People need to stop using teraflops. It is a worthless measurement. The ps4 pro has more flops than the series s. The ps4 pro isn't more powerful.
i7-13700k |
Vengeance 32 gb |
RTX 4090 Ventus 3x E OC |
Switch OLED
People need to stop using teraflops. It is a worthless measurement. The ps4 pro has more flops than the series s. The ps4 pro isn't more powerful.
i7-13700k |
Vengeance 32 gb |
RTX 4090 Ventus 3x E OC |
Switch OLED
Wii U had the most flops of any Nintendo home console! hiyoooo!
Chrkeller said: People need to stop using teraflops. It is a worthless measurement. The ps4 pro has more flops than the series s. The ps4 pro isn't more powerful. |
Its not really worthless measure, it's just series S has a vastly better CPU and everything else.
Here is my final prediction:
If GPU is Tegra T239:
The performance level to expect when docked (assuming 18-20W total power target) would be about 85%-90% (in terms of framerate) of a RTX 2050 mobile 25W.
https://www.techpowerup.com/gpu-specs/geforce-rtx-2050-mobile.c3859
The performance level to expect when portable would be slightly better than the base PS4, but of course with more modern features (i.e ray-tracing.)
Over time the Switch 2 would outperform the 2050 mobile because it would have much more video memory available.
As far as desktop GPU's are concerned, it would be somewhere between a GTX 1050ti and GTX 1650.
IF GPU is a Lovelace Tegra or if power target of the T239-based platform is loosened to 25W-30W total:
The performance level to expect when docked (25W-30W power target) would be somewhere between an RTX 2050 mobile 25W and RTX 3050 mobile 35W 4GB, likely closer to the latter.
https://www.techpowerup.com/gpu-specs/geforce-rtx-3050-mobile.c3788
A roughly comparable desktop GPU would be the RX 570 or a GTX 1650.
This would put it at about (or very slightly below) the level of a Series S when docked, with the Series S more like a GTX 1650 super/GTX 1060/RX 580.
When portable, I'd expect between PS4 and PS4 Pro performance.
EDIT: DLSS isn't magic, but it can in many cases bridge the gap between a budget/performance half-tier or even tier (i.e GTX 1650 level -> GTX 1650S level; RX 570 level -> RX 580 level; GTX 1050ti level -> GTX 1650 level, etc.) Half-tiers/tiers tend to be about 5%-15% more performance than the iteration they improve upon. That's in the realm of what gains the Switch 2 would expect to get for using DLSS over FSR and slightly decreasing internal resolution (say internal 800p instead of internal 900p/1080p by using DLSS rather than FSR; 30%-50% reduction in internal pixel-count for similar image-quality target.)
EDIT 2: I'd expect most Switch 2 games to upscale from 720p-900p -> 1080p or 1440p in docked mode and from 540p-720p -> 720p-1080p in handheld mode (depending on the screen resolution mostly.) Some simple 2d/2.5d titles would do 1080p -> 1800p or 4k. VRR would be nice, but if there isn't VRR then we'll probably have a 60fps target for simpler games, and 30fps for graphically demanding ones, dependent on genre too of course.
Last edited by sc94597 - on 10 September 2023Oneeee-Chan!!! said:
Really? Besides, aren't the GPUs of PS 3 and Xbox 360 almost equivalent? If the PS 3 GPU is equivalent to 7900 GT, then which GPU is equivalent to the Xbox 360 GPU? I was thinking about x800. |
You might have heard either the rumors it was downclocked from 550 MHz to 500 MHz, or that it was once going to use two Cell processors but instead they opted for an Nvidia GPU for the graphics.
The X360 uses an Ati GPU that was manufactured specifically for the console and has no direct counterpart among consumer parts. It's like a Radeon X1800/X1900 but with lower clocks and a more modern (unified) shader architecture.
One fact remains here: None of us saw this running. Eurogamer's sources did, and they said "comparable". Not identical, not indiscernible. Comparable is subjective, but we have to remember that these are people who, presumably, know what they're talking about. If EG and VGC are reporting this, this isn't just some reporter who was privileged with catching a glimpse of it running and thought, "Yeah, that looks the same as my son's Playstation", nor is it joe-blow's blog or someone on Twitter that got something right once. Again, they didn't say that it merely looked ok or even "pretty good" They said "comparable" to how it runs on PS5/XBS, which I take to mean, under these circumstances, very impressive... and probably, much better than expected.
