12.15 - 10.28= 1.87 Teraflops difference between the XSX and PS5 (52 CU's vs. 36 CU's)

Believe what you want, and I know what I said, but nice try attempting to turn it into a food fight for console wars. Time to move on won't waste time on the forum abiout that stuff anymore. We know what the specs are now. Next is to see what the games are like and how the SSDs change things and what they can do with the I/O improvements they've made.

Can we please stop with this SSD crap? It means shit and won't make a meaningful difference.
well, you will see as soon as the games come. Lol
 
Can we please stop with this SSD crap? It means shit and won't make a meaningful difference.
well, you will see as soon as the games come. Lol

Welp you heard it guys! Fillippos says to stop talking about the SSD even thought multiple developers have come out including Digital Foundry and said that it's a game changer. You can all go home now.
 
I guess you don't even know what the DICE Awards are or that the award for Best Technical Design goes to the game with best graphics and that they have a separate award for Art Design, but it's ok champ, your ignorance is pretty apparent at this point. 👌🏾



Welp! Ya got me! John and Richard from DF are also liars and complete shills for taking Sony's money! @dark10x
I do , still is about artistry and programming not power.
 
Everyone Everywhere : "The world is fucked"

Sony : "Let's drop the PS5 bomb unleash the Cerny"
Unleash the Cerny? I understand people love to hype this guy but really.... Slow, unnatural way of talking. Explaining how ssd and 3d audio works to a gdc tech audience... ssd's are around for almost 10 years. NVME half as much.
He is just creepy even for an engineer. Carmack gives way more details in faster time
 
The XSX didn't achieve 12 Teraflops efficiently since they needed 52 CUs to achieve a mere 1.87 Teraflops difference.

Again, this thread is to start a conversation about whether the XSX's 52 CU's was efficient or not to achieve that additional 1.87 Teraflops gain or not.
You have got the answer at that point though: Using 52 CUs is not cost efficient, but it is energy efficient and better for the longevity of the hardware. Overclocking the CUs (which are, to my knowledge the exact same otherwise) strains energy usage and the material disproportionally, so I wouldn't be surprised, if PS5, in boost mode, uses more energy to achieve less performance at a higher temperature and noise.

If positive views on the PS5 architecture are very important to you, you should not focus in GPU and CPU, focus on the good bus system and the advanced SSD built into PS5 which is much faster than the Xbox equivalent (factor 2).
 
Welp you heard it guys! Fillippos says to stop talking about the SSD even thought multiple developers have come out including Digital Foundry and said that it's a game changer. You can all go home now.

Real life game changer of 4 seconds load time to 2 second load time. Absolutely game changer who cares about every other metric it gets spanked in.

If you want to see real world difference in the sex and PS5 nvme speeds look up some comparison videos on youtube of the Samsung nvme speed tests. You're talking unnoticeable seconds.
 
No my man , the new Kinect.

Let's throw my 1 year old 1500 euro Atmos soundbar out of the window ! Sony is here to save my ears.
lol good for you but you do realise to get these features as good as Sony are saying you will need full audio setup with at least 7.1 sound system to get the most out of it, remember he said Atmos was not enough lol
 
yeah. Imagine MS would boost them like Sony did with PS5. How many TF would that be? Freaking crazy.
But this also will mean that the series X will be super quiet. And PS5 a jet engine.
XSEX at 2.23GHz = 14.84TF
(52*128*2.23/1000)

XSEX currently is at 12.14GHz (52*128*1.825/1000)

TFLOPS = (TMUs x ROPS x Cc) ÷ 1000

Used an online calculator.
 
Real life game changer of 4 seconds load time to 2 second load time. Absolutely game changer who cares about every other metric it gets spanked in.

If you want to see real world difference in the sex and PS5 nvme speeds look up some comparison videos on youtube of the Samsung nvme speed tests. You're talking unnoticeable seconds.
Can't wait till launch when the ps5 will be the only one spanking. It's going to be a great console with better sales.
 
I cannot believe that Cerny and his team were able to achieve 10.28 Teraflops with just 36 CUs!

XSX needed 52 CU's (16 additional CU's) to achieve an additional 1.87 Teraflops, which is not much of a difference when you really think about it. I guess one of the senior engineers at Microsoft really loves Blink-187 and wanted to pay homage to them in a way.

The power consumption of the XSX will be massive because of its heavy reliance on 52 CU's while the PS5 were cost and power consumption effective with just 36 CU's.

This thread is to start a conversation about whether the XSX's 52 CU's was efficient or not to achieve that additional 1.87 Teraflops gain or not.

It's 10 at peak, 9 if the Cpu is doing anything, don't forget 👍🏻
 
The variability of the performance is definitely confusing to me. Let's say you've got a game like DOOM, where the developers actually give a shit about the game maintaining 60fps. Won't they kinda have to target the lowest spec it de-clocks down to? They don't want it dropping frames just because the user chose to leave their air conditioner off...
 
Its not.

Teraflops isn't the main indicator of factor that makes a console superior overall. Sure it may help hitting 4K60FPS but what else can it do to make a game genuinely better in terms if gameplay design, audio, loading times, etc.

