|
|
Post by Johnkenn on Sept 2, 2022 9:53:44 GMT -6
I have a Sandisk Extreme - Performance specifications: Up to 1050MB/s read and 1000MB/s write speed...that should be plenty fast enough. But with SD3, I'm getting dropouts at even 128 buffers. Also have a new Mac Mini M1...any thoughts?
|
|
ericn
Temp
Balance Engineer
Posts: 14,953 
Member is Online
|
Post by ericn on Sept 2, 2022 10:14:20 GMT -6
I have a Sandisk Extreme - Performance specifications: Up to 1050MB/s read and 1000MB/s write speed...that should be plenty fast enough. But with SD3, I'm getting dropouts at even 128 buffers. Also have a new Mac Mini M1...any thoughts? First thing I would try is reformat and reinstall. |
|
|
|
Post by sirthought on Sept 2, 2022 12:15:35 GMT -6
It's going to involve the wired connection. I've seen people complain with some of Apples that they hook up to some TB hubs with USB ports and get faster transfer speed than hooking up direct to the USB port. I can't explain this, but test your drive"s speed. The advertised specs are rarely for real.
|
|
|
|
Post by Johnkenn on Sept 2, 2022 12:20:11 GMT -6
yeah...have it hooked up directly to a port in the back of the mac mini...
|
|
|
|
Post by Johnkenn on Sept 2, 2022 12:22:13 GMT -6
It works st 512 and up...but I never had issues with 128 even when I had the samples on a mechanical HD. I would think the Mac Mini would have plenty of power...
|
|
|
|
Post by sirthought on Sept 2, 2022 12:25:21 GMT -6
Is it formatted in APFS?
|
|
|
|
Post by Johnkenn on Sept 2, 2022 12:47:50 GMT -6
Yeah...looks like it...is that bad? |
|
|
|
Post by wiz on Sept 2, 2022 15:23:41 GMT -6
I have a Sandisk Extreme - Performance specifications: Up to 1050MB/s read and 1000MB/s write speed...that should be plenty fast enough. But with SD3, I'm getting dropouts at even 128 buffers. Also have a new Mac Mini M1...any thoughts? SD3 loads the samples into memory….doesn’t it? |
|
|
|
Post by Johnkenn on Sept 2, 2022 19:04:04 GMT -6
Well, it’s not cached…but I think it’s better now that I changed the performance to 5 cores.
|
|
|
|
Post by wiz on Sept 2, 2022 19:10:08 GMT -6
Well, it’s not cached…but I think it’s better now that I changed the performance to 5 cores. What I meant was.... that when I load my project... SD3 loads up all the samples for the kit then.... (not if you select cached of course) so I was wondering how play back from the SSD drive could be an issue... cheers Wiz |
|
|
|
Post by Johnkenn on Sept 2, 2022 19:52:28 GMT -6
Well, it’s not cached…but I think it’s better now that I changed the performance to 5 cores. What I meant was.... that when I load my project... SD3 loads up all the samples for the kit then.... (not if you select cached of course) so I was wondering how play back from the SSD drive could be an issue... cheers Wiz No idea… |
|
|
|
Post by Johnkenn on Sept 2, 2022 19:53:41 GMT -6
I should say - I’m talking about triggering while I’m playing. There’s no problem when I select a groove or playback.
|
|
|
|
Post by wiz on Sept 2, 2022 20:05:08 GMT -6
I should say - I’m talking about triggering while I’m playing. There’s no problem when I select a groove or playback. Aaah Now I get you |
|
|
|
Post by sirthought on Sept 3, 2022 11:59:27 GMT -6
Yeah...looks like it...is that bad? No, that is good. |
|
|
|
Post by popmann on Sept 5, 2022 21:10:07 GMT -6
This is more likely the M1's single core performance in combination with how whatever specific DAW handles threading of input monitoring.
I'd bet if you input monitor enable the track (so you COULD tap in some beats)...and THEN play some of their stock loops, it should break up just the same. It's that when you're browsing their loops you don't have it input enabled like when you're tapping in a beat--thus it's put on a different part of the engine that fully multithreads.
You can disable all the audio DSP in the SD3 instance and see if that helps. Definitely don't be sending the VI track to a DAW reverb aux, as that would cause the entire reverb aux to be processed on the real time core on TOP of the VI...
