NOR Flash vs NAND Flash: How to Choose the Right SPI Memory for Your Design

W25N01GVZEIG — 1Gbit SPI NAND Flash (SMD-8)

Introduction

Serial flash is the default external memory for embedded systems, but "serial flash" covers two very different technologies: NOR flash and NAND flash. When a design needs to store firmware, a file system, data logs or firmware-update images, choosing the wrong type means either paying too much or running out of reliability headroom. This guide compares NOR and NAND SPI flash — using the W25Q128JVSIQ (128 Mb NOR) and W25N01GVZEIG (1 Gb NAND) as reference parts — and gives a practical decision process.

Key Differences at a Glance

NOR flash (W25Q128JVSIQ type)
Capacity: typically 1 Mb - 512 Mb
Interface: SPI with execute-in-place (XIP) support
Erase granularity: 4 KB sectors, 32 KB / 64 KB blocks
Read: random access, byte-addressable reads
Endurance: ~100,000 erase cycles
Price per bit: higher

NAND flash (W25N01GVZEIG type)
Capacity: typically 1 Gb - 8 Gb and beyond
Interface: SPI (also parallel NAND variants)
Erase granularity: 128 KB blocks, pages of 2 KB
Read: page-based access, no XIP
Endurance: ~100,000 program/erase cycles per block, but requires ECC and bad-block management
Price per bit: much lower

Execute-in-Place vs Page Access

The single most important architectural difference: NOR flash supports random-access reads, so an MCU can execute code directly from it (XIP). NAND flash reads whole pages into RAM first and cannot execute in place — code must be copied to internal or external RAM before running.

This drives everything else. If your application boots from external flash, NOR is the natural choice: connect it, map it into the address space, and run. If your design stores data or receives OTA images that get moved into RAM or internal flash before use, NAND works and costs far less per gigabyte.

Reliability and Management Overhead

NAND flash is not plug-and-play. It requires:

1. ECC — every page read needs error correction (on-die ECC in parts like the W25N01GV helps, but the host must still manage it).
2. Bad-block management — NAND devices ship with marked bad blocks, and new bad blocks can develop over life; the filesystem or FTL must remap around them.
3. Wear leveling — erase cycles are concentrated on hot blocks unless the controller distributes writes.

NOR flash needs none of this: write, erase, read, done. For small firmware images and configuration storage, the simplicity is worth the higher price per bit.

Selection Guide

Choose NOR flash (W25Q128JVSIQ class) when: code executes directly from external memory, the firmware image is under ~16 MB, you need simple sector erase without a filesystem driver, or the design must survive years of field updates with minimal complexity.

Choose NAND flash (W25N01GVZEIG class) when: you need high capacity (1 Gb and up) for data logging, image storage or large OTA payloads, cost per gigabyte matters, and you can afford a filesystem with ECC and bad-block handling (or a hosted FTL solution).

Applications

NOR: boot firmware for MCUs and SoCs, BIOS and configuration storage, FPGA bitstreams, code shadowing, small parameter tables. NAND: data logging in industrial and automotive black boxes, voice and image assets in consumer devices, large OTA image staging in IoT gateways, USB drive and SSD-class storage, video buffering.

Sourcing Notes

When comparing suppliers, confirm the exact density and die revision (NAND parts from different vendors differ in page size and ECC requirements), the date code, and whether the parts are original and traceable. For NAND, check whether on-die ECC is supported and how bad blocks are marked. Verify packing (tube vs tape & reel) for production quantities. All memory parts listed in our store are original, in stock and ready to ship.

Summary

NOR is for executing, NAND is for storing. Pick NOR (W25Q128JVSIQ) for boot code and simple storage; pick NAND (W25N01GVZEIG) when capacity and cost beat simplicity. Both are SPI, both are 3.3 V — the difference is how your firmware touches them.