@ronvalencia: Old video and shit game. What is the point of this thread?
1. That shit game wasn't a walk in the park for GTX 980 Ti.
2. The same Digital Foundry style argument as Xbox Scorpio reveal i.e. what PC hardware level to run the same console game that used a new GPU feature.
3. Hints for incoming Vega based GPUs for PC and effectiveness for double rate Fp16 booster.
@ronvalencia: Old video and shit game. What is the point of this thread?
1. That shit game wasn't a walk in the park for GTX 980 Ti.
2. The same Digital Foundry style argument as Xbox Scorpio reveal i.e. what PC hardware level to run the same console game that used a new GPU feature.
3. Hints for incoming Vega based GPUs for PC and effectiveness for double rate Fp16 booster.
@ronvalencia: Old video and shit game. What is the point of this thread?
1. That shit game wasn't a walk in the park for GTX 980 Ti.
2. The same Digital Foundry style argument as Xbox Scorpio reveal i.e. what PC hardware level to run the same console game that used a new GPU feature.
3. Hints for incoming Vega based GPUs for PC and effectiveness for double rate Fp16 booster.
I've heard 'incoming' Vega for so many years now
And the 5XX series is still using polaris... smh
RX-580 "Polaris 20 XTX" stays within Tonga's price segment i.e. it's just rename with minor overclock and a large memory bandwidth bottleneck..
It has local multi-player option which is good for BBQ type party games.
"Little Street Racer" is similar to Mantis Burn Racing.
I have been installing games with local multi-player features on my laptop for future long weekend BBQ events. The usage factor is similar to Switch's local multi-player modes.
@ronvalencia: Old video and shit game. What is the point of this thread?
1. That shit game wasn't a walk in the park for GTX 980 Ti.
2. The same Digital Foundry style argument as Xbox Scorpio reveal i.e. what PC hardware level to run the same console game that used a new GPU feature.
3. Hints for incoming Vega based GPUs for PC and effectiveness for double rate Fp16 booster.
I've heard 'incoming' Vega for so many years now
And the 5XX series is still using polaris... smh
RX-580 "Polaris 20 XTX" stays within Tonga's price segment i.e. it's just rename with minor overclock and a large memory bandwidth bottleneck..
VEGA is coming.The 500 series is not vega.
Vega will be called Rage(3584SP ,4GB HBM2),Rage X(3584SP, 8GB HBM2),Rage X Fury(4096SP, 8GB HBM2) and Rage X2 Fury(2 cards on 1 pcb)
Sometimes I get in trouble when I talk about Sony too much, but, the choice they made on PS4 Pro, I totally get that choice, from their perspective and what they wanted to go do. I've said it publicly and I've said it privately, I think they've built a good 2016 PS4 Pro. With the silicon that was available, they picked the parts that made sense to go and put together a console in 2016.
But the point on not wanting framerate to drop when you go to the higher box, right, if the developers want to push resolution, to say to the player 'here you bought this higher-end console, let me show you higher-end resolution,' you don't want the framerate to drop. And that was something we didn't think we could deliver with the silicon that was available in 2016.
So some of it was time, as certain things come down in price -- some of them not as quickly as we would like. And some of it was hardware capability from our silicon partners, that allowed us to go do that in 2017.
Phil Spencer statement for AMD's year 2017 silicon on Scorpio.
Polaris 10's Fp16 feature to be enabled in Shader Model 6
While support for retail DXIL has been available across Windows 10 for a while now, production drivers expose the new capability via 'experimental mode' which requires the Windows 10 Creators Update (actually any Windows 10 Insider Preview Build 15007 or later).
To use a driver in Windows 10 experimental mode, enable the Developer Mode setting.
...
The following vendors provide drivers with hardware GPU support for DXIL:
NVIDIA's new r381 drivers (r381.65 and later) provide experimental mode support for DXIL 1.0 and Shader Model 6.0. Here are the release notes, and a download link.
AMD's latest driver with support for DXIL 1.0 and Shader Model 6 in experimental mode is Radeon Software Crimson ReLive Edition 17.4.2.
Hardware Drivers:
The following vendors provide drivers with hardware GPU support for DXIL:
NVIDIA r378 drivers (r378.49 and later) provide experimental mode support for DXIL and shader model 6. This is an early beta version to enable developers to try out DXIL and the new shader model 6 features – Wave Math and int64. Only DXIL version 0.7 (beta) is accepted by the r378 driver. Experimental mode support for DXIL v1.0 will be provided in a future driver release.