You could temporarily move a library to the internal storage--rule the USB controller and drive out in one fell swoop, but I'd bet that isn't it. Certainly not with SD3, which while no--doesn't load "everything" into RAM, it loads exponentially MORE than most VIs.
Or (for troubleshooting) try a different DAW. While I know that's not the permanent solution...Cubase and Logic Studio One and Protools all buffer input a little differently. Setting it to 128, I assume you're using Protools, because Cubase doesn't need it that low...and the LUNA can't go higher, which you say 512+ works fine. Anyway--you can't run them all with the same numbers/settings...not and expect them to run ideally in terms of performance and latency.
|
|
|
|
Post by Johnkenn on Sept 6, 2022 7:20:51 GMT -6
I was just using my new M-Audio controller. Need to try Maschine to see if it does the same thing. Don’t have any issues with any other VIs.
Seems to be better since I set the threading to 5 cores in SD3.
|
|
|
|
Post by popmann on Sept 6, 2022 7:40:24 GMT -6
I cant imagine why a controller would matter. That little bit of data that is midi is what it is….
If it got better changing threading, its the M1 running out of DSP. Is thats a new SD3 setting for Apple Silicon? Never seen that.
|
|
|
|
Post by Johnkenn on Sept 6, 2022 8:11:06 GMT -6
I cant imagine why a controller would matter. That little bit of data that is midi is what it is…. If it got better changing threading, its the M1 running out of DSP. Is thats a new SD3 setting for Apple Silicon? Never seen that. I’m up to date with everything. Maybe I should move one of the libraries to a different drive and see if that works better. |
|
|
|
Post by bluesholyman on Sept 6, 2022 10:24:28 GMT -6
Is that same drive used for recording as well? Not clear if that is happening, unless I missed that above somewhere. I have my samples drive and I have a separate "session" drive that is a Samsung T7. The OS runs on its own drive. This particular machine is an i7 MBP, so not entirely the same....but....
I am just wondering if while its trying to read samples off the drive, if something else is trying to write to it. Just a thought that crossed my mind.
Carvel
|
|
|
|
Post by Johnkenn on Sept 6, 2022 12:19:41 GMT -6
Is that same drive used for recording as well? Not clear if that is happening, unless I missed that above somewhere. I have my samples drive and I have a separate "session" drive that is a Samsung T7. The OS runs on its own drive. This particular machine is an i7 MBP, so not entirely the same....but.... I am just wondering if while its trying to read samples off the drive, if something else is trying to write to it. Just a thought that crossed my mind. Carvel Separate drive specifically for samples |
|
|
|
Post by Johnkenn on Sept 6, 2022 12:20:36 GMT -6
I should clarify it’s a Mac Studio. 32gb ram. Can’t be lack of memory.
|
|
|
|
Post by Johnkenn on Sept 6, 2022 12:38:06 GMT -6
OK - the problem might be the "write-to drive" I'm using. It's an internal SSD drive, but I'm using a USB enclosure. When I measure the read write speed of the Scandisk sample drive, it's like 900mb/s. The Samsung write-to drive is like 120mb/sec.
|
|
|
|
Post by dok on Sept 6, 2022 13:42:10 GMT -6
OK - the problem might be the "write-to drive" I'm using. It's an internal SSD drive, but I'm using a USB enclosure. When I measure the read write speed of the Scandisk sample drive, it's like 900mb/s. The Samsung write-to drive is like 120mb/sec. That's the ticket. Try writing to the actual internal on the Mac Studio. Take advantage of the Unified Memory! |
|
|
|
Post by bluesholyman on Sept 6, 2022 14:36:28 GMT -6
OK - the problem might be the "write-to drive" I'm using. It's an internal SSD drive, but I'm using a USB enclosure. When I measure the read write speed of the Scandisk sample drive, it's like 900mb/s. The Samsung write-to drive is like 120mb/sec. My guess would be the enclosures interface is slowing you down, especially if its an older enclosure. |
|
|
|
Post by Johnkenn on Sept 6, 2022 17:13:19 GMT -6
Went and got a Sandisk Extreme Pro. Supposedly 2000 write 2000 read...only showing 1000 though. Anyway. Everything seems to be kosher now even at 64 buffers. One thing - I had some kind've low buffer thing checked in Pro Tools settings once I turned that off, no more dropouts.
|
|
