Forums - Sony Discussion - What do yo think will be the hardware specifications of PS5 if it arrives arround 2019-2020?

Pemalite said:

Well. Sony doesn't have the software engineering resources that Microsoft has for starters.
Not saying they couldn't do it, but it would certainly be more difficult for Sony to do.

Even if Sony had the same resources like Microsoft did they still probably couldn't get the same results because there's far more architectural asymmetry between the PS3/PS4 than there is with the One/360 ... (3 ISA's vs 2 ISA's, local memory storage/DMA controller, non-coherent cache, it was just one insanity after another) 

People thought PS3 hardware was wild but wait till they see PS2 hardware ... (mipmapping on PS2 resembles nothing like contemporary or modern hardware)

The tons of non-trivialities baked in PS3 hardware compared to the 360 made binary patching not worth it but the final nail in the coffin was performance and patent issues abound ... (although I'd be interested to see how PS4 BC with 360 would've turned out) 

Pemalite said:

Unfortunately porting isn't possible in every scenario as source code can be lost.
It's actually been a common issue with remasters, Homeworld Cataclysm being a prime example.

Sure the source code can be lost but as long as you have the copyright license to the original works then you can just simply dump the contents and reverse engineer the compiled code from the retail versions of the games like Square Enix did when they were porting Kingdom Hearts to newer Playstation systems according to digital foundry but Microsoft's solution isn't foolproof to this issue either unlike pure emulation since they require the source code too ... (Microsoft first tries to recompile the code for x86 then goes on and collects the GPU shaders for the games then translates those too sort of like GLSL to SPIR-V)

Still in the case with Sony, it's a better idea to go through that extra work than to achieve BC the same way like Microsoft did by patching binaries with very little hope of AAA games running well on their system ... 

It's also more worthwhile than ever to remaster/port games since upgraded mid gen console SKUs have enough CPU headroom to run every last gen AAA game at 60FPS ...  

People don't really care about how older content is delivered, all they care about is whether or not they can access it ... 



Around the Network
Mar1217 said:
People are basically asking for a high-end PC for cheap.

Don't be surprised if the think cost 450$-500$ on release ...

If you have followed this thread, you will see that thats not what is being asked for at all.



if they use x86 again

a much powerful cpu. we will get much higher framerates.

16 GB of ram (they will not use more than that becaue RAM is expensive and there isn't a need for it) maybe 2 GB extra for a co processor

the GPU will be at most. 3x more powerful. my guess. around 2.5X more powerful. than the PS4 PRO.


if they do not use x86 (which is very much next to impossible and doesn't make sense)

they will go with something like a powerpc CPU. that is a much faster CPU

same amount of ram as said above..

same GPU as above.

 

 

for either they will not need that much more bandwidth for RAM than we have now. if they use the newest HBM at 16 gb they will have something like 1 TB/s which is overkill. they might downclock the HBM to save power as they prob won't use all of that bandwidth. tho that amount of HBM is expensive as hell. last I heard HBM2 costs 150 dollars per 8 GB. but its probably not as high as it will be. and that number is likely inflated.

 

 

what should they be aiming for? well. they really should be going for 4k 60 fps. thats the only reasonable jump. if its not that kind of jump it will be hard to make an argument against "why can't this run on the PS4 pro at 1080P?"

Last edited by MasterThief - on 11 March 2018

Intrinsic said:
CrazyGPU said:

Radeon R9 Fury X (275W 28 nm) lauched jun 2015 and it was a 8.6 teraflop graphic card. 512GB/s bandwith.

Radeon RX vega air(295W 14 nm FinFet) launched Aug 2017 and it was a 12.7 teraflop  graphic card. 484 GB/s bandwith, but utilize beter compression tech. 

So, two years and two months,

and AMD could improve flops by 47%, 26 months later with the new architecture.

Now, PS4 pro lauched nov 2016 with 4,2 Teraflops. Let say PS5 launches nov 2020. 48 months later.

If AMD is able to keep its pace in performance improvements, and that gets harder and harder, 

and launchs a 7 nm+ gpu, we can expect:

48 months x 47%(amd improvement) /26 months = 86.7% more performance.

4.2 Tf PS4 pro x 1.867 = 7,84 teraflops. (a little more than 4 x standar PS4)

But, let say Sony try harder, with better cooling, like with Msoft One X. 

One X lauched Nov 2017. It´s a 6 teraflops machine.

Again  let say PS5 launches 2020. 3 years later.

