• 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.

Next-Gen PS5 & XSX |OT| Console tEch threaD

Status
Not open for further replies.

CyberPanda

Banned
Don't even know why all of you guys are arguing back and forth when 1) final specs won't be set until around May and 2) if you're an Xbox gamer, no PlayStation gamer is going to change that and if you're a PlayStation gamer, no Xbox gamer is going to change that. Only Microsoft/Sony can change which console brand you prefer and that's based on if they do what you're looking for better than the other brand.
Specs need to be finalized very soon, so mass production and assembly can begin around spring time, for holiday release. May is too late.
 
Klee still chopping wood after he said that he knows the specs and hasnt posted much on the other forum. This guy been around, leaked amd chips, data miner they say, and well ppl were taking his tweets as news before he dropped the bomb of a 9.2 tflop console.
What r u talking about ? He showed the specs to the mod but asked the mod not to disclose due to his friend being at risk And mod agreed.that was made clear there. So don’t make it seem like he ran away.

this is Navi 10 classification which means it can not be final form of ps5 as raytracing is not supported in Navi 10 and if this was ps5 Navi 10 identification would not be there in AMD database as it is not Navi 10.
if Sony was launching a Navi 10 console they would be launching in 2019. I know this is a wet dream for you that Sony would launch a base 2019 tech in 2020 but it ain’t happening lol
 
Last edited:
D

Deleted member 471617

Unconfirmed Member
Specs need to be finalized very soon, so mass production and assembly can begin around spring time, for holiday release. May is too late.

May is Spring. Either way, doesn't matter. Until the final specs are set, why anyone/everyone is even arguing is beyond me. Instead, do what im doing, eliminate your backlog (if applicable) before next gen. :)
 
What r u talking about ? He showed the specs to the mod but asked the mod not to disclose due to his friend being at risk And mod agreed.that was made clear there. So don’t make it seems he ran away.

this is Navi 10 classification which means it can not be final form of ps5 as raytracing is not supported in Navi 10 and if this was ps5 Navi 10 identification would not be there in AMD database as it is not Navi 10.
if Sony was launching a Navi 10 console they would be launching in 2019. I know this is a wet dream for you that Sony would launch a base 2019 tech in 2020 but it ain’t happening lol
Whats the specs?? Clearly you cant say that i should believe that bullshit or the fact he dropped that post, and didnt post for another 17 hours, saying he fell asleep because he was so tired lol. I think you are the one who wants to believe anything to suggest sony will be more powerful and well thats the last straw. You are disputing ppl like rogame and komachi , thats on you fam:
 
May is Spring. Either way, doesn't matter. Until the final specs are set, why anyone/everyone is even arguing is beyond me. Instead, do what im doing, eliminate your backlog (if applicable) before next gen. :)
Peter you know why come on now. You know me from twitter and we talked about this. Ppl go crazy over power
 

44alltheway

Member
Whats the specs?? Clearly you cant say that i should believe that bullshit or the fact he dropped that post, and didnt post for another 17 hours, saying he fell asleep because he was so tired lol. I think you are the one who wants to believe anything to suggest sony will be more powerful and well thats the last straw. You are disputing ppl like rogame and komachi , thats on you fam:
But aren’t you doing the same thing? You keep on saying you know the XSX is more powerful, but haven’t really shown anything to make that believable. I’m not picking sides, just saying that your argument for Klee not sharing the information doesn’t necessarily mean he’s full of shit.
 
