Xbox Scarlett APU Die Shot Analysis: Die Size Estimated at 401 mm2

  • 181 results
  • 1
  • 2
  • 3
  • 4
Avatar image for Grey_Eyed_Elf
Grey_Eyed_Elf

6769

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#51  Edited By Grey_Eyed_Elf
Member since 2011 • 6769 Posts

@ronvalencia: Memory bandwidth would help, Lets have a look at a example:

2080 and 2080 Ti have a 3.38 TFLOP difference which is a 31% increase in power... The Ti has a 352 Bit at 616GB/s vs 256 Bit at 448GB/s on the 2080.

22% increase in performance... Now that's with a 31% TFLOP increase and a much higher memory bandwidth difference than the rumoured PS5 and XSX, once you take that into account that 22% difference could drop to 18-19%.

So 16.5% to 19% due to the higher memory bandwidth.

Its still not night and day nor is it anywhere close to the gap between the Pro and X1X, which is my entire point.

Avatar image for BlackShirt20
BlackShirt20

1349

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#52  Edited By BlackShirt20
Member since 2005 • 1349 Posts

@Grey_Eyed_Elf: Ok. Let’s say your calculations were off and it is closer to 20%.

That’s huge. You can sit there and say it isn’t. But that actually is. XSX will run games at higher settings with better frames and more detail. It will be noticeable. Especially when you take into account the XSX has a better CPU.

Avatar image for Grey_Eyed_Elf
Grey_Eyed_Elf

6769

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#53  Edited By Grey_Eyed_Elf
Member since 2011 • 6769 Posts

@BlackShirt20 said:

@Grey_Eyed_Elf: Ok. Let’s say your calculations were off and it is closer to 20%.

That’s huge. You can sit there and say it isn’t. But that actually is. XSX will run games at higher settings with better frames and more detail. It will be noticeable. Especially when you take into account the XSX has a better CPU.

Huge?...

  • PS4 1.84 TFLOPS vs X1 1.31 TFLOPS = 28% increase
  • X1X 6 TFLOPS vs PS4 Pro 4.2 TFLOPS = 30% increase
  • PS5 9.2 vs XSX 12 TFLOPS = 24% increase in GPU
  • PS2 6.2 GFLOPS vs NGC 9.4 GFLOPS = 35% increase
  • NGC 9.4 GFLOPS vs Xbox 20 GFLOPS = 45% increase
  • PS2 6.2 GFLOPS vs Xbox 20 GFLOPS = 69% increase

We have seen bigger, you saying its huge is implying that its inconceivable of a difference when in fact its a modest difference between two consoles compared to the last 2 console releases its the lowest its been.

Aside from the 360 vs PS3 which was 5% increase in GPU power, this is nothing new and will offer little difference as WE have seen bigger difference's especially recently.

I repeat again both consoles have 8 core Zen 2 chips a one at 3.2GHz and one at 3.6GHz that's a 400MHz increase, a 8700K stock has a all core boost of 4.3GHz and my overclocked to 5GHz gives it a 700MHz increase in clocks at 4K?.. No difference in performance.

https://www.techpowerup.com/review/intel-core-i7-8700k/13.html

Stop posting things when you don't know anything.

The PS5 vs XSX rumoured spec's is one of smallest gaps we have seen and the CPU won't make a difference between the two at 4K.

If people had a hard time telling a difference between X1 and PS4 or Pro vs X1X... The difference will be even less with PS5 and XSX.

Also if that is huge to you... Then please explain why console gamers have been pretending that their is no difference between PC and X1X for the past 4 years?... Because I mean damn, you guys can't tell the difference between low and ultra in Red Dead Redemption 2 on a 2080 Ti that is 80-100% increase in GPU power to a X1X.

Avatar image for Grey_Eyed_Elf
Grey_Eyed_Elf

6769

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#54 Grey_Eyed_Elf
Member since 2011 • 6769 Posts

I just realised this being a PC gamer on this forum is losing battle... You guys exaggerated small differences between consoles then down play HUGE difference between PC daily.

I'll just take my next gen beating PC over to a corner and play games while ignoring these daft threads till those console release.

HUGE difference... But nah that 2080 Ti ain't that much more powerful than a XSX.

You guys squabble over this dumb s***. I'm out.

Avatar image for kazhirai
KazHirai

45

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 5

#55 KazHirai
Member since 2019 • 45 Posts

@Grey_Eyed_Elf said:
@BlackShirt20 said:

@Grey_Eyed_Elf: Ok. Let’s say your calculations were off and it is closer to 20%.

That’s huge. You can sit there and say it isn’t. But that actually is. XSX will run games at higher settings with better frames and more detail. It will be noticeable. Especially when you take into account the XSX has a better CPU.

Huge?...

  • PS4 1.84 TFLOPS vs X1 1.31 TFLOPS = 28% increase
  • X1X 6 TFLOPS vs PS4 Pro 4.2 TFLOPS = 30% increase
  • PS5 9.2 vs XSX 12 TFLOPS = 24% increase in GPU
  • PS2 6.2 GFLOPS vs NGC 9.4 GFLOPS = 35% increase
  • NGC 9.4 GFLOPS vs Xbox 20 GFLOPS = 45% increase
  • PS2 6.2 GFLOPS vs Xbox 20 GFLOPS = 69% increase

We have seen bigger, you saying its huge is implying that its inconceivable of a difference when in fact its a modest difference between two consoles compared to the last 2 console releases its the lowest its been.

Aside from the 360 vs PS3 which was 5% increase in GPU power, this is nothing new and will offer little difference as WE have seen bigger difference's especially recently.

Where are you getting these percentages?

  • PS4 1.84 TFLOPS vs X1 1.31 TFLOPS = 40.45% increase
  • X1X 6 TFLOPS vs PS4 Pro 4.2 TFLOPS = 42.85% increase
  • PS5 9.2 vs XSX 12.2 TFLOPS = 32.60% increase (It's speculated at 12.2 btw, not 12)

Avatar image for BlackShirt20
BlackShirt20

1349

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#56 BlackShirt20
Member since 2005 • 1349 Posts

@Grey_Eyed_Elf: It is huge. 20% is a huge difference (Even though it will be more than that). It’s a noticeable difference. The difference between the PS4 Pro and Xbox One X was about 40%. The XOX runs every game at a higher resolution and more stable frame rate. 20% or roughly 20% is more than enough to have higher settings, textures and frames advantage.

But I digress, The point is, Xbox will be the game all games are compared too. Developers will code for the Xbox than port to PS5/PC. Before your little heart freaks out dont worry, some games will be primarily developed on PC and ported Xbox. But those games are nothing special. PUBG for instance is terrible. But PC gamers love it.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#57  Edited By ronvalencia
Member since 2008 • 28521 Posts

@Grey_Eyed_Elf said:

@ronvalencia: Memory bandwidth would help, Lets have a look at a example:

2080 and 2080 Ti have a 3.38 TFLOP difference which is a 31% increase in power... The Ti has a 352 Bit at 616GB/s vs 256 Bit at 448GB/s on the 2080.

22% increase in performance... Now that's with a 31% TFLOP increase and a much higher memory bandwidth difference than the rumoured PS5 and XSX, once you take that into account that 22% difference could drop to 18-19%.

So 16.5% to 19% due to the higher memory bandwidth.

Its still not night and day nor is it anywhere close to the gap between the Pro and X1X, which is my entire point.

Data from https://www.techpowerup.com/review/amd-radeon-rx-5700-xt/28.html (for newer drivers)

On RTX 2080 Ti vs RTX 2080, not including any stealth boost modes(1)

Average results: RTX 2080 Ti has 26.8 percent better results.

