Back at work and just like my home system "Linked Source Scaling Quality" is set to full. Hmmmmmm
---In Avid-L2@yahoogroups.com, <bogdan_grigorescu@...> wrote :
could you please check and confirm Scaling/Quality under Mixdown/Transcode and Render tabs in MediaCreation settings?
on both systems.
BG
www.finale.tv
From: "bigfish@... [Avid-L2]" <Avid-L2@yahoogroups.com>
To: Avid-L2@yahoogroups.com
Sent: Tuesday, December 6, 2016 9:12 PM
Subject: Re: [Avid-L2] Difference of CPU Usage when transcoding Red to DNxHRHQX?
They are both set to Full in the linked media setting. I copied the project and brought it home so I have to assume they are both set the same way. Very interesting that my home system seems to be cranking so much faster. Could the fact that the source and target drive were each hooked up USB3 to the same card create that much of a bottleneck? I wouldn't think so given on the home system the source drive is now connected to the front usb 2 port on the Mac Pro. I would think that would hold back performance more than two usb 3 devices to the same card. At home the target drive is feed Esata from ports on my PCIe SSD in slot 4 4 lane. I would think that would suggest that the two drives are on different busses but could that account for a 1500% difference in CPU usage?
I have the Red Plug in set to Camera metadata. On the 3 clips I've transcoded to DNxHRHQX the shortest clip of 19 minutes took 33 minutes to transcode, the middle clip of 36.5 minutes took 64 minutes to transcode and the longest clip of 42 minutes took 75 minutes to transcode. Thats 1.73 x Real Time, 1.75 x Real Time and 1.78 x Real time. That seems to exceed the expectations from the Avid workflow pdf but that was written at the time of Avid 6.5 so I assume there have been further improvements in Avid's transcode engine. According to the pdf the Red Rocket with Avid only kicked in at full debayer setting and would yield 2 times real time transcodes but again that was at the time of 6.5 Avid.
I know my GTX-680 is generally a better GPU than the Radeon with more cuda cores that Resolve and Adobe like. I know some aspects of Avid are enhanced by GPUs when it comes to real time effects processing but I'm not sure that transcoding is one of them.
I have the Red Plug in set to Camera metadata. On the 3 clips I've transcoded to DNxHRHQX the shortest clip of 19 minutes took 33 minutes to transcode, the middle clip of 36.5 minutes took 64 minutes to transcode and the longest clip of 42 minutes took 75 minutes to transcode. Thats 1.73 x Real Time, 1.75 x Real Time and 1.78 x Real time. That seems to exceed the expectations from the Avid workflow pdf but that was written at the time of Avid 6.5 so I assume there have been further improvements in Avid's transcode engine. According to the pdf the Red Rocket with Avid only kicked in at full debayer setting and would yield 2 times real time transcodes but again that was at the time of 6.5 Avid.
I know my GTX-680 is generally a better GPU than the Radeon with more cuda cores that Resolve and Adobe like. I know some aspects of Avid are enhanced by GPUs when it comes to real time effects processing but I'm not sure that transcoding is one of them.
---In Avid-L2@yahoogroups.com, <bogdan_grigorescu@...> wrote :
what are the debayer settings on each system?(in media creation)
BG
www.finale.tv
From: "John Moore bigfish@... [Avid-L2]" <Avid-L2@yahoogroups.com>
To: Yahoogroups <avid-l2@yahoogroups.com>; Yahoogroups <editing-list@yahoogroups.com>
Sent: Tuesday, December 6, 2016 6:33 PM
Subject: [Avid-L2] Difference of CPU Usage when transcoding Red to DNxHRHQX?
On my latest Red shot 4K project I did some testing on transcoding to DNxHRHQX on my work system. A mac pro Tower mid 2010 OS 10.9.5 upgraded to 12 core 3.33GHz 24GB of ram Radeon 5770 GPU Avid 8.5.2. On that system the clips transcoded at 5.6 times real time. So a 40 minute clip took around 4 hours to transcode. When I looked at the Activity monitor the CPU usage for the background transcode dithered around 450%, so around 4 and half cores. When I tried a normal foreground transcode the CPU usage was around 550%.
I took the drives home to my system a mac pro tower mid 2012 OS 10.9.5 12core 2.66GHz 64GB ram nvidia GTX 680 Avid 8.5.3. Doing a foreground normal transcode on my home system the CPU usage is dithering around 2,100% so about 21 cores. Avid projects the first clip will be done in just over an hour so that seems to be a lot faster than at work.
At work the source and target drives were connected USB 3 hooked to the same card. At home the source drive is connected to USB 3 but plugged into the Mac Pro Tower front USB 2.0 port and the target Graid drive is hooked to an Esata port that is part of my PCIe SSD drive in slot 4 4lane on the Mac Pro.
I have seen my work system crank up to 2,000 percent ish cpu when transcoding Panasonic Varicam 35 4K AVC-Intra but this is the first time I've worked with Red in a 4K world.
So why is my home system cranking sooo much faster?
I read parts of Paul Sampson's Red Workflow Avid guide and it said that with a Red Rocket card Avid can be set to full debayer quality and the Red Rocket will kick to provide about 2 times realtime Red Raw transcode. My home system seems to be outperforming that with 4K material. I'm curious why there is such a performance difference between the two seemingly similar systems.
John Moore Barking Trout Productions Studio City, CA bigfish@...
__._,_.___
Posted by: bigfish@pacbell.net
| Reply via web post | • | Reply to sender | • | Reply to group | • | Start a New Topic | • | Messages in this topic (5) |
Have you tried the highest rated email app?
With 4.5 stars in iTunes, the Yahoo Mail app is the highest rated email app on the market. What are you waiting for? Now you can access all your inboxes (Gmail, Outlook, AOL and more) in one place. Never delete an email again with 1000GB of free cloud storage.
this is the Avid-L2
.
__,_._,___
No comments:
Post a Comment