No. You can't. It's inaccurate... Which my Radeon RX 5500 example proves without a doubt.
Not only is the Radeon RX 5500 a newer and more efficient architecture... It also has more flops than the Radeon RX 590. - Yet cannot decisively out benchmark it.

Ergo. Flops is useless.

Another example is the Geforce 1030... The DDR4 variant with a 250mhz core overclock has 1.076Teraflops of performance (1402Mhz core * 2 Instructions per clock * 384 Cuda cores) - So it should decisively beat the GDDR5 variant which is 0.942 Teraflops, right? Wrong.

Even though they are based on the same GPU, the GDDR5 wins hands down, almost doubling in some instances.
https://www.techspot.com/review/1658-geforce-gt-1030-abomination/

Or lets take the Radeon 5870 and 7850? The Radeon 5870 has 2.72 Teraflops or performance verses the 7850's 1.76Teraflops, you would assume that the Radeon 5870 with an almost extra 1 Teraflop of performance would be the easy winner, right? Again. Wrong.
The 7850 is the clear victor. - https://www.anandtech.com/bench/product/1062?vs=1076

Same thing with the Radeon 7750. The DDR3 vs GDDR5 variants, same amount of Gflops, the DDR3 version even has twice the amount of Ram.
Yet the GDDR5 is the clear victor.
https://www.goldfries.com/computing/gddr3-vs-gddr5-graphic-card-comparison-see-the-difference-with-the-amd-radeon-hd-7750/

So basically in dot point form...

* Newer GPU with more flops = Doesn't mean it's faster.
* Same GPU's, one with more flops. = Doesn't mean it's faster.
* Same GPU with identical flops. = Doesn't mean it's faster.

So on what planet can we even assume that Flops has any kind of relevancy or accuracy when every example blows that assumption out of the water?

Please. Do tell. I am all ears.

You can scaling down games engine

But you will possibly lose the benefit of future hardware and more powerful hardware can do. The tech will stagnant like how COD use the same engine from PS3 era or Bethesda on every Fallout Games. 

Dude 5500 have 5TF, where you getting your number from?

Navi 50% faster than gcn in Xbox one and PS4, 20% faster than Polaris, (590= polaris)

You are aware that you are comparing half precision to single precision in that example, right? RX 5500XT has 4.7-5.2 Teraflops in single precision, significantly less than an RX 590 yet almost at the same performance. RX 5700 beats RX Vega despite having less TFlops. Which also shows that teraflops can only be compared within same architecture (and even then, with limitations) and can't be used otherwise as a yardstick.

I agree with the rest of what you're saying, just wanted to point out that inconsistency there.

CoD has made some great strides in visuals over the years, despite being based on a derivative of the Quake 2 engine.
It's just they have different design goals, low latency, high framerate is generally the goal for those engines...

For example with Modern Warefare 2, they introduced texture streaming which meant that texturing wasn't limited to the tiny pool of DRAM that the Playstation 3 and Xbox 360 had, resulting in a substantial uptick in visual fidelity, especially on the texturing side.

https://community.callofduty.com/t5/Call-of-Duty-Modern-Warfare-2/Modern-Warfare-2-Texture-Streaming/ba-p/9900744
https://www.eurogamer.net/articles/digitalfoundry-modern-warfare-2-face-off

We saw increases with Advanced Warefare when it brought a new lighting model and much improved post-process pipeline.
https://www.eurogamer.net/articles/digitalfoundry-2014-call-of-duty-advanced-warfare-face-off

Call of Duty Modern Warefare reboot saw great strides in geometry, improved lighting, shadowing and more.
https://www.eurogamer.net/articles/digitalfoundry-2019-cod-modern-warfare-delivers-series-most-advanced-visuals-yet


Every time they updated the engine they improved the core visuals. -
https://en.wikipedia.org/wiki/IW_engine

Call of Duty 2 (IW 2.0 engine): - Normal Maps, Bloom, Improved Shadowing.
Call of Duty  4 (IW 3.0 engine): Improved Lighting, Particles, Self Shadowing.
Call of Duty: Modern Warfare 2 (IW 4.0 engine): Texture and Mesh Streaming for better textures and models.
Call of Duty: Modern Warfare 3 (IW 5.0 engine): Improvements to streaming for larger environments, Improved Shadowing, Improved Reflections.
Call of Duty: Ghosts (IW 6.0 engine): Model geometry subdivision, HDR Lighting, Displacement Mapping, Tessellation, Specular Particles.
Call of Duty: Infinite Warfare (IW 7.0 engine): Physically based rendering for more accurate materials and lighting.

