pipersr 1.0.1

First super-resolution model designed for Apple Neural Engine. 2x upscale, real-time, on-device.

PiperSR

First super-resolution model designed and optimized for Apple Neural Engine.

2x upscale. 453K parameters. 928 KB. Real-time video at 44.4 FPS. Runs entirely on the ANE in every Mac, iPhone, and iPad — zero GPU, zero cloud.

---

At a Glance

	PiperSR
Parameters	453,000
Model size	928 KB (CoreML FP16)
Scale factor	2x
PSNR (Set5)	37.54 dB
PSNR (Set14)	33.21 dB
PSNR (BSD100)	31.98 dB
PSNR (Urban100)	31.38 dB
FPS (360p → 720p)	44.4 FPS (M2 Max)
ANE latency	22.5 ms/frame
Compute	Apple Neural Engine only
CPU/GPU fallback ops	Zero
Precision	FP16
Training cost	~$6 (RunPod A6000)

---

How It Compares

Quality vs. Parameter Count (2x upscale)

Models sorted by parameter count. PSNR measured on standard academic benchmarks (bicubic degradation).

Model	Params	Size	Set5	Set14	BSD100	Urban100	Year
FSRCNN	12K	~50 KB	36.94	32.54	31.73	—	2016
ESPCN	~20K	~100 KB	~36.7	~34.5	~34.3	—	2016
SPAN-S	426K	~1.7 MB	38.06	33.73	32.21	32.20	2024
PiperSR	453K	928 KB	37.54	33.21	31.98	31.38	2026
SPAN	498K	~2 MB	38.08	33.71	32.22	32.24	2024
OmniSR	792K	3.3 MB	38.12	33.70	32.22	32.38	2023
SwinIR-light	878K	~3.5 MB	38.14	33.86	32.31	32.76	2021
EDSR-baseline	1.5M	~6 MB	38.02	33.57	32.12	—	2017
SwinIR	11.9M	~48 MB	38.42	34.46	32.53	33.81	2021
Real-ESRGAN	16.7M	64 MB	—	—	—	—	2021
HAT	20.8M	~80 MB	38.63	34.86	32.62	34.45	2023

Real-ESRGAN is GAN-trained for perceptual quality on real-world degradations, not bicubic benchmarks — PSNR comparison is not applicable.

Size Comparison

PiperSR     928 KB  ██
SPAN-S     ~1.7 MB  ███
SPAN       ~2.0 MB  ████
OmniSR      3.3 MB  ██████
SwinIR-lt  ~3.5 MB  ███████
EDSR-base  ~6.0 MB  ████████████
SwinIR     ~48  MB  ████████████████████████████████████████████████████████████████████████████████████████████████
Real-ESRGAN  64 MB  █████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████

PiperSR is 69x smaller than Real-ESRGAN and 4x smaller than SwinIR-light.

Speed: Apple Silicon (2x upscale, 360p → 720p)

No other model in this parameter class was designed for ANE. These are real measurements, not theoretical throughput.

Model	Hardware	FPS	Compute	Real-time?
PiperSR	M2 Max	44.4	ANE	Yes (1.5x real-time)
SPAN	M2 Mac (ailia)	~7	GPU	No
Real-ESRGAN	M2 Mac (ailia)	~0.3	GPU	No

SPAN and Real-ESRGAN measurements from ailia SDK M2 Mac benchmarks. PiperSR is ~7x faster than SPAN and ~160x faster than Real-ESRGAN on the same hardware family.

The Tradeoff

PiperSR is 0.5 dB below SPAN on Set5 (37.54 vs 38.06). That's the cost of building for ANE instead of GPU. Here's what you get in return:

	PiperSR	SPAN-S
Set5 PSNR	37.54 dB	38.06 dB
Model size	928 KB	~1.7 MB
Apple Silicon FPS	44.4 (ANE)	~7 (GPU)
Real-time video	Yes	No
GPU usage	0%	100%
Runs on iPhone/iPad	Yes (ANE)	Requires GPU

0.5 dB is below the perceptual threshold for most content. 44.4 FPS vs ~7 FPS is the difference between real-time video and a slideshow.

---

Quick Start

pip install pipersr

CLI:

pipersr -i photo.png -o photo_2x.png

Python:

from pipersr import upscale

result = upscale("photo.png")
result.save("photo_2x.png")

Model weights (928 KB) are bundled in the package. No downloads, no API keys, no accounts.

Download

The CoreML model (PiperSR_2x.mlpackage, 928 KB) is included in this repo. Clone and run — no separate download needed.

Also available on ModelPiper.com.

---

Architecture

Input (H×W×3, FP16)
  │
  ├─ Head: Conv 3×3 (3 → 64 channels)
  │
  ├─ Body: 6 Residual Blocks
  │    ┌──────────────────────────────┐
  │    │  Conv 3×3 (64ch)             │
  │    │  BatchNorm                    │
  │    │  SiLU                         │
  │    │  Conv 3×3 (64ch)             │
  │    │  BatchNorm                    │
  │    │  + Residual                   │
  │    └──────────────────────────────┘
  │
  ├─ Tail: Conv 3×3 (64 → 12ch) → PixelShuffle(2)
  │
  Output (2H×2W×3, FP16)

453K parameters. 5 unique MIL ops. Every operation — conv, batch_norm, silu, add, pixel_shuffle — is native to the Neural Engine instruction set. Zero ops fall back to CPU or GPU.