Last edited:
But aren’t you doing the same thing? You keep on saying you know the XSX is more powerful, but haven’t really shown anything to make that believable. I’m not picking sides, just saying that your argument for Klee not sharing the information doesn’t necessarily mean he’s full of shit.
No because well we are pretty sure this is sony apu that was leaked, and its clocking in around 9.2 where i believe i was told this number not too long ago by someone in industry. Well a range. Everything is matching up from komachi saying oberon is 300mm2 to 350mm2 for arden “xbox”. Todays rogames tweets, multiple sources from everyone guys in this forum quote or fawn over, things such as the form factor. Sony will never be going with that design because well they arent doing 2 sku approach and well they have a much more global market. Um being directly told in 2018 that xbox wont ever lose the power crown @ e3. Lots of things make me confident and thats why i was confident for every single post here that Xbox will be more powerful. Vs someone who even ppl at MS call the fake insider “klee”. Aside from a miracle, sony aint taking the power crown. Digest that all. Im done for the night. Goodnight
 

bitbydeath

Member
Specs need to be finalized very soon, so mass production and assembly can begin around spring time, for holiday release. May is too late.

Even Microsoft have admitted that the specs mentioned at the game awards are what they’re aiming for and therefore not what currently exists.

Some here like to think what Sony had in 2018 is final though.
 
Last edited:
A teraflop is a teraflop. When there were efficency increases from GCN 1.0 and 2.0, were people talking about if it's GCN 1.0 or 2.0 tflops? No, because there isn't a distinction. Flops are calculated by amount of cores etc in the GPU.
And how exactly would they reach an efficency percentage to put into a "GCN" flop vs a "RDNA" flops?
Test 50 games, get the average FPS increase? Maybe it would take 100 games? But what about DX12 games vs Open GL? What about the ability of different developers?
Or maybe they benchmark test a GCN card with the same clock and bandwidth etc to see what the increase is? But what program should they use as the default to get the official efficency ratio? Maybe they use all of them, and get an average?

Please just stop th silliness.
A Flop is a flop. They are calculated the same way. There is no such thing as a GCN or RDNA flop.

Totally inaccurate. This current situation is unlike previous GCN generation updates. RDNA represents the most significant modification of AMD's core architecture since the introduction of GCN. The changes are dramatic. Even the compute unit setup is vastly different. Even the way the process operations has been doubled in some instances inside the inner workings of each compute unit. So, yes, in this case, an RDNA teraflop is almost certainly superior to a GCN teraflop. AMD has measured for this, and game benchmarks for their only Navi chips on the market bear this out.

Why does AMD's 9.7 teraflop Navi GPU spank their 12.6 teraflop VEGA 10 GPU? The same 295w TDP Vega GPU running a 2048-bit bus of HBM2 ram. So, no, an AMD teraflop isn't an AMD teraflop in this new post RDNA architecture world. Performance per clock is 1.25x better and performance per watt is 1.5x better. AMD doesn't make these claims lightly because they haven't made them so boldly before.
 

Proelite

Member
Btw tidbit, RT is done by an accelerator on the APU that's outside of the GPU. Not hybrid RT but nifty like the powerVr solution.

If RDNA 2 RT sucks then it's a good choice on Sony's part.
 
Last edited:
A Teraflop is a unit of measurement. It measures the peak theoretical compute a GPU (or CPU) is capable of. A TF is a TF is a TF. Actual gaming performance is not directly measured by floating point operations. Why? Because not all game code saturates all the SIMDs in an optimal manner.
A GCN Tflop is the same as an RDNA Tflop. Actual performance is higher with RDNA due to higher SIMD utilisation and higher clockspeeds.
 
A Teraflop is a unit of measurement. It measures the peak theoretical compute a GPU (or CPU) is capable of. A TF is a TF is a TF. Actual gaming performance is not directly measured by floating point operations. Why? Because not all game code saturates all the SIMDs in an optimal manner.
A GCN Tflop is the same as an RDNA Tflop. Actual performance is higher with RDNA due to higher SIMD utilisation and higher clockspeeds.

No. at the same exact clock speeds, RDNA is 1.25 faster than GCN. And at the exact same wattage, RDNA is 1.5x faster.

Navi-Slide-20.jpg



Navi-Slide-21.jpg


As you can see, frequency accounts for the least of the improvement in performance over GCN. The biggest improvement comes from raw performance per clock enhancements, meaning the architecture is clearly superior to GCN. Clock speed accounts for less than 20% of the improvement over GCN whereas the architectural changes account for nearly 60%.
 
