ROWID is of alias of an integer primary key:

CREATE TABLE test1(id INTEGER, b TEXT, PRIMARY KEY(id))
INSERT INTO test1 (id, b) VALUES (5, 'five');
INSERT INTO test1 (id, b) VALUES (6, 'six');
INSERT INTO test1 (b) VALUES ('seven');
SELECT rowid, * FROM test1;

SQLite utilizes up to 90% of the my SSD disk when SHIB coin riches its local peaks with more than 10.000 – 18.000 trades/min that is 166 – 300 trades/second. My C++ app inserts 64-bit values to tables like this:

CREATE TABLE trades (time INT PRIMARY KEY NOT NULL, price REAL, volume REAL) WITHOUT ROWID;

so it is only 3984 – 7200 bytes/second. Totally across all markets there can be about 30.000 trades/min that is 500 trades/second and only 12.000 bytes/second and it is enough to slow down my SQLite database.

I use WAL with 4GB cache:

PRAGMA journal_mode = WAL;
PRAGMA cache_size = -4 * 1024 * 1024;

and do all the inserts in a transaction once a 5 seconds into 500+ tables (a separate table for each market) with a prepared statement.

See the screenshots of my app below:

I did a quick Google search on “SQLite performance” and found the following:

• What are the performance characteristics of sqlite with very large database files? (The performance degrades significantly when the database size riches 7GB).
• Some Russian article on habr.com. (To boost the insertion we insert by batches in a separate transactions, set synchronization mode to OFF or NORMAL and probably disable indices).
• Some old article on Database Speed Comparison.
• Appropriate Uses For SQLite (An SQLite database is limited in size to 140 terabytes)
• Faster bulk inserts in sqlite3? (You can also try tweaking a few parameters to get extra speed out of it. Specifically you probably want PRAGMA synchronous = OFF;)
• A trivial C++ example.

Then to benchmark SQLite performance by myself I used the following C++ code that inserts 1000 batches of 1000 000 rows to a single table with an integer primary key: