Sharper shadows are normally less compute intensive and less realistic than soft shadows.
But that should really be a very tiny difference
Last edited:
Digital Foundry Article Technical Discussion [2023]
- Thread starter BRiT
- Start date
- Status
But that should really be a very tiny difference
snc
Veteran
Gaf user found answer comparing to pc settings using config file, cascade settings higher on ps5 (r.Shadow.CSM.MaxCascades 4 vs 3), perf. not big diff so maybe they will fix it on xsx later
Oh I was just making an objective observation and not a qualitative one.
Great video separating the different technologies coming together to create the product.
If you’re and enthusiast of progression there’s a lot here to be hopeful for, on the future of graphics. A quick run down:
Brute for ray tracing capability has been doubling in performance for each generation from the last
New hardware based techniques to further performance arriving every other generation.
New software innovations on how to handle RT happening every year, something that is applicable to all
New reconstruction techniques able to offer significant performance advantages every other generation.
New AI techniques to assist with ray tracing arriving possibly every other generation.
All of these combined really shows how quickly we can get there. The next generation of consoles will be part of this new generation of games where there is RT only lighting and the difference will come down to bounces and diffuse in terms of settings.
It’s looking great. Lots to learn and see, a whole new era of optimizations and hardware to support this as well as interesting novel techniques.
I honestly like these videos much better than the console comparisons which have gotten stale by now. Usually, SX and PS5 perform 99% the same and only console warriors nitpick the differences to declare a "winner". This is what DF should do more often. Tech deep dives, dev interviews, videos about rendering techniques, etc. But I guess it wouldn't generate as many clicks as comparisons which is a shame.
I'm a bit surprised listening to this part of the video. Alex explained the difference in performance cannot be measured because there is no way to disable SER. Why not compare GPUs of different architectures (since SER is locked to Ada) that are delivering similar raster performance like a 4070 vs a 3080 and then measure the performance difference in Overdrive? If SER really makes a difference, the 4070 should be noticeably faster.
This has already been tested and the 4070 is just on par with a 3080 in Cyberpunk overdrive. Therefore we can assume this feature is more of a marketing gimmick rather than anything else.
I liked the video and I get Alex is very excited about this technology (I am too) But I would like a bit more critical thinking next time.
By the way, Path tracing still does not fix one of the major flaws the game's presentation has. Objects and NPCs still appear like they are floating when lit from sunlight.
Last edited:
davis.anthony
Veteran
SER is a lot like VRS where it only gives you a performance boost in certain circumstances.
If your ray tracing is bandwidth bound then SER isn't going to do a fat lot.
davis.anthony
Veteran
We have more than one.
Portal RTX uses it and has been noted to give 10-20% performance increase.
But that's a much simpler game that CP2077.
Comparing SER on and off is the only way to actually know its performance differential.
Comparing 2 completely separate architectures, hardware, and clocks and one having SER and the other does not, and using that differential to judge performance is the definition of a lack of critical thinking.
That doesn't matter to the enduser. If the 2023 $599 4070 cannot push above its weights against the three year old $699 3080 in an ideal scenario (path tracing with SER enabled), then that feature is worthless.
Alex was talking about how these modern cards are better adapted to path tracing workloads than Ampere, but we've not seen that based on the data we have (being, the 4070 is just about as fast as the 3080).
Last edited:
What test are you referring to? I know Richard used the two in his video, but from memory, the result there was that the 4070 just edged out the 3080 in CP2077 PT but fell slightly behind in other Ray traced games. So while the difference isn't as transformational as one might have expected in this game at least, it arguably has been shown to make some difference, perhaps as much 10% or so?
Flappy Pannus
Veteran
I also prefer in-depth looks at technologies like these too, but ultimately DF exists because it's covering technologies that are being delivered in commercial products. This type of coverage will scale when the technology is available to more consumers through shipping products that the majority of their viewerbase can afford.
Most games just don't really warrant these kind of tech deep dives because they're restricted in what they can incorporate due to the platforms they target and their development budget. Still, games like TLOU:RM and Horizon:Forbidden West have gotten these 30min-1hour deep looks as well even though they're not necessarily breaking new rendering ground, at least to the degree that 2077 Overdrive is. Admittedly though I also sometimes have some reservations with those types of videos as well as they can skirt close to 1 hour infomercials for the game in question at points.
Your
No one does comparisons in that way. You want to know the value of a singular component all other features must be isolated. A 4070 has 33% less physical hardware than a 3080. I’m not sure what you’re banging on about SER, if it has comparable or better performance than a 3080 with 33% less hardware in RT games like this one, I would say the new features combined are making the difference.
No one does comparisons in that way. You want to know the value of a singular component all other features must be isolated. A 4070 has 33% less physical hardware than a 3080. I’m not sure what you’re banging on about SER, if it has comparable or better performance than a 3080 with 33% less hardware in RT games like this one, I would say the new features combined are making the difference.
snc
Veteran
To be fair, there were plenty of df pc gpu comparison vs consoles using top notch cpu on market ;d
Yes, which is doing exactly what @iroboto said. Isolating the GPU as the point of comparison by remove any possibility of a CPU bottleneck on the PC side. Obviously the chance still exists that there is a CPU bottleneck on the consoles side (although a well developed game, particularly if it's using DRS, should balance the usage of the two components pretty well in a console) but at least by removing any potential bottleneck on the PC side you're reducing the number of unknowns from 2, to 1.
snc
Veteran
With possibility with ending with wrong idea that you dont need to have strong gpu when majority will not have as powerful cpu and will be more limited in this aspect. Wrote it many times so will not repeat myself but using mid range gpu with high end cpu can create false representation of what real performance will be on mid range pc.
- Status
Similar threads