Why These Choices

Decision	Why
BatchNorm (not RMSNorm)	RMSNorm decomposes into 4 ops that ALL fall back to CPU. Switching to BatchNorm gave a 2.5x throughput increase — the single biggest optimization in the project
SiLU (not SwiGLU)	Single ANE-native op. SwiGLU adds complexity for marginal gain at this model scale
No attention	At 128×128 tiles, attention = 16K tokens. Adds compute without clear quality benefit for a conv SR model at 453K params
PixelShuffle (not transposed conv)	Deterministic reshape, no checkerboard artifacts, maps to simple memory ops on ANE
64 channels	ANE tile-aligned. 48ch (used by SPAN) wastes ~25% of each ANE tile on padding
.cpuAndNeuralEngine (not .all)	.all is 23.6% slower — CoreML silently misroutes pure-ANE ops onto the GPU

Video Pipeline (Double-Buffered)

For real-time video, PiperSR uses a double-buffered pipeline where each hardware unit works on a different frame simultaneously:

Frame N:    [CPU convertIn] → [ANE predict: 22.5ms] → [Metal GPU convertOut]
Frame N+1:                    [CPU convertIn] ─────── → [ANE predict: 22.5ms] → ...

Resolution	Input → Output	ANE Predict	Streaming FPS	Real-time?
360p	640×360 → 1280×720	22.5 ms	44.4 FPS	1.5x real-time
480p	854×480 → 1708×960	32.7 ms	30 FPS	1.0x real-time
720p	1280×720 → 2560×1440	71.6 ms	14 FPS	No

Available in ToolPiper for real-time video upscaling on macOS.

---

Benchmarks

Reproducible. Run python benchmark.py on your own hardware.

# PSNR evaluation (requires Set5 dataset with LR/ and HR/ directories)
python benchmark.py --dataset /path/to/Set5

# FPS benchmark at 360p
python benchmark.py --resolution 640x360 --iterations 200

Full PSNR Results

Dataset	PiperSR	Bicubic	Delta
Set5	37.54 dB	33.66 dB	+3.88 dB
Set14	33.21 dB	30.24 dB	+2.97 dB
BSD100	31.98 dB	29.56 dB	+2.42 dB
Urban100	31.38 dB	26.88 dB	+4.50 dB

Inference Speed by Hardware

Hardware	Mode	FPS	Latency
M2 Max	Full-frame 360p	44.4	22.5 ms
M2 Max	Full-frame 480p	30.0	32.7 ms
M2 Max	Full-frame 720p	14.0	71.6 ms
M2	Tiled 128×128 (static weights)	125.6	7.96 ms
M2	Tiled 128×128 (dynamic weights)	44.5	22.5 ms

Static (baked) weights are 2.82x faster than dynamic — the main bottleneck is weight loading, not ANE compute.

---

Why ANE?

Every Apple Silicon chip has a Neural Engine — a dedicated inference accelerator that most ML models completely ignore. The entire ML ecosystem targets CUDA. Models get "converted" to CoreML as an afterthought, and the conversion is usually bad: misaligned tensor shapes cause pipeline stalls, unsupported ops silently fall back to CPU, and full-frame predictions can't exploit ANE parallelism.

PiperSR was designed from scratch for ANE, not converted from a GPU model. Every dimension, every operation, every data type was chosen based on measured ANE hardware characteristics. The result: real-time super-resolution that uses zero GPU. Your GPU stays free for rendering, compositing, or whatever else you're doing.

Supported Hardware

Any Apple Silicon device:

Platform	Chips
Mac	M1, M1 Pro/Max/Ultra through M5
iPhone	A15+
iPad	M1+

Performance scales with ANE generation. M2 Max benchmarks shown above.

For Video & Real-Time

PiperSR is an image model. For real-time video super-resolution with double-buffered frame scheduling, Metal GPU output conversion, and streaming — see ToolPiper, which integrates PiperSR into a production video pipeline.

---

Model Details

• Task: Single-image 2x super-resolution
• Architecture: 6-block residual CNN, 64 channels, BatchNorm + SiLU, PixelShuffle upsampling
• Input: RGB image, any resolution (CoreML flexible input)
• Output: 2x upscaled RGB image, FP16
• Format: CoreML .mlpackage
• Compute: Apple Neural Engine via .cpuAndNeuralEngine
• Training data: Standard SR datasets (DIV2K, Flickr2K)
• Training cost: ~$6 total on RunPod A6000 instances

For the full model card, see MODEL_CARD.md.

License

The model weights are CC BY 4.0 — fully permissive. Use them for anything: personal, academic, commercial. Ship them in your app, build a product, make money with it. Seriously, that's allowed and encouraged.

The only thing we ask: link back to ModelPiper.com as attribution, so others can find the project.

Powered by PiperSR from ModelPiper — https://modelpiper.com

• Model weights: CC BY 4.0 — permissive, commercial use encouraged, attribution = a backlink to modelpiper.com
• Source code: AGPL-3.0

Links

• ModelPiper.com — Download PiperSR, browse benchmarks, explore on-device models
• ToolPiper — Local macOS AI toolkit with real-time video upscaling
• Ben Racicot — Author

Citation

@software{pipersr2026,
  author = {Racicot, Ben},
  title = {PiperSR: Real-Time Super-Resolution for Apple Neural Engine},
  year = {2026},
  url = {https://github.com/ModelPiper/PiperSR},
  note = {453K parameters, 928 KB CoreML. Available at https://modelpiper.com}
}