@ronvalencia: Not a lie, I just assumed you knew the 88% figure came from the PC max settings. And if the particle system isn't in the tech demo, it isn't in the tech demo and wouldn't be in the PC max settings either. Rain isn't the only particle in the game.
Digital Foundry and the Xbox Scorpio - Discussion Thread ?
@asylumni said:
@ronvalencia: Not a lie, I just assumed you knew the 88% figure came from the PC max settings. And if the particle system isn't in the tech demo, it isn't in the tech demo and wouldn't be in the PC max settings either. Rain isn't the only particle in the game.
Particle system wasn't active during the grid line up.
The context for particles from http://www.eurogamer.net/articles/digitalfoundry-2017-forza-motorsport-on-project-scorpio-the-full-story
"With this scene, it's good that you point that out because we don't have a few of the components that we have within the budget, like particles. I'd say it would probably be 14ms but with this type of a grid it might be closer to 15ms," Tector replies - effectively suggesting that GPU utilisation in the mid-60s on Project Scorpio hardware would translate to 84 to 90 per cent running the same content at 1080p on Xbox One.
On XBO, particle system is about 10 to 14 percent. You got to be nutz arguing with 10 to 14 percent.
The math
XBO's 1.31 TFLOPS x 90 percent = 1.179 TFLOPS (GCN 1.1).
About 0.131 TFLOPS (GCN 1.1) reserve for particle system. That's 0.524 TFLOPS scaled by 4X resolution.
Scorpio's mid 60 percent = 4K with XBO settings and 4K texture assets, hence 1.179 TFLOPS x 4X = 4.716 TFLOPS with GCN 1.1.
65 percent of 6 TFLOPS = 3.9 TFLOPS
Effective CU IPC gain
4.716 TFLOPS GCN 1.1 / 3.9 TFLOPS GCN 1.3+ = 1.209X.
Polaris CU has 1.15X IPC gain which nearly pointless since it's improvements are memory bandwidth bound i.e. the half-ass-ed upgrade.
Scorpio's 6 TFLOPS GCN ("year 2017 silicon") ~= 7.2 TFLOPS GCN 1.1 (year 2013 silicon) effective.
Log in to comment