Welcome to our community

Be a part of something great, join today!

  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Mac Pro 8 core 2.93 Transcode times in Compressor/Qmaster

Thanks a lot. This is driving me crazy.

What format were you attempting to export to? I had no problem exporting to 4k tif/dpx. Let me know what your output settings are and I will try to duplicate the same.
 
Wow. How can you guys afford that new 2.93 tower? Expensive isn't it?
Perhaps the way to go is to combine two of the older 2.8 Ghz towers linked
with Qmaster to do computational etc. I think you could buy (2) 2.8 Ghz towers
for the price of (1) new 8-core 2.93 chipset tower.
 
Wow. How can you guys afford that new 2.93 tower? Expensive isn't it?
Perhaps the way to go is to combine two of the older 2.8 Ghz towers linked
with Qmaster to do computational etc. I think you could buy (2) 2.8 Ghz towers
for the price of (1) new 8-core 2.93 chipset tower.

I'm a professional gambler and part time prostitute.
 
Nehalem Mac Pro arrived today so the first thing I did was a transcode in Compressor utilizing Qmaster clusters. The transcode was RED footage (4096x2304 Res) to Apple ProRes 422 (1920x1080). The Nehalem did the job in 2 minutes 15 seconds. The 2007 Mac Pro 3GHz in 3 minutes 42 seconds. 2007 Mac Pro 8 core 3GHz has 16GB RAM and RAID 5 array with 4x750GB drives. This is an amazing trouncing. The Nehalem only has 6GB and no RAID.

The 2007 Mac Pro 8 core 3GHz was setup with at first 4 instances and then 8 instances. Time for 4 instances was 03:55 and time for 8 instances was 03:42. The Nehalem was set for 16 instances.

Screen capture: http://farm4.static.flickr.com/3645/3361114864_009f031f5e_o.png

Can't wait to do the same test with 16GB of RAM in the Nehalem & RAID 5.

Agree! This was also the first major thing I did when my 2.93Ghz 12GB rolled in was to perform a series of multipass segmented transcodes with Compressor.app and QMASTER. It is indeed impressive over the MAC PRO 2007 8xCOre I had before.

In short:
  • my source and targets where on two seperate SAS Disk arrays to insure there was no latancy at all - yes we all know the I/O component is not a big deal in transcoding, however I wanted to make sure that this was not an issue
  • initially use the R.O.T of 2 cores / compressor instance
  • fed compressor a variey of R3D, PRORES & DVCPROHD in various states transcode into H.264 Mulipass in various resolutions and bit rates (old jobs that had been slow on the previus machine)

Found the 16 cores were running only just over 50-60% so I cranked up the number of instanced of compressor to 14 (fourteen) just to see what it would do and allowing 2 x virtual cores available so my MACPRO would not lock up (as it used to when used too many instances).

To be sure, the old R.O.T of two cores per compressor instance does not stand anymore with these new machines.. at leats for the model I have.

Well.. all I can say is Wah! (amazingly fast)

NO accurate times yet but these observations are so far:
  • multipass segented transcoding is approximately 2-3 times faster than my previosu machine (same as as the one used by Facetmedia)
  • all cores being utilised between 95%-100%
  • during these transcodes, MACOS and the pother .apps was very responsive and I could edit freely in FCP including doing reference movie exports to be ready for compressor and do other chores while all this work was going on
  • havenot tried all 16 instances of compressor yet.
  • have not tried SHAKE.app with 16 instances yet - will post
  • will try cpu bound operations today (SHAKE smoothcam etc etc)
  • will try with some other QMASTER nodes
  • also with some output/package to lossless motion jpeg2000 - just curious on the encode and then the playout for LMJP2K too.

For me this was certainly worth the money for the reduced workflow time!
fwiw

w
 
Warwick,

Great to hear your feedback. I'll have to install Shake tonight and give it a spin. I was playing around in AE (rendering out 4k DPX's/TIFs)... having only 6Gb is affecting performance per core. The good news is that OWC dropped the prices of their 4GB modules by a few hundred dollars. I will get 6x4GB modules to maintain triple channel performance once they start shipping the modules... can't wait. I don't think I've ever been so excited about computing and the RED camera before. I know this year will be even more revolutionary.

O' BTW... do you have CS4? What's performance per core looking like? Here's a screen grab from my little transcode out to 4k dpx: http://farm4.static.flickr.com/3596/3368131583_0b6b6971db_b.jpg
 
hi Andrae cool! regarding cs4. no I have not yet upgraded from cs3. sorry mate. I am also curious for this too

further. I am quite curious about the increased efficiciencies of snowleopard's opencl and 'grand central' to make the utilisation of all cores accessible to all .apps. this trully I trully intrigues me. I am certain we will we further workflow efficiencies when sn hits us

please post your results and timings and details for the stuff you do

I will do the same

I realise there are many doubters, cynics and those unsure whether to move to these new mac pros.

sadly many still see CPU clock speed as the defining point for running their business rather than the actual workflow efficiencies that help profitability and business growth (my rant from my iPhone! )

w
 
What format were you attempting to export to? I had no problem exporting to 4k tif/dpx. Let me know what your output settings are and I will try to duplicate the same.

