Sign in
AppMarketSpy › Education › GPU Performance Lab
GPU Performance Lab icon

GPU Performance Lab

CUDA puzzles and real GPU labs
GPUFlight · Education
View on App Store ↗☆ Watch
★ 5.00average rating
1ratings
+0new ratings, last 7 days
—not in top charts now

GPU Performance Lab is an iOS Education app by GPUFlight, released on Nov 6, 2026. The current version, 1.1, came out 4 days ago. It has 1 ratings with an average of 5.0 stars. It is free to download and offers 1 in-app purchase priced $6.99.

Ratings breakdown

★★★★★
100%
★★★★
0%
★★★
0%
★★
0%
★
0%
🔒 Rating history14 days of daily rating and rating countUnlock free — sign up

App info

Developer
GPUFlight
Category
Education, Developer Tools
Released
Nov 6, 2026
Version
1.1 (Nov 17, 2026)
Price
Free
Size
5.8 MB
Requires
iOS 17.4
Age rating
16+
Languages
English

In-app purchases

Practice$6.99
Current prices, App Store US.
🔒 Price history5 snapshots: when prices changed and new plans appearedUnlock free — sign up

Keyword rankings

Where this app shows up in App Store search (US).
gpuscore: in vitro 2#25▲2

Recent reviews

A sample of the latest reviews we captured on the App Store.
★★★★★ Fun way to learn CUDA · Nov 6, 2026
This is a great app to dive into CUDA and sharpen your skills especially if you're into GPU performance and profiling.

About GPU Performance Lab

GPUs are expensive, and AI runs on them. Teams that run their own models need developers who can make GPU code fast. That skill starts with CUDA. Take the first step on your phone: short CUDA puzzles, endless practice sets, then labs on real cloud GPUs that work up to softmax, SwiGLU and attention, the kernels inside LLMs. LAB PATHS: FROM A CPU LOOP TO LLM KERNELS 20 labs in 4 paths, all open on the free plan: • Elementwise Kernels: your first CUDA kernel, grid-stride loops, launch shape, coalescing, then a fused SwiGLU. • Reductions: atomics, a block tree and warp shuffles, up to stable and online softmax, the trick attention kernels are built on. • Matrix Multiplication: from a CPU loop to shared-memory tiling at N = 4096. • Decoder Block: RMSNorm, a packed QKV projection, RoPE, grouped-query attention and causal attention. Small sizes for teaching; no real model weights. HOW A LAB…
🔒 Developer intelligencelegal entity, account ID and every app from this sellerUnlock free — sign up

Top free Education apps right now

Full chart → · All Education

Other Education apps released around the same time