And I can keep going. Point is, it doesn't make sense to reinvent the wheel... Because these days, developers do not rebuild engines from scratch anymore.

Take the creation engine for instance, yes it looks dated today... But when it debuted with Skyrim in 2011 it was a decent looking engine that showcased what the 7th gen consoles could do with an open world environment... And that engine is based upon Oblivion's Gamebryo Engine from 2006... Which in turn is based on Morrowind Net Immerse engine from 2002.

Again, that engine spans multiple console generations and has scaled across hardware really well. - Elder Scrolls 6 is likely to be built on those same engine foundations and look absolutely stunning doing it.
Here is the jump from Morrowind on Xbox to Skyrim on Xbox 360.

Generational leap, no?

Even Unreal Engine is based on the engines that came before it, Unreal Engine 3.0 and 4.0 still contains code from the original Unreal Engine from the late 90's for example. Again. - Why reinvent the wheel?

https://en.wikipedia.org/wiki/Unreal_Engine

Essentially that engine technology has scaled from (In terms of hardware power) the Dreamcast right up to the Xbox One X.

The latest Battlefield being a graphical powerhouse (Especially on PC) is based on the Frosbite engine, the same engine which debuted with Battlefield: Bad Company on the Xbox 360 back in 2008. - The engine has had some massive overhauls since with Frostbite 1.5, 2.0, 3.0, 4.0... And I wouldn't be surprised if code from refractor still lingered somewhere.

https://en.wikipedia.org/wiki/Frostbite_(game_engine)

I think the reason why there is this misconception that older game engines hold back newer platforms is because people do not fundamentally understand what a game engine is or what it actually does.

A game engine is NOT just the "thing" that draws the pretty pictures on your display... A Game engine is a bunch of components working under a "framework" - Be it Audio, Physics, Networking and more. - Thus you can rewrite a part of the engine like with say... The lighting and materials shaders and take advantage of a newer platforms graphics capabilities.

But you can bet your ass that most game engines today are derived from technology of yesterday, because they scale.

God dammit.

A flop is a flop.

Navi doesn't have 1.4x more "theoretical flops" than Graphics Core Next. It just doesn't.

Flops is based on the number of Stream Processors * Instructions Per Clock * Clock Rate.

Navi gets 1.4x more performance, not because of flops, but because of everything else in the GPU, if you were to throw a purely compute task at Graphics Core Next, it would be able to achieve some impressive flop numbers, often higher than Navi. - But when it comes to gaming, games need more than just flops, ergo Navi is able to pull ahead.

Thus a flop on Navi is the same flop as Graphics Core Next. - Navi is just more efficient.

Engines are simply scalable, that is all there is to it, that doesn't change when new console hardware comes out with new hardware features that gets baked into new game engines.

You can turn effects down/off, you can use different (less demanding) effects in place of more demanding ones and more, which is why we can take a game like Doom, The Witcher 3, Overwatch, Wolfenstein 2 which scales from high-end PC CPU's, right down to the Switch... A game like the Witcher 3 still fundamentally plays the same as the PC variant despite the catastrophic divide in CPU capabilities.

Scaling a game from 3 CPU cores @ 1ghz on the Switch to 6 CPU cores at 1.6Ghz on the Playstation 4 to 8+ CPU cores @3.4Ghz on the PC just proves that.

The Switch was certainly not the baseline for those titles, the Switch didn't even exist when those games were being developed, yet a big open world game like the Witcher 3 plays great, game design didn't suffer.

I mean, I get what you are saying, developers do try and build a game to a specific hardware set, but that doesn't mean you cannot scale a game downwards or upwards after the fact.

At the end of the day, things like Ray Tracing can simply be turned off, you can reduce geometric complexity in scenes by playing around with Tessellation factors and more and thus scale across different hardware.

You have misconstrued my statements.

The Single Precision Floating Point numbers being propagated around are NOT including the Ray Tracing capabilities of the part, because the FLOPS are a function of Clockrate multiplied by functional CUDA/RDNA/GCN shader units multiplied by number of instructions per clock. - It excludes absolutely everything else, that includes Ray Tracing capabilities.

You can have identical flops with identical chips and still have half the gaming performance.

