PE-505 RFC 2544 Test Report
The Brainboxes PE-505 5-port embedded Fast Ethernet switch was benchmarked in-house against RFC 2544, the industry-standard methodology for network device performance testing. Testing was performed on 2026-08-17 over 2 hours 49 minutes using the Xena2544 v2.97 automated test suite across all 5 ports in a full-mesh, bidirectional topology at 100 Mbps per port (500 Mbps aggregate), using the seven IEEE default frame sizes from 64 to 1518 bytes.
The PE-505 passed all four RFC 2544 sub-tests — Throughput, Latency/Jitter, Frame Loss Rate, and Back-to-Back — at every one of the seven frame sizes and at both offered rates. Throughput converged to 100.00% of line rate at every frame size, and not a single frame was lost anywhere in the run: zero loss in the throughput test, zero loss in the frame loss test at both 95% and 100% rates, and zero loss across all 20 bidirectional flows throughout.
| Test | Result |
|---|---|
| Throughput | Pass |
| Latency and Jitter | Pass |
| Frame Loss Rate | Pass |
| Back-to-Back | Pass |
The latency, frame loss, and back-to-back tests were each run twice — once at 95% and once at 100% of line rate — to show behaviour both just below saturation and at full wire speed.
Because the throughput test converged to 100.00%, the 100% figures are a genuine full-line-rate measurement rather than a deliberate overload. Both rates returned zero frame loss, so the two columns differ only in latency and jitter, not in delivery.
Test Configuration
| Parameter | Value |
|---|---|
| Customer Service ID | PE-505 |
| Test Date | 2026-08-17, 14:02 |
| Test Duration | 02:49:08 (h:m:s) |
| Test Tool | Xena2544 v2.97, running on Microsoft Windows 10 Enterprise |
| Chassis | C-0 — Abyssinian (serial 2343912), 192.168.0.39:22606 |
| Module | M-0-6 — Odin-10G-5S-6P-CU (serial 650338) |
| Topology | Mesh |
| Direction | Bidirectional |
| Frame Size Type | IEEE Default |
| Frame Sizes | 64, 128, 256, 512, 1024, 1280, 1518 bytes |
| Port Count | 5 |
| Port Speed | AUTO negotiation, 100 Mbps port rate per port |
| Aggregate Capacity | 500 Mbps (5 ports × 100 Mbps) |
| Bidirectional Flows | 20 (5 ports × 4 peers) |
| Iterations per Test | 3 |
| Test Duration per Iteration | 60 seconds |
| Rates Tested (L/J, FL, B2B) | 95.00% and 100.00% of line rate |
Port Configuration
| Port ID | Interface | Speed | Rate |
|---|---|---|---|
| P-0-6-0 | 10GBASE-T | AUTO | 100 M |
| P-0-6-1 | 10GBASE-T | AUTO | 100 M |
| P-0-6-2 | 10GBASE-T | AUTO | 100 M |
| P-0-6-3 | 10GBASE-T | AUTO | 100 M |
| P-0-6-4 | 10GBASE-T | AUTO | 100 M |
10GBASE-T is the interface type of the Xena test module's ports, not of the PE-505. Each test port was auto-negotiated down to 100 Mbps to match the PE-505's Fast Ethernet ports. The PE-505 is a 5-port 10/100 Mbps switch.
Testing was pure Layer 2 — the protocol segment profile was Ethernet only, with no IP addressing configured — so these results measure frame forwarding, not routing or IP-level behaviour.
Throughput Test
The throughput test determines the maximum offered rate at which the device under test can forward frames with zero loss. Each trial ran for 60 seconds across 3 iterations, starting at an initial rate of 100.000% of line rate and binary-searching between a maximum of 100.000% and a minimum of 0.000% with a resolution of 0.500%. Acceptable loss was set to 0.0000% — zero frame loss permitted — and the rate result scope was Common Result, so the reported throughput is the single highest rate all five ports sustained loss-free at the same time.
The PE-505 converged to 100.00% offered rate at every frame size. The binary search never had to step down: the initial 100.000% rate was sustained with zero loss on the first attempt, at all seven sizes. Peak aggregate throughput was 499.99 Mbps at Layer 1 against a 500 Mbps theoretical maximum — 99.998% of line rate — making the PE-505 a non-blocking, wire-speed switch across the whole IEEE default frame range.
