-
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 Weekly: If Xbox Series X is more powerful, why do some PS5 games run better? We finally have some answers.
- Thread starter Thick Thighs Save Lives
- Start date
- Hardware
Mr Moose
Member
That's 28%.
Edit:Wait I might've done this wrong.
Edit: It's 44%?
2560 × 1440
3072 x 1728
Last edited:
diffusionx
Gold Member
DF wasn't wrong with their initial analysis of the specs. They lacked a holistic picture of the situation but they're not devs, and devs were fine with letting their games do the talking. Yes, Cerngod's approach was right but for obvious reasons DF should not have taken what he said at face value.
If DF made any sort of editorial mistake, it was that they didn't write this article 3 years ago, and instead chose to continue to shill for MS and say BS like it's the tools.
What PC titles are on Xbox but not on PS5, outside of the ones made by Microsoft? PS4 games also work fine on PS5, it turned out much better than people feared before release.
There's no reason why the PS6, iterating upon the PS5, cannot be fully backwards compatible.
Last edited:
winjer
Member
But they were wrong with their initial specs, because they only looked at the TFLOPs.
Meanwhile there people talking about rasterization, culling and geometry throughput of the PS5.
Lysandros
Member
Anyone still looking at those culpable propagandist clowns for an 'explanation' for this entirely made up "mystery" at this point (four years into the generation) is the very root reason for perpetuating this retracted PR fuelled lie to this day. The explanations were in plain sight from the first hour and we were writing about those for four years in this very site.
Last edited:
Does make me curious at the level of performance Sony can get from the PS5 Pro with PSSR and the ‘secret sauce’ Cerny is working on. Increasingly it is all about software solutions and efficiency.
I can see Alex getting quite flustered when he see’s what PC hardware is required to match PS5 Pro!
I can see Alex getting quite flustered when he see’s what PC hardware is required to match PS5 Pro!
JackMcGunns
Member
The higher clock rate we knew about, that's not an advantage, it's a trade-off. Less compute units = less heat => higher clock vs More compute units = more heat => lower clock.
Higher clocks can help certain engines while having more Compute Units could help in others, such as Unreal Engine 5. It's a wash really when it comes to whether higher clock or higher compute is better, but the real world differenciator is the more efficient compiler, that's key in marking a difference.
It will be around a 45% performance improvement that we saw on the leaked document. PSSR won't make much of a difference from FSR 2.X, 3.X, or other TAA custom implementations from a performance perspective. It has a huge 2 ms cost on frametime to reconstruct a 1080p image to 2160p, when FSR 3.0 cost is in the low 1.X ms on a GPU with the same performance level. It is clearly a solution to obtain better image quality with a similar performance to other techniques.
And whatever 'Secret sauce' Sony is doing it will only be usefull if they are trying to alleviate potential bottlenecks with hardware customization and fixed function automatization, otherwise devs won't take care if they put more optional dedicated processors to offload some task.
Topher
Identifies as young
Mark Cerny talked about the advantage higher clocks bring to the table in Road to PS5. Now DF and devs they have talked to are echoing what Cerny said. So no, it isn't a wash according to people who would know.
timestamped
Last edited:
TheThreadsThatBindUs
Member
DF are disingenuous hacks....:
DF when comparing PC GPUs: "Don't just look at TFLOPs because these are BS theoretical sustained floating point operations that are almost never seen in real software workloads."
Also DF, when comparing XSX with PS5: "On paper, the XSX is a considerably more powerful console [because it has 18% higher SP GPU FLOPS]"
Make up your mind, DF. According to you, max. theoretical single precision FLOPs only seem to matter when Xbox has a bigger number.
DF when comparing PC GPUs: "Don't just look at TFLOPs because these are BS theoretical sustained floating point operations that are almost never seen in real software workloads."
Also DF, when comparing XSX with PS5: "On paper, the XSX is a considerably more powerful console [because it has 18% higher SP GPU FLOPS]"
Make up your mind, DF. According to you, max. theoretical single precision FLOPs only seem to matter when Xbox has a bigger number.
diffusionx
Gold Member
Like I said, the people who knew let the games do the talking. The problem for DF was that they pulled out the immature tools excuse almost immediately. It's impoissible to find anything a few years old on the internet anymore, but The Verge wrote an article saying as much that leans heavily on DF's analysis:
/cdn.vox-cdn.com/uploads/chorus_asset/file/21990372/vpavic_4261_20201023_0068.jpg)
Why is the PS5 outperforming the ‘world’s most powerful console’?