Last edited:
D

Deleted member 471617

Unconfirmed Member
Peter you know why come on now. You know me from twitter and we talked about this. Ppl go crazy over power

Power matters but it's not the be all end all. Excellent exclusive games matter a lot more than power because if you're not getting the exclusives that are taking advantage of that power, it doesn't matter. If anything, that power will just be wasted.

Outside of Gears 5, there's not a single exclusive on Xbox One that can match Uncharted 4, Lost Legacy, Horizon Zero Dawn, Spider Man and God of War in terms of just pure quality. Now, YOU can like Sea of Thieves for example more than all those games but is it actually better? No.

I love Mad Max. For me, it's a 9/10 but realistically and favoritism aside, it's no better than a 7/10 but I love the game. Days Gone is an excellent 9.5/10 for me and my 2019 goty. But realistically, it's no better or worse than an 8/10. Everyone can love, favor and prefer what they want but can everyone actually say what is legitimately better?

I love Days Gone but is it better than REmake2, Sekiro, Gears 5, Control, Jedi Fallen Order? For me, it is but in general and it being my open world dream freaker game come true aside, I can admit that it's not because I know the game's flaws and negatives.

As for Xbox 4 (yes, im calling it Xbox 4, lol) vs PlayStation 5 in regards of power, im expecting Xbox 4 to be more powerful but in the range of 52%-48%. It won't be a big enough gap to make a noticeable difference. What will be noticeable though is what all the respective first party studios can get out of those consoles. In my eyes, Sony's first party studios overall are simply far better. I only put CDPR and Rockstar in the same class.

Microsoft's first party studios are mostly newly acquired and obviously have to prove themselves as a first party exclusive development studio for Microsoft. The big 3 are established but as much as I enjoyed Gears 5, The Coalition needs a new IP before Gears 6 in my opinion. Same for 343 after Halo Infinite.

I'm a huge fan of Ninja Theory but Hellblade 2 while I will play through it shouldn't have been their next major game. It should have been a new IP. Create a new IP, establish it into a new franchise and go from there. Playground remaking/rebooting Fable to me isn't the answer. A new fantasy action RPG IP would have been better if it is indeed Fable.

Perfect example is this - during Xbox 360/PlayStation 3 generation, I was an Xbox 360 guy because they gave me more exclusives that I wanted to play than PlayStation 3 did. I had literally no interest in Killzone, Infamous, Resistance, etc. whatsoever. PlayStation 3 gave me 5 exclusives. 4 of them from Naughty Dog with the 5th being Dead Nation from Housemarque (love Housemarque!!!) but otherwise, nothing.

This gen, while Guerrilla released Killzone Shadow Fall, they went on to release 10m+ seller in Horizon Zero Dawn. What an excellent action RPG. Not only does Guerrilla go from first person to third person, they go from shooter to action RPG and go from linear to open world and hit a home run with what is a new IP. Majority including myself was not expecting that. Insomniac (now owned by Sony) gave me Spider Man which was great overall and way better than anything they did on PS3. Santa Monica resurrected God of War. Naughty Dog disappointed me with Uncharted 4 but got me back with Lost Legacy. Bend Studio gave me a new IP in Days Gone. Sucker Punch is going to give me a new IP with Ghost of Tsushima.

Point being is that Sony did this gen for me what they didn't do last gen - give me exclusives that I actually want to play and this is what im hoping Microsoft can do with XBOX 4 compared to Xbox One. Give me exclusives that I actually want to play.

They're off to a better start with Halo Infinite and eventually Hellblade 2. No interest in either Forza series. Rare is basically an online multi-player focused studio which isn't for me. Undead Labs disappointed me greatly with SOD 2 despite the fact that I very much enjoyed the original SOD on 360 but they need to show me something major with SOD 3 to get me interested again. 343/Coalition are basically just Halo/Gears which is great and all but I do wish they would create a new IP in between their next installments.