Layer 2 throughput climbs from 380.95 Mbps at 64 bytes to 493.29 Mbps at 1518 bytes. That rise is expected arithmetic rather than a change in switch behaviour: each frame carries a fixed 20 bytes of preamble and inter-frame gap, so as frames grow that overhead shrinks as a proportion of wire time. Layer 1 utilisation stays flat at ~500 Mbps throughout, which means the switch saturates its backplane at every frame size — including minimum-length 64-byte frames, the hardest case for any switch. The lower Layer 2 figure at 64 bytes is the Ethernet framing overhead floor, not a shortfall in the PE-505.
| Frame Size (bytes) | Offered Rate | Tx Rate (L1) | Tx Rate (L2) | Tx (Frames) | Rx (Frames) | Loss | Result |
|---|---|---|---|---|---|---|---|
| 64 | 100.00% | 499.99 Mbps | 380.95 Mbps | 44,642,400 | 44,642,400 | 0 | Pass |
| 128 | 100.00% | 499.98 Mbps | 432.41 Mbps | 25,336,800 | 25,336,800 | 0 | Pass |
| 256 | 100.00% | 499.98 Mbps | 463.75 Mbps | 13,586,400 | 13,586,400 | 0 | Pass |
| 512 | 100.00% | 499.99 Mbps | 481.20 Mbps | 7,048,800 | 7,048,800 | 0 | Pass |
| 1024 | 100.00% | 499.95 Mbps | 490.37 Mbps | 3,591,600 | 3,591,600 | 0 | Pass |
| 1280 | 100.00% | 499.82 Mbps | 492.13 Mbps | 2,883,600 | 2,883,600 | 0 | Pass |
| 1518 | 100.00% | 499.79 Mbps | 493.29 Mbps | 2,437,200 | 2,437,200 | 0 | Pass |
The slight decline in Layer 1 rate at the largest frame sizes — 499.99 Mbps at 64 bytes to 499.79 Mbps at 1518 bytes — is rate-quantisation error in the traffic generator, not switch loss. With fewer, larger frames per second, the generator's minimum rate step becomes a marginally larger fraction of the total. All seven points sit within 0.05% of the 500 Mbps line rate, and loss is zero at every one.
All 20 bidirectional flows (5 ports × 4 destinations each) showed zero loss at every frame size, with per-flow received frames matching transmitted frames exactly. Traffic distribution across the mesh was perfectly even: each port carried exactly one fifth of the aggregate (8,928,480 of 44,642,400 frames at 64 bytes), and within each port the four egress flows carried exactly one quarter of that port's traffic (2,232,120 frames each). No port ran hot and no flow was starved.
Per-Port Throughput Results
All five ports are byte-identical at every frame size — same frame counts, same L1 and L2 rates, same frames per second — so no port is favoured or starved. Port P-0-6-0 is shown as representative.
Port P-0-6-0
| Frame Size (bytes) | Tx (Frames) | Tx Rate (L1) | Tx Rate (L2) | Tx Rate (fps) | Rx (Frames) |
|---|---|---|---|---|---|
| 64 | 8,928,480 | 100.00 Mbps | 76.19 Mbps | 148,808 | 8,928,480 |
| 128 | 5,067,360 | 100.00 Mbps | 86.48 Mbps | 84,456 | 5,067,360 |
| 256 | 2,717,280 | 100.00 Mbps | 92.75 Mbps | 45,288 | 2,717,280 |
| 512 | 1,409,760 | 100.00 Mbps | 96.24 Mbps | 23,496 | 1,409,760 |
| 1024 | 718,320 | 99.99 Mbps | 98.07 Mbps | 11,972 | 718,320 |
| 1280 | 576,720 | 99.96 Mbps | 98.43 Mbps | 9,612 | 576,720 |
| 1518 | 487,440 | 99.96 Mbps | 98.66 Mbps | 8,124 | 487,440 |
Ports P-0-6-1, P-0-6-2, P-0-6-3 and P-0-6-4 recorded values identical to P-0-6-0 at every frame size.