36 months x 47%(amd improvement) / 26 months = 65%

6 Tf XBoX X x 1.65 = 9,9 teraflops. 

So I really don´t understand on what basis one would expect 15 Teraflops on an APU!.

I said I´m expecting 10-12 Teraflops for 2020 and I´m being optimistic.

of course that performance can go higher if they go discrete. 

 

 

Edit: I was checking Radeon RX Vega 56 

Specs 10.5 Teraflops. 410 GB/s Bandwith. 210 W. Lauched August 2017. Mostly a 4k 30fps graphic card.

Now, XBox one X peaks 172 W in Gears of War 4.

https://www.anandtech.com/show/11992/the-xbox-one-x-review/6

Take away CPU side consumption, fans, disks, and what´s for the GPU side of the apu? 100w?

Now if they can reduce Vega consumption by half in 3 years, you can have a 10 Teraflop GPU on PS5. Or Sony can go the Nvidia route, which is much more complicated.

 

 

 

Thats because you are basing your calculations off the wrong constants. Look at it this way instead.

Forget whatever one GPU or the other has achieved and how, instead look at the individual compute units and their clocks.

So start with the PS4 OG.
PS4OG = 20CU (2 deactivated)@800mhz = 1.8TF (28nm)
PS4pro = 40CU (4 deactivated)@911mhz = 4.2TF (16nm)

What does that tell you? Exactly doubling the GPU Compute units should theoretically give you 3.6TF. Assuming the clocks stays identical. However, because the GPU has been upclocked from 800mhz to 911mhz that TF value went up from 3.6TF to 4.2TF. And this is before any architecture inprovements have been taken into account.

Now lets use the XB1X as the base.

XB1X = 44CU (4 deactivated)@1172mhz = 6TF (16nm)
PS5/XB2 = 88CU (8 deactivated) @1172 = 12TF (7nm)

Now I am even lowballing this, because

  1. one constant going from higher to lower fabrication processes is being able to clock higher because of better thermal efficiency. So at the very least the next gen GPUs should be clocked higher than the 1172mhz seen in the XB1X. 
  2. I am assuming that as much as 8 CUs will be deactivated to improved yields.
  3. We are not taking any architectural design improvements into account.
I do not know why you are using the standards you are using for your theory, when you have very accurate data from the already existing PS4pro and XB1X to see exactly what you can expect from a direct fabrication shrink and upclock when they went from 28nm to 16nm. 
 
As I have said before and keep saying, the next generation of consoles will not come until 7nm chip fabrication is available and maybe even mature. They will wait. But what I have just pointed out is exactly what to expect from a APU that is 360mm sq as we have in the XB1X built using the 7nm process. 
 
Lastly, by how much do you think the system can be up-clocked? Can we go up to 1300mhz? 1500mhz? Whatever it is; just know that for every 117mhz clock increase you add an addition 1.2TF of performance. So if its clocked at `1300+mhz you are looking at a 14TF+ GPU. See why i expect it may even be as high as 15TF?
 
Now if we start talking discrete CPU/GPU? then its a totally different ball game. With that they could have a 300mm sq chip as the GPU alone, keep the clock locked to 1172mhz, but increase CUs from 88 to as much as 140!!! And thats how you get to ~19TF. 

 

CrazyGPU said:
Intrinsic said:

Thats because you are basing your calculations off the wrong constants. Look at it this way instead.

Forget whatever one GPU or the other has achieved and how, instead look at the individual compute units and their clocks.

So start with the PS4 OG.
PS4OG = 20CU (2 deactivated)@800mhz = 1.8TF (28nm)
PS4pro = 40CU (4 deactivated)@911mhz = 4.2TF (16nm)

What does that tell you? Exactly doubling the GPU Compute units should theoretically give you 3.6TF. Assuming the clocks stays identical. However, because the GPU has been upclocked from 800mhz to 911mhz that TF value went up from 3.6TF to 4.2TF. And this is before any architecture inprovements have been taken into account.

Now lets use the XB1X as the base.

XB1X = 44CU (4 deactivated)@1172mhz = 6TF (16nm)
PS5/XB2 = 88CU (8 deactivated) @1172 = 12TF (7nm)

Now I am even lowballing this, because

  1. one constant going from higher to lower fabrication processes is being able to clock higher because of better thermal efficiency. So at the very least the next gen GPUs should be clocked higher than the 1172mhz seen in the XB1X. 
  2. I am assuming that as much as 8 CUs will be deactivated to improved yields.
  3. We are not taking any architectural design improvements into account.
I do not know why you are using the standards you are using for your theory, when you have very accurate data from the already existing PS4pro and XB1X to see exactly what you can expect from a direct fabrication shrink and upclock when they went from 28nm to 16nm. 
 