Memory bandwdith: RTX 2080 Ti has 37.5 percent higher memory bandwdith

ROPS: RTX 2080 Ti has 33 percent higher ROPS count

GPC count: Both SKUs has the same six GPC unit count <---- front-end bottleneck potential which includes geometry inputs.

L2 cache size: RTX 2080 Ti has 50 percent higher

Both GPUs has similar clock speeds.

X1X has semi-custom design advantage over PS4 Pro i.e. X1X is not a simple PS4 Pro scaled to 6 TFLOPS.

My RTX 2080 Ti GX Trio (paper spec +1755 Mhz e.g. 1900Mhz) and GTX 1080 Ti FE can boost beyond the paper specs clock speeds.

https://www.techpowerup.com/review/msi-r9-390x-gaming/30.html

R9-390X (effectievly dual Pitcairn R9-270X in very area) vs R9-270X

R9-290X Hawaii XT has 89 percent scaling boost over R9-270X

R9-390X Hawaii XT2 has 115 percent scaling boost over R9-270X

Hawaii XT has nearly double everything from Pitcairn XT including the geometry engines i.e. quad vs dual

We do not know Shader Engine configuration for XSX's GPU e.g.

1. Dual shader engine (SE) with 30 CU for each SE (64 ROPS). Four geometry inputs per SE with 8 total geometry inputs for dual SE units.

OR

2. Three shader engine (SE) with 20 CU for each SE (96 ROPS). 12 total geometry inputs for three SE units.

NAVi 14's single shader engine has 24 CUs, hence RDNA SE can scale to 24 CU.

RX 5600 XT shows NAVI 10 36 CU with 196 bit bus hence AMD's RDNA can copy NVIDIA's 192 bit bus wdith configs.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#58  Edited By ronvalencia
Member since 2008 • 28521 Posts

NAVI 10 vs Hawaii GCN on raster power vs TFLOPS IPC relationship.

https://www.techpowerup.com/review/amd-radeon-rx-vega-64/31.html

R9-390 Hawaii GCN (5.1 TFLOPS) still has superior raster vs TFLOPS IPC ratio when compared to Polaris and Vega

Scale R9-390's 5.1 TFLOPS results to RX 5700 XT's 9.75 TFLOPS and it would land on near RX 5700 XT's results. RX 5700 XT is effectively Hawaii GCN designed for high clock speed.

R9-390's 5.1 TFLOPS scaled to 6.0 TFLOPS lands near GTX 980 Ti.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#59  Edited By ronvalencia
Member since 2008 • 28521 Posts

@jahnee said:
@ronvalencia said:

@Grey_Eyed_Elf said:

Both... Bigger means more cores on it but also means hotter, more power hungry, more expensive.

Also everyone is letting the numbers get to their heads 9.2 TFLOPs vs 12 TFLOPs on the same GPU architecture is not night and day, the difference in real world performance is less than 10%.

If this thing costs $100 more than a PS5 II just don't see it selling.

Also people need to take into account that ray tracing cores will take up space on the die and will add to the heat, the die size also doesn't really indicate how many cores will be active... 56 CU rumour could have 4 or 8 CU's disabled. Who knows.

All we know for now is that the new larger and non conventional case design means its hot.

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

When comparing against PS4 Pro's 4.2 TFLOPS with 256 bit GDDR5-7000 memory bus, X1X's 6 TFLOPS GPU is backed by 384 bit GDDR5-6800 memory bus and ROPS's 2MB render cache and difference between X1X and PS4 Pro is more than 10 percent.

I agree with you that pure TFLOPS arguments are nearly pointless without backing extra TFLOPS with extra memory bandwidth.

XSX's external memory bus is greater than 256 bit bus and software tiled compute programming tricks with fusion link enables data to remain in L2/L3 cache and reduce CPU's external memory access. There is programming pathway for game consoles to reduce CPU's external memory bus access.

XSX's GDDR6-14000 memory modules and memory bus greater than 256 bits has been confirmed from E3 2019 reveal.

Do you speculate the differences between the XsX and PS5 to be similar to X1X and PS4pro?

TFLOPS arguments without I/O arguments are nearly useless.

Increase TFLOPS power should scale with increase raster power and we don't know XSX GPU's configuration while Techpowerup speculated 96 ROPS and there's no comment on geometry input increase.

Avatar image for Grey_Eyed_Elf
Grey_Eyed_Elf

6769

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#60 Grey_Eyed_Elf
Member since 2011 • 6769 Posts

@kazhirai said:
@Grey_Eyed_Elf said:
@BlackShirt20 said:

@Grey_Eyed_Elf: Ok. Let’s say your calculations were off and it is closer to 20%.

That’s huge. You can sit there and say it isn’t. But that actually is. XSX will run games at higher settings with better frames and more detail. It will be noticeable. Especially when you take into account the XSX has a better CPU.

Huge?...

  • PS4 1.84 TFLOPS vs X1 1.31 TFLOPS = 28% increase
  • X1X 6 TFLOPS vs PS4 Pro 4.2 TFLOPS = 30% increase
  • PS5 9.2 vs XSX 12 TFLOPS = 24% increase in GPU
  • PS2 6.2 GFLOPS vs NGC 9.4 GFLOPS = 35% increase
  • NGC 9.4 GFLOPS vs Xbox 20 GFLOPS = 45% increase
  • PS2 6.2 GFLOPS vs Xbox 20 GFLOPS = 69% increase

We have seen bigger, you saying its huge is implying that its inconceivable of a difference when in fact its a modest difference between two consoles compared to the last 2 console releases its the lowest its been.

Aside from the 360 vs PS3 which was 5% increase in GPU power, this is nothing new and will offer little difference as WE have seen bigger difference's especially recently.

Where are you getting these percentages?

  • PS4 1.84 TFLOPS vs X1 1.31 TFLOPS = 40.45% increase
  • X1X 6 TFLOPS vs PS4 Pro 4.2 TFLOPS = 42.85% increase
  • PS5 9.2 vs XSX 12.2 TFLOPS = 32.60% increase (It's speculated at 12.2 btw, not 12)

Sorry was tipsy... Got the math wrong. Essentially did it backwards.

Point still stands... The XSX vs PS5 is one of the smallest gaps we have ever had:

  • PS4 1.84 TFLOPS vs X1 1.31 TFLOPS = 40% increase
  • X1X 6 TFLOPS vs PS4 Pro 4.2 TFLOPS = 42% increase
  • PS5 9.2 vs XSX 12.2 TFLOPS = 32% increase in GPU
  • PS2 6.2 GFLOPS vs NGC 9.4 GFLOPS = 51% increase
  • NGC 9.4 GFLOPS vs Xbox 20 GFLOPS = 112% increase
  • PS2 6.2 GFLOPS vs Xbox 20 GFLOPS = 222% increase

All this huge difference narrative people are throwing out is false, we have had MUCH bigger gaps in performance from consoles.

You can get a bigger increase going from a 2060 to a 2070 Super which is 39%.

A 32% increase in power is not huge. Its smaller than a PS4 to X1 and X1X to Pro.. By 10%.

Avatar image for Grey_Eyed_Elf
Grey_Eyed_Elf

6769

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#61 Grey_Eyed_Elf
Member since 2011 • 6769 Posts
@BlackShirt20 said:

@Grey_Eyed_Elf: It is huge. 20% is a huge difference (Even though it will be more than that). It’s a noticeable difference. The difference between the PS4 Pro and Xbox One X was about 40%. The XOX runs every game at a higher resolution and more stable frame rate. 20% or roughly 20% is more than enough to have higher settings, textures and frames advantage.

