1 person found this helpful
We need more information: Source material, codec in use, export settings, background processes, tuning performed, etc.
1 person found this helpful
You may be "drive" bound as your system sounds pretty strong for so few hard drives.
I suggest you run the PPBM5 benchmark (PPBM5.com) and report the results back here to validate where your system is using a known reference.
The source is NTSC 29.97fps HDV footage rendering (and scaling) to H.264 25Mbits at 1920 x 1080.
Using: Maximum Render Quality, 2 pass, progressive, Audio 224kbps
Effects from Magic bullet looks have been applied.
I don't have anything major running in the background: Outlook and skype.
The only tuningis that I have overclocked the processor from 3.4 to 4.4 which runs faster, but whether overclocked or not, the issue of not utilizing the full CPU remains.
I just ran the ppbm test and interestingly, during the MPEG2-DVD and H.264 tests the processor did go as high as 95% CPU usage averaging approx 85% going down to about 75%. I realised after, that the project by default had all scratch discs set to run solely from the SSD though so the other drives were not tested.
How do I report the results of the PPBM test? The output.txt have numbers in them but not sure how to interpret them.
I have a bit more homework to test the other scratch discs. Too late for tonight. I will check back tomorrow sometime.
I have already seen your results. Your disk results are mediocre and your H.264 results are very disappointing. The MPEG and render results are quite good. I'll process them in the next couple of days.
The low CPU usage during export to H.264 that you noticed must be the reason of your dissappointing H.264 score, which is near the bottom of the list. You likely have some background processes that mess things up, probably due to not having tuned your system. Have a look here, which also applies to Win7: Adobe Forums: Guide for installing and tuning a Vista...
My initial guess it anti-virus, anti-something, indexing, disk defragging, mDNSresponder, jusched or similar processes that cause these. Process Explorer is a good tool to analyse those processes.
1. I just ran the ppbm test and interestingly, during the MPEG2-DVD and H.264 tests the processor did go as high as 95% CPU usage averaging approx 85% going down to about 75%.
2 I realised after, that the project by default had all scratch discs set to run solely from the SSD though so the other drives were not tested.
1. Different encoding formats use more or less CPU and GPU utilization so there probably is nothing you can do on any specific run to use more CPU unless as Harm said you have a disk or interfering process bottleneck.
2. That is the reason that we use AME directly from a project as opposed to Exporting directly from Premiere, it avoids using the default cache on the OS drive. The optimum place to have and run all the PPBM5 files is your fastest disk drive and specifically if you have a RAID array.
I have older twin of your PC. A months ago I assembled the same CPU, motherboard, size of RAM, OS... Small deviation is in my GPU, I have ASUS nVidia Ge Force GTX 460 1GB only. Bigger deviation is in hard drives. OS is installed on SATA3, source files on Adaptec RAID10 SATA2 and output is going on RAID0 SATA3. I tried to balance power of whole chain. Usage of CPU and speed of encoding depends on many things and even with the same setup I have various values, nevertheless I am reaching up to 90% with single encoding. Better efficiency I am reaching by paralel encoding (2 different files) with Media Encoder and Premiere. CPU usage is touching in such case 100%.
Hi a few days ago I had thought to post it, I have the same computer as the teammatebut a GTX460 @ 4.7 and MPE using colorful II if I use my video in the timeline onlystutters, and a cpu usage of 30 %.
This is normal?
I have two discs, one for OS and one for the project, and the preview footage files.
discs can be problems?
Thanks again for all your precious time.
Below is the essence of my PPBM5 score:
Total Time (Sec.) Relative Performance Index Disk I/O test CPU test MPEG2-DVD CPU test H.264 BR CPU / GPU MPE ON CPU / GPU MPE OFF CPU / GPU Result MPE gain Q1 534.0 Med 320.3 Q1 163.0 D9 33.0 D1 333.0 D9 5.0 Q3 83.0 D9 5.0 16.6 X
It seems my system has a very low score on the H264 BR test (red D1 333.0) which is to do with the exchange of data between the CPU and the memory. As far as I can tell both should be up to scratch. Would you have advice on improving this?
Could it be HDD speed that is somehow interfering with this ?
I was looking to get a WD 2TB black drive, however if much increased HDD speed is what is needed, maybe a pair of 1Tb blacks striped might be of greater asisstance.
To be used a scratch disc perhaps?
I am still trying to understand all this, but I will get there with a little help
woops, also forgot you mentioned I may have backgrounds processors that may be slowing things.
