Zen 2 was not the greatest of gaming CPUs on release or anything, it was just better than its predecessors. CCX latency is not friendly to games using the full core count and leads to performance being underwhelming in spite of paper specs or how it fairs in synthetic MT benches. On Console you also have to remember that they take away the larger cache of the desktop models... then you are left with something that is not exactly great.It’s wild to me that an 8-core 3.5Ghz Zen 2 struggles to hit 60fps in any game.
Zen 2 was not the greatest of gaming CPUs on release or anything, it was just better than its predecessors. CCX latency is not friendly to games using the full core count and leads to performance being underwhelming in spite of paper specs or how it fairs in synthetic MT benches. On Console you also have to remember that they take away the larger cache of the desktop models... then you are left with something that is not exactly great.
It’s wild to me that an 8-core 3.5Ghz Zen 2 struggles to hit 60fps in any game.
I think a fair bit of the CPU power is being used to simplify development.The 3700x is good for 120fps in windows. Even with lower clocks and reduced cache it’s hard to understand why 60fps is challenging on a closed box console.
But in business it is. Why choose to make less money? $50 each for 10 million users if $500 million. The only reason a company would turn their noses up at half a billion is if they could turn that into more money elsewhere, such as selling to a larger audience.... losing 50$/€ per unit is not that big of a deal, or at least it shouldn't.
https://www.gamesindustry.biz/playstation-5-now-represents-50-of-sonys-active-console-baseBut in business it is. Why choose to make less money? $50 each for 10 million users if $500 million. The only reason a company would turn their noses up at half a billion is if they could turn that into more money elsewhere, such as selling to a larger audience.
Exactly this. They are just preoccupied about getting the higher bonuses just before they leave. It perfectly worked for Ryan as he left just before the Concord trainwreck. Those people are not interested into building something for the future of Playstation (+5-10 years) and the next generation of executives."The firm revealed revenue per console is $731 for PS5 vs $580 for PS4, which is partially due to price rises (and higher tiers) on games, accessories and services this generation."
I'm sure they are making their calculations, but from the outside, the strategy of losing a bit more on the hardware to gain users has worked in the past, but right now it feels that both Sony and Microsoft are more preoccupied with appeasing investors in the short term instead of selling a vision for a more profitable future.
Even steam is not really taking advantage of the situation honestly, they could be a bit more proactive in the hardware space. A steam desktop box sold in stores would make Sony and Microsoft sweat a little.
Even steam is not really taking advantage of the situation honestly, they could be a bit more proactive in the hardware space. A steam desktop box sold in stores would make Sony and Microsoft sweat a little.
Now that proton is working well, and steam os is good enough thanks to the experience with steam deck, a new steambox made by valve could be a success.Steam tried in the past with steambox but it was a bit messy and didn't go anywhere.
Valve is doing more than well with their software stack so why bother. And handhelds have unexpectedly started selling in the PC space esp as iGPU's finally stop sucking.
Ah, right. I'd completely forgotten about that. Even though it's the first post in this thread lol ¯\_(ツ)_/¯The Sony developer documents that leaked out earlier this year, or maybe it was last year. 40-50% more general GPU perf and up to 2-4x in RT operations.
Now that proton is working well, and steam os is good enough thanks to the experience with steam deck, a new steambox made by valve could be a success.
At 2.35ghz (Sony numbers) the GPU should be 18 TF, the same way PS5 GPU is 10.3 TF. The 45% improved rendering is a developer estimation from Sony, this is the improvement they expect in most games without RT or PSSR. It's logical considering the bandwidth weak gains (28%) and GPU performance that doesn't improve linearly with clocks (most expect around 75% raw GPU TF boost). For reference PS4 Pro GPU was 128% more powerful than PS4 and most games got their resolution improved by only 77%.Ah, right. I'd completely forgotten about that. Even though it's the first post in this thread lol ¯\_(ツ)_/¯
With the PS5 being (128*36*2.23=) 10.28TF a full 60CU Pro at the same clockspeed and 17.13TF doesn't seem likely. 30WGP total with the same amount of deactivation for yield brings us to a 54CU model and 15.41TF.
I'm still pretty surprised that Sony went the route of a larger chip though. I was really expecting them to stick with 40CU's, maybe 44, and push the clockspeed as high as possible towards 3GHz. Enough to take 40fps modes to a slightly rocky 60fps and so on.
Then, any imperfect chips would still likely have sufficient functional CU's to go in the cloud as base model PS5's for streaming.
I suppose PSSR hardware has rendered that entirely unfeasible though - there'd just be a bunch of NPU hardware sat twiddling its thumbs in the cloud, and at that point you may as well play it safe with a lower clockspeed and wider GPU.
Does an NPU need to be incorporated into the GPU? Or could it theoretically be its own chip connected by IF much like CPU's are in other AMD APU's?
I'm just wondering if that's the kind of thing we might see next generation as wafer prices increase and presumably the desire to not waste any would too.
You could make a game that averages 15fps on the 7800X3D if you really wanted to.It’s wild to me that an 8-core 3.5Ghz Zen 2 struggles to hit 60fps in any game.
You could make a game that averages 15fps on the 7800X3D if you really wanted to.
It's entirely on how developers use the hardware. There is nothing about any given processor that inherently allows it to hit some specific performance figure in any given game. 60fps has existed in games for many decades, so it's not about the hardware. It's a developer choice. Everything on the PS4/XB1 could have been 60fps if developers really thought it was a priority. There were plenty of 60fps titles in that generation, for instance.
This was discussed a couple of years ago:Now that proton is working well, and steam os is good enough thanks to the experience with steam deck, a new steambox made by valve could be a success.
I don't know enough about how latency sensitive these workloads are to give an answer.Ah, right. I'd completely forgotten about that. Even though it's the first post in this thread lol ¯\_(ツ)_/¯
With the PS5 being (128*36*2.23=) 10.28TF a full 60CU Pro at the same clockspeed and 17.13TF doesn't seem likely. 30WGP total with the same amount of deactivation for yield brings us to a 54CU model and 15.41TF.
I'm still pretty surprised that Sony went the route of a larger chip though. I was really expecting them to stick with 40CU's, maybe 44, and push the clockspeed as high as possible towards 3GHz. Enough to take 40fps modes to a slightly rocky 60fps and so on.
Then, any imperfect chips would still likely have sufficient functional CU's to go in the cloud as base model PS5's for streaming.
I suppose PSSR hardware has rendered that entirely unfeasible though - there'd just be a bunch of NPU hardware sat twiddling its thumbs in the cloud, and at that point you may as well play it safe with a lower clockspeed and wider GPU.
Does an NPU need to be incorporated into the GPU? Or could it theoretically be its own chip connected by IF much like CPU's are in other AMD APU's?
I'm just wondering if that's the kind of thing we might see next generation as wafer prices increase and presumably the desire to not waste any would too.