Tim Sweeney says photorealism may be achieved at 40 teraflops

RIP CPU's continuing to advance in leaps and bounds with regards to processing power like it was at the time he made that prediction. In the years leading up to 1999, CPU processing power was advancing at an incredible rate.

If CPU's hadn't hit a wall with regards to processing power at reasonable power consumption his prediction may have come true.

Regards,
SB
 
Which is why GPUs have evolved. Not saying Sweeney was right, but he also wasn't particularly wrong either. The concept of the software renderer and voxels and REYES is a real thing. What he got wrong was the idea that the CPU would be so fast it could drive this. CPU evolution hit a wall. GPU evolution has been able to extend into the programmability Sweeney was talking about. So right concept, wrong processor and wrong timeline. ;)

Incidentally, what were the CPus of 2007? Core 2 Duo and Athlon 64? Cell was the closest to Sweeney's vision. In that regard, explicit prediction, he was very wrong!
 
Incidentally, what were the CPus of 2007? Core 2 Duo and Athlon 64? Cell was the closest to Sweeney's vision. In that regard, explicit prediction, he was very wrong!

Yup, that was around the time when Intel realized that they could no longer chase increasing clockspeed as they had basically hit a power wall. So they switched gears towards increasing perf/watt versus focusing purely on performance. 2006 was when Intel switched from the NetBurst microarchitecture (Pentium 4) to the successor (Core 2 series and beyond) of the P6 microarchitecture (Core series) which was used in the previous Pentium M series which was developed simultaneously with the NetBurst microarchitecture. NetBurst being for desktop use with the enhanced P6 microarchitecture being used on mobile. That led to the development of the Core microarchitecture (Core 2) which is currently in use.

AMD was caught out as they couldn't easily switch to a more perf/watt oriented architecture. They were all in on the desktop market and couldn't switch to a simultaneously developed mobile CPU variant like Intel could. And hence when the power wall hit them, it hit them much harder than Intel as they couldn't just terminate one development path and switch to the other.

Regards,
SB
 
Another fresh prediction from Sweeney
"Slowly, over the next 5 years, they will force-patch Windows 10 to make Steam progressively worse and more broken. They’ll never completely break it, but will continue to break it until, in five years, people are so fed up that Steam is buggy that the Windows Store seem like an ideal alternative."

Not simply bugging steam, but removing with a patch Win32 dlls from all computer
 
Last edited by a moderator:
Another fresh prediction
"Slowly, over the next 5 years, they will force-patch Windows 10 to make Steam progressively worse and more broken. They’ll never completely break it, but will continue to break it until, in five years, people are so fed up that Steam is buggy that the Windows Store seem like an ideal alternative."

Not simply bugging steam, but removing with a patch Win32 dlls from all computer
tim-sweeney-crackpot.jpg
 
What? Breaking compatibility with ALL win32 programs on ALL the computers?
- Open this document
- I can't... after the lat upgrade office is no more compatible with windows

He better stop at management and bloody musicals
If there is a will, there is a way.
 
Tim sweeney says a lot. In the last years he reminds me of John McAfee. First a really intelligent guy, but after a few years he suffers from paranoia.

Now @topic
every few years, someone claims that we get photorealism @realtime in the next few years. We heard that a lot of times and it was never achieved. Yes, something like uncharted looks really cool, but it is not even close to photorealism. Maybe in a few scenes it looks almost real but not in a whole game.
But, who want's that? Games should look like fun. Photoreal graphics aren't fun. For some simulations, yes it might be a cool thing (like flight simulators) but for everything else, not. The other problem with almost photorealistic games is, they aging really bad, while others that don't even try to look as realistic as possible are in a much better state to even look good after a few years.
Developers shouldn't try to make games more realistic, they should use the power for more and better cool looking effects and for better gameplay.
 
Every few years someone claims we will have realtime toystory level gfx and ...
well we now do
Look, I will always remember Hannah's hair. :) I still have to see that in a videogame, though... Maybe the closest thing is TressFX hair, but still...

Toy Story thread spawn incoming...
That would be a great idea, actually.
 
Last edited:
Back
Top