Higher Resolution Codecs and Striped SSDs

[SourceChild]

A point is that larger slower SSDs are cheaper than smaller faster ones.

I mention SSD raid For capacity reasons as opposed to speed. The speed part I can make back up by using a PCI raid controller. I am referring to Raid controllers cards with on board processing.

I'll say I'm jumping the gun a little based on rumors of the up and coming release of the new Mac Pros which are supposed to boast Xeon 5600 6 core chips, USB 3.0, and SATA r3.

As far as audio though, I have never seen a situation where an HD video file didn't drop frames at some point coming from a software based media server. Yes, it's negligible and most people will never notice but it's still not satisfactory for broadcast.

[samsc]

I dont believe there is sufficient performance difference between 'cheaper' and 'more expensive SSD' to make any difference- 3 years ago there was - these days I havent seen anything in any 3rd party performance testing to indicate it would make a difference-
Run test G with different drives and email me-

I cant imagine any PCI raid controller being at all useful with SSD -
these things are designed to deal with things that simply do not occur with SSD- There are no access times with SSD
PCI raid controllers are designed to deal with access time latencies

----

I count drop frames in all my proper testing my performance graphs count drop frame - my testing HUD counts drop frames- i suggest you take a look at performance G - i have done very long term testing-

And run it to see where you are getting dropped frames-

There are a number of Operating system things ( the OS can go away and do something for a while - some house keeping ) - this is minimised by turning off everything you can like spotlight - on all discs

synchronisation things- VGA/DVI display rates not being the same as broadcast -

And performance things - with slower discs- performance is usually the lowest factor these days

A point is that larger slower SSDs are cheaper than smaller faster ones.

I mention SSD raid For capacity reasons as opposed to speed. The speed part I can make back up by using a PCI raid controller. I am referring to Raid controllers cards with on board processing.

I'll say I'm jumping the gun a little based on rumors of the up and coming release of the new Mac Pros which are supposed to boast Xeon 5600 6 core chips, USB 3.0, and SATA r3.

As far as audio though, I have never seen a situation where an HD video file didn't drop frames at some point coming from a software based media server. Yes, it's negligible and most people will never notice but it's still not satisfactory for broadcast.

[SourceChild]

I'm getting some new gear when I get back in May. I plan on doing a great deal of testing.

Run test G with different drives and email me

I currently only have 5 different types of SSD drives.

I cant imagine any PCI raid controller being at all useful with SSD

My hope is that offloading the CPU etc will allow for smaller capture buffers in the future, hopefully getting to less than two frame latency.

i have done very long term testing

I'm hoping to do a big push with some of the broadcast clients. Believe me, I'll be testing thoroughly as well. I hope to collaborate with you on this.

And run it to see where you are getting dropped frames-...
synchronisation things- VGA/DVI display rates not being the same as broadcast

Converting DVI to HD-SDI is always a culprit. Maybe someday there will be reasonably priced HD-SDI video graphics cards for OS X.

As far as operating system, I have scripts to turn dozens of things off but I've found myself shot in the foot for their functionality later.

[samsc]

what you mean 2 frame latency here?

what you referring to?

---

dvi doesnt convert nicely to hdsdi and vice versa... unless you can program dvi that work exactly at hdsdi frame rates...

I'm getting some new gear when I get back in May. I plan on doing a great deal of testing.



I currently only have 5 different types of SSD drives.



My hope is that offloading the CPU etc will allow for smaller capture buffers in the future, hopefully getting to less than two frame latency.



I'm hoping to do a big push with some of the broadcast clients. Believe me, I'll be testing thoroughly as well. I hope to collaborate with you on this.



Converting DVI to HD-SDI is always a culprit. Maybe someday there will be reasonably priced HD-SDI video graphics cards for OS X.

As far as operating system, I have scripts to turn dozens of things off but I've found myself shot in the foot for their functionality later.

[ajmaudio]

What kind of issues are you guys having converting from dvi to hdsdi? I think I will more than likely be doing this in the not so distant future and since you mentioned issues I thought I would ask. On a side note I have noticed that matrox is releasing some new sdi output cards for mac but I dont know if they support openGl etc.

