NI Data Link Test Framework System Benchmarks

VST Acquisition to Coprocessor Streaming Benchmarks

An FPGA VI was designed to count the time from when the RFSA Start Trigger is received on the coprocessor to when the first valid sample is received at the Read Endpoint. The VI was used to generate a bitfile for each supported Read Endpoint for each supported FPGA target.

  • Latency Test—The test starts with an RFSA Start Trigger on the VST under test. That trigger is exported to the coprocessor and used to start a counter that increments until the rising edge of the Output Valid indicator on the Read Endpoint is detected. The counter value is converted to seconds and reported in the following tables.
  • Throughput Test—The test starts with the same conditions as the latency test and runs for a set duration of time. The throughput is measured multiple times on the coprocessor by checking the number of valid samples that have been received during a measurement window. The measured values are reported in the following tables.
    • Figure 26. VST Acquisition to Coprocessor Streaming Benchmarks
    VST acquisition-to-coprocessor streaming benchmark setup showing an RFSA software trigger, VST-to-coprocessor MGT data path, read endpoint, and latency measurement from trigger receipt to first valid sample.

    PXIe-5842 to PXIe-7903

    Table 62. PXIe-5842 to PXIe-7903 Device Test Conditions
    Condition Value
    RF Cable RG-402, 50 Ω SMA to SMA Cable (1 m length)
    MGT Cable Mini-SAS zHD-Mini-SAS zHD HSS Passive Cable (1 m length), NI part number 788927-01
    Latency Test Iterations 100
    VST Bitfile VST MGT Streaming
    Coprocessor CLIP dmsc_7903_01
    Table 63. PXIe-5842 VST Acquisition to PXIe-7903 Coprocessor Streaming Benchmarks
    Read Endpoint RFSA IQ Rate Minimum Latency Maximum Latency Average Latency Throughput Average
    U32 1SPC 156.25 MS/s 4.475 µs 4.512 µs 4.495 µs 156.25 MS/s
    U32 2SPC 312.50 MS/s 4.025 µs 4.087 µs 4.047 µs 312.50 MS/s
    U32 4SPC 625.00 MS/s 3.756 µs 3.806 µs 3.785 µs 625.00 MS/s
    U32 8SPC 1.25 GS/s 3.587 µs 3.631 µs 3.607 µs 1250.00 MS/s
    U32 16SPC 2.50 GS/s 3.525 µs 3.581 µs 3.541 µs 2500.00 MS/s

    PXIe-5841 to PXIe-6594

    Table 64. PXIe-5841 to PXIe-6594 Device Test Conditions
    Condition Value
    RF Cable RG-402, 50 Ω SMA to SMA Cable (1 m length)
    MGT Cable Nano-Pitch Male to Nano-Pitch Male OCuLink x4 Cable (1 m length), NI part number 785486-01
    Latency Test Iterations 100
    VST Bitfile VST MGT Streaming
    Coprocessor CLIP dmsc_6594_00
    Table 65. PXIe-5841 VST Acquisition to PXIe-6594 Coprocessor Streaming Benchmarks
    Read Endpoint RFSA IQ Rate Minimum Latency Maximum Latency Average Latency Throughput Average
    U32 1SPC 156.25 MS/s 2.100 µs 2.144 µs 2.122 µs 156.25 MS/s
    U32 2SPC 312.50 MS/s 1.737 µs 1.781 µs 1.757 µs 312.50 MS/s
    U32 4SPC 625.00 MS/s 1.500 µs 1.544 µs 1.523 µs 625.00 MS/s
    U32 8SPC 1.25 GS/s 1.425 µs 1.475 µs 1.451µs 1.25 GS/s

    PXIe-5841 to PXIe-7915

    Table 66. PXIe-5841 to PXIe-7915 Device Test Conditions
    Condition Value
    RF Cable RG-402, 50 Ω SMA to SMA Cable (1 m length)
    MGT Cable Nano-Pitch Male to Nano-Pitch Male OCuLink x4 Cable (1 m length), NI part number 785486-01
    Latency Test Iterations 100
    VST Bitfile VST MGT Streaming
    Coprocessor CLIP dmsc_7915_00
    Table 67. PXIe-5841 VST Acquisition to PXIe-7915 Coprocessor Streaming Benchmarks
    Read Endpoint RFSA IQ Rate Minimum Latency Maximum Latency Average Latency Throughput Average
    U32 1SPC 156.25 MS/s 2.056 µs 2.094 µs 2.075 µs 156.25 MS/s
    U32 2SPC 312.50 MS/s 1.687 µs 1.725 µs 1.710 µs 312.50 MS/s
    U32 4SPC 625.00 MS/s 1.462 µs 1.494 µs 1.478 µs 625.00 MS/s
    U32 8SPC 1.25 GS/s 1.387 µs 1.425 µs 1.406 µs 1.25 GS/s