The PS5 is punching well above its weight
www.theverge.com
Mr.Phoenix
Member
And you are still making the same mistake DF made all those years ago. Trying to compartmentalize everything into one neat tidy box. Whereas, as DSF should have at least known... consoles don't work that way.
A good way to look at consoles and PCs, is that consoles are Formula One cars Fitted with a 1.6L V6 engine that produces up to 850HP. But a PC, is like a run-off-the-mill consumer vehicle that would need a V8-V12 engine that weighs twice as much to produce the same amount of power.
The point here, is that consoles are typically far more optimized, with every single thing designed and working together to make the whole. From the compute units, to the cache scrubbers, to the compiler, to the IO stack, to the SSD... even down to how the blocks or buses in RAM is addressed. Everything works to make the system what it is. You... and DF make the mistake of forgetting (or not knowing?) this, DF focused on the TFs. To them, its simply more TFs = More performance, you... are focusing on the compiler...
what I am saying, and to be fair a lot of others have said the same on this forum over the years... it's a lot more than that. Everything works together. MS simply made a less efficient system. Be that from compute units, memory architecture, tools.... whatever. The literally made a 12TF machine that performs like a 10TF machine, and sony made a 10Tf machine that can match or outperform a 12Tf machine.
JackMcGunns
Member
Yes, we understand that, but it's been discussed that the higher compute unit count in Series X does have its benefits over higher clocks in certain areas.
In an apples to apples comparison with both systems having the same theoretical performance, where one has more compute units and lower clocks vs another with higher clocks and less compute units, it's without a doubt more beneficial to have higher clock speeds, but in the specific case between PS5 and Series X, there's an 18% gain on Series X side because it has a considerably higher CU count, more so than just matching the higher clocked PS5; it exceeds it. So again, both systems will perform relatively the same, that's what I meant by it being a wash, not that the higher clock doesn't help. What is certain is that the major crux in the Series X side is not having the best compiler.
Geometric-Crusher
"Nintendo games are like indies, and worth at most $19" 🤡
12 teraflops is the new 64 bit
Microsoft always follows strategies that are proven to fail. Rumored next Xbox will have reference specs for third party manufacturers.
Last edited:
Topher
Identifies as young
Fair enough. Not sure if its the "crux" but the compiler and API are certainly both factors according to DF.
TFs can be used to make ballpark comparisons within the same architecture, because the ratio between compute and the other theoretical rates is generally kept pretty constant. In this case the PS5 has a higher fillrate due to the much higher clock speed and also some architectural differences.
Lysandros
Member
Not only fill rate, every single GPU fixed function metric not tied to CUs including geometry generating/culling and command processing is higher on PS5. Even things like async compute scheduling are faster on it not to mention lower latency caches, higher bandwidth L1 etc.
Last edited:
TheThreadsThatBindUs
Member
Not really when the clockspeed difference is so significant.
Clockspeed impacts the entire GPU pipeline, e.g. front-end, caches etc., so just looking at TFLOPs (which only apply to the ALUs) and thinking that that provides a meaningful picture of the combined GPU performance differential between the two systems in real games is misguided.
Katatonic
Member
Yeah man, just wait...
skyfinch
Member
Been waiting for the past year. I gave up and bought a Nintendo Wii.
TheRedRiders
Member
Do people realise that by the time compute heavy engines become the standard, the Series X/S will have fallen into irrelevance.
SlimySnake
Flashless at the Golden Globes
What a bunch of tools. 40-50 more frames per second. 30% instead of 18%. 9 tflops instead of 10 tlfops. STAGGERING! Get the fuck out of here.
Last edited:
marquimvfs
Member
Well, I'm prepared to be shocked. When it will happen?
Last edited:
Physiognomonics
Member
The problem should not be why PS5 is quite ahead in some games (like Hogwart Legacy, Dragon Dogma 2 or Elden ring, all 3 using 3 different engines) with a 20% smaller APU, it's why XSX performs so badly compared to similarly specced PC. And what do compiler mostly do? They optimize caches hits. Here XSX is quite disadvantaged compared to PS5 and RDNA2 GPUs. Maybe some games also "compile" better on PS5 because the cache architecture is better.