I am yet to go through the tuning guide link you posted previously. Sorry, will get onto this next.
Since your other scores are quite good I would suggest an incorrect setting or some minor error running the H.264 test. Re run the test and double check your settings. Or some other background process kicked off and ruined this test.
1. Your 33 seconds MPEG2-DVD encoding (CPU intensive) score is great.
2. Your Disk intensive score of 163 seconds is probably as good as a two disk system can do. If you really want to improve this score you would have to go to a RAID 0 two drive system.
3. Your GTX 470 is doing its job properly as indicated by the Render Timeline MPE on score.
If you really want to see how good your results could be sort on one of your best numbers like MPEG2-DVD encoding. See how far you jump up the rankings and you will also see theat you are clustered with a bunch of i7-2600K systems and from this I would conclude that your system should score about 70-80 seconds on the H.264 test.
Message was edited by: Bill Gehrke
I plan to run the PPBM CS5 test again. The instructions say to leave the scratch disk location as same as the project ...right ?
Would it not be more optimum for me to put it on an unused drive?
I suggest you try both approaches. The benefit is that you can try different disk setups and see the effect it has on your results. Only make sure that you use the same location for your preview (rendered) files and the export location.
Is there any Benchmark for mac users?
I can't imagine how your sata drives could handicap you in your rendering speed test, 333 seconds is so far
away from your hard disk speeds. the entire PPBM 5 source files are only 120MB of data to be read and any modern hard drive can read all those test clips in under 2 seconds, then they are in your computers cache and should not need to even be accessed again.
you certainly have a strange problem, i would start with your gpu, turn it on or off and see if it changes, maybe it is on and your drivers are doing something strange and slowing down an effect that the gpu is rendering, then your cpu would be just sitting there patiently waiting for its turn which could be your issue.
which nvidia drivers are u using?
Is there any Benchmark for mac users?
Sorry but it is difficult to have a dual platform benchmark specifically when neither Harm of myself have access to a Mac Since before the Mac version was availble we based our disk intensive test on writing out a Microsoft DV AVI file which is not native to Mac and the result data is gathered by a VBS script, again not native to Mac. I suspect that four of the five tests might be possible but I am sure the project files would probably have to be translated, and I know where to find the scores on two of those tests but I have no way of knowing how to get the other two results.
I found this thread exploring a similar problem. My export out of Premiere via Adobe Media Encoder is running WAY too slow (and barely utilizing any system resources).
The machine is a Bloomfield 950 CPU at 3.07GHz with 64-bit Win7 and 12GB of RAM. The harddrive is a 7TB RAID5 Gtech GSpeed esPro (x2) with an SAS connection that regularly reports read/write speeds upwards of 800MB/sec (obviously the disk read/write in this case isn't the issue, as has been suggested in several previous posts).
I have a 102 minute feature in an HD Premiere project sequence. It is simply a clip linked to an uncompressed file (YUV 10bit 4:2:2) and a "title" on a second video timeline to burn in a watermark (thus the "title" is the only thing actually being rendered, per se). When I export to MPEG-DVD using AME, the render takes around 42 hours (what the what?!). When I use the render cue in After Effects with the exact same MPEG-DVD settings, it only takes around 3 hours.
In After Effects, I can adjust how many CPU's are being utilized, what percent of each core, how much RAM, etc.
I don't see anywhere in Adobe Media Center where you can adjust the resources being utilized. This seems to be a massive problem (see afforementioned 42 hour render times).
Is there really no way to adjust the system resources being utilized by AME?
Is it in any way acceptable that an export would take 42 hours via Adobe Media Encoder, but only 3 hours via After Effects?
PP and AME automatically use all your cores as best they can. AFX sets the number of cores as it actually launches another copy of AFX for each core you give it and renders a frame seperately on each core.
my only thought is that it seems you are rendering from HD footage down to SD -DVD.
when i know i am going to a dvd i make my sequence SD and scale down my HD footage to fit the frame. It seems to render much faster than exporting an HD sequence to DVD. Not sure why, i think that premiere is processing the frame in full HD rez then scaling it down to SD then encoding the export. premiere seems to be much faster at scaling the HD video into a SD sequence, then it only has to process the effects and titles and whatever your doing in SD rez so it is much faster then it just encodes it.
You did not mention of you have a MPE card but it should handle the scaling by itself and free up your cpu for the rest of the processsing.