Aren't those standard on RDNA2?
Weird presentation for sure, but surely there are some assumptios that are safe to make considering it is, after all, an RDNA2 GPU.
Playstation 5 [PS5] [Release November 12 2020]
- Thread starter BRiT
- Start date
Weird presentation for sure, but surely there are some assumptios that are safe to make considering it is, after all, an RDNA2 GPU.
Having to devote compute resources to 3D audio has always put it first on the chopping block. When push comes to shove I'm sure that'll be the case this generation also, especially if there's enough of a performance delta between ps5, seriesX, and PC where they're having to throw every last bit of grunt to game logic and graphics to make them look roughly the same.
So with Sony PS5 having ability for consumer M2 NVME drives that meet as yet to be determined threshold, will those too suffer from the heat related slowdowns that DF covered on the SeriesX deep-dive? If so, what will be their throughput after hitting the slowdowns?
I was quite amused by the fact after his long (10, 15? minute) chat on audio, and all the individual mapping they were doing, he said that a number of people couldn't actually tell the difference.
I also felt it was a weird presentation, too lightweight and waffly for a dev presentation, too dry for a general consumption one.
Ho hum.
I also felt it was a weird presentation, too lightweight and waffly for a dev presentation, too dry for a general consumption one.
Ho hum.
It was there. You just couldn't see it behind the raytraced cardboard cutouts. LOL
Tommy McClain
Variable load times for PS5.
It felt like a pretty boilerplate GDC keynote. Very technical-sounding for a casual audience, but not entirely useful for any developer. Basically a marketing-oriented talk.
That was interesting, especially reading the comments on here, at the same time as I was looking at the stream.
My initial impression is that Sony looked at what will make a better game experience now that graphics (within a reasonable price and power envelope) has plateaued somewhat, while MS just went as high and wide in every direction they could.
Maybe artisan vs brute force approach.
Will be interesting to see if Cerny got it right (last time 18 CU more balanced system) or if the POWAH approach from MS is the better choice. I guess it comes down to how easy it easy to eak out performance from each console and how fast to market the games can come.
I have no clue in regards to boost clock or reliability, I defer to experts on here, but it seems nobody on here likes Cerny's "boost solution" to make the TF gap smaller on paper.
But what is cool, we have similar consoles, with some different approaches / priorities, be interesting to see what works best.
My initial impression is that Sony looked at what will make a better game experience now that graphics (within a reasonable price and power envelope) has plateaued somewhat, while MS just went as high and wide in every direction they could.
Maybe artisan vs brute force approach.
Will be interesting to see if Cerny got it right (last time 18 CU more balanced system) or if the POWAH approach from MS is the better choice. I guess it comes down to how easy it easy to eak out performance from each console and how fast to market the games can come.
I have no clue in regards to boost clock or reliability, I defer to experts on here, but it seems nobody on here likes Cerny's "boost solution" to make the TF gap smaller on paper.
But what is cool, we have similar consoles, with some different approaches / priorities, be interesting to see what works best.
Depending on how they design the expansion bay, I think. They could put hestsink on the cradle or something.
It's as dynamic as the system.
Wanted to point out that the MS SSD expansion may well not be that proprietary. Seagate was etched on the clip and it was being discussed, not by MS, that maybe MS would simply allow other manufacturers to build to the spec. Meaning - price may not be some ridiculous point set by MS to make money ala' a peripheral.
I saw one move at some point. Guy on the left picked his nose or something. Thought they were a CGI loop of some kind.
I remember that dev doc from Bungie years ago. The sound engineer mentioned how "Feels like you got shat on." and "Oh it's just audio. They will figure it out in the end."
When the camera zoomed back I kept looking at the silhouettes of the 'people'. I thought I going mad but they sure as hell looked like cutouts.
I saw one move at some point. Guy on the left picked his nose or something. Thought they were a CGI loop of some kind.
I remember that dev doc from Bungie years ago. The sound engineer mentioned how "Feels like you got shat on." and "Oh it's just audio. They will figure it out in the end."
For rough, realistic reflections, it's definitely the case that you can trace broad GI more cheaply. Rough reflections require multiple rays and evaluation of surface shaders (shader LOD will help). GI can be very coarse and doesn't need sophisticated surface shading for traced surfaces. You can also use more hacks in traced GI. So I'm siding with Cerny on this one. I think he's reported on what they've found in their evaluation and he's communicated that to the dev community about what they can hope to achieve.
So they don’t yet know what drives will be able to handle the system, or even fit (his own words). So. How do they know how big of an expansion port they will need?
Do they..... DO THEY STILL NOT KNOW WHAT THE PS5 LOOKS LIKE?
Do they..... DO THEY STILL NOT KNOW WHAT THE PS5 LOOKS LIKE?
https://forum.beyond3d.com/posts/2110892/
It must be the absolute max at 2.23, they probably tried 2.26ghz (i was told over a month ago already). They must have adjusted to lower.
Now maybe people learn that people who really have the info would never tell anyone about it.
D
Deleted member 13524
Guest
It might be very close to 1/1 in GPU performance due to 22% faster back-end.
Biggest difference should be memory bandwidth. That's the one instance where the Series X has a very clear advantage. Sony has the advantage of having all its memory using the whole bus, but I think the SeriesX's approach is higher performant in the end. Cerny claims the L1 and L2 caches at higher clocks will have an impact in effective bandwidth, which is probably true, but I doubt it'll have as much impact as a significantly higher system memory bandwidth.
The faster SSD could also mean they get to stream more stuff without caching it on the GDDR6 pool too, so that could make some small difference.
I wonder if just using 2GB chips on all the 10 channels would have been much of a difference in BoM. The Series X getting 20GB of GDDR6 at 320bit would have put a lot more pressure onto Sony. Perhaps more than the TFLOPs difference.
DavidGraham
Veteran
Maybe, but how can GI be less intensive than Shadows?
It's dynamic. Like a chameleon. It was actually in the presentation but you couldn't see it.
And that 30% produces same looking games, with a resolution advantage on X.
What do we have now? 15% or so? Meh.
They’re basically the same, one will run at imperceptibly higher res. The other will load games and levels imperceptibly faster.
Boring.
Can we talk games now?
Similar threads
- Locked
- Deleted member 7537
- Games Industry
- Sticky