Embearded
Member
These tweets aged like milk.
These people are complete ignorants when it comes to computer architecture and programming.
And i said it before, DF is a nice pixel comparing channel and that's it. They also don't understand how the machines work, they just repeat whatever they can find online.
They also don't have experience on game development, not most of them anyway and surely not enough to call themselves experts.
These people are complete ignorants when it comes to computer architecture and programming.
And i said it before, DF is a nice pixel comparing channel and that's it. They also don't understand how the machines work, they just repeat whatever they can find online.
They also don't have experience on game development, not most of them anyway and surely not enough to call themselves experts.
Bernoulli
M2 slut
this crazy guy is still on it
What a bunch of tools. 40-50 more frames per second. 30% instead of 18%. 9 tflops instead of 10 tlfops. STAGGERING! Get the fuck out of here.
his account is proof of mental illness
S0ULZB0URNE
Member
What's sad is people follow them.
Look, people are saying the 20% Teraflop difference is one of the smallest differences we have had in years. just in terms of simple teraflops too so that's before you even get into the idea that the CUs are more difficult to keep busy vs lower CUs at higher clocks on PS5. The PS5 was just well designed and that "20%" rarely showed in 4 years because that higher GPU clock they went with gave advantages that beat XSX at other things too beyond just theoretical compute. They have a different philosophy and those came with their own challenges, ie they had to spend money to cool the machine better (I remember the dumb fanboy FUD about PS5s running too hot), they had to provide low level access and break possible compatibility. It's ok to accept that Cerny got it right and concentrated on creating a better balanced machine that's easier to take advantage of because those fewer CUs are faster.Do you know how much theoretical compute advantage a 4090 has over a PS5... and yet that doesn't stop console people from saying the games "look basically the same".... and in most cases they do... because they aren't designed to take advantage of the hardware... which has been my entire point.
Congrats on walking right into that. Perhaps you're the one who should take a break from this thread all things considered
I'm not sure why you're dragging in the 40% difference between the PS5 Pro and SX here either. That's double the difference, no? Why are you trying to make a completely theoretical "20%" seem huge but a 40% difference (plus a completely different/newer architecture) advantage between PS5 Pro and SX seem unsubstantial?
You got a warning in the other PS5 Pro thread where you were talking about midrange GPUs being "far more powerful" than PS6 and your GPU "smoking it" for years before it releases. Then going on to say they shouldn't release the Pro because they should stick to one console for 7years.
Which I would have given the benefit of the doubt about both for claiming midgen cards will beat PS6 and for supporting a single console if you weren't boasting about the Xbox One X pushing the envelope and doing things only todays "highest end cards" can do.
That's when people can see through the bullshit.
Last edited:
Bernoulli
M2 slut
i don't know what his job is but he posts dozens tweets everyday, he has more than 100K tweets all about playstation
Ev1L AuRoN
Member
Given that they want the XSS version to be as close as possible, I'd imagine that the bulk of the optimizing time and QA are spent on the weaker machine, and they just expect the series X to brute force his way, and by doing that they end up subtilizing their stronger machine. Not a surprise given that most Xbox users are on XSS.
Last edited:
Mr.Phoenix
Member
And you still don't know what his job is?
What I find crazy is that these fuckers would still jump on the next MS announcement with a straight face claiming its the best thing since sliced bread. Not a chance they arent on a payroll.... no one can be that stupid.
What a bunch of tools. 40-50 more frames per second. 30% instead of 18%. 9 tflops instead of 10 tlfops. STAGGERING! Get the fuck out of here.
marquimvfs
Member
They're deliberately pushing the scale to favor Xbox. To me, the bias is obvious even in the way they're writing the pieces, deliberately downplaying Playstation feats here and there.
See a little of what I mean here:
Now, have same sentence written this way:
It certainly hits you different when they both say the same thing, right? To me, one praises the Xbox power, when the second could praise the PS5 feats of matching the same results.
Some might say that it is innocently done, that I'm overreacting and so, but, I say that when it is done repeatedly over the text, there's some kind of steering of the narrative toward one of the competitors.