    PXIe-5841 to PXIe-7903

    Table 68. PXIe-5841 to PXIe-7903 Device Test Conditions
    Condition Value
    RF Cable RG-402, 50 Ω SMA to SMA Cable (1 m length)
    MGT Cable Nano-Pitch Male to Mini-SAS HD Male OCuLink x4 Cable (1 m length), NI part number 786215-01
    Latency Test Iterations 100
    VST Bitfile VST MGT Streaming
    Coprocessor CLIP dmsc_7903_00
    Table 69. PXIe-5841 VST Acquisition to PXIe-7903 Coprocessor Streaming Benchmarks
    Read Endpoint RFSA IQ Rate Minimum Latency Maximum Latency Average Latency Throughput Average
    U32 1SPC 156.25 MS/s 2.087 µs 2.131 µs 2.111 µs 156.25 MS/s
    U32 2SPC 312.50 MS/s 1.725 µs 1.769 µs 1.741 µs 312.50 MS/s
    U32 4SPC 625.00 MS/s 1.487 µs 1.531 µs 1.507 µs 625.00 MS/s
    U32 8SPC 1.25 GS/s 1.413 µs 1.462 µs 1.436 µs 1.25 GS/s

    Coprocessor to VST Generation Streaming Benchmarks

    An FPGA VI was designed to download a known waveform into memory and play it back into the Write Endpoint on the coprocessor when an RFSG Start Trigger is received. The VI was used to generate a bitfile for each supported Write Endpoint for each supported FPGA target.
  • Latency Test—The test starts by configuring the Zero Waveform Length (Samples) Register on the VST under test. This register configures the VST under test to write zeros to the RF out IP chain for the requested number of samples, after which it switches to using samples from the MGT buffer. If there is no data in the MGT buffer, the VST reports an underflow. After configuration, an RFSG Start Trigger is sent to the VST under test. The RFSG Started Event is then exported to PXI_Trig0, which starts the PXIe-5842 RFSA Acquisition and the Coprocessor waveform playback. The latency is determined by increasing the Zero Waveform Length value until the VST does not report an underflow for 100 consecutive runs. Expected latency is calculated by dividing Zero Waveform Length by the RFSG IQ Rate. A PXIe-5842 VST is used to measure the latency by measuring the time from when the start trigger was received to when the downloaded waveform signal was detected. The results are reported in the following tables.
  • Throughput Test—The test starts with the same conditions as the latency test and runs for a set duration of time. The throughput is measured multiple times on the coprocessor by checking the number of valid samples that have been sent during a measurement window. The measured values are reported in the following tables.
    • Figure 27. Coprocessor to VST Generation Streaming Benchmarks
    Coprocessor-to-VST generation streaming benchmark showing waveform data streamed from a coprocessor to a VST through MGT links and triggered by an RFSG software trigger. Latency is measured using a configurable zero-waveform delay before waveform playback.

    PXIe-7903 to PXIe-5842

    Table 70. PXIe-7903 to PXIe-5842 Device Test Conditions
    Condition Value
    RF Cable RG-402, 50 Ω SMA to SMA Cable (0.5 m length)
    MGT Cable Mini-SAS zHD-Mini-SAS zHD HSS Passive Cable (1 m length), NI part number 788927-01
    Latency Test Iterations 100
    PXIe-5842 RFSA IQ Rate 2.5 GS/s
    VST Bitfile VST MGT Streaming
    Coprocessor CLIP dmsc_7903_01
    Table 71. PXI-7903 Coprocessor to PXIe-5842 VST Generation Streaming Benchmarks
    Read Endpoint Zero Waveform Length RFSA IQ Rate Expected Latency Minimum Latency Maximum Latency Average Latency Throughput Average
    U32 1SPC 496 156.25 MS/s 3.174 µs 3.157 µs 3.157 µs 3.157 µs 156.25 MS/s
    U32 2SPC 912 312.50 MS/s 2.918 µs 2.906 µs 2.906 µs 2.906 µs 312.50 MS/s
    U32 4SPC 1728 625.00 MS/s 2.765 µs 2.749 µs 2.749 µs 2.749 µs 625.00 MS
    U32 8SPC 3392 1.25 GS/s 2.714 µs 2.699 µs 2.699 µs 2.699 µs 1250.00 MS/s
    U32 16SPC 6592 2.5 GS/s 2.637 µs 2.623 µs 2.623 µs 2.623 µs 2500.00 MS/s