AMD has released its first driver with experimental mode support for DXIL and Shader Model 6: Radeon Software Crimson ReLive Edition 17.3.1, which can be downloaded from support.amd.com It uses and requires DXIL v1.0. Using this driver, all of the execution tests pass except for the 2 wave intrinsics tests. The next driver release, 17.3.2, has fixes so all of the execution tests pass.
1. That shit game wasn't a walk in the park for GTX 980 Ti.
2. The same Digital Foundry style argument as Xbox Scorpio reveal i.e. what PC hardware level to run the same console game that used a new GPU feature.
3. Hints for incoming Vega based GPUs for PC and effectiveness for double rate Fp16 booster.
I've heard 'incoming' Vega for so many years now
And the 5XX series is still using polaris... smh
RX-580 "Polaris 20 XTX" stays within Tonga's price segment i.e. it's just rename with minor overclock and a large memory bandwidth bottleneck..
VEGA is coming.The 500 series is not vega.
Vega will be called Rage(3584SP ,4GB HBM2),Rage X(3584SP, 8GB HBM2),Rage X Fury(4096SP, 8GB HBM2) and Rage X2 Fury(2 cards on 1 pcb)
Polaris 10 and 20 hasn't solved Pixel Engine to memory controller path bottleneck with Pixel Engine to L2 cache design.
The shit game wasn't a walk in the park for GTX 980 Ti. The goal for benchmarks are to assess common workload performance on different platforms not for art work critic which is subjective.
Vega Rage SKUs are from http://wccftech.com/vega-teaser-slides-leak-nda/ and it's wccftech.
R9-Fury X, R9-Fury Pro, R9-Fury Nano and R9-Fury Pro Duo has to be replaced.
Looks like PC gamers are going to be in 3rd place powerwise, with tears streaming down their faces as they play the small selection games publishers decide to toss them from much better performing and superior consoles.
Even the 1080ti is weak, compared to PS4 Pro and Scorpio. But at least it shows a nice number on 3Dmark, which is all PC gamers care about.
Vega will be called Rage(3584SP ,4GB HBM2),Rage X(3584SP, 8GB HBM2),Rage X Fury(4096SP, 8GB HBM2) and Rage X2 Fury(2 cards on 1 pcb)
Polaris 10 and 20 hasn't solved Pixel Engine to memory controller path bottleneck with Pixel Engine to L2 cache design.
The shit game wasn't a walk in the park for GTX 980 Ti. The goal for benchmarks are to assess common workload performance on different platforms not for art work critic which is subjective.
Vega Rage SKUs are from http://wccftech.com/vega-teaser-slides-leak-nda/ and it's wccftech.
R9-Fury X, R9-Fury Pro, R9-Fury Nano and R9-Fury Pro Duo has to be replaced.
@ronvalencia: Does it matter lol GTX980TI plays better 90% of other games. Who cares when PS4PRO does something right meanhwile underperforming in most cases compared to GTX980Ti. From a consumer perspective I care who plays Battlfield 1 better, not some micromachines's game. Does PS4 plays Battlefield at 4k, ultra quality and 60 fps ? Nop.
This is s shitty non impressive game run at least Horizon in true 4k using FP16 and i eat all that crap.
FP16 will not double your power FP16 can't be use on every part of the game.
Not even i make a thread about this and i am a cow.
But the GTX 980 ti can trade blows with the GTX 1080 and this clearly shows the PS4 Pro does better. So since Ron keeps insisting the Scorpio is between GTX 1070 and GTX 1080, that must mean the PS4 Pro is better than the Scorpio... (since one single tech demo vs one game is enough proof to prove Scorpio performance)...
Looks like PC gamers are going to be in 3rd place powerwise, with tears streaming down their faces as they play the small selection games publishers decide to toss them from much better performing and superior consoles.
Even the 1080ti is weak, compared to PS4 Pro and Scorpio. But at least it shows a nice number on 3Dmark, which is all PC gamers care about.
@ronvalencia: Does it matter lol GTX980TI plays better 90% of other games. Who cares when PS4PRO does something right meanhwile underperforming in most cases compared to GTX980Ti. From a consumer perspective I care who plays Battlfield 1 better, not some micromachines's game. Does PS4 plays Battlefield at 4k, ultra quality and 60 fps ? Nop.
When a GPU design includes making existing 3D engine faster with minimal bottlenecks, the result will be different e.g. Scorpio's Forza 6 example without platform specific optimizations.
PS4 Pro's GPU wasn't designed to run existing 3D engines faster with minimal bottlenecks.
This is s shitty non impressive game run at least Horizon in true 4k using FP16 and i eat all that crap.