InXile looks promising as Wasteland 3 will be the first game I play from them. Very interested in it and hoping they can get off to a great start for me with that game. Obsidian is great but im hoping their next gen game is NOT a Fallout type game and even more so, does NOT have a silent protagonist because that's two negatives for me right off the bat. As for Compulsion, we talked about them a few days ago. I'll just put that studio in the wait and see category until their next game. Playground's second studio is probably going to be Fable which im hoping will be more mature focused as opposed to teen focused even though I would prefer a new IP instead. The Initiative has a lot of great talent but with all that talent comes extremely high expectations. Even if their first game doesn't appeal or interest me at all, they need to hit a 90+ period because with all that talent, anything less would be disappointing and a letdown.

Not bashing Xbox or their first party studios. They're simply where Sony's first party studios were last gen for me. Developing and releasing games (great or not) that just don't appeal to me enough to where im interested in playing them. Nevertheless, barring a major screw up by Sony/Microsoft, im day one for both consoles and should have them pre-ordered once that becomes available after E3.

Damn, I wrote a long ass post. Sorry about that. This definitely isn't a "tweet". lmao.
 

sinnergy

Member
Reality is finally sinking in @ the Sony side... X is the monster / Lockhart will take on the 9TF PS5. I told you if PS5 doesn’t have the same size, it is not on par with series X.

Btw specs are set at least a year before launch, right about now. Only changes that can be made are clocks / ram, if it fits on the bus.
 
Last edited:
Microsoft's first party studios are mostly newly acquired and obviously have to prove themselves as a first party exclusive development studio for Microsoft. The big 3 are established but as much as I enjoyed Gears 5, The Coalition needs a new IP before Gears 6 in my opinion. Same for 343 after Halo Infinite.

I'm a huge fan of Ninja Theory but Hellblade 2 while I will play through it shouldn't have been their next major game. It should have been a new IP. Create a new IP, establish it into a new franchise and go from there. Playground remaking/rebooting Fable to me isn't the answer. A new fantasy action RPG IP would have been better if it is indeed Fable.

Super long post so let me address these first, starting with Hellblade 2. How do you know that a AAA Hellblade with Microsoft support won't be many times the game the original was, and in essence WILL be like a new IP for them that goes far further with its ambition. You telling me that world isn't one worth more deeply exploring no longer tethered by whatever limitations their 16-20 man team had with their $10 million budget? Hellblade AAA as a Microsoft First Party is essentially effectively like a brand new IP.

Halo Infinite is the right game for 343i at exactly the right time, no project could be more perfect for them. They are making what they were meant to and I believe the game will be incredible. Coalition should ABSOLUTELY immediately follow the amazing Gears 5. I guess they could try something new at some point, but this isn't that time after Gears 5. That game was fucking unbelievable.
 
Totally inaccurate. This current situation is unlike previous GCN generation updates. RDNA represents the most significant modification of AMD's core architecture since the introduction of GCN. The changes are dramatic. Even the compute unit setup is vastly different. Even the way the process operations has been doubled in some instances inside the inner workings of each compute unit. So, yes, in this case, an RDNA teraflop is almost certainly superior to a GCN teraflop. AMD has measured for this, and game benchmarks for their only Navi chips on the market bear this out.

