Microsoft's "Warp 10" Will Allow CPU-Accelerated DirectX 10

[Rob Williams]

From our front-page news:

Here's one I didn't quite see coming. Microsoft is working on allowing Direct3D 10 and 10.1 to run off of the processor, rather than require integrated or discrete graphics. The idea is to use the CPU to push out ample graphics performance to allow basic operation of Windows, in addition to light gaming, and I do mean light.

What that means for Microsoft is a potential end to the Vista-capable debacle. If Aero, or whatever the Windows 7 equivalent is called, could run off of the CPU, then the problem simply vanishes. According to them, this "WARP 10" system can run off of a CPU with an 800MHz frequency, which technically shows that the faster the CPU, the better the graphics performance.

According to the article, this technology proved to be even faster than Intel's current integrated offering, but the results are a little hard to settle on. The test machine used was an eight-core Core i7 machine, which is a lot of power. If WARP 10 took full advantage of the spare CPU cycles and still only achieved a 2FPS increase over Intel's integrated solution, then I don't think there's a reason to get excited right now.

The most important thing would be to gain enough graphics power to enable the special OS features. As long as that happens, then we can honestly say Microsoft made a smart development move.

(Picture this, but at -1 FPS)​

Of course, software rendering on a single desktop CPU isn’t going to be able to compete with decent dedicated 3D graphics cards when it comes to high-end games, but Microsoft has released some interesting benchmarks that show the system to be quicker than Intel’s current integrated DirectX 10 graphics.

Source: Custom PC
​

 

[Kougar]

Dunno, but I still find it highly amusing that a CPU can out render Intel's IGP by a fair margin. Would wonder how threaded the application was, it might not be running eight threads.

Interesting idea though... could build a quadcore laptop without any kind of GPU? Although it kind of defeats the purpose, trading a CPU core for a GPU core.

 

[Rob Williams]

Dunno, but I still find it highly amusing that a CPU can out render Intel's IGP by a fair margin. Would wonder how threaded the application was, it might not be running eight threads.

There's the problem... what we're told is far too vague. Does the timedemo take advantage of all threads? This is something I should test out later today... because if so, it would essentially mean that it took two Core i7 GPUs to match the performance from Intel's IGP. Unless of course they are talking about THREADS and not CORES, then it might only be a one CPU machine.

Although it kind of defeats the purpose, trading a CPU core for a GPU core.

Well, the article said that it only took 800MHz to deliver enough graphics power to make the PC usable, so it might be about 33% of a single core on a current Quad-Core CPU. It's an interesting concept, but one I still don't find very exciting. Even at 33% for a core, that's a lot of horsepower required, and in the end, that 33% of a single core might draw more power than Intel's IGP would.

I'm still looking forward to see where this is headed, because it seems like it's something that should be done. I'm surprised it took Microsoft so long, though.