Assuming my cheap calculations. If the PS5 GPU would had 42 CU or clocked at 2,670mhz, it could have achieved 12TFlops similar to Xbox series X. If the PS5 GPU were to have 52 CU as Xbox series X then it would have 14.82 Tflops. By contrast if XBox series X GPU were to be clocked similar to the PS5 GPU 2.23Ghz (2,283.5 Mhz ), then it would had 14.44 Tflops. What do you thinks guys?

The announced specs only makes sense if they are targeting a $400 price range. Perfectly for 4K gaming at 60fps, but not more than that. Also I'm sad they didn't talk about BC features for ps3/ps2/ps1.

Could the ps5 gpu be having several disabled CU and holding them until release? Is not unprecedented or unheard of.
In addition Cerny talked about using a GPU CU as a processing unit for audio. So...either games will use 35 CU or the PS5 GPU has more than 36 CU to be used.

Forget about the flopped word. Look at what all the Sony studios gave us with 1 Tflop. We have some of the best exclusives this generation so just imagine the games they'll give us with 10.3 tflops.

Get ready to get your minds blown tf open y'all

ram was used to "store" data you needed quick access too, you couldnt affoard to wait on your slow mechanical hard drive to get.

Thats no longer a issue, with how quick (esp sony's) SSD is.
This drastically reduces the amount of actual ram needed, when you run games.

It might even effect sizes of games, where they might have duplicates of data,
thats mixed in with differnt parts, so they "could" be loaded like this into ram.
Stuff like that will be gone with next gen.

These SSD are almost like virtual ram in themselves, you just go directly to the SSD to load stuff, isntead of first loading it into your ram.

^ atleast I think thats what cerny was saying.

You are reading to much into it.

There is likely some NAND reserved as spare area for various tasks to maintain performance that isn't accounted for.

I think you have placed far to much emphasis on flops rather than capabilities.

Me and CGI called it years ago.

Our original predictions were... 16GB of Ram... Yet we still had people thinking 128GB or more. Was pretty funny at the time!

Ram is still used to "store" data that is needed for quick access... Because we still cannot afford to wait on a slow SSD for data.

Ram is 500GB/s or more. SSD's are 1/10th of that.

It's still an issue, it's just much less of an issue.
You are right it should reduce the need for more RAM, but it doesn't remove the need for it entirely.

We will be RAM starved this generation, especially later on in the console cycle.

They didn't duplicate data on a mechanical disk, that occurred with optical disks due to their much lower seek times.

Not really. It's just "Virtual Memory". - Which is a technology that has existed for the last 30 years in various forms. Fuck. Even the Original Xbox used it!

To me, the most important thing about a gaming system is the library of games that becomes available for it, then the price tag, then the tech specs.

I think differences in tech specs between consoles have to be huge to make a difference in the minds of most people if they're honest. If you have to do side-by-side comparisons and squint to notice a difference in say the graphical capabilities of two systems, for example, that's a small enough distinction that most people I think will tend to buy whichever of those two machines is priced lower.

I wonder if SmartShift actually came from SNY. Cerny mentioned that some SNY stuff ends up in RDNA period if AMD see's use in it elsewhere, and AMD has already said it will be using that tech for Laptops/PC's earlier this year.

I don't remember reading or hearing anything about SmartShift for XBSX though, and if it originally came from SNY, well. Could that explain why it's not in XBSX and why PS5 has variable clocks while XBSX has locked clocks? Makes me wonder about Vega RPM and how Pro had it but not XB1X.

I do emphasize in capabilities. By analyzing performance strategy we can assume the final product.

For example, to be able and run bc ps3 games, CPU must be at least 3.2 ghz. PS5 over delivered on that but...

The problem I see is GPU logical assets for it being nvidia in nature and not AMD. You can't freely create a 100% legacy interpreter without their support. So, the only option that remains is  to brute force performance to achieve compatibility. The more teraflops the gpu has, the better performance and stability it can give for running games in a "ps3" emulated OS. Similarly this is the main reason og games weren't 100% supported (461/1000) in xbox 360, and less supported (41/1000) in xbox one/x. So, even taking into account that 1 tflops from a ps5 may had 1.5-2 times more performance than 1 tflops from a ps4 architecture. I don't see to much sense on what has been shown to us. He didn't want to focus in teraflops, because his specs were lower. Simple as that.

I assure you that if they would have a ps5 with 13tflops he would have talked laser focused for over 1 hour about the teraflops and 2 minutes about the audio.

If they intend to sell the ps5 at $500 or over it, then I'm hesitant to believe these will be the final specs nor that backwards compatibility is for ps4 games only. Such strategy doesn't make sense. Specially when he Cerny talked more time about the audio solution than the GPU solution.

It's fishy.

Why not give price then, if all is set in stone? Not much can be done, so much closer to launch window.

To me only two options emerge. Either the PS5 have higher hidden specs or, sony want the ps5 as a cheap solution entry level platform for the new generation and will bring a PS5 pro later on.

I'm a sony guy, all my life has been one. Probably im already over 100 ps4 games by now. I skipped xbox one this generation.  And surely I will buy ps5 day one.

But for the offered full capabilities for gaming like 4k 60+ fps, 8k or ps3/ps2/ps1 bc, 10.28 Tflops are not enough.

**double post

Both have high speed SSDs, both have eight-core processors running neck and neck, but PS5's SSD is outrageously faster.

Seems like PS5 is Usain Bolt, and XSX is Eliud Kipchoge.

They’re both impressive. The TFLOPS calculation gives the advantage to the XBSX, but the memory speed in the PS5 is insane. It will be interesting to see how both behave in real world situations. The PS5 seems to have been built with a focus on eliminating most if not any bottlenecks, while the XBSX goes for full brute force.

Time will tell which one is the more efficient approach.

But I’ll end by saying this one thing.

Nintendo may innovate when it comes to player inputs, but there’s certainly no denying that MS and Sony also innovate, but from a hardware approach.