• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

PS6 coming in 2027

Feels like there's always something holding the top end consoles back. For a while it was because all the new releases had to have last-gen versions, too. In the case of the Xbox, they need a Series S version. Now it's a portable. Normally I wouldn't care as much, but with Sony backing away from the PC, this is frustrating. Give me that clean break like the NES to SNES to N64. Move on or miss out.
 
path tracing is an extremelly loose term and console dynamic resolution can go down to 700p, i.e. of course a PS6 could run "path tracing" at ~60 fps, PS5 pro could has well.

Do only some reflection and 0.25 rays per pixel at 900p for Restir GI, that F1 2025 demo shows that even a PS5 pro is close to it at full feature unified path tracing (direct, indirect lights, shadows, reflections and ambient occlusion), algo are getting better and better, denoiser getting really good and need less and less data to work well.

Handheld for some game could mean for dev they still have to make traditionnal pre-baked map and rendering pass, removing a lot of the incentive to use the pathtracing render path, but that would have been true for non-exclusive release in general that target mass audience at least at console first few years of launch, exclusive could still put the work on to do both, as they still know at least vast majority of the time the game will run on a system that support it and will be mostly judged on it.
 
Last edited:
Marees said:
sony non-commital on PS6 release date.

may even release the handheld first, imo
Click to expand...

It’s probably going to be a game of chicken with MS on who will wait out the memory price bubble the longest. I doubt either of them want to be a loss leader by releasing when prices are still high, but will be forced to release if the other one does.

Supposedly some banks have been offloading ai debt at a discount.
 
zehoo said:
It’s probably going to be a game of chicken with MS on who will wait out the memory price bubble the longest. I doubt either of them want to be a loss leader by releasing when prices are still high, but will be forced to release if the other one does.

Supposedly some banks have been offloading ai debt at a discount.
Click to expand...
apparently inference will turn profitable shortly but training is still extremely capital intensive
 
  • Like
Reactions: zehoo
like this
Marees said:
apparently inference will turn profitable shortly but training is still extremely capital intensive
Click to expand...
Inference without amortizing the training has been profitable for a very long time relative to that space (~2024, once the A100->H100 giant efficacy bump occur, say the talk was in the 2024 OpenAI spent ~2 billions on inference with 3.7 billion revenues, it is the 5 billions in R&D-training compute that were putting them in the red).

https://epoch.ai/data-insights/openai-compute-spend#:~:text=Data for OpenAI's overall compute,and The New York Times.

Now with the 44 billions ARR on an Anthropic (50% more than not so long ago), inference is extremely profitable when on Blackwell (or the latest amazon/google chip, those people take very reasonable off the top skim on it for compute -20/30%), gross margin in the ~70% overall for an Anthropic, (for rexample for opus 4.7, SemiAnalysis estimates that the true blended cost of Opus 4.7 for agent tasks is approximately $0.99 per million tokens, significantly lower than the listed price of $5 per million input tokens.)

Rubin should be an other step function, if open model does not destroy pricing and openAI/anthropic do not engage in a price war, it would get quite profitable.
 
Last edited:
Decko said:
These consoles are more expensive, with the added costs of chips these days I'd be willing to be these consoles last close to a decade as opposed to the typical life cycle. I'm betting on 2029 or 30.
Click to expand...
playstation consoles are not the major problem. its the games who get more and more expensive
 
Hedgehog_77 said:
playstation consoles are not the major problem. its the games who get more and more expensive
Click to expand...
And take forever to be made. Like during the PS1 day we go a new Final Fantasy game every year. Now we get one every decade almost. How long has GTA 6 been in productions. 5+ years is way to long to produce a game
 
vegeta535 said:
And take forever to be made. Like during the PS1 day we go a new Final Fantasy game every year. Now we get one every decade almost. How long has GTA 6 been in productions. 5+ years is way to long to produce a game
Click to expand...
yeah right, that certainly adds to it. sometimes I think we have to take into account the different times we lived in: i drive a bmw e34 from 91' for example. this car was absolutely beautiful and very well thought out/ engineered ("built for decades"). if you look at todays productions, everyone knows that there is a clock built inside, where you can estimate how long certain parts may function until they break. the models from different car brands also look very similar today, its just about maximizing the profit today. i think thats the same with many industries.
its not a console game but I watched a youtube video about the making of Gothic 1 (PC) which was my first real RPG. The new studio who developed it in the late 90s did almost everything for the very first time to put such a great game together. it was absolute chaos if you watch the video and highly stressful, but they wanted to create something that lasted - and it did.
 
vegeta535 said:
And take forever to be made. Like during the PS1 day we go a new Final Fantasy game every year. Now we get one every decade almost. How long has GTA 6 been in productions. 5+ years is way to long to produce a game
Click to expand...
GTA6 is an outlier, though, as Rockstar has very different timelines than even its biggest competitors. GTA5 and GTA Online have been genuinely popular for over a decade; the studio can afford to take its time because it knows GTA6 could be similarly popular. Even Activision-Blizzard, EA, and Ubisoft can't count on their fall blockbuster releases remaining popular for more than a couple of years.
 
You must log in or register to reply here.
Back
Top