-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
DF: Control PS5 Vs Xbox Series X Raytracing Benchmark
- Thread starter AnotherOne
- Start date
alabtrosMyster
Banned
I'm fairly certain that in the medium/long run they will make more money on the DE clients, these are completely locked in and and they don't get to sell/buy used games-ever.
DJ12
Member
Exactly, speaking to devs just means df don't even bother to investigate. How'd that work out here.
DJ12
Member
Depends if there's a photomode that the series x 'wins' I guess.
Fans are thirsty for their 3rd win, however meaningless it is.
Concern
Member
AgentP
Thinks mods influence posters politics. Promoted to QAnon Editor.
I don't want to wade into this nonsense thread, but I push back on this claim. Hitman 3 is not evidence for a power difference. You would need to have both systems with the same setting to draw any conclusions. With the PS5 set lower and not dropping frames and the XSX set higher and dropping frames, you cannot say anything other than they are different. It's like a car in a 60mph zone doing 60mph and another in a 70mph zone and sometime dropping to 65mph. Is one car faster than the other? You simply do not know.
phil_t98
#SonyToo
except its 99% of the time higher res and 60fps, 1 section which is a bug
Rentahamster
Rodent Whores
Speculation should only be used sparingly. DF deals with measurable facts first and foremost. They don't use their tech knowledge to guess as much as you think they do. They shouldn't be producing a theory if it's just going to be talking out their ass.
Bernd Lauert
Banned
There is one cutscene where PS5 drops to 37 fps and XSX drops to 32 fps. So PS5 has 15% higher frame rate, but XSX is pushing 44% more pixels.
Applying some simple math, the XSX is 24% ahead in this instance.
ethomaz
Banned
I believe that too... the more consumers wanted the DE and the more it the production got cheap the DE could increase in production in comparison with the normal version.
Sony probably thought in the long run.
ethomaz
Banned
Like I said that is something you expect from a Technical Analysis.
DF is lacking too much.
FrankWza
Member
And they put out 3 articles about Control in just a few days. You would think they would leave no questions.
ethomaz
Banned
Even more in the supposed first "real" GPU RT Benchmark comparison.
Workload/power draw should a be thing to be analysed in each scene, no? I mean to see where the GPU is working more or less.
You can even extend that comparing with the gameplay data if they cared.
Last edited:
SlimySnake
Flashless at the Golden Globes
i dont think they have ever measured workload/power draw when comparing games.
ethomaz
Banned
I remember they did in the past... there are articles about the power draw of PS3/360 games in different scenes.
It used to have that info the the Face Offs (but they stopped to use the word face off).
Anyway PC benchmarks articles have that info... some more detailed others more superficial.
Last edited:
Fredrik
Member
Tbh I don’t think many care about power consumption, noise and file size...
Maybe input latency is more important but I’m playing it on Geforce Now and that should be a worst case scenario even if I have 15ms ping to servers. Plays great! So I can’t see that being a problem.
Md Ray
Member
It's not a mistake. One of the biggest benefits of going narrow and fast apart from profitability is also back-compat performance.
Remember Cerny's 36 CUs vs 48 CUs example?
I'll elaborate on this a bit more.
Let's say he had went with 48 CUs. In order to reach 10.28 TFLOPS, he'd have to run the GPU at 1673 MHz frequency. So...
48 CUs @ 1673 MHz = 10.28 TF
Now, in back-compat (PS4 Pro legacy mode), games would've have had access to just 36 CUs, not the full 48 CUs on PS5.
So the computational power available to BC games in this case would have been paltry 7.7 TF, not 10.28 TF. Do the math.
36 CUs @ 1673 MHz = 7.7 TF
So, just an 83% uplift in TFLOPS when going from PS4 Pro's 4.2 TF to 7.7 TF in BC mode as opposed to the 145% (4.2 TF to 10.28 TF) uplift we have now.
This means you wouldn't have seen those 2x or sometimes even over 2x boost to frame-rates (from ~29fps to 60fps locked) on PS5, when running PS4 Pro versions of the games like The Last Guardian (disc version), Ghost of Tsushima like you do now.
Last edited:
AgentP
Thinks mods influence posters politics. Promoted to QAnon Editor.
Ah, now the level with the flowers and the sniping is a bug. Not a performance issue. Check. My statement stands, you cannot draw a conclusions about power if you never push the PS5 beyond it's rendering limits. when the Ps5 "wins" it is the XSX tools, then the XSX "wins" it is the POWER! LOL.
FrankWza
Member
I don’t know. I would think that if they’re really trying to show us something groundbreaking they would leave no doubt. It appears that even outside of this thread there seems to be doubt in the depth of their work in this instance and it’s now being questioned by pros.