Take this other line as an example:
Let's rewrite it to flip the bias:
Can you spot the difference? Every win that the competition has over their "favourite" platform is, somehow, downplayed and burried under lesser adjectives. Every "win" and bigger characteristics of their preferred platform is mentioned trough hell, even if in real world it means nothing, like the teraflops measure difference.
Now, don't get me wrong, I'm not against a publication have it's own bias, certainly everyone out there does, it's journalism 1-0-1 that "neutrality" doesn't exist, it is called journalism angle, and every publication, the editors, the sponsors have one and it can certainly affect a piece written by someone. The sad part, and what bothers me much is when a publication like that does sell itself as neutral. You can 100% say that, no matter how fair the comparison are done, they will always present the results with bias, and that say a lot about their own bias. Better is to present themselves as an Xbox geared crowd, like many many others do over the internet, be it Sony or Microsoft biased.
I don't have a problem reading something interesting in an Xbox publication, but when I know that it really is an Xbox publication, I can ajust my perception to balance the possible bias. It's only natural, for everyone.
Last edited:
SonGoku
Member
Thats what im telling you, going by theoretical specs in the best case scenario for a game made from the ground up for SX all it would take for PS5 to run the exact same game is to drop the resolution by 20% which again is not a considerable difference
20% compute advantage at best would translate into a 20% resolution advantage IF there is 100% CU utilization and scaling. at thats the best case ideal CU utilization scenario
Thats subjective and very arbitrary to draw the line where a resolution "feels" like 4k. Objectively its only a 20% difference in resolution and 20% is not considerable especially after factoring in image reconstruction, up sampling and other post process effects
Thanks, my bad and thats the ideal hw utilization for SX which is less on real game scenarios where engines dont always enjoy perfect CU utilization and theres other bottlenecks to contend with. Even in the best case scenario for xbox, the spec gap is is not considerable
Last edited:
GHG
Member
Noteworthy that the 'new feat.' was accessible on PS4 a decade ago (courtesy of GNM just being a 1:1 GPU assembly basically).
Giallo Corsa
Member
Blood Borne
Member
No console has sold and dominated like the PS2, yet multi platform games on Xbox and GameCube consistently performed better, hence I’m not really buying the whole “PS5 is the lead console therefore devs are optimising it better than XSX”.
sendit
Member
That's called a moving goal post.
marquimvfs
Member
Well, one could argue that those machines were substantially different, and that the PS2 also deserved a port instead of being left to dry, because it sold way more.
One proof of it is that, sometimes, devs simply didn't bother to make Gamecube versions. Or that PS2 gained ports from games of the newer generation, even if fundamentally different, like Wolverine. Even if it was hard, they couldn't ignore.
Nowadays it's easier to make ports for everything, the machines are more similar and streamlined, so a dev could develop for everything he desires without putting so much effort like the times you're referring to, and that lack of effort could cause some disparity.
I'm sure that this is not the only problem, to do such affirmation is ridiculous, but i'm reasonable enough to recognize that it can be part of the big picture, for sure.
Last edited:
Self_Destructive
Member
You're confused... I'm not the one saying a 20% isn't substantial... and I'm certainly not saying 40% isn't substantial... quite the opposite. I mean, I can't even begin to understand how you came to that conclusion. The argument made against what I'm saying is "well just drop the resolution, problem solved"... and yet that's not really the case.
Oh was THAT the reason I got warned? Because some people can't stand to hear that I believe midrange GPUs of the time will completely smoke the PS6 when it releases due to self imposed power limits of consoles? Is it really hard to hear that and simply disagree without claiming I'm being antagonizing? Cmon for crying out loud.
solidus12
Member
They always carefully choose their words to put Xbox in a positive light.
They don’t want to lose access to the perks that Xbox can offer, i.e., trip to Xbox HQ for exclusive access to One X and Series X among others.
That’s why they’re always downplaying PlayStation.
Last edited:
Self_Destructive
Member
Probably had it before.
XBox to PS2/GC 'power' delta was comparable to that of X1X to the original PS4, but even more dramatic in some ways (HDD built in etc).
GC to PS2 is a different story - but between lack of data (noone was doing real analysis back then), wildly different featuresets and power-differential being skewed to PS2 more than GC, it really says nothing about modern console comparisons.
Last edited:
Insane Metal
Member
Well, it has been out of sight. No one's buying it.
Self_Destructive
Member