Thanks, After a lot of testing, reinstalling, trashing plugins I was able to export 4K footage. AE CS4 still is a little buggy in my opinion, and I don't like very much the multiprocessor and how it handles the 16 CPUs. Hopefully with snow leopard will be better.

One plugin that won't render 4K properly in my Mac is the Neat Video. It only exports if the temporal adjustment is set to 0. DO you have that plugin to test?

But for now I can live with that. Waiting for my 16 GB Ram from OMc. Hopefully that will make a difference in the multiprocessing as well compared to my actual 6GB. 32 is way too expensive for me now.

Thanks for your help.

IC
 
But for now I can live with that. Waiting for my 16 GB Ram from OMc. Hopefully that will make a difference in the multiprocessing as well compared to my actual 6GB. 32 is way too expensive for me now.

Thanks for your help.

IC

No I don't have the plugin. I'm in the same boat as you regarding memory. I'm attempting to get 6x4GB because from reports... filling up all 8 slots may slow your memory:

http://barefeats.com/nehal04.html

It's great that OWC dropped their prices. Wish they had it available now though.
 
No I don't have the plugin. I'm in the same boat as you regarding memory. I'm attempting to get 6x4GB because from reports... filling up all 8 slots may slow your memory:

http://barefeats.com/nehal04.html

It's great that OWC dropped their prices. Wish they had it available now though.

Well I didn't know that. But 3,000$ for 24GB RAM I definitively can't do it.

And hopefully 16RAM will be better than my actual situation.

Memory.jpg


I still don't get how AE doesn't have a better system of administrating the cores and use them all for rendering. Maybe I am missing something here. This is my AE setting.

AE-settings.jpg
 
I'm getting the same results in AE CS4 as shown by my earlier screen grab in this thread. Either its memory related or not optimized.

3368131583_0b6b6971db.jpg
 
Andrae and Warwickt,

Can you tell me what video card you got with the Mac Pro? Did you get the ATI Radeon HD 4870 512MB or wait for a third party option?

I'm looking to get one soon, but would like some advice.

Thanks!

Jim
 
Andrae and Warwickt,

Can you tell me what video card you got with the Mac Pro? Did you get the ATI Radeon HD 4870 512MB or wait for a third party option?

I'm looking to get one soon, but would like some advice.

Thanks!

Jim

ATI4870... from my test these cards are not optimized as of yet for REDCine & Color. Actually they are slower than the ATI X1900XT. As usual we will have to wait for software to catch up with hardware. When I first got a 2007 Mac Pro it was slower in Qmaster doing transcodes than my friends Quad Mac Pro. Took months for Apple to fix it. Hopefully this won't be the same repeat. Well at least the new Mac Pro Nehalems are screaming fast in Qmaster... fast playback in color/REDCine is the problem.
 
Andrae and Warwickt,

Can you tell me what video card you got with the Mac Pro? Did you get the ATI Radeon HD 4870 512MB or wait for a third party option?

I'm looking to get one soon, but would like some advice.

Thanks!

Jim

Hi Jim, I have the RADION 4870 graphics card in this MACPRO 2009. I ordered it withe MAC PRO as a BTO from the Applestore in Hong Kong.
w
 
Thanks to both of you for responding. Andrae, you mentioned you didn't think the card was optimized. Does this mean it wasn't worth the extra money, or do you think this will be resolved with an update?

I don't want to spend an extra $450 on the 4870 if I'm just going to replace it with a 3rd party a month later.

Any final thoughts on whether it was worth it?
 
Thanks to both of you for responding. Andrae, you mentioned you didn't think the card was optimized. Does this mean it wasn't worth the extra money, or do you think this will be resolved with an update?

I don't want to spend an extra $450 on the 4870 if I'm just going to replace it with a 3rd party a month later.

Any final thoughts on whether it was worth it?
G'day Jim. If you can wait for a few months to make a final judgement until SNOW LEOPARD and THEN the rumoured new FCS3 comes out soon after to see if there is more obvious utilisation of the Nehalem MACPRO Graphics card's GPU components in the rendering and transcoding (and other) processes in Compressor, Motion , QMaster and what ever else is in there (and the possible shake replacement - rumour).

Just a thought.

For me the AT RAdion HD4870 was worth the money.

As always is a business decision.

hth
w
 
Thanks to both of you for responding. Andrae, you mentioned you didn't think the card was optimized. Does this mean it wasn't worth the extra money, or do you think this will be resolved with an update?

I don't want to spend an extra $450 on the 4870 if I'm just going to replace it with a 3rd party a month later.

Any final thoughts on whether it was worth it?

It's worth the money for the apps that can utilize it now. For example Motion is benchmarked as being significantly faster with the 4870. As for the rest we'll have to wait for FCS3, SL... perhaps there will be a minor update like 10.5.7 or 10.5.8 with graphic card improvements.
 
Back
Top