Based on what I've read about SATA, my Western Digital 1 TB Black Caviar drives should be able to read and write faster than 50 MB/sec. But that's the best I've gotten from anything. I thought I could expect 300 MB/sec, or 150 at the very least. What's the deal? This is a 2008 8 core 2.8 Mac Pro, by the way.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Mac Pro SATA copying speeds seem too slow
- Thread starter clamnectar
- Start date
- Sort by reaction score
Where did you read 300 MB/sec? Perhaps in a RAID configuration.
according to Hardware Canucks "Darn near 97MB/second read speed"
http://www.hardwarecanucks.com/foru...tal-caviar-black-1tb-hard-drive-review-6.html
What program are using to test the performance?
Well, there's various speeds a typical drive can deliver:
Sustained I/O
Burst I/O
Cache I/O
Random I/O
Copying many small files from one drive to another is an example of Random I/O - Usually about 5 ~ 50 MB/s.
Copying a large file or loading one is an example of Sustained I/O - Usually 70 ~ 120 MB/s
Cache I/O is reading or loading anything already in the drive's cache - can be as high as 2 GB/s (mine typically hits 1.3 ~ 2 GB/s)
And I'm not too sure when or why burst kicks in nor what speeds are achieved - others might know.
There are I/O accelerators but I don't think Mac OS X supports them. They're usually around 500 GB or less (optionally battery backed) and have their own processor. It's a bit like adding a massive L2 cache to the SAS or SATA bus.
Sustained I/O
Burst I/O
Cache I/O
Random I/O
Copying many small files from one drive to another is an example of Random I/O - Usually about 5 ~ 50 MB/s.
Copying a large file or loading one is an example of Sustained I/O - Usually 70 ~ 120 MB/s
Cache I/O is reading or loading anything already in the drive's cache - can be as high as 2 GB/s (mine typically hits 1.3 ~ 2 GB/s)
And I'm not too sure when or why burst kicks in nor what speeds are achieved - others might know.
There are I/O accelerators but I don't think Mac OS X supports them. They're usually around 500 GB or less (optionally battery backed) and have their own processor. It's a bit like adding a massive L2 cache to the SAS or SATA bus.
50 sounds about right for a single drive. It's unlikely to give you 150 mb/s sustained transfer in real world apps. I only get about 150 mb/s on a 3 drive RAID 0 array.
Wow! That's pretty bad! I'm getting between 250 MB/s and 400 MB/s tho usually on the lower end. (3 Drive RAID 0)
this is time machine backing up / duplicating a single 4 GB file. From the sounds of it, I'm not too far off the curve, so I guess everything's working correctly.
my iMac gets 80MB sustained read/write, no RAID. does that count??
well there you have it. you can't go off Time machine benchmarks because it runs REALLY really slowly (has to update the database, runs as a lower priority task blaablaa). copy a large file to another folder or something and see how you go.
duplicating a 4 gigabyte file: zigzagging between 60 MB/sec and 40 MB/sec, as measured by activity monitor. the graph looks like a mountain range. duplicated the same file three times, always the same speed.
So, I just want to understand the question. Your assumption was that the hard drive you purchased would preform better in a mac pro than it would in test machines that yielded 50MB/sec, and you want to know "what gives"?
Also the numbers you mention 300MB/s and 150 sound suspiciously like you're anticipating your drive throughput to be a direct product of the SATA bandwidth.
Also the numbers you mention 300MB/s and 150 sound suspiciously like you're anticipating your drive throughput to be a direct product of the SATA bandwidth.
my guess is because its is copying to the same HD, has to work a lot harder.. maybe try to copy to another disk/external drive?
Yeah, I think that's about right for "duplicating" a file on the same HDD.
Knowing BSD tho I bet there's a system variable hiding out somewhere that will allow the buffer size to be increased. This might speed it up a bit. Maybe zig-zag between 60 and 90 MB/s or something.
Also a file that size and you can time it pretty accurately with a stopwatch.you're indeed right. i'm not sure why i didn't think of it, but duplicating is slower because the drive heads are reading and writing. copying that same 4 gig file to one of the other drives gives me 100 MB/sec. definitely a solid speed.
thanks for clearing up some of my misconceptions.
thanks for clearing up some of my misconceptions.
I managed to get 105Mb/sec between two WD Blacks 1TB.
They were totally empty, I was seeing how quick I could get them
12Gb H264 file, peaked at 106.4, min at like 87 towards the end.
Blazing fast!
Even managed to get 66Mb/sec across Gigabit
They were totally empty, I was seeing how quick I could get them
12Gb H264 file, peaked at 106.4, min at like 87 towards the end.
Blazing fast!
Even managed to get 66Mb/sec across Gigabit
Kewl! Data type also matters as I believe all HDDs these days are RLE. RLE is a kind of compression. So for example if you're doing that with a 12GB uncompressed AVI of nothing but a solid color it should be considerably faster than if your 12GB AVI is of already compressed data or contains temporal deltoid change.