What's funny here is that it was also Eurogamer's sources that created a stir before WiiU arrived too; they saw it and said not to expect a sizable jump over PS360. This was after we knew the specs of WiiU and YT was flooded with video claiming there was no way this was true based on what we knew. I remember a specific question to Eurogamer asking how this could be the case based on WiiU's advantage in RAM, GPGPU, out of order instruction sets... whatever. And the response was basically, "Yeah, we know, but despite the potential for advancements in those areas, they're not seeing anything that is leaps above the 7th gen console's capabilities." In fact, ports of 7th gen games performed worse. As it turns out, despite all of the naysayers, Eurogamer's sources were largely correct, and history could repeat itself here, just in Nintendo's favor this time.
I don't see anything from them claiming that anyone is going to want to give up their PS5 and start buying all of their games on Switch 2, as there will certainly be differences, but what I do take away from this is, if true, Switch 2 will be able to hold its own with multiplat performance much better than Switch did during its tenure. I really don't care how a Matrix demo runs or how much better it would look on the PS5 (comparable is perfectly fine with me), as neither are relevant to how the Mushroom Kingdom, Hyrule or Tallon IV will look on this device, and based on what these sources say, I couldn't be happier.
deerox said: Correct me if I'm wrong, but doesn't DLSS look terrible when used on a 30fps game? |
Terrible for whom? For die-hard sharp-eyes tech guys yes, for regular guy/games no.
On the topic of diminishing returns, the reason why generational leaps don't seem to be happening anymore for console gamers, as they did before, is because of the mid-gen refreshes. If there weren't a PS4 Pro and One X, then console gamers would've felt a bigger leap between the last generation and the current.
HD 7850 -> RX 6600xt level performance is a huge leap.
HD7850 -> RX 470 (bottlenecked by the CPU) -> RX 6600xt level of performance is much less so. That is more along the lines of PC gaming generations, which always felt more incremental than generational.
Consoles had essentially become closed-platform PC's for the last decade. Hell, and now the platforms are starting to open with GamePass and Playstation exclusives being available on PC's.
Nintendo is the main exception to this trend in that they didn't go the Switch Pro route with the Oled model and their games are only officially playable on their hardware.
Last edited by sc94597 - on 10 September 2023archbrix said: One fact remains here: None of us saw this running. Eurogamer's sources did, and they said "comparable". Not identical, not indiscernible. Comparable is subjective, but we have to remember that these are people who, presumably, know what they're talking about. If EG and VGC are reporting this, this isn't just some reporter who was privileged with catching a glimpse of it running and thought, "Yeah, that looks the same as my son's Playstation", nor is it joe-blow's blog or someone on Twitter that got something right once. Again, they didn't say that it merely looked ok or even "pretty good" They said "comparable" to how it runs on PS5/XBS, which I take to mean, under these circumstances, very impressive... and probably, much better than expected. |
Having a Gamecube CPU certainly didn't help Wii U and those 7th gen ports were rushed anyway. It's when we got Xenoblade X and BOTW that showed far how Wii U could truly go.
archbrix said: One fact remains here: None of us saw this running. Eurogamer's sources did, and they said "comparable". Not identical, not indiscernible. Comparable is subjective, but we have to remember that these are people who, presumably, know what they're talking about. If EG and VGC are reporting this, this isn't just some reporter who was privileged with catching a glimpse of it running and thought, "Yeah, that looks the same as my son's Playstation", nor is it joe-blow's blog or someone on Twitter that got something right once. Again, they didn't say that it merely looked ok or even "pretty good" They said "comparable" to how it runs on PS5/XBS, which I take to mean, under these circumstances, very impressive... and probably, much better than expected. |
FFVII Remake Intergrade, max settings, runs at 110+ fps on an RTX 2060. Native 1080p, no DLSS required.
Even if the Orin GPU is heavily underclocked even in docked mode, it's not that heavy of a game, TBH.
You can also run the Matrix Demo at 30 fps with DLSS in a 2060.