Are PS4 Pro level graphics good enough for this gen?

You are ahead of the average player : 1080/30 fps and call it a day :)

All the rest is SW material ^^

It looks good enough. If I want high ultra over 9000 level graphics, I'd build a pc. If the games are lacking, what is the use of those graphics?

@PAL360: yeah it blew my mind when reports came out that these same ps4 games still weren't getting 60fps on the pro. For me performance and gameplay come first, slightly better graphics can only do so much if the game is crap or runs choppy.

@vaidream45: Some games do improve performance on the Pro (Battlefield 1, Tomb Raider, Infamous SS, etc), but that should be the norm, not the exception.

While 4k is indeed great, 1080p60 is simply more balanced and clever than 4k30, and lets be honest here, native 1080p with good AA is still pretty bice to look at.

I love my PC graphics, but I have no problems with PS4 graphics. PS4 Pro is an added bonus and honestly many titles still impress me. The upscaled 4K looks amazing and I game at 4k on my PC natively for most titles.

@Commiesdie:

The 360 had a superior gpu, but the ps3 had a superior cpu. In fact, the ps3's cpu had way more of an advantage over the 360's cpu than the 360's gpu over the ps3's gpu. It's not that Sony wouldn't allow their exclusives on the 360, it's that MS console couldn't duplicate what Sony was doing in their console. Motorstorm, and Uncharted are but a few examples.

In all this hoopla over textures and resolutions, we continue to see atrocious AI and level design.

I think the graphics at their level now is fine, so I wish more resources would be dedicated to making games more memorable. Tired of the one-and-done-flash-over-substance style games.

@ermacness: yeah ur retarded It"s crystal clear Sony would,not allow it bcuz it is exclusive and htf do u know 360 could not handle it I guess be ur logic Zelda can only be done on switch. I have nothing more to say to u be gone child

@commander: The PS1 was not more powerful than the N64. It was better designed yes, but it wasn't more powerful like you said. Get your facts straight. The textures on the PS1/N64 era were so low quality and could easily fit in cartridges. Square Enix chose CDs because of their FMVs not because of textures. Compression during that era worked well enough that they could make music, textures, and all the files they needed to fit into cartridges. The only thing they would have had problems with is Final Fantasy's FMVs. Developers and companies decided they didn't want to compress because it was far easier and cheaper to put stuff on CDs and that's why they chose PS1.

Arguing with you is like arguing with a rock. You're so stubborn and you continue to spout wrong information and make up your own facts. Scorpio isn't a next gen system unless MS says so. Everything we've seen of the specs suggest it's an extension to the Xbone. It's still too underpowered compared to modern gaming PCs. Think about it, the entire power of the Scorpio (CPU + GPU combined) isn't up to processing power on a GTX 1070 card. If I add a modern PC CPU to the equation the gap becomes even more embarrassing. If it was next gen it would be the smallest leap in power for MS ever between gens.

@Commiesdie:

If it could've been done in the 360, it would've. Of course I'm not speaking of Sony bringing exclusives to the 360, but the 360 had exclusives, and out of all those exclusives most of them was outdone visually and technically by Sony's variations of their exclusives, e.g. (for the shallows) KillZone 2 vs any of the Halos, R&C vs B&K N&B.

Dude, you're wasting your time. Unless you like to be entertained?

This.

To be honest I'm even happy with the Wii U.

@BIOKILLER123: You're right, I won't bother with him anymore lol. Let him live in ignorance if he wishes.

You act like fmv's weren't important, but back then a lot games were using it. Besides final fantasy wasn't the only game that wasn't released on the n64 due to problems with storage and there were other differences with multiplats as well. Sound and video was better on the playstation because compressing did lose quality on the n64. Some media was completely removed from the n64 titles, like voice acting for instance. Environmental textures and backgrounds were nerfed on the n64 to stay within the storage limit.

