Loose lips sink ships, but drunken orgies gets code out the door.I'm fine with knowing only one side of the specs, the other side will inevitably leak at GDC from all the drunk devs.
It's quite the difference, PSVR is possibly the lowest end form of VR today. That might change with PSVR2, but that wont last long either. It all depends, if you want 'it just works' i guess it wont matter no.
7nm wafers and SSDs are expensive, so I don't expect this toy to cost below $500, regardless of a 36 CUs or 56 CUs die.
Turns out it's a lost in translation due to that person is using google translate.
Plus, you might call PSVR low end, but that's still the platform with most sales, well ahead of the others. If it was that bad, it wouldn't sell well
Yes, its the cheapest and worst at the same time. And no it is not bad, its the perfect entry for a first time casual VR experience. I have PCVR and PSVR but the former does a better job if you have the hardware for it. Astrobot is a nice exclusive though.
Quest is closing the gap fast !
https://storage.googleapis.com/stat...r-ventes-q4-2019-vr4player-img01-1140x641.jpg
If there are only 36 CU's I totally agree with what you posted above.... It all makes sense to me. The Github leak talked about 36 CU, but you can get to 9,216 TF other ways that make more sense IMO.
The other tweet from GDC that I posted just said 9,216 TF so it matched the exact number from the Github leak,but did not say how many CU's or the clock that it would take to get that number.
I don't know what your definition of casual is, but PSVR is anything but casual. It has many of the same great games you get on PC VR with the exception of Oculus exclusives and it has its own exclusives as well that are not on PC like RE7 or Firewall Zero Hour. PSVR stands on its own no matter what you say.
Phil Spencer math = XBSX is 8 X XB1 and 2 X XB1X. Not sure if anyone has already said this but if you use the XB1S as the base it's 1.4 TF X 8 = 11.2 TF vs the normal XB1 which is 1.3 TF X 8 = 10.4 TF. The math makes more sense if you use the XB1S. It's closer to the 12TF and 2 X XB1X. Just a thought.
Also the XB1S is the base now where they no longer make the XB1. It also is just confusing enough to leave some gray area and confuse people.Phil Spencer math = XBSX is 8 X XB1 and 2 X XB1X. Not sure if anyone has already said this but if you use the XB1S as the base it's 1.4 TF X 8 = 11.2 TF vs the normal XB1 which is 1.3 TF X 8 = 10.4 TF. The math makes more sense if you use the XB1S. It's closer to the 12TF and 2 X XB1X. Just a thought.
40@1.8GHz = 9.216TFlops
36@2.0GHZ = 9.216TFlops
It's possible Ariel was a non-cut down version of the chip and was used in an early dev kit. Might have been what was meant by a "early low-speed version of the dev kit" referenced in that Wired Article....which would have been technically true if it was running at 1.8GHz relative to 2.0GHz in Oberon...
Casual for being VR. PSVR is the cheapest and easiest way to jump into VR. It is far from the best though.
40@1.8GHz = 9.216TFlops
36@2.0GHZ = 9.216TFlops
It's possible Ariel was a non-cut down version of the chip and was used in an early dev kit. Might have been what was meant by a "early low-speed version of the dev kit" referenced in that Wired Article....which would have been technically true if it was running at 1.8GHz relative to 2.0GHz in Oberon...
To me, it sounded like he claimed they were more powerful than 2080ti because he added "the stuff you can get on the market today" and he said it in a way so that it sounded like he cut himself off before saying "ti".I listen the podcast it is before the TGA the 5th december. From 25 minutes Jason Schreirtalk about the fact Lockhart is not cancelled finally and they heard complain about the fact to release on two SKUs. And at 38 minutes he told PS5 and Xbox are very powerful and it compares it positively form what he heard from dev on the GPU side talking about Anaconda and PS5 as more powerful than RTX 2080. Interesting podcast.
https://podcasts.google.com/?feed=a...CAUQieUEegQIARAE&ep=6&at=1575634615373http://
When they say "low speed kit" I always assumed they meant the 1Ghz gonzalo setup that was unearthed some time ago. I don't think the difference between 1.8 and 2.0 is meaningful enough to call one slow vs the other. Or it was simply still a zen/vega setup. Would those silver PC towers have APUs in them or still desktop components?40@1.8GHz = 9.216TFlops
36@2.0GHZ = 9.216TFlops
It's possible Ariel was a non-cut down version of the chip and was used in an early dev kit. Might have been what was meant by a "early low-speed version of the dev kit" referenced in that Wired Article....which would have been technically true if it was running at 1.8GHz relative to 2.0GHz in Oberon...