03-04-2010 11:50 AM
Oh, right, understood now, thanks.
Yes, the callback seemed the easiest way to process the buffer in a different thread, and therefore a different CPU, from the thread that's writing to screen and disc. But that's because I thought it would be hard pressed to keep up. Nice to see it at just 40% of the CPU. And now that the other thread calls Sleep() when it's finished with what the callback has stored, it's down to 30% of the CPU.
In PCIe-6537 Count sample clock in ANSI C you said that you've taken 50MS/s simultaneously from multiple 6537s, so I must be misconfigured. Would my configuration problem be a cheapo motherboard (in which case I just live with it), or is there something else I can set up on the card?
Frank
03-04-2010 12:03 PM
Keith
I saw your post after I replied to Andy, and you've pretty much answered what I was asking him.
There are certain instances (depending on the maximum packet size allowed by your computer's chipset) where a PCIe-6537 cannot acquire 32 channels of data at 50MHz. This is strictly a limitation of the PC.
I cannot change that now, for this test. But if I am doing this again, I can ask for a higher spec PC. Do you know off-hand exactly what I should look for in the PC spec which would mean the PC is suitable for this card running at full speed?
Thanks
Frank
03-04-2010 02:51 PM
Frank,
Try running a finite acquisition examples at 50 MHz for a few megs of data. If this fails on that system, there is no way you'll ever get 200 MB/s (50 MHz) on that computer. If it works, there may be some tweaks that we can do maximize performance.
Playing with buffer sizes may get you better performance, for example setting your buffer size to 8M and then fetching data every 500K may help give more room to the board to move the data from the card to the software buffer.
Another thing to try is switching to different PCIe slot. Some slots designed for video have poor input performance. Typically best to plug the device in a x1 slot.
I hope this helps,
Juan Carlos
03-05-2010 04:27 AM
Thanks, Juan Carlos
Due to looming deadlines, the card has now been taken from me and will be used at 40MS/s. So I'm off the case. I've flagged Keith's post, being the first which said that the BIOS and PCIe chipset could make the sampling impossible, as the solution because that's as far as I'm going to get, at least this month.
I did try getting a reduced number of samples, and it still failed. So I'm assuming the PC is at fault.
Thanks again
Frank