There's a reason why nintendo came with the gamepack that added storage for the cartridges but by then the damage was already done. What was sometimes better though was the main 3d models since the n64 had more processing power, but it was only a slight difference since the n64 had other bottlenecks too.

Of course we can start discussing semantics but what does that matter if the end result is lower in quality. Storage capacity opened up a whole new world back then and because of that the n64 was considered a weaker system. Yet you cherry pick some facts from over the internet to prove me wrong, maybe you should look at some comparison videos and keep in mind that this after a lot of optimization that never happened on the playstation.

Why does ms need to say that the scorpio is a next gen system? because ms doesn't have any exlusivity deals for 2017-2018. If there is a large enough userbase you can be sure exclusives will happen.

Also, haven't you heard, the cpu is not that important anymore, and 6k will be more than enough for a new gen.

The FLOPS gap between console generation with Microsoft's Xbox based devices

1. Xbox 360 to Xbox One

240 GFLOPS to 1310 GFLOPS emulated FP16 and FP32 is 5.5X

2. Xbox One to Xbox Scorpio

1310 GFLOPS FP32 to 6000 GFLOPS FP32 is 4.5X

1310 GFLOPS emulated FP16 to 12000 GFLOPS native FP16 (via double rate feature) is 9X.

Phil Spenser doesn't want AMD's year 2016 road map half gen jump.

GTX 1070 is facking expensive relative to Polaris 10 232 mm^2 size GPUs and that's not including the CPU. The chip size GP104 is 314 mm^2 without Northbridge, Southbridge and CPU.

BUILD NVIDIA GeForce GTX 1070 based gaming PC for $399 to $450 USD which includes UHD Blu-ray drive.

The only stubborn fool is you. I'll take you on.

Xbox Scorpio is a generation jump with hardware based backwards compatibility.

Xbox One has enough forward compatibility with Xbox Scorpio e.g. Jaguar's 256 bit AVX runs on SSE's 128 bit hardware hence AMD built enough forward compatibility with 256 bit AVX enabled software into Jaguar.

The performance gap between RX-480 and 1070 is about 33 percent and it's mostly effective memory bandwidth. FP32 TFLOPS difference between GTX 1070 is RX-480 is 1.06X.

Scorpio's memory bandwidth is 1.25X over RX-480's physical memory bandwidth.

When Shader Model 6's FP16 optimised game arrives, current desktop PC DX12 GPUs will have accelerated aging.

GTX 1070/1080/Titan X Pascal has shit FP16 mode and has emulated FP16 shaders on FP32 shaders. Zero performance benefits when running emulated FP16 shaders on FP32 shaders.

R9-280/R9-380X/Fury X/RX-470/RX-480 has native FP16 mode at single rate with FP32 mode. Some performance benefits when running emulated FP16 shaders e.g. reduce memory bandwidth and reduce register storage.

Double rate FP16 feature has large performance benefits.

Both AMD Vega 10 and PS4 Pro's IGP has future double rate FP16 feature support and has better support for future Shader Model 6.

Scorpio is being design to fully support Shader Model 6 in hardware.

It is not.

To start a new generation of consoles, it needs to be at least 10 times more powerful, as per PS3 to PS4

When shader workloads are combined for PS3's SPEs+RSX, the FP32 FLOPS gap between PS4 and PS3 is less than 10X.

For FP16 compute workloads, 8 core AMD ZEN's octal packed 16FP SIMD 256bit AVX mode per CPU core + Scorpio's 12 TFLOPS FP16 compute mode actually exceeds 10X over Xbox One's FP16 compute mode.

For once we had the chance to upgrade mid generation, I welcome the change. The Pro and Scorpio will make rest of the gen less painful than it was with PS3/X360.

@ronvalencia: ? Wall of text of garbage. Scorpio isn't a next gen console you undercover lem.

I bet the official name when MS announces it will be Xbone Pro or some silly shit. It's a turbocharged Xbone, nothing more.

lol you're delusional if you think that power jump won't trigger a new gen.