Why does AMD's 9.7 teraflop Navi GPU spank their 12.6 teraflop VEGA 10 GPU? The same 295w TDP Vega GPU running a 2048-bit bus of HBM2 ram. So, no, an AMD teraflop isn't an AMD teraflop in this new post RDNA architecture world. Performance per clock is 1.25x better and performance per watt is 1.5x better. AMD doesn't make these claims lightly because they haven't made them so boldly before.
So does AMD market their GPUs using a flop count which factors in the 1.25 % improvement in performance per clock? No they don't.
Navi Radeon 5700
36 CUs, Max clock 1725, 7.95tflops. That's AMDs own numbers.
1725 x 2 x 2304 = 7.95tflops
Thats the exact same way they calculates tflops for Vega, Polaris, etc etc. There is only ONE way tflops is expressed, and there is no "GCN 12tflops" vs "RDNA 12tflops"
If they are stating 12tflops, it's RDNA, which is the same as GCN. Whether it's more efficient has nothing to do with how they calculates the tflop number.
So people here saying that the 12tflops is a GCN rating and therefore is only 9flops of RDNA are just delusional.
 
Reality is finally sinking in @ the Sony side... X is the monster / Lockhart will take on the 9TF PS5. I told you if PS5 doesn’t have the same size, it is not on par with series X.

Btw specs are set at least a year before launch, right about now. Only changes that can be made are clocks / ram, if it fits on the bus.

Not being the same size doesn't mean it's not on par. Xbox One soc was much bigger than PS4's last I checked, and yet PS4 was the stronger SoC.
 

Imtjnotu

Member
Reality is finally sinking in @ the Sony side... X is the monster / Lockhart will take on the 9TF PS5. I told you if PS5 doesn’t have the same size, it is not on par with series X.

Btw specs are set at least a year before launch, right about now. Only changes that can be made are clocks / ram, if it fits on the bus.
How does a 4tf machine take on a 9tf machine 😂
 
So does AMD market their GPUs using a flop count which factors in the 1.25 % improvement in performance per clock? No they don't.
Navi Radeon 5700
36 CUs, Max clock 1725, 7.95tflops. That's AMDs own numbers.
1725 x 2 x 2304 = 7.95tflops
Thats the exact same way they calculates tflops for Vega, Polaris, etc etc. There is only ONE way tflops is expressed, and there is no "GCN 12tflops" vs "RDNA 12tflops"
If they are stating 12tflops, it's RDNA, which is the same as GCN. Whether it's more efficient has nothing to do with how they calculates the tflop number.
So people here saying that the 12tflops is a GCN rating and therefore is only 9flops of RDNA are just delusional.

I knew you would say this, which is why I had this ready as a response. Yes, AMD DOES. In the official RDNA white paper in fact. Page 21 and 22.


Can I rest my case? AMD does advertise that the flops performance is superior in RDNA compared to GCN.






x4DPTrE.jpg



wyTUG0B.jpg


wyTUG0B.jpg


wyTUG0B.jpg


There's a lot going on under the hood. This ain't just frequency my friend. RDNA literally demolishes GCN at an architectural level. The FLOPS are better in RDNA, and AMD says exactly that in the white paper.
 
Last edited:
Could someone give me a brief summary of what's just happened in here? Seems that something occurred about 6 hours ago and everyone's been losing their shit ever since.

Best I've been able to pick out is that:

PS5 = 9TF
Xbox = 12TF
Lockhart = ???

Where did the info come from? Only a few days back we were being told (by industry insiders) that the PS5 was ~10% better than Xbox, with peeps then saying we're looking at a 13TF console. Now it's 9TF? That's quite a climb down.
 

ZywyPL

Banned
So, everything is pointing out that the consoles will indeed have a 5700/XT equivalent? And those 12TF are still yet to be confirmed.

So people here saying that the 12tflops is a GCN rating and therefore is only 9flops of RDNA are just delusional.

It's not delusional, it just means an RDNA-based GPU needs less TFlops to get the job done. Which has been already confirmed in hundreads of benchmarks out there.
 
Last edited:
For anyone still thinking an RDNA and GCN flop are the same, please look at page 21 and 22 of the RDNA whitepaper to see why that's just not the case. There's even more cache bandwidth available internally on a per flop basis and the raw performance numbers of a 40CU RDNA chip outdoes a mighty 64 CU GCN chip. Can you even imagine what a higher CU count RDNA chip, much less RDNA 2 can do compared to GCN? These two architectures are not the same. A TFLOPS in basic measurements are the same, but it is clear that an RDNA chip is plain better than a GCN chip not largely due to process or clock speed, but largely due to the fundamental architectural changes under the hood.
 