    PXIe-7903 to PXIe-5841

    Table 72. PXIe-7903 to PXIe-5841 Device Test Conditions
    Condition Value
    RF Cable RG-402, 50 Ω SMA to SMA Cable (1 m length)
    MGT Cable Nano-Pitch Male to Mini-SAS HD Male OCuLink x4 Cable (1 m length), NI part number 786215-01
    Latency Test Iterations 100
    PXIe-5842 RFSA IQ Rate 2.5 GS/s
    VST Bitfile VST MGT Streaming
    Coprocessor CLIP dmsc_7903_00
    Table 73. PXIe-7903 Coprocessor to PXIe-5841 VST Generation Streaming Benchmarks
    Read Endpoint Zero Waveform Length RFSA IQ Rate Expected Latency Minimum Latency Maximum Latency Average Latency Throughput Average
    U32 1SPC 264 156.25 MS/s 1.690 µs 1.663 µs 1.663 µs 1.663 µs 156.25 MS/s
    U32 2SPC 480 312.50 MS/s 1.536 µs 1.514 µs 1.514 µs 1.514 µs 312.50 MS/s
    U32 4SPC 904 625.00 MS/s 1.446 µs 1.427 µs 1.427 µs 1.427 µs 625.00 MS/s
    U32 8SPC 1736 1.25 GS/s 1.389 µs 1.371 µs 1.371 µs 1.371 µs 1.25 GS/s

    PXIe-7915 to PXIe-5841

    Table 74. PXIe-7915 to PXIe-5841 Device Test Conditions
    Condition Value
    RF Cable RG-402, 50 Ω SMA to SMA Cable (1 m length)
    MGT Cable Nano-Pitch Male to Nano-Pitch Male OCuLink x4 Cable (1 m length), NI part number 785486-01
    Latency Test Iterations 100
    PXIe-5842 RFSA IQ Rate 2.5 GS/s
    VST Bitfile VST MGT Streaming
    Coprocessor CLIP dmsc_7915_00
    Table 75. PXIe-7915 Coprocessor to PXIe-5841 VST Generation Streaming Benchmarks
    Read Endpoint Zero Waveform Length RFSA IQ Rate Expected Latency Minimum Latency Maximum Latency Average Latency Throughput Average
    U32 1SPC 256 156.25 MS/s 1.634 µs 1.612 µs 1.612 µs 1.612 µs 156.25 MS/s
    U32 2SPC 464 312.50 MS/s 1.485 µs 1.464 µs 1.464 µs 1.464 µs 312.50 MS/s
    U32 4SPC 888 625.00 MS/s 1.421 µs 1.402 µs 1.402 µs 1.402 µs 625.00 MS/s
    U32 8SPC 1688 1.25 GS/s 1.350 µs 1.333 µs 1.333 µs 1.333 µs 1.25 GS/s

    PXIe-6594 to PXIe-5841

    Table 76. PXIe-6594 to PXIe-5841 Device Test Conditions
    Condition Value
    RF Cable RG-402, 50 Ω SMA to SMA Cable (1 m length)
    MGT Cable Nano-Pitch Male to Nano-Pitch Male OCuLink x4 Cable (1 m length), NI part number 785486-01
    Latency Test Iterations 100
    PXIe-5842 RFSA IQ Rate 2.5 GS/s
    VST Bitfile VST MGT Streaming
    Coprocessor CLIP dmsc_6594_00
    Table 77. PXIe-6594 Coprocessor to PXIe-5841 VST Generation Streaming Benchmarks
    Read Endpoint Zero Waveform Length RFSA IQ Rate Expected Latency Minimum Latency Maximum Latency Average Latency Throughput Average
    U32 1SPC 264 156.25 MS/s 1.690 µs 1.665 µs 1.665 µs 1.665 µs 156.25 MS/s
    U32 2SPC 472 312.50 MS/s 1.510 µs 1.491 µs 1.491 µs 1.491 µs 312.50 MS/s
    U32 4SPC 896 625.00 MS/s 1.434 µs 1.417 µs 1.417 µs 1.417 µs 625.00 MS/s
    U32 8SPC 1728 1.25 GS/s 1.382 µs 1.367 µs 1.367 µs 1.367 µs 1.25 GS/s