Here is a link: http://www.matrox.com/video/en/press..._multi_io_mac/

If you read the article it would seem that these devices are targeted toward what we do with catalyst. You guys have any input on these?

There are also input cards and cards that do i/o.... would be amazing if these were actually suitable? your thoughts Richard?

BTW... these fit in an xserve supposedly

[SourceChild]

What kind of issues are you guys having converting from dvi to hdsdi?

The biggest issue is having consistent color. Using a scope to test the HD-SDI out from a converted mac shows tons of illegal colors. Also, matching the color is a pain. Sync is an issue as well as frame delay. Since a mac is 60Hz, the multiple is 30fps which is not 59.94 or 29.97 so every 10 seconds is a subtle frame skip. It works but if your director takes Cat as a direct broadcast feed, then the producers won't be happy.

....matrox is releasing some new sdi output cards for mac

I'll probably get and test the first HD-SDI solution that does 59.94 with broadcast color as an output option.

Still though, for all the companies using HD-SDI as their primary pipeline to the projection sources, a video card that works would be great.

BTW... these fit in an xserve supposedly

A card that fits in an xServe is great compared to trying to shoehorn a card that doesn't and having apple reject support because of it.

[samsc]

normal hdsdi - is a lower quality output signal than rgb 4:4:4 from a dvi or vga monitor-

describing an RGB monitor as displaying illegal colour in YUV colour space is
the wrong way round-

normal HDSDI cannot display a full resolution RGB 4:4:4 signal - whatever the refresh rate-
-

HDSDI is YUV encoded and 4:2:2 - same old broadcast bandwidth limiting exercise-
the conversion from RGB to YUV is never lossless - hdsdi is downsampled- its worse than DVI/VGA not better

DVI/vga is higher quality than hdsdi...

broadcast colour is lower quality than RGB 444- i think you have been taking the wrong broadcast based medicine...

http://en.wikipedia.org/wiki/Serial_digital_interface

For all serial digital interfaces (excluding the obsolete composite encodings), the native color encoding is 4:2:2 YCbCr format. The luminance channel (Y) is encoded at full bandwidth (13.5 MHz in 270 Mbit/s SD, ~75 MHz in HD), and the two chrominance channels (Cb and Cr) are subsampled horizontally, and encoded at half bandwidth (6.75 MHz or 37.5 MHz). The Y, Cr, and Cb samples are co-sited (acquired at the same instance in time), and the Y' sample is acquired at the time halfway between two adjacent Y samples.

the use of 29.97 and all the insane resulting confusing is just so confusing to everyone. its not an advantage of hdsdi - its an albatross.

The biggest issue is having consistent color. Using a scope to test the HD-SDI out from a converted mac shows tons of illegal colors. Also, matching the color is a pain. Sync is an issue as well as frame delay. Since a mac is 60Hz, the multiple is 30fps which is not 59.94 or 29.97 so every 10 seconds is a subtle frame skip. It works but if your director takes Cat as a direct broadcast feed, then the producers won't be happy.



I'll probably get and test the first HD-SDI solution that does 59.94 with broadcast color as an output option.

Still though, for all the companies using HD-SDI as their primary pipeline to the projection sources, a video card that works would be great.



A card that fits in an xServe is great compared to trying to shoehorn a card that doesn't and having apple reject support because of it.

[ajmaudio]

The biggest issue is having consistent color. Using a scope to test the HD-SDI out from a converted mac shows tons of illegal colors. Also, matching the color is a pain. Sync is an issue as well as frame delay. Since a mac is 60Hz, the multiple is 30fps which is not 59.94 or 29.97 so every 10 seconds is a subtle frame skip. It works but if your director takes Cat as a direct broadcast feed, then the producers won't be happy.



I'll probably get and test the first HD-SDI solution that does 59.94 with broadcast color as an output option.

Still though, for all the companies using HD-SDI as their primary pipeline to the projection sources, a video card that works would be great.



A card that fits in an xServe is great compared to trying to shoehorn a card that doesn't and having apple reject support because of it.

I havent heard back from the sales engineer I tried to contact regarding those cards... you know anything about them? Or any other solution for that matter. That sync issue you speak of is killing me as well. A broadcast friendly output from catalyst needs to happen.