If you're not trolling you must be genuinely having a hard time coping right now.
 
XSEX at 2.23GHz = 14.84TF
(52*128*2.23/1000)

XSEX currently is at 12.14GHz (52*128*1.825/1000)

TFLOPS = (TMUs x ROPS x Cc) ÷ 1000

Used an online calculator.
128 ROPS? You sure , that's a lot .

Looks like PS5 also has 128 ROPS.
36x128x2.23/1000 = 10.3.

So pixel filtrate the same for both?
 
Last edited:
Unleash the Cerny? I understand people love to hype this guy but really.... Slow, unnatural way of talking. Explaining how ssd and 3d audio works to a gdc tech audience... ssd's are around for almost 10 years. NVME half as much.
He is just creepy even for an engineer. Carmack gives way more details in faster time

You have massively misinterpreted my post. The order of magnitude is startling.
 
The PS4 advantage over Xbox One was alot more impressive tbh, and that didn't equate to much besides 900p vs 1080p

99% will look indistinguishable without DF lenses and specialized equipment.
 
Last edited:
Sony/Cerny has always emphasized balance in a console, so maybe they realized the ZEN 2 cores could in many instances be with idle threads so they decided on dynamically providing the extra voltage to the GPU thus keeping things in balance and avoiding any CPU-GPU related bottlenecks.
 
I'd recommend the OP look at the original PS4 and Xbox One GPU specs. Then come back with the argument they are pushing.
 
128 ROPS? You sure , that's a lot .

Looks like PS5 also has 128 ROPS.
36x128x2.23/1000 = 10.3.

So pixel filtrate the same for both?
Honestly I haven't a clue, I found a online calculator that asked for three values and only one that got to 12.14 was 128 ROPs. This stuff is over my head for the most part :messenger_tears_of_joy:
(the other 2 values we know so must be 128 ROPs)

 
Last edited:
PS5 have a crazy overclock.
2.23 is crazy even for pc standard.
With the same tech, XSX would reach a 14.8TF, insane.

I wonder if Microsoft will respond bumping their GPU clock before the launch. Reaching 2ghz should be feasible, this means +13TF.
 
onehitflunder said:
Are you going to buy a weak PS with 9Tflops, only 3 more than XBone One X in the year 2021?
Oh.my god....
People is usually so stupid at neogaf ?

Tf RDNA2 =/= Tf GCN
 
Last edited:
So the xsx is a whole ps4 more powerful than the ps5!

Wow!

The ps4 has some unbelievably good looking games that run at 1080p 60fps so add that on and that has to be a noticeable difference
 
Could be 104*64*1.825/1000 = 12.14 that more realistic for ROPs?
You have 52 CUs, so no. 52 CUs is a given.

But pixel fill rate is connected to clock I think.

So Sony has a slight edge in this area it seems , like PRO vs X, but in use we hardly see this back in games.
 
Last edited:
The 1.87 tflop difference is the minimum that you gonna get. Sony provides a system that has variable performance (9-10tflop) while MS is a rock solid 12 tflop monster. But it's not only that. Higher memory bandwidths, faster cpu, instant game switching and most importantly the xbox live ecosystem (game pass, game streaming, etc) will most likely give MS a big boost next gen. People have to admit that microsoft has done huge steps forward comparing to 2013.
 
Woke up from the nightmare that was the PS5 reveal.
My life is over basicly. MS won the nextgen and I cant see this being the case for the coming 6/7 years. Nightmare fuel.

I might as well get Corona.
 
Sony/Cerny has always emphasized balance in a console, so maybe they realized the ZEN 2 cores could in many instances be with idle threads so they decided on dynamically providing the extra voltage to the GPU thus keeping things in balance and avoiding any CPU-GPU related bottlenecks.

Especially how they balanced PS4 Pro to sound like a jet engine on many demanding games.

The whole "10.2" TF isn't fooling even the most supportive Sony fans. It's a marketing ploy to push what is really a 9+ TF console into 10+ territory. A 3 TF difference is a good chunk of performance (30%).
In the end Sony's brand is more popular right now and I'm not sure your average consumer gives a shit about TFs. Sony may lose a few core gamers but likely they will be the cheaper console with the better exclusives so unless they screw things up with having the PS5 overheat or break often (eg. RROD), they will still come out ahead.
 
Especially how they balanced PS4 Pro to sound like a jet engine on many demanding games.

The whole "10.2" TF isn't fooling even the most supportive Sony fans. It's a marketing ploy to push what is really a 9+ TF console into 10+ territory. A 3 TF difference is a good chunk of performance (30%).
In the end Sony's brand is more popular right now and I'm not sure your average consumer gives a shit about TFs. Sony may lose a few core gamers but likely they will be the cheaper console with the better exclusives so unless they screw things up with having the PS5 overheat or break often (eg. RROD), they will still come out ahead.

In this case and in real life scenario for a AAA title the GPU will never go over 2GHz clocks and never have overheating issues.
 
Top Bottom