    VST Acquisition to Coprocessor to VST Generation Streaming Benchmarks

    An FPGA VI was designed to connect a Read Endpoint to a Write Endpoint, which are both connected to the same VST. The VI was used to generate a bitfile for each supported Write Endpoint for each supported FPGA target.
  • Latency Test—The test starts by configuring the Zero Waveform Length (Samples) Register on the VST under test. This register configures the VST under test to write zeros to the RF out IP chain for the requested number of samples, after which it switches to using samples from the MGT buffer. If there is no data in the MGT buffer, the VST reports an underflow. The PXIe-5842 is then configured to generate a known waveform using the RFSG Write Arbitrary Waveform method. After configuration, an RFSG Start Trigger is sent to the VST under test. The RFSG Started Event is then exported to PXI_Trig0, which starts the VST under tests RFSA Acquisition, the PXIe-5842 RFSA Acquisition, and the PXIe-5842 RFSG Generation. The latency is determined by increasing the Zero Waveform Length value until the VST RFSG Generation does not report an underflow for 100 consecutive runs. Expected latency is calculated by dividing Zero Waveform Length by the RFSG IQ Rate. To measure the latency, the recorded waveforms from both VST acquisitions are used to find the rising edge of the signal. The latency is the difference between when the signal was detected on the PXIe-5842 and when the signal was detected on the VST under test. The results are reported in the following tables.
  • Throughput Test— The test starts with the same conditions as the latency test and runs for a set duration of time. The throughput is measured multiple times on the coprocessor by checking the number of valid samples that have been received and/or sent during a measurement window. The measured values are reported in the following tables.
    • Figure 28. VST Acquisition to Coprocessor to VST Generation Streaming Benchmarks
    Block diagram of a VST–coprocessor loopback benchmark. Waveform data is streamed from the coprocessor to the VST for generation, looped back through RF connections, and streamed back to the coprocessor for acquisition. PXI_Trig0 synchronizes RFSG and RFSA triggers, and the red path indicates the measured data flow.

    PXIe-5842 to PXIe-7903

    Table 78. PXIe-5842 to PXIe-7903 Device Test Conditions
    Condition Value
    RF Cable: RF OUT of VST Under Test to PXIe-5842 RF IN RG-402, 50 Ω SMA to SMA Cable (0.5 m length)
    RF Cable: RF IN of VST Under Test to PXIe-5842 RF OUT RG-402, 50 Ω SMA to SMA Cable (1 m length)
    MGT Cable Mini-SAS zHD-Mini-SAS zHD HSS Passive Cable (1 m length), NI part number 788927-01
    Latency Test Iterations 100
    PXIe-5842 RFSA IQ Rate 2.5 GS/s
    PXIe-5842 RFSG IQ Rate 2.5 GS/s
    VST Bitfile VST MGT Streaming
    Coprocessor CLIP dmsc_7903_01
    Table 79. PXIe-5842 VST Acquisition to PXIe-7903 Coprocessor to PXIe-5842 VST Generation Streaming Benchmarks
    Read Endpoint Zero Waveform Length RFSA IQ Rate Expected Latency Minimum Latency Maximum Latency Average Latency Throughput Average
    U32 1SPC 1008 156.25 MS/s 6.451 µs 6.434 µs 6.530 µs 6.501 µs 156.25 MS/s
    U32 2SPC 1808 312.50 MS/s 5.786 µs 5.773 µs 5.818 µs 5.790 µs 312.50 MS/s
    U32 4SPC 3360 625.00 MS/s 5.376 µs 5.362 µs 5.382 µs 5.371 µs 625.00 MS
    U32 8SPC 6448 1.25 GS/s 5.158 µs 5.149 µs 5.155 µs 5.154 µs 1250.00 MS/s
    U32 16SPC 12560 2.5 GS/s 5.030 µs 5.022 µs 5.022 µs 5.022 µs 2500.00 MS/s