As I have said before and keep saying, the next generation of consoles will not come until 7nm chip fabrication is available and maybe even mature. They will wait. But what I have just pointed out is exactly what to expect from a APU that is 360mm sq as we have in the XB1X built using the 7nm process. 
 
Lastly, by how much do you think the system can be up-clocked? Can we go up to 1300mhz? 1500mhz? Whatever it is; just know that for every 117mhz clock increase you add an addition 1.2TF of performance. So if its clocked at `1300+mhz you are looking at a 14TF+ GPU. See why i expect it may even be as high as 15TF?
 
Now if we start talking discrete CPU/GPU? then its a totally different ball game. With that they could have a 300mm sq chip as the GPU alone, keep the clock locked to 1172mhz, but increase CUs from 88 to as much as 140!!! And thats how you get to ~19TF. 

 

First of all , great post. I see your point, it´s another way to see it and a good one. 

I will add that also, in doing what you are explaining, they can perfectly run 1/4 th of the cores and mantain compatibility with PS4 if the hardware is close enough. As they did with PS4 pro. I think this time it will be a brand new architecture.

The only thing that we need to see is:

if they go for premium cooler and stuff like Xbox one X, If they do, and if 7 nm is ready, they can perfectly reach 12 Teraflops as you explained up there.

If they go for a PS4 pro kind of replacement, it could be 4,2Tf x 2(double the cores) = 8,4 teraflops. Let´s give a little more clock , (like 911Mhz pro / 800 MHz base = 13,9%)

then you get 8,4 x 1.139% = 9,57 Teraflops. With a new architecture, probably more.

Let´s give the same treatment for 12 Teraflops Xbox one X like machine. x 1.139 = 13,67 Teraflops.

And the magic marketing numbers are 10, for being the first 10 Tflop machine or more than 12, that´s 2 times Xbox One X.

So my first estimation in 2014 was 12-14 Tf. My actual one is 10-12 Tf, but with your calculation my old estimation can be achievable too. I still think that 15  high, it also depend on the bandwith to feed it. A wide bus translates in more power consumption. But I see your reasons now and I think is perfectly doable.

I hope they get there. A 14 Tf PS5 would be 7,6 times more powerfull than original PS4. A nice jump for 4k and VR.

Intrinsic said:
CrazyGPU said: 

if they go for premium cooler and stuff like Xbox one X, If they do, and if 7 nm is ready, they can perfectly reach 12 Teraflops as you explained up there.

then you get 8,4 x 1.139% = 9,57 Teraflops. With a new architecture, probably more.

Let´s give the same treatment for 12 Teraflops Xbox one X like machine. x 1.139 = 13,67 Teraflops.

And the magic marketing numbers are 10, for being the first 10 Tflop machine or more than 12, that´s 2 times Xbox One X.

So my first estimation in 2014 was 12-14 Tf. My actual one is 10-12 Tf, but with your calculation my old estimation can be achievable too. I still think that 15  high, it also depend on the bandwith to feed it. A wide bus translates in more power consumption. But I see your reasons now and I think is perfectly doable.

I hope they get there. A 14 Tf PS5 would be 7,6 times more powerfull than original PS4. A nice jump for 4k and VR.

Thanks. Now about them going for the PS4pro double..... I think that is strategically impossible. Whatever the case here, we are looking at at least double the PS4pro or XB1X. Now it would just be a flat out bad business decision to double the PS4pro when we all know that MS will at the very least double the XB1X. And if we use your calculations, we are talking about at leat 9.5TF vs 13.6TF. Thats almost 3TF worth of muscle being left on the table. That would be marketing and design suicide for sony. 

I do agree with you though that 15TF is a little high, I mean there are a number of factors that could prevent them from pushing their designs that far... reliability and longevity being two of the most important amongst them. So its possible that even if the hardware can hit 15TF, the could settle for something like 12-13TF.

But this comes down to what MS is gonna do. Neither of them is going to let the other have any meaningful performance advantage out of the gate this time around. MS knows what it cost them this gen, sony knows how it favored them. 

So in 2014 I expected that the GPU was going to be 12-14 teraflops. We didn´t know about NAVI architecture. 