Latency and Jitter Test
The latency and jitter test measures one-way frame transit time at a given offered rate. Each frame size was tested for 60 seconds across 3 iterations using Last-to-Last latency mode, relative to the throughput-determined rate. In the summary tables, average latency and average jitter are the mean across all 5 ports; minimum and maximum are the worst case observed on any port. The report also records a minimum jitter row, which is 0.000 µs on every port at every frame size — 35 of 35 cells — so it is omitted here.
Results at 95% Rate
| Frame Size (bytes) | Avg Latency (µs) | Min Latency (µs) | Max Latency (µs) | Avg Jitter (µs) | Max Jitter (µs) | Result |
|---|---|---|---|---|---|---|
| 64 | 12.330 | 7.150 | 26.755 | 0.698 | 13.470 | Pass |
| 128 | 20.493 | 12.266 | 39.055 | 0.317 | 23.036 | Pass |
| 256 | 33.805 | 22.497 | 54.771 | 0.307 | 23.052 | Pass |
| 512 | 75.586 | 42.984 | 168.261 | 0.035 | 80.663 | Pass |
| 1024 | 133.048 | 83.933 | 248.339 | 0.333 | 83.846 | Pass |
| 1280 | 160.362 | 104.428 | 312.734 | 0.212 | 104.982 | Pass |
| 1518 | 189.829 | 124.120 | 372.277 | 0.346 | 123.561 | Pass |
At 95% of line rate — 475.00 Mbps aggregate, 95.00 Mbps per port — average latency rises smoothly with frame size, from 12.330 µs for a 64-byte frame to 189.829 µs for a 1518-byte frame. That is the expected signature of a store-and-forward switch: each frame must be received in full before it is forwarded, so transit time scales with frame length. Worst-case maximum latency across all ports and frame sizes was 372.277 µs, on port P-0-6-2 at 1518 bytes.
Average jitter is highest at the smallest frame size — 0.698 µs at 64 bytes — because that is where the switch handles the greatest number of frames per second, 744,040 across the five ports. It falls to 0.035 µs at 512 bytes and stays at or below 0.346 µs for every size above that.
Per-Port Latency and Jitter at 95% Rate
Unlike the throughput and frame-loss results, latency varies from port to port: the ports contend for the same egress queues, so identical offered load produces measurably different transit times. All five ports are shown in full.
Port P-0-6-0
| Frame Size (bytes) | Avg Latency (µs) | Min Latency (µs) | Max Latency (µs) | Avg Jitter (µs) | Max Jitter (µs) |
|---|---|---|---|---|---|
| 64 | 12.402 | 7.375 | 21.487 | 0.605 | 13.470 |
| 128 | 19.824 | 12.498 | 37.708 | 0.189 | 12.829 |
| 256 | 35.958 | 22.730 | 54.771 | 0.331 | 23.052 |
| 512 | 88.387 | 43.216 | 168.261 | 0.042 | 80.663 |
| 1024 | 130.074 | 84.166 | 206.219 | 0.272 | 51.188 |
| 1280 | 149.684 | 104.660 | 238.453 | 0.199 | 104.333 |
| 1518 | 169.796 | 124.489 | 301.805 | 0.412 | 122.839 |
Port P-0-6-1
| Frame Size (bytes) | Avg Latency (µs) | Min Latency (µs) | Max Latency (µs) | Avg Jitter (µs) | Max Jitter (µs) |
|---|---|---|---|---|---|
| 64 | 11.567 | 7.150 | 17.919 | 0.269 | 7.064 |
| 128 | 21.310 | 12.266 | 39.055 | 0.406 | 23.036 |
| 256 | 35.532 | 22.497 | 53.592 | 0.342 | 21.777 |
| 512 | 77.859 | 42.984 | 127.183 | 0.031 | 79.380 |
| 1024 | 167.318 | 83.933 | 248.339 | 0.338 | 83.846 |