    PXIe-5841 to PXIe-7903

    Table 80. PXIe-5841 to PXIe-7903 Device Test Conditions
    Condition Value
    RF Cable: RF OUT of VST Under Test to PXIe-5842 RF IN RG-402, 50 Ω SMA to SMA Cable (0.5 m length)
    RF Cable: RF IN of VST Under Test to PXIe-5842 RF OUT RG-402, 50 Ω SMA to SMA Cable (1 m length)
    MGT Cable Nano-Pitch Male to Mini-SAS HD Male OCuLink x4 Cable (1 m length), NI part number 786215-01
    Latency Test Iterations 100
    PXIe-5842 RFSA IQ Rate 2.5 GS/s
    PXIe-5842 RFSG IQ Rate 2.5 GS/s
    VST Bitfile VST MGT Streaming
    Coprocessor CLIP dmsc_7903_00
    Table 81. PXIe-5841 VST Acquisition to PXIe-7903 Coprocessor to PXIe-5841 VST Generation Streaming Benchmarks
    Read Endpoint Zero Waveform Length RFSA IQ Rate Expected Latency Minimum Latency Maximum Latency Average Latency Throughput Average
    U32 1SPC 542 156.25 MS/s 3.469 µs 3.484 µs 3.490 µs 3.485 µs 156.25 MS/s
    U32 2SPC 960 312.50 MS/s 3.072 µs 3.060 µs 3.066 µs 3.060 µs 312.50 MS/s
    U32 4SPC 1752 625.00 MS/s 2.803 µs 2.786 µs 2.793 µs 2.788 µs 625.00 MS/s
    U32 8SPC 3344 1.25 GS/s 2.675 µs 2.660 µs 2.666 µs 2.662 µs 1.25 GS/s

    PXIe-5841 to PXIe-7915

    Table 82. PXIe-5841 to PXIe-7915 Device Test Conditions
    Condition Value
    RF Cable: RF OUT of VST Under Test to PXIe-5842 RF IN RG-402, 50 Ω SMA to SMA Cable (0.5 m length)
    RF Cable: RF IN of VST Under Test to PXIe-5842 RF OUT RG-402, 50 Ω SMA to SMA Cable (1 m length)
    MGT Cable Nano-Pitch Male to Nano-Pitch Male OCuLink x4 Cable (1 m length), NI part number 785486-01
    Latency Test Iterations 100
    PXIe-5842 RFSA IQ Rate 2.5 GS/s
    PXIe-5842 RFSG IQ Rate 2.5 GS/s
    VST Bitfile VST MGT Streaming
    Coprocessor CLIP dmsc_7915_00
    Table 83. PXIe-5841 VST Acquisition to PXIe-7915 Coprocessor to PXIe-5841 VST Generation Streaming Benchmarks
    Read Endpoint Zero Waveform Length RFSA IQ Rate Expected Latency Minimum Latency Maximum Latency Average Latency Throughput Average
    U32 1SPC 534 156.25 MS/s 3.418 µs 3.432 µs 3.433 µs 3.432 µs 156.25 MS/s
    U32 2SPC 936 312.50 MS/s 2.995 µs 2.989 µs 2.989 µs 2.989 µs 312.50 MS/s
    U32 4SPC 1712 625.00 MS/s 2.739 µs 2.727 µs 2.727 µs 2.727 µs 625.00 MS/s
    U32 8SPC 3264 1.25 GS/s 2.611 µs 2.601 µs 2.601 µs 2.601 µs 1.25 GS/s

    PXIe-5841 to PXIe-6594

    Table 84. PXIe-5841 to PXIe-6594 Device Test Conditions
    Condition Value
    RF Cable: RF OUT of VST Under Test to PXIe-5842 RF IN RG-402, 50 Ω SMA to SMA Cable (0.5 m length)
    RF Cable: RF IN of VST Under Test to PXIe-5842 RF OUT RG-402, 50 Ω SMA to SMA Cable (1 m length)
    MGT Cable Nano-Pitch Male to Nano-Pitch Male OCuLink x4 Cable (1 m length), NI part number 785486-01
    Latency Test Iterations 100
    PXIe-5842 RFSA IQ Rate 2.5 GS/s
    PXIe-5842 RFSG IQ Rate 2.5 GS/s
    VST Bitfile VST MGT Streaming
    Coprocessor CLIP dmsc_6594_00
    Table 85. PXIe-5841 VST Acquisition to PXIe-6594 Coprocessor to PXIe-5841 VST Generation Streaming Benchmarks
    Read Endpoint Zero Waveform Length RFSA IQ Rate Expected Latency Minimum Latency Maximum Latency Average Latency Throughput Average
    U32 1SPC 542 156.25 MS/s 3.467 µs 3.484 µs 3.484 µs 3.484 µs 156.25 MS/s
    U32 2SPC 960 312.50 MS/s 3.072 µs 3.063 µs 3.063 µs 3.063 µs 312.50 MS/s
    U32 4SPC 1752 625.00 MS/s 2.803 µs 2.788 µs 2.788 µs 2.788 µs 625.00 MS/s
    U32 8SPC 3360 1.25 GS/s 2.688 µs 2.676 µs 2.676 µs 2.676 µs 1.25 GS/s