FP16 will not double your power FP16 can't be use on every part of the game.
Not even i make a thread about this and i am a cow.
But the GTX 980 ti can trade blows with the GTX 1080 and this clearly shows the PS4 Pro does better. So since Ron keeps insisting the Scorpio is between GTX 1070 and GTX 1080, that must mean the PS4 Pro is better than the Scorpio... (since one single tech demo vs one game is enough proof to prove Scorpio performance)...
I don't support reference GTX 980 Ti can trade blows with the reference GTX 1080 FE view point.
My MSI Gaming GTX 980 Ti factory OC is not like super-clock editions with higher paper spec guarantee numbers.
GTX 980 Ti factory OC with paper spec higher than MSI Gaming GTX 980 Ti factory OC paper spec examples are
1. MSI GeForce GTX 980 Ti Lightning
2. ASUS GTX 980 Ti STRIX OC
3. Zotac GeForce GTX 980 Ti Amp Extreme <-------- This is a very fast GTX 980 Ti card and very close to GTX 1080 FE.
@asylumni said:
So since Ron keeps insisting the Scorpio is between GTX 1070 and GTX 1080, that must mean the PS4 Pro is better than the Scorpio
Don't put words in my mouth.
My MSI Gaming GTX 980 Ti proven to be faster than PS4 Pro with a game that used Vega style double rate Fp16 and native Fp16.
Zotac GeForce GTX 980 Ti Amp Extreme is faster than my mild factory OC MSI Gaming GTX 980 Ti.
This is s shitty non impressive game run at least Horizon in true 4k using FP16 and i eat all that crap.
FP16 will not double your power FP16 can't be use on every part of the game.
Not even i make a thread about this and i am a cow.
But the GTX 980 ti can trade blows with the GTX 1080 and this clearly shows the PS4 Pro does better. So since Ron keeps insisting the Scorpio is between GTX 1070 and GTX 1080, that must mean the PS4 Pro is better than the Scorpio... (since one single tech demo vs one game is enough proof to prove Scorpio performance)...
I don't support reference GTX 980 Ti can trade blows with the reference GTX 1080 FE view point.
My MSI Gaming GTX 980 Ti factory OC is not like super-clock editions with higher paper spec guarantee numbers.
GTX 980 Ti factory OC with paper spec higher than MSI Gaming GTX 980 Ti factory OC paper spec examples are
1. MSI GeForce GTX 980 Ti Lightning
2. ASUS GTX 980 Ti STRIX OC
3. Zotac GeForce GTX 980 Ti Amp Extreme <-------- This is a very fast GTX 980 Ti card and very close to GTX 1080 FE.
@asylumni said:
So since Ron keeps insisting the Scorpio is between GTX 1070 and GTX 1080, that must mean the PS4 Pro is better than the Scorpio
Don't words in my mouth.
My MSI Gaming GTX 980 Ti proven to be faster than PS4 Pro with a game that used Vega style double rate Fp16 and native Fp16.
Zotac GeForce GTX 980 Ti Amp Extreme is faster than my mild factory OC MSI Gaming GTX 980 Ti.
Though I did misread the Battlefield results, not noticing they changed the order, they are pretty close on some tests.
So you aren't saying that project Scorpio will outdo a 1070 now?
And yes, I know the Project Scorpio will notably faster than the PS4 Pro. This is simply a commentary on how you've been taking one test and running with it, creating some elaborate BS.
This is s shitty non impressive game run at least Horizon in true 4k using FP16 and i eat all that crap.
FP16 will not double your power FP16 can't be use on every part of the game.
Not even i make a thread about this and i am a cow.
But the GTX 980 ti can trade blows with the GTX 1080 and this clearly shows the PS4 Pro does better. So since Ron keeps insisting the Scorpio is between GTX 1070 and GTX 1080, that must mean the PS4 Pro is better than the Scorpio... (since one single tech demo vs one game is enough proof to prove Scorpio performance)...
I don't support reference GTX 980 Ti can trade blows with the reference GTX 1080 FE view point.
My MSI Gaming GTX 980 Ti factory OC is not like super-clock editions with higher paper spec guarantee numbers.
GTX 980 Ti factory OC with paper spec higher than MSI Gaming GTX 980 Ti factory OC paper spec examples are
1. MSI GeForce GTX 980 Ti Lightning
2. ASUS GTX 980 Ti STRIX OC
3. Zotac GeForce GTX 980 Ti Amp Extreme <-------- This is a very fast GTX 980 Ti card and very close to GTX 1080 FE.
@asylumni said:
So since Ron keeps insisting the Scorpio is between GTX 1070 and GTX 1080, that must mean the PS4 Pro is better than the Scorpio
Don't words in my mouth.
