at around 1920Mhz in the scene that was demonstrating drops. According to TPU the average gaming clock for the standard 2070 is around 1800Mhz so we're looking at around a 6.8% boost on the core. But a reduction in the memory clock for some reason. I'd wager that GPU is closer in overall performance to a stock 2070 than a stock 2070S. The thing with these heavily O/C's GPU's is that they're still constrained by power and thermal limits and the stock GPU's still try to O/C within those same or similar limits anyway, hence the end game clock isn't that different.
snc
Veteran
yeah so possbile 6% faster than 2070 and 7% slower than super, so closer to 2070, huge difference ;d
- Status
Similar threads
- Locked
