• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Sony Reveals PS5 Hardware: RDNA2 Raytracing, 16 GB GDDR6, 6 GB/s SSD, 2304 GPU Cores

:kookoo:

I have got nothing that is Sony, excuse me!
And I am not planning to get anything Sony soon!

You are in love with Cell and all the tech slides. Ownership is not a requirement to being a fan.

Sorry, "fanboy" may have been a bit harsh.
 
You are in love with Cell and all the tech slides. Ownership is not a requirement to being a fan.

Sorry, "fanboy" may have been a bit harsh.

Everyone praised the Cell in 2005-2006. All the tech media, newspapers wrote very positive articles about it.
I just respect the uber state-of-the-art technology that is Cell and its high potential.
 
Everyone praised the Cell in 2005-2006. All the tech media, newspapers wrote very positive articles about it.
I just respect the uber state-of-the-art technology that is Cell and its high potential.

In 2005-2006 it was a great way to eek the most out of a limited silicon process. Don't get me wrong, Cell was marvelous back then, if you could code for it.
 
I mean it's clear your a Sony fanboy, but the thing isn't even out yet to properly compare to NVMe. I wouldn't bet on it beating a proper MLC drive, personally.

If your suggesting that it's not going to be the fastest thing in the world, yes, you'd be right, but it will be pretty close and for a ~500$ console that's pretty damn good. Sony's solution goes beyond just the actual storage, there's a lot of custom hardware in their chip that isn't present in normal PCs.
 
I don't think the QoS tiers of the Sony SSD has a very major effect, but the dedicated decompression hardware and DMA processor on the other hand sound like major performance enhancements. The less these data-juggling processes bog down the CPU, the faster the system will be overall - especially if/when said hardware is faster than using a CPU core for the same workload.
 
You are in love with Cell and all the tech slides. Ownership is not a requirement to being a fan.

Sorry, "fanboy" may have been a bit harsh.

So? Cell is not a Sony technology....
 
I am sure SMT will be a standard for the PS5 with an option to perhaps turn off or keep on? Just like the Xbox Series X.
Another gaming console looking at a price of about $499 MAX just like the XBox Series X. Unless of course they plan on selling none with a higher price tag.

XBOX FTW, can hold 12TOPS continuously while PS5 boosts up to 10.3 which means you most of the time will get under well under 10TOPS from PS5

XBOX will have aprox 20-25% faster GFX

ALSO XBOX can hold 3.8GHz continuously while PS5 boosts up to 3.5 GHz which means you get maybe 3.2GHz in average from PS5

XBOX will have aprox 15-20% faster CPU

Keep in mind what HIDDEN POWER XBOX holds
if they decide to unlock boost, then you would see up to a total of 40% faster GFX and 40% faster CPU than PS5

PS5 saved money on silicon and cooling solution, sure maybe it will be a little cheaper but who cares about 50 or 100 bucks difference when you going to have that console for years to come.

This is EZZZZZ, i am going with XBOX
Xbox Series X looks like the better performer, and I agree. But with consoles, Developers try and utilize the entire system as a whole as efficiently as possible. Every single component will be efficiently utilized in both of these gaming consoles.

If your suggesting that it's not going to be the fastest thing in the world, yes, you'd be right, but it will be pretty close and for a ~500$ console that's pretty damn good. Sony's solution goes beyond just the actual storage, there's a lot of custom hardware in their chip that isn't present in normal PCs.
And that may be the differentiating factor between the PS5 & the XBox Series X, excluding game exclusives and such.
 
Im waiting for the Pro version of it. having this main IO chip on 12nm, and separate GPU CPU, means that whenever 5nm shrink of is available sony will integrate it very quickly. XSX can't do that, it is stuck on 7nm.

According to TMC, the 7nm node can be shrunk to 6nm without changing anything and gain 10-15% performance for the same power. Stop being a PS fanboy and go read about technology that you fail to understand...

 
Im waiting for the Pro version of it. having this main IO chip on 12nm, and separate GPU CPU, means that whenever 5nm shrink of is available sony will integrate it very quickly. XSX can't do that, it is stuck on 7nm.
Next Generation Gaming Consoles coming Christmas 2020. If Sony chooses to launch an Pro version of a unannounced 5nm, you are looking at a 2023-2024 release? 7nm is more than enough, buy either have fun.
 