But I digress, The point is, Xbox will be the game all games are compared too. Developers will code for the Xbox than port to PS5/PC. Before your little heart freaks out dont worry, some games will be primarily developed on PC and ported Xbox. But those games are nothing special. PUBG for instance is terrible. But PC gamers love it.

What?!...

Stop it.

X1X runs Forza Horizon at 4K/30FPS on a mixture of settings and was developed for that console, but the PC version looked better and ran a 4K/60 on a 1080 Ti when it launched.

You talk about HUGE percentages then talk absolute nonsense when it comes to PC.

When ever there is a multi-platform game they compare the consoles TO PC to see where the console versions are lacking in settings, not the other way around its what digital foundry has been doing since the PS4/X1 launched.

All the consoles are X86 platforms now so there is little to no advantage for consoles and the high majority of games scale with PC hardware.

All gibberish.

I gaurentee you I will see you in forums when those consoles release and the games are running at 4K/30FPS and PC's running higher settings 4K/60 you will say "oh but look at the premium you have to pay to get that".

I know your type.

You tried to compare a 12.2 TFLOP XSX GPU to a RTX 2080 Ti simply because of the TFLOP count.

Avatar image for xantufrog
xantufrog

12213

Forum Posts

0

Wiki Points

0

Followers

Reviews: 2

User Lists: 5

#62 xantufrog  Moderator
Member since 2013 • 12213 Posts

@BlackShirt20: I don't know how to break it to you, but I tested this systematically a while back and posted the FH4 benchmarks on here to discuss - an old i5 6500 and GTX970 can almost run the game as well as an X1X at 4K (hard to explore far, given the RAM usage is pegged on the poor 970 at 4K), and can run the game BETTER than an X1X at 1080p.

Avatar image for tormentos
tormentos

29646

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#63 tormentos
Member since 2003 • 29646 Posts

@ronvalencia:

Yeah I remember how MGS was 720p on Xbox and 1080p on PS4 with dynamic skies on PS4 while lacking it on Xbox,that is more than 100%,several other games showed gaps as big as that and you say nothing.

Is normal nothing out of this world for the gap in power,even if the spec end like that the series X will have a smaller gap over the PS5 than xox had over the pro,and the PS4 over the Xbox one.

Avatar image for comp_atkins
comp_atkins

36186

Forum Posts

0

Wiki Points

0

Followers

Reviews: 3

User Lists: 0

#65 comp_atkins
Member since 2005 • 36186 Posts

@pyro1245 said:

Meanwhile in the world of machine learning.....

https://singularityhub.com/2019/08/26/this-giant-ai-chip-is-the-size-of-an-ipad-and-holds-1-2-trillion-transistors/

lol

good luck yielding anything.

Avatar image for BlackShirt20
BlackShirt20

1349

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#66 BlackShirt20
Member since 2005 • 1349 Posts

@Grey_Eyed_Elf: oh my friend. You are special. I was referring to the XSX. A lot of games play better on console.

See, the beautiful thing about consoles is, you download and play. It just works. PC if your settings are wrong can screw everything up. With console, everything is simple and easy. With these new consoles the gap between PC and console will be minimal at best. At least for awhile.

Nearly all 3rd party games will be coded for Xbox. Than ported to the others. That’s just how it works.

The performance of the Xbox will be comparable to the best rigs today.

Avatar image for fedor
Fedor

5973

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#67 Fedor
Member since 2015 • 5973 Posts

@BlackShirt20 said:

@Grey_Eyed_Elf: oh my friend. You are special. I was referring to the XSX. A lot of games play better on console.

Name one.

See, the beautiful thing about consoles is, you download and play. It just works. PC if your settings are wrong can screw everything up.

This is false. If You think different provide evidence.

With console, everything is simple and easy.With these new consoles the gap between PC and console will be minimal at best.At least for awhile.

By the time the consoles release Ampere and RDNA2 will be on the market. The consoles can't even beat current tech, it will be completely trounced come release.

Nearly all 3rd party games will be coded for Xbox. Than ported to the others. That’s just how it works.

That is not how it works. If you think it is, provide a source.

The performance of the Xbox will be comparable to the best rigs today.

This is false.

Avatar image for BlackShirt20
BlackShirt20

1349

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#68 BlackShirt20
Member since 2005 • 1349 Posts

@Grey_Eyed_Elf: @fedor:

https://www.techpowerup.com/gpu-specs/xbox-series-x-gpu.c3482

https://www.techpowerup.com/gpu-specs/playstation-5-gpu.c3480

https://www.techpowerup.com/gpu-specs/geforce-rtx-2080-super.c3439

I rest my case.

Avatar image for lundy86_4
lundy86_4

54871

Forum Posts

0

Wiki Points

0

Followers

Reviews: 2

User Lists: 0

#69 lundy86_4
Member since 2003 • 54871 Posts

@BlackShirt20 said:

@Grey_Eyed_Elf: @fedor:

https://www.techpowerup.com/gpu-specs/xbox-series-x-gpu.c3482

https://www.techpowerup.com/gpu-specs/playstation-5-gpu.c3480

https://www.techpowerup.com/gpu-specs/geforce-rtx-2080-super.c3439

I rest my case.

With what evidence? We don't have access to PS5/Xbox specs.

Avatar image for fedor
Fedor

5973

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#70  Edited By Fedor
Member since 2015 • 5973 Posts

@BlackShirt20 said:

@Grey_Eyed_Elf: @fedor:

https://www.techpowerup.com/gpu-specs/xbox-series-x-gpu.c3482

https://www.techpowerup.com/gpu-specs/playstation-5-gpu.c3480

https://www.techpowerup.com/gpu-specs/geforce-rtx-2080-super.c3439

I rest my case.

2 unconfirmed spec lists and an Nvidia GPU who's architecture produces far better performance than RDNA? What is your case exactly?

I notice you also avoided answering any questions or providing any sources to your claims, probably because you simply have no clue what you're talking about. How does it feel being ignorant?

Avatar image for BlackShirt20
BlackShirt20

1349

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#71 BlackShirt20
Member since 2005 • 1349 Posts

@fedor: @lundy86_4:

I have spoken.

Avatar image for lundy86_4
lundy86_4

54871

Forum Posts

0

Wiki Points

0

Followers

Reviews: 2

User Lists: 0

#72 lundy86_4
Member since 2003 • 54871 Posts

@BlackShirt20 said:

@fedor: @lundy86_4:

I have spoken.

Incorrectly...

Avatar image for BlackShirt20
BlackShirt20

1349

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#73  Edited By BlackShirt20
Member since 2005 • 1349 Posts

@lundy86_4: This is a warning.

That’s 1. Stop while you can.

Avatar image for lundy86_4
lundy86_4

54871

Forum Posts

0

Wiki Points

0

Followers

Reviews: 2

User Lists: 0

#74 lundy86_4
Member since 2003 • 54871 Posts

@BlackShirt20: You may as well gimme 2 and 3. You posted a big nothing-burger, bud.

Avatar image for fedor
Fedor

5973

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#75 Fedor
Member since 2015 • 5973 Posts

@BlackShirt20 said:

@fedor: @lundy86_4:

I have spoken.

Yes, and you were wrong... Again.

Avatar image for BlackShirt20
BlackShirt20

1349

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#76 BlackShirt20
Member since 2005 • 1349 Posts

@lundy86_4: You sure? You just say the word and I’ll talk to a mod.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#77  Edited By ronvalencia
Member since 2008 • 28521 Posts

@Grey_Eyed_Elf said:

Sorry was tipsy... Got the math wrong. Essentially did it backwards.

