Ericbowen
Forum Replies Created
-
I linked the Cineraid enclosures since they use hardware LSI SAS raid on the 8 bay units. Software raid will always be slower than Hardware raid especially in parity raids and rebuild times. The 4 Bay OWC unit is TB2 which is an advantage over the 4Bay Cineraid unit which is TB1. I would recommend looking at the 8 bay unit though and just putting in 4 drives to start if the budget is more limited. Desktop HDD’s are not a good idea for parity raids due to them lacking the timeout recovery feature enterprise drives have. Often times a drive will take to long in a repair operation and the controller will prematurely mark it out of the raid since it doesn’t respond during repair operations. Enterprise drives are more expensive but have the timeout recovery feature. Parity raids such as raid 5 and 6 show this problem far more than Raid 0, 1, or 10. Keep that in mind regardless of what unit you look at. Software raids will take long periods of time to rebuild degraded arrays.
Tero is correct regarding 4K compressed versus uncompressed. A 4 Drive raid 5 will normally net you 450 to 500MB/s average transfer rate. Burst may be over 600MB/s but that drops as soon as the cache is full. The Cineraid units also have 1GB onboard cache which significantly improves performance over time with read and write operations. The lower the cache the quicker the burst speed performance drops.
Eric-ADK
Tech Manager
support@adkvideoediting.com -
-
That system is not really worth upgrading at this point especially since even the lowest end Haswell system whether mac or PC is several times faster. I really suggest you take a look at new system options. Nothing added to that system is going to improve your workflow enough based on your information so far to justify any money spent.
Eric-ADK
Tech Manager
support@adkvideoediting.com -
Do not get any CPU under 2.6GHz for realtime editing applications. The latency of the processing for media is to high to handle realtime playback well. I would suggest the I7 6 Core versus the Dual Xeons with the applications you mention especially considering the Xeons being considered. If you want a Xeon then make sure the clock speed is over 2.6Ghz and preferably 3Ghz and above.
Eric-ADK
Tech Manager
support@adkvideoediting.com -
What export codecs are you testing with? Are you exporting to PNG? What is your CPU load with Multiprocessing on and exporting to AVI Lossless? What media are you exporting from Premiere and to what codec?
Eric-ADK
Tech Manager
support@adkvideoediting.com -
Look online for one of the Mac flashed 700 series Nvidia cards or get a 680GTX for Mac.
Eric-ADK
Tech Manager
support@adkvideoediting.com -
Yes you need to delete that file in the Premiere and Adobe Media Encoder directory. What Nvidia driver are you running?
Right click on the Desktop and select the Nvidia Control Panel. Then select help and then system information.
Eric-ADK
Tech Manager
support@adkvideoediting.com -
If you use C4D then I would highly suggest you take a look at Octane for GPU acceleration.
Eric-ADK
Tech Manager
support@adkvideoediting.com -
Comps of that many layers are really going to require some CPU threads. A 6 Core at 4.4GHz would be the minimum I would suggest on a X79 board with 64GB of ram. A 10 Core 2690 V2 Xeon would be better. I would definitely suggest a 780Ti card over that 580 card. I would not bother getting another 580 at this point. I suggest starting with the Asrock Extreme 9 for boards. Just so you know Ray Tracer is Nvidia only. Stay away from AMD cards.
Eric-ADK
Tech Manager
support@adkvideoediting.com -
The Drive is still handling read requests. Remember the Hardware MPE player is buffering frames ahead based on how much ram you have, GPU in the system, and codec of media played along with how many layers. When you hit play it has to start buffering realtime where it left off the last time it cached. It doesn’t wait to cache till the previous cached amount is played through. It immediately starts caching as you play. Other wise your video would stutter every time the cache frame data was played through and it had to completely re-cache again. It’s much more efficient to cache data as you go based on where it left off with the original cache of frames. When you jump the transport to a new area it has to re-cache again completely which is often why it takes up to a few seconds before the program monitor reflects the change to the new start frame. The entire MPE engine ie hardware acceleration player is creating a new cache frame list and building data buffers for the GPU acceleration data. This process takes time and you will see it. Once you hit play though as long as the cache ahead is far enough to give enough time for new frame buffers to be created the editor wont see any drops in playback as this process continues until playback stops.
Eric-ADK
Tech Manager
support@adkvideoediting.com