Im waiting for the Pro version of it. having this main IO chip on 12nm, and separate GPU CPU, means that whenever 5nm shrink of is available sony will integrate it very quickly. XSX can't do that, it is stuck on 7nm.
There are no separate chips, both consoles are powered by monolithic APUs. Don't let block diagrams showing different components within a monolithic chip fool you into thinking they are separate pieces of silicon. Unless Sony plans on making a bunch of different versions (as in tens of versions or more) going chiplet based rather than monolithic is much more expensive. And, well, this is a console, so there will be a single version, with a possible Pro down the line, but if so it will be a new monolithic design. The reason chiplets are cheaper for AMD in the PC space is that they can use them across a wide range of designs - from desktop to HEDT to server - saving the cost of designing many different pieces of silicon. This is not the case for a console.
 
It is a Sony technology, as much as it is an IBM one and a Toshiba one.
Sony couldn't continue with CELL evolution without R&D resources from IBM.
 
Everyone praised the Cell in 2005-2006. All the tech media, newspapers wrote very positive articles about it.
I just respect the uber state-of-the-art technology that is Cell and its high potential.

That's all good but its a 15 year old chip by now :)
 
PS4 until the Xbox One X came out was significantly stronger than the original Xbox One. Original Xbox One was bandwidth starved with that DDR3 (NOT GDDR3) memory and a super weak GPU.

Overall PS4 sales since launch of both consoles is still higher.

Both Microsoft and Sony usually sell their systems at a loss initially. They make it up with accessories, game licensing, subscriptions to services.

Xbox users wont have full access to that 1TB though.

LOL do you know how TFLOPS are calculated? Do you also know it only takes into consideration one operation by a compute chip? It doesnt tell the whole story. Its also primarily marketing for consoles since average console player doesnt know what the hell it is. Its rarely used as an actual way of marketting a chip when it comes to PCs, etc.

Overall system performance between the 2 systems, Xbox is really only about 10-15% more powerful. There's no "unlocking" boost on Xbox. They are already pushing thermals at this point.

Also making parts of your posts in bold, doesnt make you sound any smarter.
RDNA CU includes ALU, TMU, TFU, SRAM and RT cores. XSX's GPU TFLOPS increase and additional PC CPU node were matched by memory bandwidth increase.

Sony loaded PS5's 448 GB/s bandwidth (the same as 5700/5700 XT memory bandwidth) with additional PC CPU node and slightly higher GPU TFLOPS.

Try again.

Yup, to further illustrate the point, Nvidia historically has worse TFLOPS but more FPS, GTX 1080 Ti for example has 11.34 TFLOPS and the (comparable to Vega 64) GTX 1080 has just 8.873 TFLOPS (both theoretical).
GTX 1080 Ti FE has 1,777 Mhz average clock speed, which yields about 12.74 TFLOPS.

In 2005-2006 it was a great way to eek the most out of a limited silicon process. Don't get me wrong, Cell was marvelous back then, if you could code for it.
3215211-1545338557-30514.jpg

GeForce 8800 is better.

CELL CPU = half-assed in-order Atom like CPU.
CELL SPU = DSP, half-assed wannabe GPU

CELL = master of none.

I'm game for another round of PS3 vs Core 2 + GeForce 8800 debates
 
GeForce 8800 is better.

Which was also pretty good around the turn of the milenia... you are missing the point. My point was old hardware is old.

I'm game for another round of PS3 vs Core 2 + GeForce 8800 debates

I'll raise you something modern and kick all your butts.
 
RDNA CU includes ALU, TMU, TFU, SRAM and RT cores. XSX's GPU TFLOPS increase and additional PC CPU node were matched by memory bandwidth increase.

Sony loaded PS5's 448 GB/s bandwidth (the same as 5700/5700 XT memory bandwidth) with additional PC CPU node and slightly higher GPU TFLOPS.

Try again.


GTX 1080 Ti FE has 1,777 Mhz average clock speed, which yields about 12.74 TFLOPS.


View attachment 149201
GeForce 8800 is better.

