Let's set this straight - No one can do 24p consistently well - Page 19

You must have missed Damian's responses to my questions in which he explained that when the offset counter shown in the XBMC debug window rolls over without a frame drip that it is the result of locking back to a sync frame in the audio video stream.

So, what's the latest status regarding whether Intel Ivy Bridge can do proper 23.976 Hz video output?

See this article:

http://www.anandtech.com/show/5773/intels-ivy-bridge-an-htpc-perspective/5

Where the following is claimed:

"The good news is that Intel is claiming that this issue is fully resolved in the latest production BIOS on their motherboard. This means that BIOS updates to the current boards from other manufacturers should also get the fix. Hopefully, we should be able to independently test and confirm this soon."

So has the issue been resolved yet? Are we still waiting for the BIOS update? Or did the BIOS update not solve the 23.976 Hz video output issue on Ivy Bridge after all ?

No.

Rene, Good that you are back! When you say "No", could you expand on that at all?

To give you two examples of possible interpretations of your "No":

1. Intel released a BIOS that they claimed would fix the problem (the "latest production BIOS" referred to in the Anandtech review), but this has been tested and it didn't fix anything.

2. Nobody is quite sure whether Intel has released the BIOS Anandtech referred to, but in any case the problem is still apparent with the latest BIOS on all 7-series Intel motherboards regardless of CPU.

Have you personally done any testing, or can you point me to other published testing - I haven't really seen anything much since the Anandtech review?

I tried a couple of IVB+Intel 7 series chipset mb with the latest BIOS and graphics driver. The refresh rate is still the same as SNB+Intel 6 series chipset mb, ~23.9728Hz. Mb are not Intel brand, but I guess if Intel released such a BIOS for its mb, then somebody would report it immediately.

 

I hardly call it a *bug* or *issue*. Just a frame drop every 4 minute or so and almost nobody notices it. Rather non-fixing "full" RGB color space is a more serious issue for some people.

Edited by renethx - 6/11/12 at 12:29am

Regardless of how severe this problem is in practice, I think it is poor that Intel made statements saying this issue would be improved with the 7-series motherboards, when in actual fact there has been no improvement. When this was noticed during Anandtech's early review, Intel then told Anandtech there was a new "production" BIOS that really did fix things. Yet, six weeks later, there is still no sign of any fix.

Like it or not, this has become a totemic issue. Intel has come up with all kinds of excuses over the years, yet despite repeated promises, they have failed to deal with the problem.

Why should Intel have to make excuses when hardly any other video card, be it from Nvidia or AMD, have, for the most part the same problem? You read it yourself, hardly anybody notices this problem (apparently just those of us with big ass TVs that do 24P).

Well, how poorly anyone else does it is no excuse IMO. I notice it. And I, as well as all my customers have "big ass TV's". Mine it 56" tall, 11 feet wide and is what I consider "medium" size.

I shouldn't have to be a super-freak-geek and open source geek-ware expert to get this thing to work, without issue. But alas, you pretty much have to dig in to work on it. Or live with what it spits out. Our use a standalone that relatively speaking, always works flawlessly. If not for the convenience of the MyMovies package I wouldn't even consider using one over my now older Oppo '83. If the Oppo would play BRD over the network, like it will on attached drives, I'd drop the PC in a heartbeat, but it does not. The MyMovies software convenience overall has outweighed the pain of the HTPC and server it runs on. So far.

Ok, so that nobody else fixes it either is no excuse for not fixing it, but why does it seem to be a "totemic issue" only for Intel? Do you write the same thing blasting AMD and nVidia?

I understand people being upset that this issue persists. I don't understand people acting like Intel is the only one who keeps dropping the ball.

Have you had a chance yet to determine how HD2500 and 4000 graphics fit into your MadVR chart? That is, what combinations of cpu and memory speed are required to get full quality with the Ivy Bridge chips?