Point still stands... The XSX vs PS5 is one of the smallest gaps we have ever had:

  • PS4 1.84 TFLOPS vs X1 1.31 TFLOPS = 40% increase
  • X1X 6 TFLOPS vs PS4 Pro 4.2 TFLOPS = 42% increase
  • PS5 9.2 vs XSX 12.2 TFLOPS = 32% increase in GPU
  • PS2 6.2 GFLOPS vs NGC 9.4 GFLOPS = 51% increase
  • NGC 9.4 GFLOPS vs Xbox 20 GFLOPS = 112% increase
  • PS2 6.2 GFLOPS vs Xbox 20 GFLOPS = 222% increase

All this huge difference narrative people are throwing out is false, we have had MUCH bigger gaps in performance from consoles.

You can get a bigger increase going from a 2060 to a 2070 Super which is 39%.

A 32% increase in power is not huge. Its smaller than a PS4 to X1 and X1X to Pro.. By 10%.

One would assume AMD didn't replicate Fury/Vega IPC degradation by adding CU and skiping I/O scaling.

Classic GPU hardware mostly deals with preparing and feeding data to the CU, hence it's geometry-raster I/O which covers items such as geometry input (back face culling), TMU(texture read/write + filter processing) and ROPS (color and z-buffer/depth buffer read/write + blending and MSAA processing).

PS4 Pro vs X1X gap is already indicated by RX-470 vs R9-390X results

R9-390X beaten RX-470 by 48 percent.

RX-470 has 4.9 TFLOPS

R9-390 has 5.1 TFLOPS

R9-390X has 5.9 TFLOPS

R9-390X has 20 percent higher TFLOPS when compared to Polaris based RX-470.

R9-390's 5.1 TFLOPS defeats RX-480's 5.8 TFLOPS. LOL

X1X is based on Hawaii GCN with Polaris and Vega IP improvements, hence Microsoft selected highest raster to TFLOPS ratio IPC from GCN family. Microsoft focus on improving ROPS, memory bandwidth and lower graphics pipeline latency.

Raja Koduri joined AMD during Hawaii GCN's launch year in 2013 and PC GPU development focused on server bias, hence higher TFLOPS priority over improving classic GPU power i.e. hardware area that matter for games.

Intel Xe GPU also has server bias under Raja Koduri i.e. it's not gamer 1st.

Polaris and Vega R&D deveopments are clusterfu*ked. RX-5700 XT effectively returned to Hawaii GCN and improved upon it.

When Vega 56 at 1700 Mhz with 12 TFLOPS can beat Vega 64 at 1590Mhz with 13 TFLOPS, then there's something wrong with AMD's GPU priorities i.e. it's "Bulldozer" server focus.

It's important to purge out who pushed worst IPC Polaris and Vega projects.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#78  Edited By ronvalencia
Member since 2008 • 28521 Posts
@tormentos said:

@ronvalencia:

Yeah I remember how MGS was 720p on Xbox and 1080p on PS4 with dynamic skies on PS4 while lacking it on Xbox,that is more than 100%,several other games showed gaps as big as that and you say nothing.

Is normal nothing out of this world for the gap in power,even if the spec end like that the series X will have a smaller gap over the PS5 than xox had over the pro,and the PS4 over the Xbox one.

Prior to transparent split render buffer software update, XBO has additional problems with 32 MB ESRAM usage and it's use it's not transparent i.e. manual programming effort to use it.

Later in XBO's life cycle, transparent split render buffer software update was applied and game programmers has frame buffer setup that transparently balance between DDR3 and 32 ESRAM memory pools. This software update doesn't fix lower TFLOPS problem when compared to PS4 GPU's 1.84 TFLOPS i.e. XBO remains inferior like W5000 when compared to R7-265 (PS4 like).

RDR2's resolution scales with GCN TFLOPS from XBO to PS4 then to PS4 Pro, but there's a large resolution jump with X1X.

A reminder, Hawaii XT GCN will crush RX-470 with gaps almost as large PS4 Pro vs X1X.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#79  Edited By ronvalencia
Member since 2008 • 28521 Posts
@xantufrog said:

@BlackShirt20: I don't know how to break it to you, but I tested this systematically a while back and posted the FH4 benchmarks on here to discuss - an old i5 6500 and GTX970 can almost run the game as well as an X1X at 4K (hard to explore far, given the RAM usage is pegged on the poor 970 at 4K), and can run the game BETTER than an X1X at 1080p.

For FM7 at 4K, GTX 970 get KO'ed with wet tracks. FH4 has tone down alpha blending with different developers.

X1X was designed for Digital Foundry's XBO resolution gate i.e. higher rendering resolution for $$ was the priority for X1X.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#80  Edited By ronvalencia
Member since 2008 • 28521 Posts
@fedor said:
@BlackShirt20 said:

@Grey_Eyed_Elf: @fedor:

https://www.techpowerup.com/gpu-specs/xbox-series-x-gpu.c3482

https://www.techpowerup.com/gpu-specs/playstation-5-gpu.c3480

https://www.techpowerup.com/gpu-specs/geforce-rtx-2080-super.c3439

I rest my case.

2 unconfirmed spec lists and an Nvidia GPU who's architecture produces far better performance than RDNA? What is your case exactly?

I notice you also avoided answering any questions or providing any sources to your claims, probably because you simply have no clue what you're talking about. How does it feel being ignorant?

Besides RTX 2080 Ti's 13 to 15 TFLOPS advantage, Turing TU102 has better classic GPU hardware when compared to RX-5700 XT and RX Vega II.

Classic GPU hardware mostly deals with preparing and feeding data to the CU, hence it's geometry-raster I/O which covers items such as geometry input (back face culling), TMU(texture read/write + filter processing) and ROPS (color and z-buffer/depth buffer read/write + blending and MSAA processing).

When you buy TU102 class GPU, it includes industry leading classic GPU hardware to feed it's high TFLOPS power!

Mining workloads bypass the classic GPU hardware.

How does it feel being the real ignorant?

PS, MSI GeForce RTX 2080 Ti GX TRIO has 15 TFLOPS+ with clock speeds +1900Mhz and I'm not overclocking it i.e. it's automated.

Buying TU102 class GPU gives the user the leading edge solution for math power and raster graphics pipeline power.

RX Vega II has the math power with RX-5700 XT level raster graphics pipeline power. It's a half baked solution.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#82 ronvalencia
Member since 2008 • 28521 Posts

@fedor said:
@ronvalencia said:

Besides RTX 2080 Ti's 13 to 15 TFLOPS advantage, Turing TU102 has better classic GPU hardware when compared to RX-5700 XT and RX Vega II.

Classic GPU hardware mostly deals with preparing and feeding data to the CU, hence it's geometry-raster I/O which covers items such as geometry input (back face culling), TMU(texture read/write + filter processing) and ROPS (color and z-buffer/depth buffer read/write + blending and MSAA processing).

When you buy TU102 class GPU, it includes industry leading classic GPU hardware to feed it's high TFLOPS power!

Mining workloads bypass the classic GPU hardware.

How does it feel being the real ignorant?

PS, MSI GeForce RTX 2080 Ti GX TRIO has 15 TFLOPS+ with clock speeds +1900Mhz and I'm not overclocking it i.e. it's automated.

Buying TU102 class GPU gives the user the leading edge solution for math power and raster graphics pipeline power.

RX Vega II has the math power with RX-5700 XT level raster graphics pipeline power. It's a half baked solution.

Hey Ronald! Happy to see ya again. Nothing You wrote there in that wall of spam negates my statement. Fancy that huh, you spamming a wall of crap that has no bearing what so ever on my post. 😄 you're a special guy, Ronald. Still waiting for that MS quote about Vega BTW 😄 have a good one.