No. at the same exact clock speeds, RDNA is 1.25 faster than GCN. And at the exact same wattage, RDNA is 1.5x faster.

Navi-Slide-20.jpg



Navi-Slide-21.jpg


As you can see, frequency accounts for the least of the improvement in performance over GCN. The biggest improvement comes from raw performance per clock enhancements, meaning the architecture is clearly superior to GCN. Clock speed accounts for less than 20% of the improvement over GCN whereas the architectural changes account for nearly 60%.
I know that. That's why I said it's a combination of higher SIMD utilisation and clockspeed.

Higher SIMD utilisation is due to the shift to a 32-wide wavefront executed over 1 cycle using a 2*32-wide SIMD core, instead of a 64-wide wavefront executed over 4 cycles using a 4*16-wide SIMD core. There is also a better cache hierarchy.
However, Navi being clocked higher than Vega due to 7nm, is also a factor at play.

At the end of the day though, a floating point operation is a floating point operation. Doesn't matter how you slice it. If Vega 64 was crunching code that was designed optimally for its SIMD core design, then at the same frequency, it would outperform Navi.

In fact in some applications, it's almost strictly the clock (and bandwidth) limitations keeping Navi ahead. In some compute workloads in fact Vega 10 (both XL and XT) outperform Navi. Why? Because they have higher peak compute.
 

CyberPanda

Banned
For anyone still thinking an RDNA and GCN flop are the same, please look at page 21 and 22 of the RDNA whitepaper to see why that's just not the case. There's even more cache bandwidth available internally on a per flop basis and the raw performance numbers of a 40CU RDNA chip outdoes a mighty 64 CU GCN chip. Can you even imagine what a higher CU count RDNA chip, much less RDNA 2 can do compared to GCN? These two architectures are not the same. A TFLOPS in basic measurements are the same, but it is clear that an RDNA chip is plain better than a GCN chip not largely due to process or clock speed, but largely due to the fundamental architectural changes under the hood.
RDNA is similar to Turing right?
 
I knew you would say this, which is why I had this ready as a response. Yes, AMD DOES. In the official RDNA white paper in fact. Page 21 and 22.


Can I rest my case? AMD does advertise that the flops performance is superior in RDNA compared to GCN.






x4DPTrE.jpg



wyTUG0B.jpg


wyTUG0B.jpg


wyTUG0B.jpg


There's a lot going on under the hood. This ain't just frequency my friend. RDNA literally demolishes GCN at an architectural level. The FLOPS are better in RDNA, and AMD says exactly that in the white paper.
You are just trolling at this point.
 
So, everything is pointing out that the consoles will indeed have a 5700/XT equivalent? And those 12TF are still yet to be confirmed.



It's not delusional, it just means an RDNA-based GPU needs less TFlops to get the job done. Which has been already confirmed in hundreads of benchmarks out there.
I see I am wasting my time.
 

Lort

Banned
Xbox one s 720p - 900 p 30 fps ( no rt res textures)
Ps4 900p-1080p 30 fps ( no rt low res textures)
Ps4 pro 1440p or 4kcb 30 fps ( no rt low res texture)
Xbox one x 4k 30 fps ( no rt medium res textures)
Xbox lockhart 1080p 60 fps ( rt medium res textures)
Ps5 1440p 60 fps (rt high res textures)
xbox anaconda 4k 60 fps (rt high res textures)

For latest cross gen games .. rt ray tracing will prob run at 30 fps but game at 60 ( on some consoles).
 
Last edited:

Lone Wolf

Member
Could someone give me a brief summary of what's just happened in here? Seems that something occurred about 6 hours ago and everyone's been losing their shit ever since.