My MSI Gaming GTX 980 Ti proven to be faster than PS4 Pro with a game that used Vega style double rate Fp16 and native Fp16.
Zotac GeForce GTX 980 Ti Amp Extreme is faster than my mild factory OC MSI Gaming GTX 980 Ti.
Though I did misread the Battlefield results, not noticing they changed the order, they are pretty close on some tests.
So you aren't saying that project Scorpio will outdo a 1070 now?
And yes, I know the Project Scorpio will notably faster than the PS4 Pro. This is simply a commentary on how you've been taking one test and running with it, creating some elaborate BS.
GTX 1080 FE is faster than GTX 1070 FE and reference GTX 980 Ti.
@asylumni said:
So you aren't saying that project Scorpio will outdo a 1070 now?
WTF? Scorpio's Forza 6 demo doesn't have platform specific optimizations while PS4 Pro Mantis Burn Racing 4K has platform specific optimizations e.g. native Fp16 with double rate.
You didn't even notice the small and important specific details.
If you apply 1.3X memory bandwidth scale on RX-480's Battlefield 1 35.8 fps, it lands on 46.54 fps which is in GTX 1070's 46.6 fps range which backs Digital Foundry's Scorpio GPU being in GTX 1070 and Fury X range claim.
Scorpio's GPU is the first AMD GPU with larger scale 44 CU design with 384 bit memory bus and memory delta compression.
Sometimes I get in trouble when I talk about Sony too much, but, the choice they made on PS4 Pro, I totally get that choice, from their perspective and what they wanted to go do. I've said it publicly and I've said it privately, I think they've built a good 2016 PS4 Pro. With the silicon that was available, they picked the parts that made sense to go and put together a console in 2016.
But the point on not wanting framerate to drop when you go to the higher box, right, if the developers want to push resolution, to say to the player 'here you bought this higher-end console, let me show you higher-end resolution,' you don't want the framerate to drop. And that was something we didn't think we could deliver with the silicon that was available in 2016.
So some of it was time, as certain things come down in price -- some of them not as quickly as we would like. And some of it was hardware capability from our silicon partners, that allowed us to go do that in 2017.
Phil Spencer statement for AMD's year 2017 silicon, hence your argument for Scorpio being yet another RX-480 is dead.
4.8 TFLOPS R9-290's similar frame rate result with RX-480 is showing 5.8 TFLOPS is being bottle-necked by memory bandwidth.
@asylumni said:
And yes, I know the Project Scorpio will notably faster than the PS4 Pro. This is simply a commentary on how you've been taking one test and running with it, creating some elaborate BS.
The only BS is your post, you don't notice the small and important details.
Besides Forza 6 demo, I used F1 2016, BattleField 1, SW BattleFront, Doom Vulkan benchmarks to double check Scorpio's GPU being in GTX 1070 and Fury X class GPUs.
I'm with careful selecting PC game benchmarks that wouldn't cause tessellation bottlenecks since we don't know tessellation hardware behavior on Scorpio with only the hint is LOD system being disabled for the 15 cars. Ultimately, tessellation bottlenecks are red herrings as NVIDIA Gameworks BS wouldn't exist on Xbox One.
Notice the old R9-Fury X was also affected by Scorpio's Forza 6 demo results.
The old R9-390X and year 2016 RX-480 hasn't equated GTX 1070 and Scorpio is the first AMD GPU to rivaled GTX 1070 on TFLOPS efficiency.
Your "you've been taking one test" is a less-elaborate BS.
PS; I can not post more than two pictures due to admin restriction. My arguments are based from Beyond3d. Who are you? Memory bandwidth are very important for high TFLOPS PC GPUs.
NVIDIA has recently allowing AIB to install faster memory modules for Pascal refresh e.g. GDDR5-9000 and GDDR5X-11000 (outside of GTX 1080 Ti model).
@ronvalencia: Does it matter lol GTX980TI plays better 90% of other games. Who cares when PS4PRO does something right meanhwile underperforming in most cases compared to GTX980Ti. From a consumer perspective I care who plays Battlfield 1 better, not some micromachines's game. Does PS4 plays Battlefield at 4k, ultra quality and 60 fps ? Nop.
Using the same game as a benchmark, out-of-the-box GeForce 980 Ti has defeated PS4 Pro's version that used double rate Fp16. One can't underestimate memory bandwidth limiting GPU's high TFLOPS.
Log in to comment