PS: The 2080S is a TU104. Maybe if you learned to read you would have caught that. I await your next wall of spam, Buddy.

1. I'm not "Ronald".

2. Your posted argument is the real spam. This topic is about GPU hardware. You can't handle beyond3d forum style debates.

3. Your Vega comment and about me are the real spam in this topic.

https://www.eurogamer.net/articles/digitalfoundry-2017-the-scorpio-engine-in-depth

According to Goossen, some performance optimisations from the upcoming AMD Vega architecture factor into the Scorpio Engine's design, but other features that made it into PS4 Pro - for example, double-rate FP16 processing - do not. However, customisation was extensive elsewhere

  • MS claims Vega IP into Scorpio
  • MS claims one of Vega IP feature wasn't selected i.e. RPM.
  • Full Vega IP has improvments more than just a single RPM feature e.g. ROPS with MB cache design which X1X ROPS has 2MB render cache which doesn't exist for Polaris GCN.
  • I'm glad MS didn't select full Vega IP since it's IPC to inferior to Hawaii GCN.

Avatar image for fedor
Fedor

5973

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#83  Edited By Fedor
Member since 2015 • 5973 Posts

@ronvalencia said:
@fedor said:
@ronvalencia said:

Besides RTX 2080 Ti's 13 to 15 TFLOPS advantage, Turing TU102 has better classic GPU hardware when compared to RX-5700 XT and RX Vega II.

Classic GPU hardware mostly deals with preparing and feeding data to the CU, hence it's geometry-raster I/O which covers items such as geometry input (back face culling), TMU(texture read/write + filter processing) and ROPS (color and z-buffer/depth buffer read/write + blending and MSAA processing).

When you buy TU102 class GPU, it includes industry leading classic GPU hardware to feed it's high TFLOPS power!

Mining workloads bypass the classic GPU hardware.

How does it feel being the real ignorant?

PS, MSI GeForce RTX 2080 Ti GX TRIO has 15 TFLOPS+ with clock speeds +1900Mhz and I'm not overclocking it i.e. it's automated.

Buying TU102 class GPU gives the user the leading edge solution for math power and raster graphics pipeline power.

RX Vega II has the math power with RX-5700 XT level raster graphics pipeline power. It's a half baked solution.

Hey Ronald! Happy to see ya again. Nothing You wrote there in that wall of spam negates my statement. Fancy that huh, you spamming a wall of crap that has no bearing what so ever on my post. 😄 you're a special guy, Ronald. Still waiting for that MS quote about Vega BTW 😄 have a good one.

PS: The 2080S is a TU104. Maybe if you learned to read you would have caught that. I await your next wall of spam, Buddy.

1. I'm not "Ronald".

2. Your posted argument is the real spam. This topic is about GPU hardware.You can't handle beyond3d forum style debates.

You can't handle system wars style debates, Ronald... Guess where you are?

3. Your Vega comment and about me are the real spam in this topic.

https://www.eurogamer.net/articles/digitalfoundry-2017-the-scorpio-engine-in-depth

According to Goossen, some performance optimisations from the upcoming AMD Vega architecture factor into the Scorpio Engine's design, but other features that made it into PS4 Pro - for example, double-rate FP16 processing - do not. However, customisation was extensive elsewhere

  • MS claims Vega IP into Scorpio
  • MS claims one of Vega IP feature wasn't selected i.e. RPM.
  • Full Vega IP has improvments more than just a single RPM feature e.g. ROPS with MB cache design which X1X ROPS has 2MB render cache which doesn't exist for Polaris GCN.
  • I'm glad MS didn't select full Vega IP since it's IPC to inferior to Hawaii GCN.

From your link:

"Similarly, Scorpio's Radeon graphics core has features from AMD's latest Polaris architecture - but there is no equivalent part to it in the PC space."

"We have Polaris features in Scorpio that we've picked up. Some of the big ones are delta colour compression, so that helps us out on our bandwidth, both for 4K textures and 4K rendering solutions to achieve that," says Goossen."

"but other features that made it into PS4 Pro - for example, double-rate FP16 processing - do not."

Your quote is the only time Vega is mentioned in the entire article and it's a throwaway line. LMAO!!!

So it's Polaris... Also, in fact it has less Vega features than PS4 pro. Basically it's insignificant, and you know this, don't you?

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#84  Edited By ronvalencia
Member since 2008 • 28521 Posts

@fedor said:
@ronvalencia said:

1. I'm not "Ronald".

2. Your posted argument is the real spam. This topic is about GPU hardware.You can't handle beyond3d forum style debates.

You can't handle system wars style debates, Ronald... Guess where you are?

3. Your Vega comment and about me are the real spam in this topic.

https://www.eurogamer.net/articles/digitalfoundry-2017-the-scorpio-engine-in-depth

According to Goossen, some performance optimisations from the upcoming AMD Vega architecture factor into the Scorpio Engine's design, but other features that made it into PS4 Pro - for example, double-rate FP16 processing - do not. However, customisation was extensive elsewhere

  • MS claims Vega IP into Scorpio
  • MS claims one of Vega IP feature wasn't selected i.e. RPM.
  • Full Vega IP has improvments more than just a single RPM feature e.g. ROPS with MB cache design which X1X ROPS has 2MB render cache which doesn't exist for Polaris GCN.
  • I'm glad MS didn't select full Vega IP since it's IPC to inferior to Hawaii GCN.

From your link:

"Similarly, Scorpio's Radeon graphics core has features from AMD's latest Polaris architecture - but there is no equivalent part to it in the PC space."

"We have Polaris features in Scorpio that we've picked up. Some of the big ones are delta colour compression, so that helps us out on our bandwidth, both for 4K textures and 4K rendering solutions to achieve that," says Goossen."

"but other features that made it into PS4 Pro - for example, double-rate FP16 processing - do not."

So it's Polaris... Also, in fact it has less Vega features than PS4 pro. Basically it's insignificant, and you know this, don't you?

1. I'm not "Ronald".

2. Your posted argument is the real spam. This topic is about GPU hardware. You can't handle beyond3d forum style debates.

You can't handle system wars style debates, Ronald... Guess where you are?

3. Your Vega comment and about me are the real spam in this topic.

Full quote without your omission

"We have Polaris features in Scorpio that we've picked up. Some of the big ones are delta colour compression, so that helps us out on our bandwidth, both for 4K textures and 4K rendering solutions to achieve that," says Goossen. "It's typically quite easy for the developers to integrate and then also more transparently we picked up some geometry and quad-scheduling improvements AMD has done in the Polaris architecture."

According to Goossen, some performance optimisations from the upcoming AMD Vega architecture factor into the Scorpio Engine's design, but other features that made it into PS4 Pro - for example, double-rate FP16 processing - do not. However, customisation was extensive elsewhere. Microsoft's GPU command processor implementation of DX12 has provided big wins for Xbox One developers, and it's set for expansion in Scorpio.

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

From Polaris Architecture

  • delta colour compression
  • some geometry
  • quad-scheduling
  • 2MB L2 cache (from Polaris 10/20)

Polaris Architecture doesn't include Render Backend improvments hence it's old school like the previous GCN generation!