Then in february 2018 I thought we were going to get arround 10-12, but Intrinsic was thinking arround 12-15 based on the number of cores achivable with 7 nm. 

Anyway, we now know that ps5 is going to run NAVI and good rumors point to NAVI 10, arround a RX 5700, which has arround 8 Teraflops but is 4% faster than the GCN based VEGA 64 which was 12,6 Teraflops. So we are witnessing at a machine comparable with GCN at 13 Teraflops. We were right on the performance target, but not with the core number and architecture. 

On the CPU side, things look bright, it´s almost certain that we´re going to get ZEN 2+ 8 cores 16 Thread with 3.2 Ghz. That´s way better than my prediction on 2014 which was 2x PS4, but hey, Ryzen didn´t exist back then. 

On the bandwith Intrinsic was right in 2018 thinking in arround 500 GBps if the NAVI 10 rumors are correct. 

On The HD side, things look better than I thought, cause the SSD is confirmed and its going to be blazing fast. 

Things are getting very interesting, we´ll get a PS5 with no less than 4x the CPU performance, 6 times the GPU performance and SSD. 

We still have to wait for final confirmations on: 

memory bandwith, 

GPU and memory frequency,

main memory size, 

how RAY TRACING is implemented in hardware.

most people go for 16 GB of RAM, I would like to get 24 if the machine is spected to go until 2027.

I don´t expect the highest jump of any generation, but it seems that it will be a very good machine for 4k gaming, some games will run 30 fps others at 60, but for 500 USS it will be cheaper than any high end PC. 

I expect Microsoft Scarlet to be very close this time, faster or slower. But I´ll go for Sony for 2 reasons. 1st I have a high end PC with a 1070 and a 2k monitor. and 2nd I love Sony exclusives. 



What we're gonna see is around 2x more GPU performance compared to Xbox One X on PS5 and 6x more CPU performance. Compared to base PS4 it's like 5x GPU and 8x CPU which we can consider a proper generational leap. Especially when having NVME SSD and faster RAM bandwith added to the mix



29/08/14:

PS4 will sell 25 million units before 31st of March!

Edit: I didn't state what year! HAHA, so I'm still right in some sense

Around the Network
elazz said:
What we're gonna see is around 2x more GPU performance compared to Xbox One X on PS5 and 6x more CPU performance. Compared to base PS4 it's like 5x GPU and 8x CPU which we can consider a proper generational leap. Especially when having NVME SSD and faster RAM bandwith added to the mix

It´s more than that. Base PS4 was 1.84 Teraflops and GCN architecture. And Navi RX 5700 which is what i would spect from next gen, is something comparable to a 13 Teraflops VEGA 64 GPU, so, even considering that the performance would be comparable to a cpu which performance is arround 12 Teraflops on a GCN VEGA 64, that´s 6.5x the GPU. On the CPU side I would think about at least 4x the performance.



CrazyGPU said:

So in 2014 I expected that the GPU was going to be 12-14 teraflops. We didn´t know about NAVI architecture. 

Then in february 2018 I thought we were going to get arround 10-12, but Intrinsic was thinking arround 12-15 based on the number of cores achivable with 7 nm. 

Anyway, we now know that ps5 is going to run NAVI and good rumors point to NAVI 10, arround a RX 5700, which has arround 8 Teraflops but is 4% faster than the GCN based VEGA 64 which was 12,6 Teraflops. So we are witnessing at a machine comparable with GCN at 13 Teraflops. We were right on the performance target, but not with the core number and architecture. 

On the CPU side, things look bright, it´s almost certain that we´re going to get ZEN 2+ 8 cores 16 Thread with 3.2 Ghz. That´s way better than my prediction on 2014 which was 2x PS4, but hey, Ryzen didn´t exist back then. 

On the bandwith Intrinsic was right in 2018 thinking in arround 500 GBps if the NAVI 10 rumors are correct. 

On The HD side, things look better than I thought, cause the SSD is confirmed and its going to be blazing fast. 

Things are getting very interesting, we´ll get a PS5 with no less than 4x the CPU performance, 6 times the GPU performance and SSD. 

We still have to wait for final confirmations on: 

memory bandwith, 

GPU and memory frequency,

main memory size, 

how RAY TRACING is implemented in hardware.

most people go for 16 GB of RAM, I would like to get 24 if the machine is spected to go until 2027.