CELL CPU = half-assed in-order Atom like CPU.
CELL SPU = DSP, half-assed wannabe GPU

CELL = master of none.

I'm game for another round of PS3 vs Core 2 + GeForce 8800 debates

Try again?
 
I think a new definition of spreadsheet heroes is slowly taking shape here.
 
Try again?
It's for your following statement

LOL do you know how TFLOPS are calculated? Do you also know it only takes into consideration one operation by a compute chip? It doesnt tell the whole story. Its also primarily marketing for consoles since average console player doesnt know what the hell it is. Its rarely used as an actual way of marketting a chip when it comes to PCs, etc.

Your argument mirrors Mark Cerny's PS5 defense argument, but an increase in CU count also increases TMU, TFU, SRAM and RT cores.

XSX already scaling into RTX 2080 level results with two weeks raw Gears of War 5 port's built-in benchmark at PC's Ultra settings, hence TFLOPS is scaling.
 
It's for your following statement

LOL do you know how TFLOPS are calculated? Do you also know it only takes into consideration one operation by a compute chip? It doesnt tell the whole story. Its also primarily marketing for consoles since average console player doesnt know what the hell it is. Its rarely used as an actual way of marketting a chip when it comes to PCs, etc.

Your argument mirrors Mark Cerny's PS5 defense argument, but an increase in CU count also increases TMU, TFU, SRAM and RT cores.

XSX already scaling into RTX 2080 level results with two weeks raw Gears of War 5 port's built-in benchmark at PC's Ultra settings, hence TFLOPS is scaling.
Based on what has been revealed by both Microsoft & Sony, the Xbox Series X is the faster console versus the PS5. Will that make a difference in actual gaming performance and visual quality? To me, it seems Microsoft took the brute force method of achieving high end performance without image quality loss, where as Sony took the non brute force method, which may require some clever development techniques to somehow utilize the entire platform to gain high end performance without image quality loss.

Not saying that they won't cleverly design games to fully utilize the Xbox Series X too, but at the end of all this, the XSX is the stronger console, Microsoft went all out, and seems a lot more serious over Sony this time around. And I don't blame them for such a decision, they need to gain as much market share as possible, get back to the Xbox 360 sales figures or beyond.

The only thing that will determine the success of the XSX is price. If they get that wrong, the SP5 is going to clober it.
 
Based on what has been revealed by both Microsoft & Sony, the Xbox Series X is the faster console versus the PS5. Will that make a difference in actual gaming performance and visual quality? To me, it seems Microsoft took the brute force method of achieving high end performance without image quality loss, where as Sony took the non brute force method, which may require some clever development techniques to somehow utilize the entire platform to gain high end performance without image quality loss.

Not saying that they won't cleverly design games to fully utilize the Xbox Series X too, but at the end of all this, the XSX is the stronger console, Microsoft went all out, and seems a lot more serious over Sony this time around. And I don't blame them for such a decision, they need to gain as much market share as possible, get back to the Xbox 360 sales figures or beyond.

The only thing that will determine the success of the XSX is price. If they get that wrong, the SP5 is going to clober it.
I kind of disagree - IMO pushing clocks for a limited size die is more of a brute force approach than sizing the die sensibly for the workload at hand. Regardless of Sony's fancy talk of balancing power draw etc., that they have such a high peak clock tells us quite clearly that they at some point in development (beyond the point of no return for hardware designs) realized they were significantly behind in power and decided to boost clocks to compensate.

As for Cerny's argument that a smaller, higher clocked GPU is "more nimble" - that's nonsense, plain and simple. If that was indeed the case, overclocking PC GPUs would yield higher-than-linear results, and lower-tier OC cards would outperform stock-clocked higher tier cards. What actually happens is that gains from OC's are almost universally far lower than the clock increase would indicate, with the typical recent examples being 2-3% average performance increases from 8-10% clock increases and 20-30% power increases. Sony is trying to put a positive spin on putting their money on a weaker chip.

Now, the difference between the consoles in compute power isn't massive by any means, and the PS5 will no doubt have excellent looking games, but cross-platform games will look better and/or sustain frame rates better, just like on the XOX vs. PS4 Pro.

