Macroblocking with new sandy bridge build, dxva2 enabled.

@phanos. Nothing irreconcilable. It just took a little time to figure out how to turn overscan off and that I had to turn it off for every resolution and framerate I intended on using.

Ok Thanks . For a moment I was worried that it was not working for ATI cards.

phanos Wrote:Hi Spyhop,

which ATI related headaches are you referring? I am thinking of buying the same car to solve the i3 issue Thanks.

May I suggest the Sapphire Radeon HD6570 1GB. It has an HQV test score of 204.
(I've that card)..

It's also a lot more money, has a fan on it, uses more power, and unavailable in low profile.

Spyhop Wrote:I joined to post a solution on another thread and wanted to give a shout out on this one. I was one of the many who used a Sandy i3 after a buddy insisted, "his worked awesome"

Mine didn't. Same issues listed in this thread.
-i3 Software rendering had significant frame drops
-i3 dxva didn't work on Dharma
-i3 dxva ALMOST worked on the latest nightlies, but since the HD2000 on the i3 doesn't support 23.976 it lurched every 5 minutes or so as it re-synced

My solution was to bite the bullet and get a discrete Radeon 6450. Totally solved the problem. It also introduced some catalyst related headaches, but I got those ironed out. Now everything works great. I've been struggling with thist build all weekend. Now I'm just going to back away and never meddle with it again. lol

Anyway, thanks again for this thread. I learned a lot.

I'm still holding out hope that there will be a software solution for our issues.

I'll give it a couple more months than take your route...

Cfishy Wrote:I'm still holding out hope that there will be a software solution for our issues.

I'll give it a couple more months than take your route...

The issue is related to the hardware capability of the iGPU in the intel SB CPUs (24.000 fps vs 23.976 fps) and I don't think that any software is going to correct that. Now that h/w acceleration appears to be working with the nightlies you have three options:

1. Live with your system as is
2. Install a discrete graphics card that supprots playback at 23.976 fps
3. Upgrade to Ivy Bridge when it comes out in 2012 as Intel indicated that they would fix the 24 fps issue in the Ivy Bridge release
   

Sounds like we've gotten off topic with the 24p talk......Has the macroblocking issue been resolved with the most recent nightly builds??

In my post I said, "Now that h/w acceleration appears to be working with the nightlies..."

If you recall, the macroblocking issue was a result of enabling h/w acceleration. I have not confirmed this myself but several users have posted in this thread that using the nightly build instead of the lastest stable release (Dharma) results in proper h/w acceleration hence no macroblocking. So now because it appears that the macroblocking issue is gone we can get back to complaining about the 24p issue.

len45 Wrote:ok I spoke too soon. I am actually seeing the playback get choppy and frames dropped even with HW acceleration turned off. Seems like the i3 without HW acceleration can't decode all that smoothly in highly complex/high bitrate scenes. The way I gauged this is the "o" on-screen display in XBMC using a clip with fast moving CGI (the DTS trailer below). There's a point where the bitrate gets high and the "o" screen shows a "dropped" count that ticks up. And the screen framerate looks a bit jerky for a second there.

Is there a place we can check to see if the devs are working on fixing HW acceleration with Intel graphics in XBMC 11?

http://www.demo-world.eu/trailers/redire...ssless.rar

I'll test it out tonight. If HW acceleration is indeed fixed I think I can live with teh 24 frames a second...

Spyhop Wrote:It's also a lot more money, has a fan on it, uses more power, and unavailable in low profile.

I was mistaken, I've the Sapphire Radeon HD6570 1GB (not the HD6750)..

I was afraid of fan-noise myself, but the CPU fan spins faster when I pull the card out, so in reality my HTPC makes less noise when the card is in.