Ms stated quite clearly that eventually , when the userbase is big enough, exclusives will be made, but now at the moment it's focus is on 4k tv's and steady framerates.

Your post is garbage.

Stoping being hypocrite i.e. you asserted "Scorpio isn't a next gen system unless MS says so" and I addressed it with a statement from Phil Spenser.

http://www.cinemablend.com/games/

I'm not a big fan of Xbox One and a half. If we're going to move forward, I want to move forward in big numbers, […] For us, our box is doing well. It performs, it's reliable, the servers are doing well. If we’re going to go forward with anything, like I said, I want it to be a really substantial change for people--an upgrade

FLOPS jump from Xbox One to Scorpio is comparable to FLOPS jump from Xbox 360 to Xbox One.

Against your "Think about it, the entire power of the Scorpio (CPU + GPU combined) isn't up to processing power on a GTX 1070 card" assertion, the main magic sauce for Pascal/Maxwell GPUs are mostly with very good delta memory compression not with ALU shaders.

GTX 1070 has 6.4 TFLOPS shader compute FP32 and lacks proper double rate FP16 mode.

Again, the FLOPS gap between console generation with Microsoft's Xbox based devices

1. Xbox 360 to Xbox One

240 GFLOPS to 1310 GFLOPS emulated FP16 and FP32 is 5.5X

2. Xbox One to Xbox Scorpio

1310 GFLOPS FP32 to 6000 GFLOPS FP32 is 4.5X

1310 GFLOPS emulated FP16 to 12000 GFLOPS native FP16 (via double rate feature) is 9X.

Phil Spenser doesn't want AMD's year 2016 road map half gen jump.

-----------

NVIDIA's main magic sauce i.e. effective memory bandwidth advantage via very good delta memory compression.

As for RX-480, any overclock editions will be bounded by effective memory bandwidth.

For reference RX-480

((((256 bit x 8000Mhz) / 8) / 1024) x Polaris's 77.6 percent memory bandwidth efficiency) x Polaris's compression booster 1.36X = 263.84 GB/s

--

Scorpio's "more than 320 GB/s memory bandwidth" claim.

((((384 bit x GDDR5-6900 Mhz) / 8) / 1024) x Polaris's 77.6 percent memory bandwidth efficiency) x Polaris's compression booster 1.36X = 341 .34 GB/s

PS;

((384 bit x GDDR5-6900 Mhz) / 8) / 1024) = 323 GB/s physical memory bandwidth.

((384 bit x GDDR5-7000 Mhz) / 8) / 1024) = 328 GB/s physical memory bandwidth.

Comparison.

The memory bandwidth gap between Fury X and R9-290 = 1.266X (random textures)

With Fury X, it's delta memory compression is inferior to NVIDIA's Maxwell.

The FLOPS gap between Fury X and R9-290 = 1.48X

The frame rate gap between R9-290X and Fury X is 1.19X.

Random texture memory bandwidth gap's 1.266X factor is closer to frame rate gap's 1.19X.FLOPS gap between R9-290X (5.8 TFLOPS)and Fury X (8.6 TFLOPS) plays very little part with frame rate gap.

With 980 Ti (5.63 TFLOPS), it's superior memory compression enables it to match Fury X's results.

Conclusion: When there's enough FLOPS for a particular workload, effective memory bandwidth is better prediction method for higher grade GPUs.

-------------------

Example of near brain dead Xbox One ports running PC GPUs.

Frame rate difference between 980 Ti and R9-290X is 1.31X with Forza 6 Apex

Effective memory bandwidth between 980 Ti and R9-290X is 1.38X

Forza 6 Apex is another example for effective memory bandwidth influencing the frame rate result.

Microsoft knows AMD's GPU main design problem i.e. effective memory bandwidth with weak delta memory compression.

http://gamingbolt.com/ps4-pro-bandwidth-is-potential-bottleneck-for-4k-but-a-thought-through-tradeoff-little-nightmares-dev

PS4 Pro developer single out memory bandwidth as the bottleneck for 4.2 TFLOPS GPU.