| 1280 | 172.846 | 104.428 | 312.317 | 0.169 | 104.982 |
| 1518 | 165.563 | 124.120 | 297.619 | 0.282 | 123.553 |
Port P-0-6-2
| Frame Size (bytes) | Avg Latency (µs) | Min Latency (µs) | Max Latency (µs) | Avg Jitter (µs) | Max Jitter (µs) |
|---|---|---|---|---|---|
| 64 | 10.937 | 7.423 | 19.875 | 0.761 | 7.705 |
| 128 | 22.724 | 12.538 | 38.943 | 0.409 | 12.829 |
| 256 | 31.579 | 22.770 | 51.868 | 0.290 | 22.419 |
| 512 | 62.917 | 43.264 | 130.968 | 0.027 | 42.889 |
| 1024 | 142.881 | 84.214 | 189.237 | 0.332 | 83.197 |
| 1280 | 168.455 | 104.708 | 312.734 | 0.261 | 104.341 |
| 1518 | 222.812 | 124.153 | 372.277 | 0.360 | 123.561 |
Port P-0-6-3
| Frame Size (bytes) | Avg Latency (µs) | Min Latency (µs) | Max Latency (µs) | Avg Jitter (µs) | Max Jitter (µs) |
|---|---|---|---|---|---|
| 64 | 13.320 | 7.407 | 26.755 | 0.886 | 11.522 |
| 128 | 21.192 | 12.522 | 38.141 | 0.419 | 12.829 |
| 256 | 32.416 | 22.754 | 48.396 | 0.306 | 22.394 |
| 512 | 70.541 | 43.240 | 133.502 | 0.034 | 42.897 |
| 1024 | 99.265 | 84.190 | 135.763 | 0.309 | 49.272 |
| 1280 | 150.089 | 104.684 | 299.271 | 0.231 | 104.958 |
| 1518 | 195.056 | 124.513 | 329.051 | 0.323 | 123.537 |
Port P-0-6-4
| Frame Size (bytes) | Avg Latency (µs) | Min Latency (µs) | Max Latency (µs) | Avg Jitter (µs) | Max Jitter (µs) |
|---|---|---|---|---|---|
| 64 | 13.422 | 7.238 | 23.411 | 0.967 | 12.171 |
| 128 | 17.416 | 12.354 | 28.535 | 0.164 | 12.829 |
| 256 | 33.539 | 22.585 | 50.633 | 0.267 | 21.144 |
| 512 | 78.228 | 43.080 | 137.319 | 0.039 | 42.905 |
| 1024 | 125.703 | 84.029 | 206.452 | 0.413 | 83.197 |
| 1280 | 160.738 | 104.516 | 273.011 | 0.200 | 104.974 |
| 1518 | 195.917 | 124.353 | 261.978 | 0.352 | 122.246 |
Results at 100% Rate
| Frame Size (bytes) | Avg Latency (µs) | Min Latency (µs) | Max Latency (µs) | Avg Jitter (µs) | Max Jitter (µs) | Result |
|---|---|---|---|---|---|---|
| 64 | 12.320 | 7.150 | 21.904 | 0.016 | 14.112 | Pass |
| 128 | 18.270 | 12.266 | 30.788 | 0.017 | 12.179 | Pass |
| 256 | 37.212 | 22.497 | 62.829 | 0.011 | 22.419 | Pass |
| 512 | 65.140 | 42.984 | 121.883 | 0.019 | 42.873 | Pass |
| 1024 | 141.346 | 83.933 | 260.607 | 0.022 | 83.862 | Pass |
| 1280 | 164.412 | 104.428 | 307.193 | 0.024 | 104.333 | Pass |
| 1518 | 233.243 | 124.120 | 469.627 | 0.031 | 123.545 | Pass |
At full line rate the cost of saturation appears only on large frames. From 1024 bytes upward, average latency is consistently higher than at 95%: 141.346 µs versus 133.048 µs at 1024 bytes, 164.412 µs versus 160.362 µs at 1280 bytes, and 233.243 µs versus 189.829 µs at 1518 bytes — a 1.23× increase at the largest frame size, with worst-case maximum latency rising in step from 372.277 µs to 469.627 µs (1.26×). That is the expected cost of deeper queueing when every port is saturated simultaneously. At 64 bytes the two rates are indistinguishable, 12.320 µs against 12.330 µs, and at 128 and 512 bytes the full-rate figures are actually lower — differences at those sizes fall within the run-to-run scatter rather than showing a consistent direction.
