Tesla K80 poor performance on Azure

Hi, I am moving my first steps into CUDA by using my old laptop’s GeForce 610M. I have written a couple of functions involving cublasDSYMM and on my laptop the comparison between GPU vs CPU implementation gave no such big speed improvements. Since I have the possibility to run code on an Azure’s VM, specifically the NC6, endowed with a Tesla K80, I was puzzled in seeing that execution time on the GPU implementation on the K80 was way slower than on the GeForce 610M (roughly 5secs vs 1 sec - matrix size 1600). My first thought was that this could be due to my poor ability of coding, so I have measured the execution time of deviceQuery on both machines: the picture did not change. My impression is that there is a big “latency” on the VM and don’t know why. Do you have some suggestions in order to understand why the K80 on the VM is performing so poorly?

Here are some outputs from the VM
Please note that I have resized the VM in order to see if something would have changed with a better GPU, regrettably no good news. The outputs below refer to a Tesla P100


Wed Jun  5 13:09:41 2019       
| NVIDIA-SMI 410.104      Driver Version: 410.104      CUDA Version: 10.0     |
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|   0  Tesla P100-PCIE...  Off  | 000036B6:00:00.0 Off |                    0 |
| N/A   26C    P0    25W / 250W |      0MiB / 16280MiB |      1%      Default |
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|  No running processes found                                                 |
./deviceQuery Starting...

 CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 1 CUDA Capable device(s)

Device 0: "Tesla P100-PCIE-16GB"
  CUDA Driver Version / Runtime Version          10.0 / 9.0
  CUDA Capability Major/Minor version number:    6.0
  Total amount of global memory:                 16281 MBytes (17071734784 bytes)
  (56) Multiprocessors, ( 64) CUDA Cores/MP:     3584 CUDA Cores
  GPU Max Clock rate:                            1329 MHz (1.33 GHz)
  Memory Clock rate:                             715 Mhz
  Memory Bus Width:                              4096-bit
  L2 Cache Size:                                 4194304 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
  Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 2048 layers
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  2048
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 2 copy engine(s)
  Run time limit on kernels:                     No
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Enabled
  Device supports Unified Addressing (UVA):      Yes
  Supports Cooperative Kernel Launch:            Yes
  Supports MultiDevice Co-op Kernel Launch:      Yes
  Device PCI Domain ID / Bus ID / location ID:   14006 / 0 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 10.0, CUDA Runtime Version = 9.0, NumDevs = 1
Result = PASS

And to show how long does it take to run deviceQuery

time ./deviceQuery > /dev/null

real	0m1.040s
user	0m0.004s
sys	0m0.942s

1 second is just unacceptable, don’t you think so?