Did they ever use this type of solution to come up with a way to benchmark? And, is workload and power draw something that goes hand in hand when benchmarking this component? Is it the equivalent to leaving out say, compression when comparing SSD read speeds or something? I don’t know, just asking.
Slippery slope. I know they depend on clicks but if you’re going in depth I think providing as much context as possible is kind of important. If not, maybe they rushed it out? They could have waited and offered a more comprehensive look. So it doesn’t really matter what people care about if it offers more detail in a benchmark comparison it should be included.
phil_t98
#SonyToo
Your posting in a benchmark thread where the PS5 is lagging behind lol
Bryank75
Banned
When you're producing millions of chip, 10% is still massive in terms of console output. If it's as high as 20%..... that's cutting your potential customers served from 5 million down to 4 million.
Fredrik
Member
They’ve already done a lot of videos for Control, might have more coming I guess but I think they covered what people wanted to see:
Graphics parity and on console it plays the best on PS5.
Finding the photo mode GPU benchmark was pure luck, they probably stressed it like you say, wanted to be first.
Personally I find it super interesting though since we haven’t had anything like that for console before. And for Xbox gamers it’s cool since it shows that XSX have some extra power ready to be unleashed. I don’t think this means anything for this game though, that ship has sailed, no I don’t think Remedy will update it to 60fps RT.
In the end it’s mostly something that gives some hope to those who started thinking the MS PR for the power was pure bullshit. Me included.
Now it’s up to the devs to show that they can actually use that power for something useful without it breaking something else.
SlimySnake
Flashless at the Golden Globes
I dont know if the workload or power draw is gonna tell you much. Several people in this thread including myself have measured GPU and CPU usage during the photomode and its nearly identical to gameplay mode CPU and GPU usage.
This idea that the PS5 GPU isnt performing well because well its not operating at peak speeds simply doesnt jive with the fact that we ARE seeing variable framerates going all the way up to 50 fps in photomode. That tells me that the gpu IS indeed running at max.
I am bewildered by the refusal to accept these results tbh. 16% better GPU performance is exactly what was expected based on the specs. Both are RDNA 2 GPUs. Both are Zen2 CPUs. Both are using GDDR6 RAM. The only difference is that the xsx has 18% more tflops which as we can see are resulting in a 16% boost to performance. Same as we see on PCs when comparing two cards in the same family.
Now the stuttering and framedrops during gameplay are part of some other problem on the XSX which I would love to see explored more, but this 16% average is as legit as it gets.
SlimySnake
Flashless at the Golden Globes
right i forgot about dynamic resolution.
Neo_game
Regarding Hitman 3, I meant that the devs would settle for 1800p if they cant hit native 4k consistently since it's the most obvious step down and far easier to implement than dynamic resolution. Just like how they would settle for a 30 fps cap for better performing xsx games instead of upping the settings like we saw in Control.Werewolfgrandma
Banned
It's probably more about getting clicks and making all the dumb people watch the videos 5 times to get screen grabs of a few fps advantage here and there. Also people like you keep talking about them so what they doing is working.
On The Road to PS5 presentation, At 36:50 he states the PS5 SOC looks at the activities on the chip to set the frequencies. I imagine if this is indeed the case, part of the GPU that handles the RT operation would be in use mostly and thus the chip is underutilized. That in my mind would mean to keep thermos/fan noise down the system would deem it wouldn't be necessary to give the GPU more power than what's needed. You guys need to watch the video again, everything that is happening was already stated in Cerny's video.
Last edited:
vpance
Member
I wondered if PS5 could be downclocking itself here but I'm not sure if it would in this situation. It's only supposed to do that when certain heavy ops are performed that would cause the APU to heat up more than they want. Then they downclock it to avoid those spikes in usage. It's not meant to be a prolonged downclock nor does it need to do that to maintain good temps.
If either CPU or GPU is underutilized power from either gets directed to the other so they can maintain freqs if it needs it. If a game is really simple that doesn't mean the GPU downclocks to 1GHZ or anything like that. It almost always stays at 2.23.
Could be wrong but that how I remember it.
"Rather than running a constant frequency and letting power vary based on the workload we run at essentially constant power and let the frequency band Vary based on the workload"
This is what Mark said...
ethomaz
Banned
But Bathallia said he hates console wars
Werewolfgrandma
Banned
They all like moneyBut Bathallia said he hates console wars
OverHeat
« generous god »
Normal is a PC master race kind of guyBut Bathallia said he hates console wars
JackMcGunns
Member
Right, and with an unlocked framerate, the PS5 will run the game to the best of its ability. If it can hit 60fps with RT on, it will, but it can't, therefore it will output the max possible. With both consoles locked at 30, we wouldn't know which one is performing better, because despite the overhead, the locked framerate limits us to see only 30 from both even if under the hood they can render past it, so the cool part was discovering the unlocked framerate in Photomode allowing DF the ability to compare both consoles, not that comparing photomode is something typical for benchmarkers, it just so happens that it's the only way to run the game with RT on and an unlocked framerate, but of course as logical a situation, it won't stop the warriors from formulating conspiracy theories damning DF in order to defend their precious piece of plastic.