I don´t expect the highest jump of any generation, but it seems that it will be a very good machine for 4k gaming, some games will run 30 fps others at 60, but for 500 USS it will be cheaper than any high end PC. 

I expect Microsoft Scarlet to be very close this time, faster or slower. But I´ll go for Sony for 2 reasons. 1st I have a high end PC with a 1070 and a 2k monitor. and 2nd I love Sony exclusives. 

Lol..great necrobump. Fun times indeed.

Yes, we were spot on with the GPU, and what I was saying pretty much happened; but those "architectural improvements" I mentioned was something we couldn't possibly account for or know. So we end up with a GPU that has much lower CUs, higher IPC and significantly higher clocks. But it's equivalent to the 12TF GCN GPU range that we were tossing around back then.

In some other post I actually also predicted that nextgen consoles will be using SSDs. I think some people said I was crazy and not being realistic because they were going to be prohibitively expensive.

As for RAM... I really don't see 24GB happening.. at least not in the way you suggest. I expect next-gen consoles to have anywhere between 20-24GB of total RAM. But not all GDDR6. Basically an X (where x is the amount of GDDR6 RAM) + 4GB of LPDDR4 RAM. So either 16GB GDDR6 + 4GB LPDDR4 or 20GB GDDR6 + 4GB LPDDR4. My money is on the former because consoles really do not need more than 16GB of GDDR6 for games. In truth its actually more like 15GB GDDR6 that will be available because like 500MB - 1GB will be used for th OS while the bulk of the OS sits in the 4GB LPDDR4 RAM pool.

15GB GDDR6 available for games is like 3x the amount of game available RAM this gen. And that's plenty, especially when you consider these consoles will be coming with SSDs capable of moving upwards of 3GB-5GB/s of data. That allows for better RAM management.

Lastly, I think you will be pleasantly surprised how big a generational jump this will be. It's like people forget that games like GOW, Detroit, TLOU2, RDR2...etc were made on consoles with a jaguar CPU (that next-gen CPUs are almost 8 times better than) and a 1.8TF GPU. Mind you, clock or clock that is 1.8TF GCN GPU is equivalent to a 1.3TF RDNA GPU. And next-gen we are getting like 10TF of the stuff. Can you start to imagine how GOW would have looked like if it was made on hardware that is like 7-8 times better?



elazz said:
What we're gonna see is around 2x more GPU performance compared to Xbox One X on PS5 and 6x more CPU performance. Compared to base PS4 it's like 5x GPU and 8x CPU which we can consider a proper generational leap. Especially when having NVME SSD and faster RAM bandwith added to the mix

Its a lot more than that bruh

I could show you info and details but I am too lazy.



I won't settle for anything less than 120 fps (locked) at 8k native resolution with super HDR+.



Pemalite said:

They will go with whatever is cheapest. - Even the Xbox One X isn't using the biggest, fastest mechanical disks and that was a "premium" console.

SSD's/m.2 drives still command a premium and will do so even in a few years from now.
Plus the demand for NAND is increasing, so costs could actually increase, especially if NAND fabs retool to start taking advantage of higher DRAM prices.

Turns out price plummeted thanks to the abundance of stacked TLC and QLC NAND rushing onto the market, making it a viable option.
Supply/Demand in commodity markets can be a fickle thing at times!
Happy that we aren't going to be using mechanical disks, but my trepidation on the use of NAND was for naught.

Could still be interesting how Microsoft and Sony handle expanded storage though.

Pemalite said:

DDR5 isn't going to happen. JEDEC has only just finished demonstrating the DRAM and hasn't even finished ratifying the standard... And is thus probably still a few years away from a market release.
It does promise to double the bandwidth and density of DDR4, which means it's not going to be enough for next gen anyway unless you implement a stupidly wide bus.

It will be GDDR6, it will be what will offer the greatest benefits, higher capacities and good enough bandwidth/latency.
GDDR6 is currently ramping up production right now, so costs will be leveled out by the time 2020 hits.
 

Turns out I was right.

Pemalite said:

It will use AMD's next gen GPU architecture and not Vega which is a GPU architecture that released in 2017.

And I highly doubt they will even choose Navi, but it really depends how long AMD flogs that horse.

Well. It certainly didn't use Vega like many claimed. - But it did turn out it is using Navi, albeit a deviation of Navi as it's a custom design implementing aspects from AMD's next-gen GPU uArch like hardware Ray Tracing.



--::{PC Gaming Master Race}::--