Slashdot Mirror
EA Announces Multi-Title Unreal Engine 3 License
An anonymous reader writes to mention a Gamasutra article about a surprising announcement from EA. They've made the move to license the Unreal 3 Engine for a series of next-generation titles. "The brief announcement states that EA 'employs a variety of engines, tools and technologies to best serve the needs of each game and development team', but raises interesting issues regarding the Criterion-authored Renderware engine, purchased by EA in 2004 alongside the Burnout developer, and its intended global EA rollout."
(well, someone had to reply eventually)
There used to be a time where your engine defined what you could do in a game, and the engine you choose would have a massive impact on the quality of game you could produce; I think these days are long gone. If you discount certain cutting edge graphical techniques, there are few (gameplay modifying) features that are implemented in the Unreal 3/Doom 3 engines that could not be done in an open source engine written in Java.
Personally, I think that it is time that someone focuses on generating an open source java framework that is designed around splitting a game engine into its smaller components (Graphics, Physics, Scripting and AI); this would allow for smaller (more focused) open source projects to exist which (should) produce higher quality results.
I'm not sure whether this is bad or good. I was thinking it might make future games feel generic, but then I thought... more than now? Let's hope not. But maybe the generic feel of today's FPSes is that the oft-reused game engines are not quite flexible enough, so the player "recognizes" the engine underneath. Maybe in the future they will fix that.
"Something is seriously wrong at EA if their own internal development teams can't come up with something better than the Doom3ish low poly/overly normal mapped rendering engine from Epic."
No, not at all. They know when they release their new "bumpy shiny armor" FPS it will sell x amount of copies for x amount of money making them x amount of profit. As long as X > Development costs, they will keep doing it.
AC I think you have more problems to worry about then EA though. You seem to have some weird split personality going as your next comment is flaming your original comment. Seek help.
I keep telling myself I'm not the desperate type.
I had nothing to do with previous posts, but Microsoft dropping the x86 (in particular the intel x86) had little to do with performance and much more to do with licencing costs.
The original XBox was designed much faster than any other console has ever been designed (went from an idea to a full fledged system in 12 or 16 months IIRC); the problem with this was that Microsoft had to stick to (mostly) over the counter components. Microsoft didn't bother to get decent licencing terms from either Intel or nVidia and in the long run this cost them because the licencing they were paying eventually was more then it would cost to buy the processors wholesale, and it prevented a major internal redesign that would cut costs dramatically.
Contrast this to Nintendo, who spent 4 years working with ArtX and IBM to produce the Gamecube; the licencing terms were so sweet for Nintendo that (even though the system was about as powerful as the XBox) Nintendo charged dramatically less then their competition and still turned a profit on hardware.
IBM became the default processor for all three consoles this generation because of their willingness to produce a high-quality processor at a reasonable price while maintaining a flexable licence.
As for the x86 processor, the AMD X2, and Intel Pentium 4 processors have (in practical terms) about the same performance as either the Cell or Xeon processors; the problem with both these companies is that they are not going to allow you to produce their processor for them and are going to be very reluctant to work on incorporating a graphics processor into their CPU's (as Sony did with the slimline PS2).
Parent poster is right. Intel and AMD have proven that the ISA is irrelevant to processor performance. Modern x86 chips implement a RISC superscalar out-of-order pipeline internally. Of the 100+ million transistors on the die, Intel/AMD only have to burn a million or so to implement the x86->RISC conversion (called uOps or micro-ops).
And no, architectural registers is largely irrelevant also because of techniques like register renaming.
So yes, the anti-x86 trolls should move along. x86 is here to stay and it doesn't matter.
This is simply not true. The modern x86 is designed for, and great on, general code but if you have a loop and you want to optimize it to the max you cannot obtain the same clock-for-clock performance on an x86 as you can on say the PPC and the factor is somewhere between 20% and 50%. This is partly down to register name shortage (it does matter when you can't express your algorithm without inserting several dozen extra u-ops, and those stack pushes are not optimized away). The other big reason is that PPC has predictability while what's going on inside the x86 is anybody's guess. You can't optimize for x86. Abrash's best work is pure trial-and-error.
The x86 may be "here to stay" in the desktop or server PC (er, until you want more than 2GB of RAM) but it's dead in the water in every other CPU-based industry. All that x86 to RISC costs a fortune in die size, power consumption and maximum throughput efficiency. The only reason anyone buys an x86-compatible processor today is (duh) to run x86-compatible code. If there is no legacy code to cope with, no-one uses x86. Microsoft made that mistake and lost a lot of money and market placement as a result.
Anyway the x86 ISA is a 32-bit ISA and it can't survive past the current generation of 32-bit CPUs so your point is about 5 years too late. The death of the x86 ISA is right around the corner. It will live on in software emulation just as the 6809 did.
If IA-32 was so inferior, a new ISA would be dominant today. Yes, backwards compatibility is a big deal. But if a new ISA delivers 50-100% better performance, you can bet that people would adopt it and MS would offer their products for it (e.g. NT4 was available for DEC Alpha).