Not really, without looking it up no one can name most of these timelines. They have always been determined by a NEW console that pretty much wipes the slate clean, new games, new accessories, new hardware... nes->snes->n64->gc->wii->wii u looking at nintendo generations.

With the way MS is approaching consoles, the idea of generations with xbox is quickly going to start looking more and more like phone iterations...

@ronvalencia: You're still on the FP16 nonsense? You know not everything is capable of lower precision so a lot of the numbers you're pulling out of you're ass are... complete nonsense. Unless you love weird graphical glitches you're not getting to 100% FP16 mode... ever.

Graphic is fine but fps is not

@aroxx_ab: I concur

@ronvalencia: Stop doing drugs and get back to reality. Your posts are nothing but gibberish and unconfirmed crap. MS has already stated the Scorpio is an enhanced Xbone and won't have its own exclusives, You're pulling numbers from your ass again to prove a point that is foolish at best.

@ronvalencia: Your Spencer quote doesn't prove the Scorpio is a next gen console. You are just interpreting it that way.

Who said about 100 percent FP16 mode? R9-390X can 4K/30 fps for most PC games at high/very high settings. RX-480 needs a memory bandwidth boost for improved 4K/30 fps stability.

Xbox 360 has support for 10 bit, 16 bit and 32 bit FP shaders. 10 bit FP was well known for Xbox 360's HDR. DX10 GPUs removed native support for FP16 and being restored for SM6 era GPUs.

Most optimisations for NVIDIA RSX are FP16 in nature.

10X power jump = new console generation jump was BS. I have addressed 10X nonsense with 10X compute operation jump in FP16.

Phil Spenser was talking about compute power jump between generations. Scorpio's compute power jump from XBO is similar in magnitude as generation jump from Xbox 360 to XBO.

Quote from Phil Spenser from http://digiworthy.com/2016/10/10/phil-spencer-xbox-scorpio-not-use-amd-jaguar/

About two and a half years ago we started to look at a hardware refresh that we might want to do, which in the end led to the Xbox One S and Scorpio in terms of designs. We’d looked at doing something that was higher performance this year, and I’d say the [PS4] Pro is about what we thought–with the GPU, CPU, memory that was here this year–that you could go do, and we decided that we wanted to do something different.

So we looked at Scorpio and 4K and what I thought was a bigger step in terms of performance. It was something that we wanted to focus on.

XBOX One half gen jump would have been PS4 Pro like solution.

Really? you don't get to go from 6000 to 12000 from FP32 to FP16 without it being 100%.

The drug filled fool is you. Phil Spenser was talking about generation jump in terms of compute power and Scorpio's compute power jump from XBO is in similar magnitude as the jump from Xbox 360 to XBO.

Xbox doesn't completely kill the old console generation since it has a constant backwards compatibility and AMD designed Jaguar's 128 bit SIMD hardware with forward compatibility with 256 bit AVX software.

Hardware capability is different from the usage factor.

Xbox 360 FP shaders has 10 bit, 16 bit and 32 bit. The problem with Xbox 360 is with rasteration hardware's interaction with certain data formats.

With the same graphics settings, R9-390X can be greater than 5.5X effective over XBO results.

XBox 1st to Xbox360 20gflops vs 240 gflops

Where is the jump of x12?

Why the Xbone doesnt have 2.9 Tflops?

Why Scorpio doesnt have 35 tflops?

Maybe because the 360 was the last monster console?

That made $1000 PCs cry in 2005?

Xbox 1's DX8 pixel shaders are in integers (integer 16 bit, quad pipelines) format while DX8 vertex shader are in floating point format. FLOPS refer to floating point operations.

FX5800 converted GF4 Ti/NV2A's quad pixel pipeline from integer 16 bit into FP 16 bit.