Best I've been able to pick out is that:

PS5 = 9TF
Xbox = 12TF
Lockhart = ???

Where did the info come from? Only a few days back we were being told (by industry insiders) that the PS5 was ~10% better than Xbox, with peeps then saying we're looking at a 13TF console. Now it's 9TF? That's quite a climb down.
Basically Xbox series X is 12TF, maybe RDNA TF or maybe GCN TF.
PS5 is either 9TF or 13 or maybe even 14. Also comes with some kind of ReRun based on a character from an old show called What’s Happening.
 
It should be noted that Microsoft never specified whether XSX had twice the compute performance of the XB1X, just that it was twice as powerful.

It's not unreasonable to assume that could be interpreted as twice the compute performance, but it can quite as easily mean twice the absolute gaming performance.

We will find out of course.

If Oberon is indeed the final PS5 configuration, I have to say I expected better from Sony. They went wide and slow with the PS4 and maximised bandwidth and it was a winning strategy to the point where Microsoft mimicked and extrapolated it for the 1X. Would be odd and disappointing if they went for some exotic solution.
 

pawel86ck

Banned
So does AMD market their GPUs using a flop count which factors in the 1.25 % improvement in performance per clock? No they don't.
Navi Radeon 5700
36 CUs, Max clock 1725, 7.95tflops. That's AMDs own numbers.
1725 x 2 x 2304 = 7.95tflops
Thats the exact same way they calculates tflops for Vega, Polaris, etc etc. There is only ONE way tflops is expressed, and there is no "GCN 12tflops" vs "RDNA 12tflops"
If they are stating 12tflops, it's RDNA, which is the same as GCN. Whether it's more efficient has nothing to do with how they calculates the tflop number.
So people here saying that the 12tflops is a GCN rating and therefore is only 9flops of RDNA are just delusional.
Agree. Xbox x architecture was also much more efficient (4x TFLOPS difference, yet 5.5x difference in real games) yet MS only mentioned TFLOPS. It's impossible to "do the math" when it comes to architecture gains, because each game and even test scenario within the same game will give you different results. For example Digital Foundy has compared RX 580 to 10TF Navi and RDNA1 not always was 2x faster, most of the time it was just 80% (and people should keep in mind xbox x GPU was even better than RX580, especially when it comes to memory bandwidth).

The only way you can try to measure gains related to architecture is to run dozens of different games and then calculate the average. However RDNA2 GPU's are not even out yet, so people cant even calculate anything.

If someone would want to take RDNA1 (first gen navi) and assume performance results on that, then based on average results I have seen you need at least 11 TF Navi to match 2x xbox x results 100% of time, and 8 FT Navi to match 8x xbox one consoles. Xbox one tahiti architecture was much worse than customized polaris in xbox, so that's why you would only need 8 Navi TFLOPS to match 8x xbox one results. 8 TF and 11TF are two different performance levels, and I doubt Phil would want to confuse people like that.

After Windows Central leak and xbox SX reveal (huge box) even Digital Foundry (Richart Leadbetter for sure) no longer believe xbox scarlett will be only 8-9TF (10TF barely could match 2x RX 580, not to mention 9TF). IMO some people just dont want to accept MS will deliver 12TF at the end of 2020 in their premium console. I bet when XSX will launch the same people will say 12TF is nothing in 2020, because PC GPU's are already at 18-20TF.
 
Last edited:

ZywyPL

Banned
I see I am wasting my time.

Exactly. To put it into even broader perspective - NV GPUs outperform AMD's with much lesser TF.

It's like having two cars with 200HP - same power, but one weights 2 tonnes and is as aerodynamic as a brick, while the other is 900kg, is super sleek, has semi-auto gearbox, slick tires etc. and while both indeed have the exact same power on paper, the latter will do circles around the other one once you put them on a track.

The TF numbers were a good and quite precise indicator when talking about more or less the same GCN architecture that powers all four current-gen consoles.