Jitter moves the opposite way, and dramatically so: average jitter at full rate never exceeds 0.031 µs at any frame size, against 0.698 µs at 64 bytes under the 95% rate — a more than twentyfold improvement. This is not a contradiction. At exactly 100% of line rate the generator emits frames back-to-back with the minimum legal inter-frame gap, so the offered stream has no slack in which its timing can vary; at 95% the generator must insert idle periods to hold the target rate, and the placement of those gaps is itself a source of variation. Saturation therefore makes the traffic more regular even as it makes it slower. Minimum jitter remains 0.000 µs on every port at every frame size.
Aggregate transmitted and received frame counts match exactly at every frame size — 44,642,400 frames at 64 bytes down to 2,437,200 at 1518 bytes — so there is no frame loss at full rate, and all seven frame sizes report Pass.
Per-Port Latency and Jitter at 100% Rate
Port P-0-6-0
| Frame Size (bytes) | Avg Latency (µs) | Min Latency (µs) | Max Latency (µs) | Avg Jitter (µs) | Max Jitter (µs) |
|---|---|---|---|---|---|
| 64 | 12.676 | 7.375 | 21.904 | 0.012 | 7.072 |
| 128 | 19.126 | 12.723 | 29.577 | 0.018 | 11.546 |
| 256 | 37.647 | 22.730 | 62.829 | 0.012 | 21.753 |
| 512 | 61.067 | 43.449 | 91.470 | 0.017 | 42.240 |
| 1024 | 147.686 | 84.166 | 260.607 | 0.023 | 68.491 |
| 1280 | 156.752 | 104.660 | 289.633 | 0.020 | 104.333 |
| 1518 | 210.925 | 124.489 | 356.842 | 0.031 | 122.879 |
Port P-0-6-1
| Frame Size (bytes) | Avg Latency (µs) | Min Latency (µs) | Max Latency (µs) | Avg Jitter (µs) | Max Jitter (µs) |
|---|---|---|---|---|---|
| 64 | 12.737 | 7.150 | 21.784 | 0.025 | 13.470 |
| 128 | 19.865 | 12.266 | 30.748 | 0.018 | 12.139 |
| 256 | 39.656 | 22.497 | 62.717 | 0.010 | 22.419 |
| 512 | 61.546 | 42.984 | 86.884 | 0.016 | 34.558 |
| 1024 | 141.958 | 83.933 | 208.480 | 0.022 | 83.862 |
| 1280 | 169.647 | 104.428 | 307.193 | 0.026 | 104.285 |
| 1518 | 213.225 | 124.120 | 381.899 | 0.033 | 122.895 |
Port P-0-6-2
| Frame Size (bytes) | Avg Latency (µs) | Min Latency (µs) | Max Latency (µs) | Avg Jitter (µs) | Max Jitter (µs) |
|---|---|---|---|---|---|
| 64 | 12.963 | 7.423 | 21.679 | 0.024 | 14.112 |
| 128 | 17.180 | 12.571 | 24.734 | 0.017 | 11.530 |
| 256 | 38.562 | 22.770 | 61.354 | 0.010 | 17.287 |
| 512 | 67.751 | 43.264 | 113.721 | 0.016 | 42.873 |
| 1024 | 125.593 | 84.214 | 238.749 | 0.018 | 83.838 |
| 1280 | 181.930 | 104.700 | 306.977 | 0.023 | 104.333 |
| 1518 | 243.911 | 124.537 | 437.530 | 0.032 | 123.537 |
Port P-0-6-3
| Frame Size (bytes) | Avg Latency (µs) | Min Latency (µs) | Max Latency (µs) | Avg Jitter (µs) | Max Jitter (µs) |
|---|---|---|---|---|---|
| 64 | 11.479 | 7.407 | 18.015 | 0.012 | 7.697 |
| 128 | 17.114 | 12.522 | 23.323 | 0.017 | 9.581 |
| 256 | 34.577 | 22.754 | 51.708 | 0.012 | 17.928 |
| 512 | 70.539 | 43.240 | 121.883 | 0.023 | 42.248 |
| 1024 | 135.288 | 84.190 | 239.230 | 0.023 | 83.846 |
| 1280 | 163.518 | 104.684 | 269.539 | 0.029 | 103.692 |
| 1518 | 254.248 | 124.626 | 469.627 | 0.029 | 123.537 |
Port P-0-6-4
| Frame Size (bytes) | Avg Latency (µs) | Min Latency (µs) | Max Latency (µs) | Avg Jitter (µs) | Max Jitter (µs) |
|---|---|---|---|---|---|
| 64 | 11.744 | 7.238 | 19.274 | 0.006 | 7.040 |
| 128 | 18.063 | 12.354 | 30.788 | 0.016 | 12.179 |
| 256 | 35.617 | 22.585 | 52.213 | 0.012 | 22.394 |
| 512 | 64.797 | 43.080 | 91.366 | 0.024 | 42.272 |
| 1024 | 156.205 | 84.029 | 234.724 | 0.024 | 83.846 |
| 1280 | 150.212 | 104.524 | 289.104 | 0.021 | 104.333 |
| 1518 | 243.904 | 124.345 | 413.908 | 0.028 | 123.545 |
Frame Loss Rate Test
The frame loss rate test holds a fixed offered rate for the full test duration and counts what fails to arrive. Each frame size was tested for 60 seconds across 3 iterations at 95% and 100% of line rate. No pass/fail threshold was configured (Use Pass/Fail Criteria: No), so the Pass states below record completion with the measured loss rather than compliance with a limit.
