New Mac Pro real world performance
Posted: Thu Jul 03, 2014 10:02 am
To anyone who may be interested.
Just did a round-robin i/o latency test on my Mac Pro 3 GHz 8-Core Intel Xeon E5 - 32 GB Ram 1867 MHz DDR3 ECC - DP 8.06 - RME HDSPe Madi card - iz ADA converters.
In english ..... that's a new 8 core mac pro with 32GB of ram.
Recording was done at 96Khz with a 64 samples buffer. The delay was crudely measured by line up the delayed audio (phase flipped till null) then comparing the original time stamp to the new adjusted time stamp i.e. no actual test equipment used but still as near as damn it.
Delay incurred was 166 samples i.e. 1.73ms which is pretty awesome in my book.
Latency though is one thing, but what about track count, completely separate issue I know but to those interested, I was recording 24 tracks simultaneously at 96Khz and 64 samples buffer and the Mac was fine. I was recording to an external TB "G drive" sata drive.
Gonna get a TB2 SSD when they are cheap enough, not sure what affect if any on the machines performance that will have, but given that over fragmented drives are an issue then so too is disk access speed so anything that is faster is better, (I think?).
It'll certainly make backups (to the same) a lot quicker!
For the record, I'm not a techy, just a jobbing engineer so I bow to anyone's else's superior technical knowledge on this. But this was a real world test and the delay calculation is correct.
When I do a session in anger that'll be the acid test.
This machine is a monster
Regards
Sean
Just did a round-robin i/o latency test on my Mac Pro 3 GHz 8-Core Intel Xeon E5 - 32 GB Ram 1867 MHz DDR3 ECC - DP 8.06 - RME HDSPe Madi card - iz ADA converters.
In english ..... that's a new 8 core mac pro with 32GB of ram.
Recording was done at 96Khz with a 64 samples buffer. The delay was crudely measured by line up the delayed audio (phase flipped till null) then comparing the original time stamp to the new adjusted time stamp i.e. no actual test equipment used but still as near as damn it.
Delay incurred was 166 samples i.e. 1.73ms which is pretty awesome in my book.
Latency though is one thing, but what about track count, completely separate issue I know but to those interested, I was recording 24 tracks simultaneously at 96Khz and 64 samples buffer and the Mac was fine. I was recording to an external TB "G drive" sata drive.
Gonna get a TB2 SSD when they are cheap enough, not sure what affect if any on the machines performance that will have, but given that over fragmented drives are an issue then so too is disk access speed so anything that is faster is better, (I think?).
It'll certainly make backups (to the same) a lot quicker!
For the record, I'm not a techy, just a jobbing engineer so I bow to anyone's else's superior technical knowledge on this. But this was a real world test and the delay calculation is correct.
When I do a session in anger that'll be the acid test.
This machine is a monster
Regards
Sean