If you want it to work, without a lot of fuss, then get a 6450 Radeon card. I have had zero issues since putting it in. I get HD audio and perfectly 24P performance with no dropped frames or audio dropouts to my "big ass" plasma TV.

I can actually watch the clock drifting in the XBMC debug window, then it resyncs itself before any frame or audio dropout occurs. It works (from where I sit) like a BD player does.

I used to spend hours goofing around with re-clock, custom resolutions and settings... and now I don't have to deal with any of that BS. I just program XBMC to "sync to source" and let the 6450 do its thing.

Yes, done. As the name indicates,

 

HD/HD 2000 (SNB) < HD 2500 (IVB) < HD 3000 (SNB) < HD 4000 (IVB)

 

HD 4000 is powerful enough to play back every kind of SD and HD videos in the high quality setting of madVR. DDR3-1600 is enough. None of the others has enough execution units, in particular for SD video-based contents. HD 3000 is barely good for "3 taps" in luma upsampling if GPU is overclocked and DDR3-2133 is used.

Edited by renethx - 6/11/12 at 6:33pm

I'll leave it to others to discern the intent of "totemic", not my word. I don't know that it's only an issue for Intel. But as I purchased a Intel MB and an Intel CPU I'm not really interested in what others can('t) do. I would if I had some expectation they should have it working, and I had purchased their hardware . I don't care who else can't fix it. If someone had a solution that worked, I'd buy it.

Anyone know if there is equivalent functionality in MPC-HC? Maybe this is what the EVR Sync renderer is for? Is there a way to get something like this while keeping MadVR?

I haven't been keeping up with the sync options in MPC-HC in the last year or two - maybe it has been addressed?

I wish I could help but I don't have any experience with MPC-HC, I am very happy with XBMC native video performance and expect I will be even happier when I start piping it through the Onkyo 818 that I have ordered.

The main reason I specifically mentioned Intel is that there has been interest in whether it might have made improvements in Ivy Bridge or the 7-series chipset. Those parts have just been released and are beginning to be tested.

Intel has said in the past that, for Sandy Bridge, they did not have time to re-engineer the PCH, and that this was responsible for the problem with "24p". Intel further claimed that this would be addressed in Ivy Bridge. However, that does not appear to be true on the evidence of actual testing. This is news, and therefore worthy of comment.

However, it is also the case that Intel's problem with 24p is relatively greater than the problem that AMD or nvidia has.

Anyway, I don't want to go overboard on this - I am just saying it is disappointing whenever any company promises something that it does not deliver.

Intel never claimed that it would be fixed w/ the 7-series chipset, they (or at least one person from Intel at IDF) did say however that it would be better w/ the updated PCH - which has not turned out to be the case.

According to the Anandtech review of Sandy Bridge (see http://www.anandtech.com/show/4083/the-sandy-bridge-review-intel-core-i7-2600k-i5-2500k-core-i3-2100-tested/7) the following was claimed:

"Intel has committed to addressing the problem in the next major platform revision, which unfortunately seems to be Ivy Bridge in 2012."

So far we've seen none of that...

I would guess that the reviewer misunderstood what Intel meant by "next major platform revision".

The 7-Series chipset and IVB (Tick) are minor revisions to the previous generation. Haswell (Tock), and the associated hardware ecosystem, is the next major platform revision.

OK, another at AnandTech (jakmal):

 

Note that this is not unconditionally true.
In my tests, the HD4000 does work with all content i tried in madVR, as long as i did NOT use the Intel QuickSync decoder, and instead used software decoding.
If you combine madVR with the Intel QS decoder (available in LAV or ffdshow), you get issues with some high-bandwidth content, like 720p60 from some camcorders, most likely because the memory or bus bandwidth is somehow limiting there.

If you never watch such content, and instead limit yourself to SD DVDs or HD movie content (720p24, 1080p24 or 1080i30), you should not run into trouble - or if you just picked one of the faster CPUs and can deal with software decoding without any issues.

