LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

What tools do you use to identify PC bottleknecks?

What programs do you use to identify bottleknecks which may effect DAQ rates?

 

I've looked at specs for buses and hard drive and memory bus rates but this doesn't help me see what the actual realized speeds are.

 

I've used CPU-Z, HDTach, and a few other applications before but I'm wondering if there is something more powerful and configurable or perhaps even a way to measure these things in LabVIEW itself? 

 

Thanks,

 

Dave

0 Kudos
Message 1 of 5
(2,668 Views)

After watching computer architectures change over the years to achieve higher throughput, I have a "gut feel" that I can rely on but until you get that fell for yourself...

 

Benchmarking settles all arguements.

 

If you have something more concrete, I'll check back.

 

Ben

Retired Senior Automation Systems Architect with Data Science Automation LabVIEW Champion Knight of NI and Prepper LinkedIn Profile YouTube Channel
Message 2 of 5
(2,661 Views)

Hi Ben,

 

Do you have any recommendations for benchmarking utilities or should I focus on each potential bottlekneck separately?  If the latter I'll start with RAM. 

 

I've read that some DAQ uses DMA; do you know of a tool that would differentiate between DMA and non-DMA throughput?

 

Thanks,

 

Dave

 

 

0 Kudos
Message 3 of 5
(2,650 Views)

DMA (Direct Memory Access ) is always faster than non.

 

DMA describes hardawre that is capable of transfer data directly from its unborad buffers to physical memory vai the interconnecting bus.

 

Non-DMA requires teh CPU get invloved. Under normal circumstances this means tow bus transfers, one for the I/O device to the CPU and from the CPU to memory.

 

Next!

 

Memory

 

You have a number of different types of memory, on-board memory (used by I/O devices to hold data while waiting to be put on the bus) physical memory and virtual memory. Virtual memory is just a trick the modern OS perform to make it appear as if th computer has more physical memory that it actually has. The trick is pulled off using a disk drive to swap data into and out of real physical memory. Physical memory can operate as fasts as your CPU can click. The disk drive requires moving a physical head to a new location (ages in computer time) so it is orders of magniture slower than memory.

 

Caches

 

Caches are hidden blocks of memory used to speed up trasnfers to and from something. Cahces are included on most modern chips to hold data and code that hads been accessed recently by CPUs. Disk drives are also available that have a cache to speed up disk transfers.

 

I can can go for days wit this so I'll stop and let you poke again.

 

Ben

Retired Senior Automation Systems Architect with Data Science Automation LabVIEW Champion Knight of NI and Prepper LinkedIn Profile YouTube Channel
Message 4 of 5
(2,642 Views)

In addition to DMA vs interrupt (choose DMA) the bus type is very important.  Not only does bus speed matter but compare the busses to see if they are shared.  PCI acquisition is much worse that PCIe not only because it is slower but the PCI is shared while PCIe is a guarantied bandwidth so you dont have to worry about other devices making your acquisition bandwithd less deterministic.

 

Paul Falkenstein
Coleman Technologies Inc.
CLA, CPI, AIA-Vision
Labview 4.0- 2013, RT, Vision, FPGA
0 Kudos
Message 5 of 5
(2,635 Views)