GCN version 1.1 has quad rate integer 8 bit with a single rate 16 bit integer.

PS4 Pro has double rate 16 bit for both integer and floats.

I don't know why I bother, you're example basically stated xbox one can run in fp16 mode entirely.

6000 -> 12000 is exactly double the performance if you change from fp32 to fp16.

You're not at all addressing where I called you out on bogus numbers and you're.

With the same graphics settings, a 5.1 TFLOP video card is 5.5x faster than a 1.2TFLOP card... so 6.6 ish TFLOP. Thats like only 30%, where's the other 70% from you're calculation?

the Scorpios GPU is a 6TFLOP chip with some features to help speed up certain operations. Saying it's equivalent to 12TFLOPs is essentially like a marketing team would lie about the performance.

The only bogus is your post. XBO has about 1.31 TFLOPS GPU with Kinect's resource usage minimised. On the PC, the GPU is also a shared resource. Similar shared GPU conditions.

R9-390X's 4K result vs XBO.

Scorpio's estimate from R9-390X results.

Far Cry Primal

XBO = 1440x1080 with ~30 fps. 1,555,200 pixels per frame http://www.eurogamer.net/articles/digitalfoundry-2016-face-off-far-cry-primal

46656000 pixels per second

R9-390X = 3840x2160 with ~36 fps. 8,294,400 pixels per frame. 298598400 pixels per second

R9-390X's effectiveness over XBO is about 6.4X. PC version also has higher graphics settings than XBO, hence R9-390X's effectiveness over XBO is even greater than 6.4X.

Higher TFLOPS = less time to compute a result within a frame's render time.

Higher memory bandwidth = less time to read and write within a frame's render time.

Memory write operations are dependant on compute results being available in a timely manner.

Higher TFLOPS can be memory bandwidth bound as compute results may wait for memory write opportunities.

My point, AMD can not continue to add CU without scaling memory bandwidth at appropriate levels. MS knows about AMD GPU's behavior and has increased Scorpio's physical memory bandwidth. NVIDIA's Pascal/Maxwell has excellent delta memory compression.

The graphics details from the above benchmarks are higher than XBO's graphics detail settings.

R9-390X destroys XBO's Killer Instinct results with greater than 5.5X effectiveness.

Vega's double rate FP16 feature is just another performance boost on top of Scorpio's R9-390X OC like solution.

Your point has no context with you're previous statements that scorpio is 12TFLOP card because of FP16 compatibility. you're comparing a 5.5TFLOP card with high bandwidth graphics memory with a 1.3TFLOP card with normal DDR and... to top it off you're not even looking at FP16 performance gains at all.

So sorry I got the FLOPS wrong on xbone, at 1.3...

Ps4 pro already started a new gen

Native FP16 support comes with Shader Model 6 and there's no current DirectX12 game with SM6.

12 TFLOPS via double rate FP16 is hardware capability NOT a usage factor.

R9-390X has 5.9 TFLOPS.

RX-480 OC still needs extra memory bandwidth for increase 4K/30 fps stability, but expect Scorpio to rival/beat RX-480 OC.

I expect Scorpio to fall between GTX 1060 and 1070 OC for current FP32 shader/Shader Model 5.1 enabled games.

Won't the games have a version on regular xbox and the only difference is an increase in frame rate, resolution and the removal of colour grading filters that help the original limp along at 30?

@ronvalencia said:

Native FP16 support comes with Shader Model 6 and there's no current DirectX12 game with SM6.

12 TFLOPS via double rate FP16 is hardware capability NOT a usage factor.

R9-390X has 5.9 TFLOPS.

Sorry you are wrong here. FLOPS aren't generally measured in half precision so you're spinning some benefit of a supported type. That would be like me coming in and saying it's only 3 TFLOPS because I decided to use double precision in my FLOP calculation. 12TFLOPS isn't hardware capability unless you make sure you mention loss in precision. 12TFLOPS w/ 16bit computation. And a useless measurement when the vast majority of graphics requires at least 32bit...

