While rendering to H264 BR my CPU usage is around 75-80% and takes about 1.4 x the timeline.
I only tried MPEGII so far. Which CPU do you have? The same?
Sounds like you're only using one core to render at the moment. Double-check your preferences - there's a setting for Optimize Rendering for Performance/Memory - if it's set to memory, it'll only use a maximum of two threads for rendering.
If it's already set for performance mode, it might be bottlenecking on writing preview files to disk. Where is your scratch disk set to for your preview files? If it's a slow drive, that could also be problematic.
Yes and it is showing all 8 cores at around 75-80%.
that sounds like a good idea. I searched the settings in premiere, media encoder and the rendering job settings without success. Could you give me a hint where that setting can be found?
Ok, I found the setting. But it was set to performance already...
Regarding the drive speed: 85MB/s, 6ms access
For me, it uses 2 cores at about 20% and another 2 at about 50%.
Okay, I "fixed" it. For that test, I used the auto-generated opening credits. For some reason, I ended up in the described problem. I now tried some real footage and I finally all cores are used.
I now tried some real footage and I finally all cores are used.
That is often the best way, use real footage.
I still wonder about your disk. 85 MB/s is rather average for a disk that is not fully filled, 6 ms access time is indicative of a SCSI disk. SATA disks usually have 10+ ms access time. How did you measure this?
It's a 1.5TB Samsung SATA drive. The test was done with sandra.
Could there be a mistake somewhere? I ran the Sandra benchmark on physical disks and got these results:
With 6.00 ms access time your access time would be the same as a 15K SCSI-3 disk.
All the SATA disks, including my Areca raid show access times around 13 to 16 ms.
BTW, Sandra gave my raid a drive index of 1479 MB/s. It shows that interpreting these results from different benchmarks it a difficult job. The results vary so much between programs. For instance:
SANDRA: 1479 MB/s, but indicates the size correctly as 10 TB
HD Tune Pro: 745 MB/s, but indicates the size INcorrectly as 2199 GB
HD Tach: 802 MB/s
all on the same raid array. The same applies to access times, with Sandra 13.2 ms, with HD Tune Pro 9.1 ms. All my single SATA disks show around 13.5 ms access time in HD Tune Pro. The drive index for single disks do not vary as much, both Sandra and HD Tune Pro show around 91 MB/s.
From what I have seen it looks like HD Tune Pro is the most representative benchmark for disk performance, more so than Sandra.
I just tried it again, and it shows 12.5ms now. Maybe I mixed up the dots in a hurry :-)
Thanks for the clarification. I reran the Sandra benchmark and the performance increased to 1531 MB/s.