The PE-505 lost zero frames in every single one of the 14 tests — seven frame sizes at each of two rates. Aggregate received frames equalled aggregate transmitted frames in all 14 cases, receive rate equalled transmit rate in all 14, and every one of the 160 per-flow loss counters across the two rates read zero.
Results at 95% Rate
| Frame Size (bytes) | Tx (Frames) | Rx (Frames) | Tx Rate (L1) | Loss (Frames) | Loss Rate | Result |
|---|---|---|---|---|---|---|
| 64 | 42,410,400 | 42,410,400 | 475.00 Mbps | 0 | 0% | Pass |
| 128 | 24,070,800 | 24,070,800 | 475.00 Mbps | 0 | 0% | Pass |
| 256 | 12,907,200 | 12,907,200 | 474.98 Mbps | 0 | 0% | Pass |
| 512 | 6,696,000 | 6,696,000 | 474.97 Mbps | 0 | 0% | Pass |
| 1024 | 3,411,600 | 3,411,600 | 474.89 Mbps | 0 | 0% | Pass |
| 1280 | 2,739,600 | 2,739,600 | 474.86 Mbps | 0 | 0% | Pass |
| 1518 | 2,316,000 | 2,316,000 | 474.93 Mbps | 0 | 0% | Pass |
The chart is a flat line along zero, which is the intended result. At 95% of line rate the switch forwarded 42,410,400 frames at 64 bytes down to 2,316,000 frames at 1518 bytes with received counts matching transmitted counts exactly at every size.
Per-Port Frame Loss at 95% Rate
All five ports recorded identical figures. Port P-0-6-0 is shown as representative.
Port P-0-6-0
| Frame Size (bytes) | Tx (Frames) | Tx Rate (L1) | Tx Rate (fps) | Rx (Frames) |
|---|---|---|---|---|
| 64 | 8,482,080 | 95.00 Mbps | 141,368 | 8,482,080 |
| 128 | 4,814,160 | 95.00 Mbps | 80,236 | 4,814,160 |
| 256 | 2,581,440 | 95.00 Mbps | 43,024 | 2,581,440 |
| 512 | 1,339,200 | 94.99 Mbps | 22,320 | 1,339,200 |
| 1024 | 682,320 | 94.98 Mbps | 11,372 | 682,320 |
| 1280 | 547,920 | 94.97 Mbps | 9,132 | 547,920 |
| 1518 | 463,200 | 94.99 Mbps | 7,720 | 463,200 |
Ports P-0-6-1, P-0-6-2, P-0-6-3 and P-0-6-4 recorded values identical to P-0-6-0 at every frame size.
