Send News. Want a reply? Read this. More in the FAQ.   News Forum - All Forums - Mobile - PDA - RSS Headlines  RSS Headlines   Twitter  Twitter
Customize
User Settings
Styles:
LAN Parties
Upcoming one-time events:

Regularly scheduled events

Report this Comment
Use this form to report the selected comment to the moderators. Reporting should generally be used only if the comment breaks forum rules.

69. Re: GeForce GTX TITAN Announced Feb 20, 2013, 08:26 dj LiTh
 
eRe4s3r wrote on Feb 20, 2013, 06:30:
Hen and Egg problem. If CPU's weren't slow pieces of crap compared to the GPU's we would have gameplay evolving, and not just stagnating like it does now. Currently RTS games (TOTAL WAR) can't have individual unit AI on par because there is no CPU on the planet that can handle even 10k units with good AI (and their pathfinding/Collision detection). And think about why there is so little procedural deformation in physics. If your physics engine is affecting collision it needs to be calculated entirely by the CPU.

And don't even get me started on proper neural networks and real-time raytracing. Everything is held back by the CPU and the stupid lets keep everything to a single core mandate that next gen consoles are hopefully going to break.

To give an idea how insane this performance issue is

Intel I7 = 109 gigaFLOPS (or 150 gigaflops if you push all 6 cores to the limit) Yes, that means 150 / 6 is the actual performance of your cpu in most games, because what's grinding games to a halt is when a single thread on a single core is cpu starved.

GTX TITAN = 4500 gigaFLOPS (according to the PR stuff floating around)

Or put differently, a GTX Titan is 40 times faster than the fastest CPU your money can buy. Even if there is a 50% variance, gpu's are still ahead of cpu's by nearly a decade.

At this point we should be running 1024 core cpu's

Wow, just so much wrong with that post. Where to begin... First off why dont you compare 1 cpu core to 1 gpu core if you think cpu's are so slow? Second, GPU cores are much slower than CPU cores on a thread basis. Third GPU's lack interrupts and virtual memory which is pretty usefull to OS's. Fourth GPU's cant take on complex calculations, only simple ones unlike CPU's. Fifth Tasks that cannot be paralleled your pretty much SOL on a GPU. Sixth Its fairly easy to write an x86 compatible program, and very complex to write one that takes advantage of the GPU's parallel processing.

Those are just off the top of my head.
 
Avatar 46370
 
 
Subject
  
Optional
Message
 
Login Email   Password Remember Me
If you don't already have a Blue's News user account, you can sign up here.
Forgotten your password? Click here.
 




footer

.. .. ..

Blue's News logo