Last edited:
Loxus
Member
No worries about the double post.
I was referring to the comment not the video.
I don't think you actually get it. Photo mode is heavily GPU bound. All it shows that for "Control "the engine favours XBOX SX GPU. More CU"s, more GPU compute power for RT. If you performed the test, on proper game mode with RT and uncapped frame rate results could likely be much different since your frame budget would need to include CPU and IO factors. This doesn't mean that this would swing some "points" in PS5 favour, but based on what we are seeing in the performance mode, it potentially could. We just don't know.
Loxus
Member
My main point is, DF don't take the time to look and see the PS5 has better details.
Which is why the XBSX has a higher frame rate.
Hey, it's all good and cool if you take DF word as gospel but they aren't Video Game Developers.
They just analyses pixels and frame rates.
Loxus
Member
So your trying to say XBSX has low lod and textures at a distance compared to the PS5?
ps5 doesn’t have better details. Actually XSX and PC have double reflections they’re aligned together as show below. The glass is only a inch away from the metal surface behind. The ps5 actually has the reflection aligned further apart. I would take pc the way it’s suppose to be since it’s superior in RT
XSX shots
face
Look at right hand/arm
JackMcGunns
Member
I already addressed that in previous comments. The CPU's in PS5 and XSX are basically identical, with 100Mhz more on XSX giving it a slight edge. There's no reason to believe the PS5 would have an advantage even during gameplay moments when the CPU is also involved, I mean think about it for a minute, NXGamer showed a scene where the Series X was CAPPED at 60fps which means it could go even higher, but the PS5 was hitting 52fps and 50 at times, having said that, BOTH will drop in framerate when the rest of the gameplay elements have to be processed. The PS5 would have to have a MASSIVE CPU advantage to make up and surpass Series X already sizeable advantage with RT on.
Hmmm. Once you factor in the I/O pipeline, caches performance and game engine optimizations it may end up not having the sizable advantage in RT when running the games. We'll see. It's too early to tell. I do expect SX to have an advantage in most 3rd party cross-gen titles coming out this year.
alabtrosMyster
Banned
1 - people do care about file size, if games are often 30-40% smaller on PS5 then the smaller SSD becomes less of an issue... I mean, both companies offer upgrade paths for storage, MS even created a scheme to sell their proprietor storage media and made a fuss about it.
2 - Power draw is mostly academic, to help understand PS5's power profiles (and how the Xbox handles it as well) by example. It's about as academic as locked 60fps1800p vs native 4k with drops.
Who's to say its working harder in this mode? This system is a sum of its parts and trying to isolate parts makes the system fall apart. RT is just a part of the GPU, a section if you will. That part can be stressed but not the part which handles primitives. Shit, the geometry engine is a part of the gpu and can be bypassed
Fredrik
Member
Yes, if...
The small SSD size is definitely one of the biggest issues with these consoles, and MS allowing extra storage just as fast is great even at it’s high price.
But here we’re just talking about one game. 20-something extra GB on the SSD. It’s mostly a mysterious difference at this point. But sure if it becomes a norm then I too want lots of focus on that, but we’re not there yet.
martino
Member
if you put all that in mouth only.
PlayStaytion
Member
I'm against parity but historically speaking, the gap do actually pisses fans and weakens the sales ,pc gamers actually the ones that pays the ultimate price for these types of practices that's beyond anyone's reach.
Last edited:
Loxus
Member
Bruh, LOL.ps5 doesn’t have better details. Actually XSX and PC have double reflections they’re aligned together as show below. The glass is only a inch away from the metal surface behind. The ps5 actually has the reflection aligned further apart. I would take pc the way it’s suppose to be since it’s superior in RT
XSX shots
face
Look at right hand/arm
I don't know what your showing me in those images.
Can see anything.
Both are in line for future development. MS getting on more CUs Sony on the I/O however MS is also betting on I/O albeit differently because they cater to the PC crowd also.
saintjules
Member
ToTTenTranz
Banned
For someone who claims to only care about the PC, Alex has quite the regular presence in beyond3d's console forums...
OverHeat
« generous god »
Why should you care about consoles 3rd party when you got a beefy PC beside fun and giggles that guys is right.