Results at 100% Rate
| Frame Size (bytes) | Tx (Frames) | Rx (Frames) | Tx Rate (L1) | Loss (Frames) | Loss Rate | Result |
|---|---|---|---|---|---|---|
| 64 | 44,642,400 | 44,642,400 | 499.99 Mbps | 0 | 0% | Pass |
| 128 | 25,336,800 | 25,336,800 | 499.98 Mbps | 0 | 0% | Pass |
| 256 | 13,586,400 | 13,586,400 | 499.98 Mbps | 0 | 0% | Pass |
| 512 | 7,048,800 | 7,048,800 | 499.99 Mbps | 0 | 0% | Pass |
| 1024 | 3,591,600 | 3,591,600 | 499.95 Mbps | 0 | 0% | Pass |
| 1280 | 2,883,600 | 2,883,600 | 499.82 Mbps | 0 | 0% | Pass |
| 1518 | 2,437,200 | 2,437,200 | 499.79 Mbps | 0 | 0% | Pass |
At the full 500 Mbps aggregate rate the result is unchanged: 44,642,400 frames forwarded at 64 bytes down to 2,437,200 at 1518 bytes, zero loss and a 0% loss rate at every size. This is the important figure for an embedded design — every port sending to every other port simultaneously, at full wire speed, for a full minute per frame size, without a single dropped frame.
Per-Port Frame Loss at 100% Rate
All five ports recorded identical figures. Port P-0-6-0 is shown as representative.
Port P-0-6-0
| Frame Size (bytes) | Tx (Frames) | Tx Rate (L1) | Tx Rate (fps) | Rx (Frames) |
|---|---|---|---|---|
| 64 | 8,928,480 | 100.00 Mbps | 148,808 | 8,928,480 |
| 128 | 5,067,360 | 100.00 Mbps | 84,456 | 5,067,360 |
| 256 | 2,717,280 | 100.00 Mbps | 45,288 | 2,717,280 |
| 512 | 1,409,760 | 100.00 Mbps | 23,496 | 1,409,760 |
| 1024 | 718,320 | 99.99 Mbps | 11,972 | 718,320 |
| 1280 | 576,720 | 99.96 Mbps | 9,612 | 576,720 |
| 1518 | 487,440 | 99.96 Mbps | 8,124 | 487,440 |
Ports P-0-6-1, P-0-6-2, P-0-6-3 and P-0-6-4 recorded values identical to P-0-6-0 at every frame size.
Back-to-Back Test
The back-to-back test measures the longest burst of frames the device can forward without loss. Each frame size was tested for 60 seconds across 3 iterations at 95% and 100% rates using a common result scope.
Every one of the 20 flows recorded zero receive loss at both rates and at all seven frame sizes — received frames matched transmitted frames exactly throughout.
The 100%-rate bursts are larger than the 95%-rate ones — 44,642,840 frames versus 42,410,700 at 64 bytes, about 5.3% more. That ratio is exactly 100/95, which tells you what happened: the switch never dropped a frame, so each burst simply ran for the full 60-second test duration and its size is the offered rate multiplied by that time.
In other words, the test never found the PE-505's burst limit. These figures are a floor on its back-to-back capacity, not a measurement of its buffer depth. A switch that does saturate its buffers shows the opposite pattern, with the 100% burst ending sooner than the 50% or 95% one.
Results at 95% Rate
| Frame Size (bytes) | Tx Burst (Frames) | Tx Burst (Bytes) | Max Offered Rate (fps) | Max Offered Rate (Mbps) | Result |
|---|---|---|---|---|---|
| 64 | 42,410,700 | 2,714,284,800 | 697,046 | 468.42 | Pass |
| 128 | 24,070,940 | 3,081,080,320 | 401,182 | 475.00 | Pass |
| 256 | 12,907,600 | 3,304,345,600 | 212,131 | 468.39 | Pass |
| 512 | 6,696,420 | 3,428,567,040 | 110,058 | 468.40 | Pass |
| 1024 | 3,412,340 | 3,494,236,160 | 56,083 | 468.41 | Pass |
| 1280 | 2,740,380 | 3,507,686,400 | 45,673 | 475.00 | Pass |
| 1518 | 2,316,300 | 3,516,143,400 | 38,070 | 468.41 | Pass |
At 95% of line rate the switch absorbed a 42.4-million-frame burst at 64 bytes (2.71 GB) through to 2.32 million frames at 1518 bytes (3.52 GB), without dropping a frame anywhere. Burst frame counts fall with frame size while burst bytes rise, because larger frames carry proportionally less per-frame overhead and so move more payload in the same 60 seconds.