Vega 64's improvments list

https://www.anandtech.com/show/11717/the-amd-radeon-rx-vega-64-and-56-review/3

  • Higher clocks (X1X GPU's 6 TFLOPS and TDP is about half Vega 64's 12.5 TFLOPS and 295 watts, Polaris 10/20/30 XT can't scale to Vega 64 within 295 watts budget)
  • Double rate FP16 math (Rapid Packed Math) (X1X doesn't have this)
  • HBM2 (X1X doesn't have this, X1X has 384 bit GDDR5-6800 instead, Polaris 10/20/30 doesn't have 384 bit GDDR5)
  • New memory page management for the high-bandwidth cache controller
  • Tiled rasterization (Draw Stream Binning Rasterizer) (XBO has software tiled raster support with slow 32MB ESRAM, X1X ROPS has 2 MB render cache for micro-tile renderer)
  • Increased ROP efficiency via L2 cache (X1X ROPS has 2MB render cache instead, missing on Polaris Architecture)
  • Improved geometry engine
  • Primitive shading for even faster triangle culling
  • Direct3D feature level 12_1 graphics features
  • Improved display controllers

-----

Microsoft offically didn't select Vega Rapid Packed Math for X1X GPU just as AMD's Vega 24 didn't include Vega RPM!

Radeon HD R9-290X/R9-390X’s aging RBE (Render Backend which contains ROPS) vs X1X GPU comparison.

https://www.slideshare.net/DevCentralAMD/gs4106-the-amd-gcn-architecture-a-crash-course-by-layla-mah

16 RBE (Render Backends) units with each RBE unit containing 4 ROPS units. Each RBE units has 24 KB cache.

24 bytes x 16 = 384 KB.

VS

Xbox One X’s RBE/ROPS has 2048 KB or 2 MB render cache.

Xbox One X’s RBE has 256 KB. 8x RBE = 2048 KB (or 2 MB) render cache.

Xbox One X will hold more rendering data on the chip when compared to Radeon HD 7970, R9-290X/R9-390X and Polaris 10/20/30!

X1X GPU has atleast 4MB cache from 2MB L2 cache (for TMU) + 2MB render cache (for ROPS) similar to Vega 56/64's 4MB L2 cache (for ROPS and TMU)!

Render Back Ends includes it’s own math operations i.e. “Logic Operations” and Blending which on X1X ROPS has 2MB render cache!

These features are missing from Polaris Architecture.

https://www.eurogamer.net/articles/digitalfoundry-2017-the-scorpio-engine-in-depth

We have hardware techniques for making checkerboarding very efficient.

....

Andrew Goossen tells us that the GPU supports extensions that allow depth and ID buffers to be efficiently rendered at full native resolution, while colour buffers can be rendered at half resolution with full pixel shader efficiency

X1X GPU includes

  • Hardware checkerboard (like PS4 Pro)
  • Variable Rate Shading like feature where geometry is rendered at native resolution with lower shading resolution. Read https://developer.nvidia.com/vrworks/graphics/variablerateshading
  • Microsoft has patented Variable Rate Shading. Variable Rate Shading feature appeared on Turing and RDNA 2.

These features are missing from the Polaris Architecture.

Avatar image for fedor
Fedor

5973

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#85  Edited By Fedor
Member since 2015 • 5973 Posts

Yes, Ron. It's Polaris based. Your own article supports that. It having insignificant features from Vega is meaningless. The Pro is more Vega than Scorpio is. Stop making a fool of yourself.

My original post to black shirt was 100% accurate. If You want to interject your opinion again please do it within the context of my post. And get a sense of humor, for the love of everything Holy.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#86 ronvalencia
Member since 2008 • 28521 Posts

@fedor said:

Yes, Ron. It's Polaris based. Your own article supports that. It having insignificant features from Vega is meaningless. The Pro is more Vega than Scorpio is. Stop making a fool of yourself.

My original post to black shirt was 100% accurate. If You want to interject your opinion again please do it within the context of my post. And get a sense of humor, for the love of everything Holy.

Vega RPM is real insignificant feature in most current gen games while ROPS with multi-MB cache improvments has higher usage!

Try again.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#87 ronvalencia
Member since 2008 • 28521 Posts

@fedor said:

If you continue Scorpio doesn't have Vega debate, I'll apply precedent applied by Juub1990. Removing you from this topic.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#88  Edited By ronvalencia
Member since 2008 • 28521 Posts

@SecretPolice said:

Good stuff, Captain Ron.... lol :P

I haven't done anything, except report news articles. I made my comments on this issue and I welcome other comments on this topic's subject matter.

Avatar image for fedor
Fedor

5973

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#89  Edited By Fedor
Member since 2015 • 5973 Posts

@ronvalencia: You posted a link that supported my arguments, not yours. Its staggering that you don't realize this. Scorpio is Polaris based (as your posted article supports), having insignificant features from Vega doesn't change this fact. Spamming word salad all over the place won't change it either.

"Vega RPM is real insignificant feature in most current gen games while ROPS with multi-MB cache improvments has higher usage!"

LMAO!!!

Once again, My original post to black shirt was 100% accurate. If you want to interject your opinion again please do it within the context of my post.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#90  Edited By ronvalencia
Member since 2008 • 28521 Posts

@fedor said:

@ronvalencia: You posted a link that supported my arguments, not yours. Its staggering that you don't realize this. Scorpio is Polaris based (as your posted article supports), having insignificant features from Vega doesn't change this fact. Spamming word salad all over the place won't change it either. Once again, My original post to black shirt was 100% accurate. If you want to interject your opinion again please do it within the context of my post.

1. Your argument that demishes Microsoft claim on both Polaris and Vega IPs are patently false.

2. Without Vega RPM usage, Vega CU acts like Polaris CU i.e. it's like any other GCN with degraded IPC when compared to Hawaii based R9-390X. The large gap between PS4 Pro and X1X indicates Hawaii GCN+ like IPC vs PS4 Pro's Polaris/Vega IPC. R9-390X OC (6 TFLOPS) can deliver similar large gap (>49 percent) against RX-470 (4.9 TFLOPS).

R9-390X OC has issues with heavy blends and alpha effects which X1X GPU has mitigated.

3. ROPS with hyper fast multi-MB cache is one of secret sauce for NVIDIA's Maxwell/Pascal GPUs. NVIDIA updated L2 cache design for Turing by doubling the storage size i.e. 2MB (GP104) to 4MB (TU104), 3MB (GP102) to 6MB (TU102). On GPUs, NVIDIA has higher design credibility when compared to AMD's RTG.

Avatar image for fedor
Fedor

5973

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#91  Edited By Fedor
Member since 2015 • 5973 Posts

@ronvalencia said:
@fedor said:

@ronvalencia: You posted a link that supported my arguments, not yours. Its staggering that you don't realize this. Scorpio is Polaris based (as your posted article supports), having insignificant features from Vega doesn't change this fact. Spamming word salad all over the place won't change it either. Once again, My original post to black shirt was 100% accurate. If you want to interject your opinion again please do it within the context of my post.

1. Your argument that demishes Microsoft claim on both Polaris and Vega IP is patently false.

My argument is it's Polaris based, which is true. How You do not understand this is beyond me. MS doesn't support you either. You know it's true too, that's why you cherry picked a quote from that article even though the entire write up supports it being Polaris based. Why did you even post it? Did you think you could sneak it by and I wouldn't read it?

Here I'll make it easy for you, which architecture does Scorpio share more in common with? Vega or Polaris?

The answer is Polaris, far more. You know this, trying to argue against it would only prove you a liar. So stop the nonsense (I know you won't).

2. Without Vega RPM usage, Vega CU acts like Polaris CU i.e. it's like any other GCN with degraded IPC when compared to Hawaii based R9-390X.

3. ROPS with hyper fast multi-MB cache is one of secret sauce for NVIDIA's Maxwell/Pascal GPUs. NVIDIA updated L2 cache design for Turing by doubling the storage size i.e. 2MB (GP104) to 4MB (TU104), 3MB (GP102) to 6MB (TU102). I believe NVIDIA's design direction and AMD/MS should follow NVIDIA.