Thus flops are certainly irrelevant as it doesn't account for the capabilities of the entire chip.

Even overclocked the Geforce 1030 DDR4 cannot beat the GDDR5 variant, they are the EXACT same chip, roughly the same flops.
https://www.gamersnexus.net/hwreviews/3330-gt-1030-ddr4-vs-gt-1030-gddr5-benchmark-worst-graphics-card-2018

nVidia's Ray Tracing on the 2080Ti is not the same as RDNA2's Ray Tracing coming out next year, the same technology that next-gen consoles are going to leverage, so it's best not to compare.

Plus, developers are still coming to terms on how to more effectively implement Ray Tracing, it is certainly a technology that is a big deal.

Well. They are irrelevant, it's a theoretical number, not a real world one, the relevant "Flop number" would be one that is based on actual, real-world capabilities that the chips can actually achieve.

And like the Geforce 1030 example above, you can have identical/more flops, but because of other compromises, you end up with significantly less performance.

The Switch we know exactly what it's capabilities are because Nintendo are using off-the-shelf Tegra components, we know what clockspeed and how many functional units it has as well thanks to Homebrew efforts that cracked the console open.

The WiiU is still a big unknown because it was a semi-custom chip, we do know it's an AMD Based VLIW GPU with an IBM PowerPC CPU though.

And exactly, you can't claim 4 different rumors as all being true.

No hard feelings, no forced assimilation, just worthwhile thoughts.

While certain info isn't available to make a reliable conclusion, I don't focus as much on the past. It's certainly necessary and useful, but too much focus on what was, without enough consideration about what is, will make discerning the future less likely.

I mean, who could have foreseen this?

We know very little about Sony/Microsfts ray-tracing solution, the person who leaked Prospero first says Ps5 and Xbox Series X uses completely different ray-tracing solution. I would assume Microsoft uses AMD and Sony has there own. Yes frame-rate for games could be all over the place because someone has better ray-tracing but weaker Flop performance.

Whoever has more TF will probably market there consoles as WORLD MOST POWERFUL CONSOLE :)

.. if the rumor of 9 vs 12 Tflops is true (40 vs 56 CUs) were true than the silicon budget of PS5 would have been used in other stuff instead of just giving away over 33% in power (if everything else in the consoles would give the same 9 vs 12 advantage to Series X). Because that devkit was just to0 big to have so much less power.

The current rumours are  I think 40cus at 2GHz vs 56cus at 1.8GHz, so looking at cus only, it would be a 25% difference (both are very optimistic clock rates imho).

Power comes at a price. Particularly at 7nm, every square mm adds $ to the chip cost. It's unikely that Sony uses other stuff, it's just that Sony uses a smaller chip overall to save significant costs. Rumours have the PS5 at around 310-320mm^2, while the XSX is at 360-380mm^2. That is a significantly higher cost (probably around $30 per chip) for the latter chip. If the performance difference stays within 20%, people will barely notice the difference. Both will be constraint to a certain extent by memory. 16GByte is the rumoured size for both, but if XSX uses, say, 24GBytes (the 384bit bus rumour), that would mean a significant difference in performance. 16GByte is the very bottom you can get away with for something that has to live in the 4k world for even just a few years.

Well a 30USD cost for 25% performance would be an easy expense for me =p

But looking at the size of XSX when the control is near it (much smaller than the original xbox and little bigger than X1S) I don't think PS5 will be much smaller (if so the devkit is just much much much bigger than retail version).

Well if you intend to sell 100mio consoles, that's a pretty $3billion in the end. If you think that is easy expense for you, just send me a percent of that

The volume of the XSX (is the power supply inside or external, btw?) is considerably bigger than the volume of the X1S. Noone has seen Sony's idea of the PS5 end product. Given Sony's track record of building small(er) things with jet engines inside, I don't have much hope in that area.

My mistake. I was reading my database incorrectly.
My point still stands however.

Yeah I also made a mistake, turns out Proelite over at Beyond3d was a fake insider, it was he who said the devkit for PS5 was 40CU's. But looks false.

Oberon remains a mystery, maybe it just haft to do with backwards compatibility and nothing else.

Oberon says 3 things, that the gpu is clocked at 2ghz and it have 2 backwards compatility modes. 1 mode where 18CU's is active which matches PS4 and another with 40CU's which don't match PS4 pro so you would assumed this was Boost mode with all CU's active, but the insider at beyond3d is false.