I can only speculate that this was a misunderstanding based on a marked change in accuracy during the BIOS development stage. It's also worth noting that an ASUS board was used for that review, so it's quite possible that the claim was made by ASUS and not directly by Intel (i.e. telephone effect). TBF, Intel isn't as transparent about this as they should be (e.g. referring to "29.97"), which only facilitates confusion.

Everyone I spoke with at Intel WRT the IVB BIOS noted that the numbers I got (~SNB) were as good as it was going to get with the current platform. There were never any claims that it was going to be perfect on IVB/7-Series PCH, in fact no one I spoke with has been willing to go on the record for when it will get fixed. Obviously Anandtech's reviewers have better access than I do, but if Intel was making general, public, claims about it there would be no reason for the hesitation to provide a straight answer once they have the scoop.

Have you tried QS implemented in the latest ffdshow? I tested with ffdshow rev4462 and 720p60 playback was much better than QS in LAV. Surely playing back 720p60 with a 1080p display is the most taxing for GPU. Here is a 2-minute clip I used:

 

La Traviata (2010) [2 min 720p video MKV].mkv (1280x720 60fps AVC + DTS-HD Master Audio + PGS subtitles)

 

GPU clock	GPU memory clock	Video Decoder	Rendering time	Dropped frames
1150MHz	800MHz	LAV/avcodec	14.2ms	1
1150MHz	800MHz	LAV/quicksync	14.2ms	9
1150MHz	800MHz	ffdshow/quicksync	14.2ms	1

 

14ms of the average rendering time is a bit high. So I overclocked GPU and memory. I still saw 1 dropped frame whatever decoder I use.

 

GPU clock	GPU memory clock	Video Decoder	Rendering time	Dropped frames
1350MHz	1066MHz	LAV/avcodec	12.2ms	1
1350MHz	1066MHz	LAV/quicksync	12.2ms	1
1350MHz	1066MHz	ffdshow/quicksync	12.2ms	1

Edited by renethx - 2/11/13 at 10:20am

I haven't updated the QS decoder for a while in LAV because of the new and unstable VPP features, its possible that also some other improvements were made in the meantime. The last tests i ran showed that LAV was usually around 1% faster then ffdshow due to optimized internal processing (on the same QS decoder version).
When talking about dropped frames, its important to note if they drop right at the start, or slowly over the course of the file.

I ignored dropped frames during the first few seconds at start (by pressing Ctrl+R).

Sorry if this sounds a bit off-topic but I'd like to perform frame rate tests in a similar fashion and I can't seem to find the software you're all using for this (displayed in the screenshot in the first post of this thread, showing frame rate and lots of other interesting info). Can you point me to it?

Thanks.

MPC-HC

press control + J to get the text overlay

I had no idea you could do that with Media Player Classic. Thank you for the info!

Just finished my HTPC, and it's almost perfect except for a couple of little nagging problems. My HTPC runs to a LG 47LV5500 capable of 60p/24p, and the video card is an Nvidia GTX460 with the latest drivers. I'm running XBMC. I have been testing out PowerDVD 11 and TotalMedia Theatre 5 for Blu-ray playback. I also have Anydvd HD installed to force the computer into 24p when a blu-ray is inserted into the HTPC. When I insert the blu-ray into the system, I can verify with the TV remote that the TV switches from 60p to 24p.

My problems are:

1. If I use TMT5 for playback, the blu-ray with autoplay directly into the movie, however, the TV switches back to 60p.
2. If I use PowerDVD 11, the TV stays in 24p mode, however, the blu-ray will not autoplay. I have to click "Play" once more from within the PowerDVD software.

Both are minor issues, but ultimately I would like the TV to stay in 24p mode AND autoplay the blu-ray (eliminating the second Play click). I don't have a preference on the software. Does anyone have a solution to either force TMT5 to stay in 24p mode, or to have PowerDVD go straight into blu-ray playback?

Thanks.