The effects of Sony's other hardware investments (SSD and 3D Audio) will be very interesting to see, but I sincerely doubt they'll do anything to alleviate the performance bottleneck - slightly shorter loading times is good, but unless MS has really botched their implementation the difference won't be huge, and while I'm really looking forward to games implementing a good, realistic 3D audio system, it won't be what makes or breaks a game. Of course a 15-20% performance advantage isn't likely to either, but it's more immediately noticeable. Then again Sony has such a mindshare advantage that they'll really need to botch this to not still come out ahead in terms of sales.
 
I kind of disagree - IMO pushing clocks for a limited size die is more of a brute force approach than sizing the die sensibly for the workload at hand. Regardless of Sony's fancy talk of balancing power draw etc., that they have such a high peak clock tells us quite clearly that they at some point in development (beyond the point of no return for hardware designs) realized they were significantly behind in power and decided to boost clocks to compensate.

As for Cerny's argument that a smaller, higher clocked GPU is "more nimble" - that's nonsense, plain and simple. If that was indeed the case, overclocking PC GPUs would yield higher-than-linear results, and lower-tier OC cards would outperform stock-clocked higher tier cards. What actually happens is that gains from OC's are almost universally far lower than the clock increase would indicate, with the typical recent examples being 2-3% average performance increases from 8-10% clock increases and 20-30% power increases. Sony is trying to put a positive spin on putting their money on a weaker chip.

Now, the difference between the consoles in compute power isn't massive by any means, and the PS5 will no doubt have excellent looking games, but cross-platform games will look better and/or sustain frame rates better, just like on the XOX vs. PS4 Pro.

The effects of Sony's other hardware investments (SSD and 3D Audio) will be very interesting to see, but I sincerely doubt they'll do anything to alleviate the performance bottleneck - slightly shorter loading times is good, but unless MS has really botched their implementation the difference won't be huge, and while I'm really looking forward to games implementing a good, realistic 3D audio system, it won't be what makes or breaks a game. Of course a 15-20% performance advantage isn't likely to either, but it's more immediately noticeable. Then again Sony has such a mindshare advantage that they'll really need to botch this to not still come out ahead in terms of sales.
Interesting. So you actually think Sony's announced clock speeds were not that high? Because they seen the XBox Series X specifications and went WOW, we have a problem on our hands people? We completely underestimated Microsoft's XSX specs. And since it's too late to re-design our special PS5 console, lets jack up the clocks higher and hope by the time its released, we gain an attractive performance/ energy consumption/ balance.

I AGREE. :D
 
Interesting. So you actually think Sony's announced clock speeds were not that high? Because they seen the XBox Series X specifications and went WOW, we have a problem on our hands people? We completely underestimated Microsoft's XSX specs. And since it's too late to re-design our special PS5 console, lets jack up the clocks higher and hope by the time its released, we gain an attractive performance/ energy consumption/ balance.

I AGREE. :D
Pretty much, yes. They obviously knew well before the announcement (or at least had a solid idea of ballpark performance), as they otherwise wouldn't have had time to actually test power/cooling/etc. for an announcement (disregarding the fact that it was recorded days if not weeks before, demonstrated by some media outlets being allowed to see it early). But yes, I do believe they initially ordered what they saw as a powerful chip at an acceptable price for selling the console (especially when accounting for the additional cost of a bespoke 12-channel SSD controller and several accompanying bespoke silicon designs going into the APU), got word that MS' chip was ~33% faster (12TF v. 9TF), and kind of went "Oh shi-".
 
Pretty much, yes. They obviously knew well before the announcement (or at least had a solid idea of ballpark performance), as they otherwise wouldn't have had time to actually test power/cooling/etc. for an announcement (disregarding the fact that it was recorded days if not weeks before, demonstrated by some media outlets being allowed to see it early). But yes, I do believe they initially ordered what they saw as a powerful chip at an acceptable price for selling the console (especially when accounting for the additional cost of a bespoke 12-channel SSD controller and several accompanying bespoke silicon designs going into the APU), got word that MS' chip was ~33% faster (12TF v. 9TF), and kind of went "Oh shi-".

If Sony's game library is better like it was with PS4, performance won't even matter. People with a gaming PC have almost zero reason to have an Xbox since all those games will be playable on PC. All of them.
 
Back
Top