Once again, My original post to black shirt was 100% accurate. If you want to interject your opinion again please do it within the context of my post.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#92 ronvalencia
Member since 2008 • 28521 Posts

@fedor said:
@ronvalencia said:
@fedor said:

@ronvalencia: You posted a link that supported my arguments, not yours. Its staggering that you don't realize this. Scorpio is Polaris based (as your posted article supports), having insignificant features from Vega doesn't change this fact. Spamming word salad all over the place won't change it either. Once again, My original post to black shirt was 100% accurate. If you want to interject your opinion again please do it within the context of my post.

1. Your argument that demishes Microsoft claim on both Polaris and Vega IP is patently false.

2. Without Vega RPM usage, Vega CU acts like Polaris CU i.e. it's like any other GCN with degraded IPC when compared to Hawaii based R9-390X.

3. ROPS with hyper fast multi-MB cache is one of secret sauce for NVIDIA's Maxwell/Pascal GPUs. NVIDIA updated L2 cache design for Turing by doubling the storage size i.e. 2MB (GP104) to 4MB (TU104), 3MB (GP102) to 6MB (TU102). I believe NVIDIA's design direction and AMD/MS should follow NVIDIA.

Once again, My original post to black shirt was 100% accurate. If you want to interject your opinion again please do it within the context of my post.

1. Your argument that demishes Microsoft claim on both Polaris and Vega IP is patently false.

The large gap between PS4 Pro and X1X indicates Hawaii GCN+ like IPC vs PS4 Pro's degraded Polaris/Vega IPC.

R9-390X OC (6 TFLOPS) can deliver similar large gap (>49 percent) against RX-470 (4.9 TFLOPS, let alone the lesser 4.2 TFLOPS enabled PS4 Pro GPU).

2. Without Vega RPM usage, Vega CU acts like Polaris CU i.e. it's like any other GCN with degraded IPC when compared to Hawaii based R9-390X.

The large gap between PS4 Pro and X1X indicates Hawaii GCN+ like IPC vs PS4 Pro's Polaris/Vega IPC. R9-390X OC (6 TFLOPS) can deliver similar large gap (>49 percent) against RX-470 (4.9 TFLOPS). You can't deny gaming results.

R9-390X OC has issues with heavy blends and alpha effects which X1X GPU has mitigated. ROPS with fast multi-MB render cache is NOT insignificant. You can't deny gaming results.

3. ROPS with hyper fast multi-MB cache is one of secret sauce for NVIDIA's Maxwell/Pascal GPUs. NVIDIA updated L2 cache design for Turing by doubling the storage size i.e. 2MB (GP104) to 4MB (TU104), 3MB (GP102) to 6MB (TU102). On GPUs, NVIDIA has higher design credibility when compared to AMD's RTG.

4. You are spamming my Xbox Scarlet topic with your "Scorpio is Polaris based with insignificant features from Vega" arguments.

5. Your insignificant features assertion are debatable and unproven.

Avatar image for fedor
Fedor

5973

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#93 Fedor
Member since 2015 • 5973 Posts

@ronvalencia said:
@fedor said:
@ronvalencia said:
@fedor said:

@ronvalencia: You posted a link that supported my arguments, not yours. Its staggering that you don't realize this. Scorpio is Polaris based (as your posted article supports), having insignificant features from Vega doesn't change this fact. Spamming word salad all over the place won't change it either. Once again, My original post to black shirt was 100% accurate. If you want to interject your opinion again please do it within the context of my post.

1. Your argument that demishes Microsoft claim on both Polaris and Vega IP is patently false.

2. Without Vega RPM usage, Vega CU acts like Polaris CU i.e. it's like any other GCN with degraded IPC when compared to Hawaii based R9-390X.

3. ROPS with hyper fast multi-MB cache is one of secret sauce for NVIDIA's Maxwell/Pascal GPUs. NVIDIA updated L2 cache design for Turing by doubling the storage size i.e. 2MB (GP104) to 4MB (TU104), 3MB (GP102) to 6MB (TU102). I believe NVIDIA's design direction and AMD/MS should follow NVIDIA.

Once again, My original post to black shirt was 100% accurate. If you want to interject your opinion again please do it within the context of my post.

1. Your argument that demishes Microsoft claim on both Polaris and Vega IP is patently false.

The large gap between PS4 Pro and X1X indicates Hawaii GCN+ like IPC vs PS4 Pro's degraded Polaris/Vega IPC.

R9-390X OC (6 TFLOPS) can deliver similar large gap (>49 percent) against RX-470 (4.9 TFLOPS, let alone the lesser 4.2 TFLOPS enabled PS4 Pro GPU).

2. Without Vega RPM usage, Vega CU acts like Polaris CU i.e. it's like any other GCN with degraded IPC when compared to Hawaii based R9-390X.

The large gap between PS4 Pro and X1X indicates Hawaii GCN+ like IPC vs PS4 Pro's Polaris/Vega IPC. R9-390X OC (6 TFLOPS) can deliver similar large gap (>49 percent) against RX-470 (4.9 TFLOPS). You can't deny gaming results.

R9-390X OC has issues with heavy blends and alpha effects which X1X GPU has mitigated. ROPS with fast multi-MB render cache is NOT insignificant. You can't deny gaming results.

3. ROPS with hyper fast multi-MB cache is one of secret sauce for NVIDIA's Maxwell/Pascal GPUs. NVIDIA updated L2 cache design for Turing by doubling the storage size i.e. 2MB (GP104) to 4MB (TU104), 3MB (GP102) to 6MB (TU102). On GPUs, NVIDIA has higher design credibility when compared to AMD's RTG.

4. You are spamming my Xbox Scarlet topic with your "Scorpio is Polaris based with insignificant features from Vega" arguments.

5. Your insignificant features assertion are debatable and unproven.

My argument is it's Polaris based, which is true. How You do not understand this is beyond me. MS doesn't support you either. You know it's true too, that's why you cherry picked a quote from that article even though the entire write up supports it being Polaris based. Why did you even post it? Did you think you could sneak it by and I wouldn't read it?

Here I'll make it easy for you, which architecture does Scorpio share more in common with? Vega or Polaris?

The answer is Polaris, far more. You know this, trying to argue against it would only prove you a liar. So stop the nonsense (I know you won't).

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#94 ronvalencia
Member since 2008 • 28521 Posts

@fedor said:
@ronvalencia said:

1. Your argument that demishes Microsoft claim on both Polaris and Vega IP is patently false.

The large gap between PS4 Pro and X1X indicates Hawaii GCN+ like IPC vs PS4 Pro's degraded Polaris/Vega IPC.

R9-390X OC (6 TFLOPS) can deliver similar large gap (>49 percent) against RX-470 (4.9 TFLOPS, let alone the lesser 4.2 TFLOPS enabled PS4 Pro GPU).

2. Without Vega RPM usage, Vega CU acts like Polaris CU i.e. it's like any other GCN with degraded IPC when compared to Hawaii based R9-390X.

The large gap between PS4 Pro and X1X indicates Hawaii GCN+ like IPC vs PS4 Pro's Polaris/Vega IPC. R9-390X OC (6 TFLOPS) can deliver similar large gap (>49 percent) against RX-470 (4.9 TFLOPS). You can't deny gaming results.

R9-390X OC has issues with heavy blends and alpha effects which X1X GPU has mitigated. ROPS with fast multi-MB render cache is NOT insignificant. You can't deny gaming results.