Besides, who really gives a shit if the consoles will do 4K60 on 8,9, 10 or 12TF?? Well, apparently lots of nerds here do lol. But for average, mentally sane people all that matters is that the consoles will indeed hit 4K60, something everyone way praying for, and something that all of a sudden seems to be completely irrelevant right now, it's all about TF wars now.
 
Last edited:

draliko

Member
500hp engine is still 500hp engine doesn't matter if diesel, elettric, petrol, wankel, it will behave differently (v8, v6, v12, etc) but still same metric used to misure power. Tflops are the same, don't mistake efficiency for Tflops, they are different things. Rdna is a better overall architecture so it performs better. Watch cpus, 3ghz Intel and 3ghz amd, different results but still both 3ghz nonetheless
 
Both companies have a budget for the console they have to stick with. Last gen MS spent alot of their budget on Kinect, meaning they had to gimp on the silicon. Their esram also cost a bit and took up alot of space on the die.
This time it looks like Xbox has gone all out on the silicon, while Sony may have put more money to other things such as their SSD.
There is also talk that Sonys controller is also going to be very expensive including and LCD screen on it as well.

I think that Sony isn't too worried about losing people to Xbox, as there is one thing in theirs, and MS favour, and that's back comparability. By giving customers access to their old PS4 games, and in MS case, Xbox One, 360 and OG Xbox, no one who owns a PS4 with maybe 30 games on it, is going to jump ship to MS and lose their old game library. It's one of the reasons MS pushed it so hard. The battle will be for the new customer who hasn't owned either one, and not an existing one.
 
It should be noted that Microsoft never specified whether XSX had twice the compute performance of the XB1X, just that it was twice as powerful.

It's not unreasonable to assume that could be interpreted as twice the compute performance, but it can quite as easily mean twice the absolute gaming performance.

We will find out of course.

If Oberon is indeed the final PS5 configuration, I have to say I expected better from Sony. They went wide and slow with the PS4 and maximised bandwidth and it was a winning strategy to the point where Microsoft mimicked and extrapolated it for the 1X. Would be odd and disappointing if they went for some exotic solution.

Microsoft has made clear that Xbox Series X is 4 to 5 times more powerful than Xbox One X. That's for overall raw performance jump.

For just GPUs however, purely on the numbers, it is over 8 times the xbox one GPU and twice the xbox one x GPU. He was referring to compute power, because even a 9TF navi GPU is easily more than 8 times more powerful than the xbox one GPU if we aren't talking purely teraflops.
 
Last edited:

Armorian

Banned
Xbox one s 720p - 900 p 30 fps ( no rt res textures)
Ps4 900p-1080p 30 fps ( no rt low res textures)
Ps4 pro 1440p or 4kcb 30 fps ( no rt low res texture)
Xbox one x 4k 30 fps ( no rt medium res textures)
Xbox lockhart 1080p 60 fps ( rt medium res textures)
Ps5 1440p 60 fps (rt high res textures)
xbox anaconda 4k 60 fps (rt high res textures)

For latest cross gen games .. rt ray tracing will prob run at 30 fps but game at 60 ( on some consoles).

Hahaha

1440p is 3 686 400 pixels, 4k is ‭8 294 400‬, XSX would have to be more than twice as powerful as PS5 to render same games is 4k compared to 1440p. Also, I don't think cross gen games will last more than 18 months into the next gen, games that will utilize Ryzen performance simply won't perform at acceptable framerate on Jaguar consoles.

For cross gen games:

X1 - 720p@30 fps
PS4 - 720/900p@30 fps
PS4P - 900/1080p@30 fps
X1X - 1080/1440p@30fps
XSS - 1080/1440p@60fps
XSX - 2160p@60fps
PS5 - ?? (probably 2160p@60)

MGS5 is good example on how cross gen games could look like, sub 720p@30 fps on PS360 and 1080p@60 fps on PS4.
 
Last edited:
Status
Not open for further replies.
Top Bottom