The achieved offered rate sits at 468.39–468.42 Mbps at five of the seven frame sizes rather than the nominal 475.00 Mbps, about 1.4% below target; only 128 and 1280 bytes reached a full 475.00 Mbps. Since zero frames were lost at any size, that shortfall is in what the generator was able to offer, not in what the switch could forward.
Per-Port Back-to-Back at 95% Rate
All five ports recorded identical figures. Port P-0-6-0 is shown as representative.
Port P-0-6-0
| Frame Size (bytes) | Tx Burst (Frames) | Tx Burst (Bytes) |
|---|---|---|
| 64 | 8,482,140 | 542,856,960 |
| 128 | 4,814,188 | 616,216,064 |
| 256 | 2,581,520 | 660,869,120 |
| 512 | 1,339,284 | 685,713,408 |
| 1024 | 682,468 | 698,847,232 |
| 1280 | 548,076 | 701,537,280 |
| 1518 | 463,260 | 703,228,680 |
Ports P-0-6-1, P-0-6-2, P-0-6-3 and P-0-6-4 recorded values identical to P-0-6-0 at every frame size.
Results at 100% Rate
| Frame Size (bytes) | Tx Burst (Frames) | Tx Burst (Bytes) | Max Offered Rate (fps) | Max Offered Rate (Mbps) | Result |
|---|---|---|---|---|---|
| 64 | 44,642,840 | 2,857,141,760 | 744,047 | 500.00 | Pass |
| 128 | 25,337,820 | 3,243,240,960 | 422,297 | 500.00 | Pass |
| 256 | 13,586,940 | 3,478,256,640 | 226,449 | 500.00 | Pass |
| 512 | 7,048,860 | 3,609,016,320 | 117,481 | 500.00 | Pass |
| 1024 | 3,591,940 | 3,678,146,560 | 59,866 | 500.00 | Pass |
| 1280 | 2,884,600 | 3,692,288,000 | 48,077 | 500.00 | Pass |
| 1518 | 2,438,220 | 3,701,217,960 | 40,637 | 500.00 | Pass |
At full line rate the switch absorbed a 44.6-million-frame burst at 64 bytes (2.86 GB) through to 2.44 million frames at 1518 bytes (3.70 GB), again without loss. The offered rate held a flat 500.00 Mbps at all seven frame sizes — the complete aggregate capacity of the five ports — sustained for the full 60 seconds of every burst.
Per-Port Back-to-Back at 100% Rate
All five ports recorded identical figures. Port P-0-6-0 is shown as representative.
Port P-0-6-0
| Frame Size (bytes) | Tx Burst (Frames) | Tx Burst (Bytes) |
|---|---|---|
| 64 | 8,928,568 | 571,428,352 |
| 128 | 5,067,564 | 648,648,192 |
| 256 | 2,717,388 | 695,651,328 |
| 512 | 1,409,772 | 721,803,264 |
| 1024 | 718,388 | 735,629,312 |
| 1280 | 576,920 | 738,457,600 |
| 1518 | 487,644 | 740,243,592 |
Ports P-0-6-1, P-0-6-2, P-0-6-3 and P-0-6-4 recorded values identical to P-0-6-0 at every frame size.
About This Test
RFC 2544 — Benchmarking Methodology for Network Interconnect Devices — defines the standard test procedures for measuring throughput, latency, frame loss rate, and back-to-back burst capacity of network devices in a controlled laboratory environment.
Testing was performed with the Xena Networks Xena2544 automated test suite (v2.97), which implements the binary search and timing procedures defined by the RFC. Each test sub-section ran for 60 seconds per frame size across 3 iterations; reported values are averages of those iterations. The full run took 2 hours 49 minutes of wall-clock time across all four test suites.
The report's own comment records the intent: "Brainboxes PE-505 tested to RFC2544 with frame sizes of 64 to 1518 with Xena2544."
The PE-505 is a 5-port embedded Fast Ethernet switch in the Brainboxes PE range of Pure Embedded industrial Ethernet switches. See the PE-505/508 overview for hardware features, the PE-505/508 specifications for electrical and operating data, the PE-505/508 design guide for integration guidance, and other RFC 2544 test reports for comparable results across the Brainboxes switch ranges — including the PE-508, the 8-port version of the same Fast Ethernet design.
