Benchmark clarifications

[L0laapk3]

Hi Awawa,

Nice work on all the hyperhdr related work.

I have a question regarding the benchmark section on this repo. Were these benchmarks carried out at 80MHz or 160MHz cpu frequency?

My understanding of ESP8266 is that the default CPU frequency is 80MHz, and I did not spot any board_build.f_cpu = 160000000L in the platformio.ini. Do you think the project could benefit from running at 160MHz?

Thanks!

[L0laapk3]

Hi, I attempted to run some benchmarks myself. I'm not 100% confident if my methodology is correct, so please let me know if you see any issues:

• I used your latest hyperhdr release on PC, with screen capture set to 120FPS (unfortunately it doesn't go higher, I have a 240Hz monitor..). Smoothing is disabled.
• I used your latest HyperSerialESP8266 release v11.0.0, firmware_SK6812_RGBW_COLD.bin on wemos d1 mini pro v1.1 (which has the CP2104).

I will note that I did not actually have this many leds connected, just a single one. The rest is taped to my tv as I typically use the webos build, I'm just using PC now for testing :) I did verify each time that the first led responds correctly to source image changes.

Interestingly, I got around double the performance compared to your benchmark table. That said, I saw no difference for a different cpu frequency. I tried the following configurations:

• Your latest release .bin
• compiled the sketch myself with no changes
• adding board_build.f_cpu = 16000000L
• adding board_build.f_cpu = 8000000L

I will say that for me its not clear to me how these numbers are supposed to make any sense, my understanding is that they are exceeding the theoretical framerate possible with sk6812: 800kbps / 32 bit / 300 leds = 83fps...

300 leds:

2026-03-29T13:59:48.308Z [PERFORMANCE] [INSTANCE0: FPS = 125.00, processed = 7500], [LED0: FPS = 124.82, send = 7489, processed = 7500, dropped = 11]
2026-03-29T14:00:48.308Z [PERFORMANCE] [INSTANCE0: FPS = 125.00, processed = 7500], [LED0: FPS = 124.73, send = 7484, processed = 7500, dropped = 16]

600 leds:

2026-03-29T13:55:48.308Z [PERFORMANCE] [INSTANCE0: FPS = 124.97, processed = 7498], [LED0: FPS = 78.76, send = 4726, processed = 7500, dropped = 2774]
2026-03-29T13:56:48.308Z [PERFORMANCE] [LED0: FPS = 78.77, send = 4726, processed = 7498, dropped = 2772], [INSTANCE0: FPS = 124.99, processed = 7500]
2026-03-29T13:57:48.314Z [PERFORMANCE] [LED0: FPS = 78.75, send = 4725, processed = 7500, dropped = 2775], [INSTANCE0: FPS = 124.97, processed = 7499]

1200 leds:

2026-03-29T14:02:48.318Z [PERFORMANCE] [INSTANCE0: FPS = 125.00, processed = 7500], [LED0: FPS = 39.48, send = 2369, processed = 7487, dropped = 5118]
2026-03-29T14:03:48.319Z [PERFORMANCE] [INSTANCE0: FPS = 125.00, processed = 7500], [LED0: FPS = 39.48, send = 2369, processed = 7500, dropped = 5131]
2026-03-29T14:04:48.319Z [PERFORMANCE] [INSTANCE0: FPS = 124.98, processed = 7499], [LED0: FPS = 39.48, send = 2369, processed = 7495, dropped = 5126]

[awawa-dev]

Hi
This only tests the host's sending capability. To test the actual driver capability, you should do it this way: https://www.hyperhdr.eu/2023/02/ultimate-guide-on-how-to-build-led.html#chapter9 . Or use logic analyzer.