I'm having trouble with occasional dropouts when recording 16 tracks live (44.1/24) on a Lenovo Thinkpad W540 onto its internal 5400 rpm drive with a 16GB "M.2 Solid State Drive Double", which (I think) acts like an SSD cache.  (Recording through the USB 2.0 interface on a Behringer X32 console.)
 
When I do the math, it says that each track consumes about 132KB per second.  Dividing that into 16GB (and dividing by 60 (seconds in a minute)  and 60 again (minutes in an hour) then by 16 (tracks)) implies that that 16GB "cache" would fill in about 2 hours.  In my last session (where I was very careful not to bump the laptop which makes the heads temporarily park), it dropped out about 40 minutes in saying it thought it was out of disk space.  (The spinning media still had 500GB of free space.)
 
I'm wondering if I can sidestep all of this by recording through SONAR on the laptop directly to a high speed USB 3.0 or Thunderbolt flash drive.  There are some SanDisk Extreme devices (64GB sequential writes at 220MB/s...or 128GB that writes faster than that) that might solve the problem.
 
Then again...I might not have even properly identified what my problem is.
 
Has anyone else chased this...and what was your solution?
 
...and pardon me for having posted this in the "Computers" section and subsequently hearing crickets...

There's some extra info that's needed. How longer are the files you're creating? Is this a 32- or 64-bit system (there are file size limits with 32-bit systems)? Have you turned off wi-fi and other elements that give latency spikes? What interface are you using and at how many sample buffers?
 
I would tend to think the 5400 RPM internal drive is a bottleneck. I'm also unclear about what the 16GB of memory does. If it's simply caching data on the way to the 5400 RPM, I think it still needs to write to that hard drive at some point...but I'll leave any definitive answers to the computer experts.

Are you monitoring through SONAR?
 
If not you can take liberties with your sample buffer and run at a very large setting and see if that helps.
 
When I do field recording I try to quickly hit "save" and then "save as" between songs to make sure I have what I think I already have.
 
I once had a Windows XP laptop that had a DPC latency below 20µs and it would record 16-24 tracks at 24bit/44.1kHz for hours without a single hic cup.
 
that laptop died.
 
The past 2 Win 7 laptops I have owned have been incapable of running below 200µs DPC latency and so they have been useless for reliable field recording. I drag an old desktop out for the task now.
 
Good luck.

I agree with Craig that the 5400 rpm drive will be a bottleneck. 
 
I'm also unclear about the 16G cache.  does only the audio data fill up that cache when recording?  that might be a question for Noel.  I'm not sure how much of it is only raw audio data and how much might be pointers, windows overhead, sonar overhead, etc.
 
also, does the harddrive wait until the 16G cache is full before it starts writing?  I'm not sure on that either.

This article does a good job of explaining hybrid drives http://www.pcworld.com/article/2012223/hybrid-hard-drives-how-they-work-and-why-they-matter.html Streaming audio writes would bypass the SSD portion and go straight to the platter(s).

FWIW, 5400rpm is plenty for 16 tracks of 24/44.1 if the hard drive isn't hampered by some especially bad on board controller.
 
Been there done that.

Also, we don't know what version of SONAR or system OS is being run.

mike_mccue
FWIW, 5400rpm is plenty for 16 tracks of 24/44.1 if the hard drive isn't hampered by some especially bad on board controller.
 
Been there done that.

theoretically and  probably in practice by some.  but we don't know how many other programs might be accessing his hard drive at the same time he's recording.  if this is a dedicated machine and it's lean then 5400 might be just fine.

Well, yeah....
 
 
... I'd advise checking the DPC latency and see how bad the interrupts are.
 
On the machine I mentioned that I was able to get below 20µs there were almost no interrupts to the CPU or the system devices and 5400rpm worked without a hitch. On the machines that can't get below 200µs a 7200rpm drive doesn't help.
 
I'd advise looking for a problem elsewhere and avoid the distractions of the drive as a bottleneck theory. 
 
The dialog may say something like "the drive is full"' but that is just because that is what someone typed in the dialog text line many years ago. The system isn't actually reporting why the stream stopped... it's just popping open a flash card.

SONAR X3e build 352
11 ms latency
All track and bus fx bins bypassed
Not monitoring through SONAR
Windows 7 Pro 64 bit Service Pack 1
i7-4900MQ 2.80 GHz
16 GB Memory
Norton 360 in "Silent" mode
WiFi not connected to anything
Interface - Behringer X32 via X-UF USB (2.0) ASIO driver
The laptop sits on top of a stack of SoundBlaster cards, just the way CJ recommended.
No cat pictures have been used as wallpaper.
The laptop is oriented with the screen facing South...the operator facing North.
 
I haven't checked DPC latency yet.
 
Getting back to the original question...has anyone tried recording directly to USB 3.0 connected high speed flash media?