@waahahah said:

@ronvalencia said:

Native FP16 support comes with Shader Model 6 and there's no current DirectX12 game with SM6.

12 TFLOPS via double rate FP16 is hardware capability NOT a usage factor.

R9-390X has 5.9 TFLOPS.

Sorry you are wrong here. FLOPS aren't generally measured in half precision so you're spinning some benefit of a supported type. That would be like me coming in and saying it's only 3 TFLOPS because I decided to use double precision in my FLOP calculation. 12TFLOPS isn't hardware capability unless you make sure you mention loss in precision. 12TFLOPS w/ 16bit computation. And a useless measurement when the vast majority of graphics requires at least 32bit...

You are wrong. R9-390X's 5.9 TFLOPS is for both FP32 and emulated FP16 (no performance advantage when using emulated FP16 shaders).

NV40 has performance advantage with FP16. DX10 GPUs followed ATI DirectX9b pattern and SM6's native FP16 support attempts to restore the performance boost feature when using FP16.

For raster hardware on RGBA16F vs RGBA32F, 2X performance increase. Shader support for native FP16 matches raster hardware's native data format support.

You are wrong on "And a useless measurement when the vast majority of graphics requires at least 32bit". Native FP16 shader support completes the feature set with raster hardware.

http://phoronix.com/scan.php?page=news_item&px=AMDGPU-F16-Lands-VI

AMD just added half float support to their Linux open source driver since GCN 3ºrd gen (R9 285, R9 380/380X and Fury) on November 2016.

Mobile gaming brought back the importance of half floats. What is described by Cerny here happened with the custom Apple GPU (derived from PowerVR mobile GPUs) where they, too, went the "half floats can share a float register" method and it was a major component why iPhones were so much more performant in 3d applications than comparable phones.

EDIT: Here's an article http://www.realworldtech.com/apple-custom-gpu/

They look rather close to some PC games today.

@ronvalencia said:

R9-390X's 5.9 TFLOPS is for both FP32 and emulated FP16. You are wrong.

I don't care about 390x, the point is you were showing someone else that scorpio can be 12TFLOPS if you were in FP16 ' mode' which isn't how it works and its a totally bogus number. FP16 is a type and you can set up values where reduced precision is possible, it saves on storage space, and you can use 1 regester to do 2x16 bit calculations s instead of 1 32bit. The FLOPS is generally measured in standard precision, not half or double. It's misleading nonsense.

@waahahah said:

@ronvalencia said:

R9-390X's 5.9 TFLOPS is for both FP32 and emulated FP16. You are wrong.

I don't care about 390x, the point is you were showing someone else that scorpio can be 12TFLOPS if you were in FP16 ' mode' which isn't how it works and its a totally bogus number. FP16 is a type and you can set up values where reduced precision is possible, it saves on storage space, and you can use 1 regester to do 2x16 bit calculations s instead of 1 32bit. The FLOPS is generally measured in standard precision, not half or double. It's misleading nonsense.

Mobile gaming brought back the importance of half floats. What is described by Cerny here happened with the custom Apple GPU (derived from PowerVR mobile GPUs) where they, too, went the "half floats can share a float register" method and it was a major component on why iPhones were so much more performant in 3d applications than comparable phones.

Here's an article http://www.realworldtech.com/apple-custom-gpu/

On non-mobile GPU parts, Vega 10, PS4 Pro and GF100 (the only real new Pascal SM design) supports double rate FP16 shaders.

The method for packing multiple FP16 data elements into a register is a typical SIMD method. This is similar to 8 data element FP16 extensions for X86 SSE 128 bit register.

You're an idiot. I understand how it works. You're still not getting my point about terminology and how 12TFLOPS isn't a useful metric and it's misleading. FP16 will increase efficiency *where possible* and increase FLOPS, it will never double FLOPS.

Shit, this thread is making my head hurt :(