DirectX August SDK

Saw that as well. What somewhat surprised me is that they're talking about a bump to the Shader Model to SM4.1-do you have any idea what they're changing/adding/improving?
 
Cool thanks for the link!

PS: I finally figured out what was making DXUT dirt-slow ever since the June release: gamepad input in cameras (which is enabled for all of the default camera types), even when no gamepad is plugged in! It's so slow that even if you're not rendering anything or doing anything it will slow down the app to 200Hz max. Anyways simply changing the third parameter of all of the calls to GetInput to *false* (disable gamepad input) fixes the issue, but ideally there's a more elegant solution that allows gamepads to work without the ludicrous performance hit!
 
Saw that as well. What somewhat surprised me is that they're talking about a bump to the Shader Model to SM4.1-do you have any idea what they're changing/adding/improving?
That should hardly be surprising. Every feature addition that affects the shading language, e.g. adding a new type for cube map arrays, needs a version bump.
 
but ideally there's a more elegant solution that allows gamepads to work without the ludicrous performance hit!
Unfortunately joystick ports are rather slow. They use variable resistance for the axis and I think it's measured by charging a capacitor for which the RC time constant is not particularly fast. Seems silly, but I guess 25 years ago it was a lot cheaper to have a voltage comparator than an ADC.

When I coded for the joystick way back in the day (386, Turbo Pascal), I used polling to time how long it took for the charging to happen. I think you can use interrupts too, but I don't know the details. Maybe measuring the charging time over multiple frames causes inconsistency somehow.
 
Unfortunately joystick ports are rather slow. They use variable resistance for the axis and I think it's measured by charging a capacitor for which the RC time constant is not particularly fast.
Huh, well I can believe that but I wouldn't expect it to be slow even when there are no (plugged in) devices! Furthermore when I used DirectInput for gamepads/joysticks I could still get >1000Hz, and that was for the whole game, not just the gamepad part! That said, maybe things have deteriorated since then, who knows.

If I remember right DXUT use XInput and there was some talk about query a not plugged device is slow.
Aha... indeed it does use XInput and that could certainly explain the slowdown. Do you happen to have a link to such a discussion? I'm only casually interested as I don't really need gamepad support right now, but it would certainly be nice for them to fix this for all of the upcoming XNA titles ;)
 
The August release includes Beta 1 of XAudio2. XAudio2 is a new cross-platform audio API (Windows and Xbox 360) that is based on the Xbox 360 XAudio API. XAudio2 is a low-level audio signal processing library for Windows XP and Windows Vista providing a fully modern audio pipeline...

This sounds interesting.
 
The August release includes Beta 1 of XAudio2. XAudio2 is a new cross-platform audio API (Windows and Xbox 360) that is based on the Xbox 360 XAudio API. XAudio2 is a low-level audio signal processing library for Windows XP and Windows Vista providing a fully modern audio pipeline...

This sounds interesting.

I'll say it stronger: It's about time! The DirectSound API is still based on working around the limitations of ISA soundcards. It has been quite irritating to know that the current sound hardware works pretty much they way I want to program but then I have to use an API that emulates hardware from 1995.

I'm refering to how the DirectSound buffer interface artificially ties voices to samples. Back in the bad-ol-days the ISA bus was too slow to adequately stream samples in real time so you had to upload the samples to onboard sound RAM in advance. No sound cards work like that today but DSound still works as if they do. It's not hard to work around, but it's annoying and unnecessary.
 
XAudio2 does indeed look pretty interesting. I was lucky enough to be in a chat with the PM (I think) for XA2 back in March and was pretty excited about it after an hour of hearing about it. ISTR what I saw was a version of the GDC'07 slides boiled down to a few key diagrams and lots of talking :smile:

I'm pretty sure the GDC'07 slides are up on the DX/XNA portal for anyone who wants to see more.

Jack
 
I can't believe it took the Inq to catch this:
Direct3D 10.1 Tech Preview

Direct3D 10.1 is an incremental, side-by-side update to Direct3D 10.0 that provides a series of new rendering features that will be available in an upcoming generation of graphics hardware.

* TextureCube Arrays which are dynamically indexable in shader code.
* An updated shader model (shader model 4.1).
* The ability to select the MSAA sample pattern for a resource from a palette of patterns, and retrieve the corresponding sample positions.
* The ability to render to block-compressed textures.
* More flexibility with respect to copying of resources.
* Support for blending on all unorm and snorm formats.
 
Yeah but XAudio2 can only be software accelerated (in addition it lacks the ability to support extensions like EAX). :cry: I just don't see why a develop would choose it over OpenAL. I mean what's XAudio2's mean selling point, compatibility between Vista/XP and Xbox? OpenAL already does that (with additional support for other platforms).
 
I don't think XAudio2 will be coming to XP. Especially if it's tied to any flavor of DX10.
 
* The ability to select the MSAA sample pattern for a resource from a palette of patterns, and retrieve the corresponding sample positions.

That is the one feature there that has me the most excited. Now if only developers will support this. Although I'm assuming that there will be significant pressure from Nvidia for developers not to support this as I'm assuming it will by-pass their hardware AA and do the custom AA through shaders? Similar to what was done in the CoJ DX10 Demo?

The new sound system also sounds interesting. It's about time sound joined 3D in the present rather than being stuck with the limitations of the past.

Regards,
SB
 
The release schedule for the SDK is changing, too, Microsoft moving away from bi-monthly releases to a different four-yearly schedule. November 07, March 08, June 08 and August 08 are the next four months to see an SDK release for DirectX.
From 2 months to 4 years !? Wow, Microsoft must be getting lazy. :LOL:
 
Back
Top