3. ROPS with hyper fast multi-MB cache is one of secret sauce for NVIDIA's Maxwell/Pascal GPUs. NVIDIA updated L2 cache design for Turing by doubling the storage size i.e. 2MB (GP104) to 4MB (TU104), 3MB (GP102) to 6MB (TU102). On GPUs, NVIDIA has higher design credibility when compared to AMD's RTG.

4. You are spamming my Xbox Scarlet topic with your "Scorpio is Polaris based with insignificant features from Vega" arguments.

5. Your insignificant features assertion are debatable and unproven.

My argument is it's Polaris based, which is true. How You do not understand this is beyond me. MS doesn't support you either. You know it's true too, that's why you cherry picked a quote from that article even though the entire write up supports it being Polaris based. Why did you even post it? Did you think you could sneak it by and I wouldn't read it?

Here I'll make it easy for you, which architecture does Scorpio share more in common with? Vega or Polaris?

The answer is Polaris, far more. You know this, trying to argue against it would only prove you a liar. So stop the nonsense (I know you won't).

Again,

1. Your argument that demishesMicrosoft claim on both Polaris and Vega IP is patently false.

The large gap between PS4 Pro and X1X indicates Hawaii GCN+ like IPC vs PS4 Pro's degraded Polaris/Vega IPC.

R9-390X OC (6 TFLOPS) can deliver similar large gap (>49 percent) against RX-470 (Polaris 4.9 TFLOPS, let alone the lesser 4.2 TFLOPS enabled PS4 Pro GPU).

You can't deny gaming results.

2. Without Vega RPM usage, Vega CU acts like Polaris CU i.e. it's like any other GCN with degraded IPC when compared to Hawaii based R9-390X.

The large gap between PS4 Pro and X1X indicates Hawaii GCN+ like IPC vs PS4 Pro's Polaris/Vega IPC. R9-390X OC (6 TFLOPS) can deliver similar large gap (>49 percent) against RX-470 (4.9 TFLOPS). You can't deny gaming results.

R9-390X OC has issues with heavy blends and alpha effects which X1X GPU has mitigated. ROPS with fast multi-MB render cache is NOT insignificant.

You can't deny gaming results.

3. ROPS with hyper fast multi-MB cache is one of secret sauce for NVIDIA's Maxwell/Pascal GPUs. NVIDIA updated L2 cache design for Turing by doubling the storage size i.e. 2MB (GP104) to 4MB (TU104), 3MB (GP102) to 6MB (TU102). On GPUs, NVIDIA has higher design credibility when compared to AMD's RTG.

4. You are spamming my Xbox Scarlet topic with your "Scorpio is Polaris based with insignificant features from Vega" arguments.

5. Your insignificant features assertion are debatable and unproven.

Avatar image for fedor
Fedor

5973

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#95 Fedor
Member since 2015 • 5973 Posts

@ronvalencia: LMAO!!! You really try desperately to clutter any argument you get into. It's the weakest defense mechanism I've ever seen.

"5. Your insignificant features assertion are debatable and unproven."

Yet you haven't disproved them in context of Scorpio, funny.

"4. You are spamming my Xbox Scarlet topic with your "Scorpio is Polaris based with insignificant features from Vega" arguments."

Responding to you isn't spamming. If you want me to stop talking to you, just say the word.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#96  Edited By ronvalencia
Member since 2008 • 28521 Posts

@fedor said:

@ronvalencia: LMAO!!! You really try desperately to clutter any argument you get into. It's the weakest defense mechanism I've ever seen.

"5. Your insignificant features assertion are debatable and unproven."

Yet you haven't disproved them in context of Scorpio, funny.

"4. You are spamming my Xbox Scarlet topic with your "Scorpio is Polaris based with insignificant features from Vega" arguments."

Responding to you isn't spamming. If you want me to stop talking to you, just say the word.

4. You started

@fedor said:

Hey Ronald! Happy to see ya again. Nothing You wrote there in that wall of spam negates my statement. Fancy that huh, you spamming a wall of crap that has no bearing what so ever on my post. 😄 you're a special guy, Ronald. Still waiting for that MS quote about Vega BTW 😄 have a good one.

That's spamming my thread.

The only wall of crap comes from you.

https://www.gamespot.com/forums/system-wars-314159282/systemwars-survival-guide-and-code-of-conduct-33470358/

Flaming, spamming, personal attacks and thread derailing are not allowed and will be moderated accordingly. Do not use the word retard, or any variation of retard (ex; libtard), to demean or attack anyone. It will be moderated.

You derailed my topic.

@fedor said:

PS: The 2080S is a TU104. Maybe if you learned to read you would have caught that. I await your next wall of spam, Buddy.

I already know 2080 Super Edition has TU104 with six GPCs layout.

https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/technologies/turing-architecture/NVIDIA-Turing-Architecture-Whitepaper.pdf Page 64 of 86

What's your point?

5. Why should I disprove X1X's Polaris IP when it has some good improvments e.g. compensate for missing 128 bit memory bus from Hawaii XT's 512 bit memory bus design?

Polaris IP has better compression bandwdith again when compared to Vega IP.

Effective physical bandwdith vs compressed bandwdith gain

RX 580: 47 percent gain.

Vega 64: 40 percent gain.

Vega 56: 41 percent gain.

Vega 64 2GT: 35 percent gain.

----

Scaling RX-580's 256 bit bus effective phyical memory bandwdith witth it's compression factor to X1X's 384 bit bus yields 361 GB/s.

R9-390X's 384 GB/s is reduced to 315.6 GB/s effective phyical memory bandwdith.

Conclusion: MS's Polaris IP DCC selection has mitigated and improve upon the missing 128 bit memory bus from Hawaii XT's 512 bit memory bus design.

MS has to pick the right IP for $$.

The next improvments are with ROPS and 2MB render cache. Vega ROPS has multi-MB L2 cache which is missing with Polaris ROPS.

My argument for Scorpio has been Hawaii GCN with Polaris/Vega improvements.

Again, your insignificant features arguments are debatable and unproven.

Avatar image for ronvalencia
ronvalencia

28521

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#97 ronvalencia
Member since 2008 • 28521 Posts

@BlackShirt20 said:

@ronvalencia: DF confirmed this a week ago. They said, this new console is a absolute beast.

The statement "New console is a absolute beast" is not specfic.

Avatar image for fedor
Fedor

5973

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#98  Edited By Fedor
Member since 2015 • 5973 Posts

@ronvalencia: "2 unconfirmed spec lists and an Nvidia GPU who's architecture produces far better performance than RDNA? What is your case exactly?"

I notice you also avoided answering any questions or providing any sources to your claims, probably because you simply have no clue what you're talking about. How does it feel being ignorant?"

This is how it started, actually. My post to someone else, (that's completely accurate) you then jumped in with your typical tangent.

"You derailed my topic."

Lmao!!! What an incredible reach. It's hilarious one little quote of mine (which you still haven't disproved, you can't) got you all up in arms. In fact you posted an article proving me right and have been having a meltdown ever since. No giant wall of text will change that, Ron. My post to Black shirt was 100% true, you should have just kept quiet and not looked like a lunatic.

You "derailed" your own thread.

Avatar image for tormentos
tormentos

29646

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#99  Edited By tormentos
Member since 2003 • 29646 Posts

@BlackShirt20:

Is funny how 20% is now HUGE to lemmings on 2013 40% was nothing.

But considering that on the 360 generation many lemmings make a huge deal over a 720p vs 640p resolution difference it doesn't really surprise me at all.

Lemmings 40% is nothing when the PS commands,when the Xbox command 20% is HUGE.🤷🤦

Avatar image for n64dd
N64DD

12514

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#100 N64DD
Member since 2015 • 12514 Posts

Ron, do you actually play games?