Talk to usTalk to us
Measured · same site on both stacks · no page cache

More requests a second, with the plugins on.

On a 4-vCPU Intel server, BookPress serves about 4.5 times the requests per second of a tuned WordPress on real WooCommerce and Elementor pages. On a 4-vCPU ARM server it serves over 5 times. The WordPress side runs PHP-FPM, OPcache with its JIT, a Redis object cache and MariaDB. The stacks and every number →

~4.5×
the requests per second of tuned WordPress on real WooCommerce and Elementor pages, Intel server
5+×
the requests per second on the ARM server, where BookPress runs on four real cores
~2.3×
more Elementor requests per second on the ARM server than the Intel one, on BookPress itself (2.2× for WooCommerce)
0
page caches, object caches or database servers in the request path
The hard case

We show each page on each server as a multiple of WordPress.

We take the numbers from a repeatable setup. It builds a tuned WordPress and loads the identical demo content into WordPress and BookPress. Once it has checked that both show the same page, it drives each engine to the most requests a second it serves with 99% of them under one second and none failing. BookPress renders every page fresh. The WordPress side runs its own PHP with the page cache switched off, the state a cart, a search, an account page or a logged-in session is in.

Both engines served one site

We migrated a real WordPress site across, with the Kadence theme, WooCommerce with a five-product catalogue, a built Elementor showcase page, Yoast SEO and Contact Form 7. Both stacks serve the Elementor showcase, the product page and the shop catalogue from the same seeded content. Before timing, we check that the two databases carry the same row counts.

The WordPress we compared against

WordPress 7.1.3 on PHP 8.3.35 with OPcache and the JIT enabled, a Redis object cache over a socket, MariaDB, and PHP-FPM behind nginx, with WooCommerce on its newer order storage (HPOS). We tuned it the way a competent host tunes one. We switch the page cache off for every measurement and check that it's off before timing.

Both servers

A 4-vCPU Intel server and a 4-vCPU ARM server (Google Cloud c2-standard-4 and c4a-standard-4), both 16 GB. BookPress wins on both. The ARM server has four real cores where the Intel one has two. BookPress serves about 2.3 times as many Elementor requests per second on the ARM server as on the Intel one.

Every number, as a multiple of WordPress
Each figure is BookPress against WordPress on the same server. Requests per second is each engine at its own knee; the response-time multiples are at WordPress's knee rung, the same concurrency for both.
PageServerRequests / secondp50 response timep99 response time
Elementor page Intel 4.53×162 against 36 pages/s 3.80× 1.93×
WooCommerce shop Intel 4.85×134 against 28 pages/s 4.75× 2.79×
Elementor page ARM 5.30×376 against 71 pages/s 5.30× 3.70×
WooCommerce shop ARM 5.19×297 against 57 pages/s 5.20× 2.50×

Measured on Google Cloud c2-standard-4 (Intel) and c4a-standard-4 (Arm), uncached dynamic pages, BookPress 0.2.64 (the packaged build you install) against WordPress 7.1.3 / PHP 8.3.35 with nginx, PHP-FPM, Redis object cache, OPcache with JIT and MariaDB.

Requests per second is at each engine's knee; p50 and p99 are the median and 99th-percentile response-time multiples at WordPress's knee rung. We measure WordPress without a page cache. Either setup serves a cacheable page from its cache, so a cache hit doesn't measure the engine.

A site with no plugins runs no PHP interpreter. There BookPress serves 112.9 times the requests per second of WordPress on the Intel server (5,234 against 46) and 125 times on the ARM server (11,436 against 91).

Memory

We measured the same site's memory.

We took the memory figures earlier, in a separate run on BookPress 0.2.57. Under full load on the Intel server the same site took about 0.6 GB on four BookPress workers, against 0.55 GB for a right-sized WordPress and 1.8 GB for a tuned 64-child pool. Thirty seconds after the load stopped, BookPress released to about 0.3 GB while the WordPress pools held their 0.43 to 1.8 GB. The density page has the full memory figures →

We measured memory on BookPress 0.2.57, 13 October 2026, and counted memory shared between processes once.

Where we are careful

Two limits apply to every number here.

We publish the stacks so that you can check them. The numbers above don't cover two cases.

Page caches are a separate layer

We measure WordPress without a page cache. Either setup serves a cacheable page from its cache, so a cache hit doesn't measure the engine. Every figure here is an uncached dynamic render on both stacks, the state a cart, a search or a logged-in session is in.

Classic PHP themes are newer to BookPress

WordPress's own classic templates are lighter than its block themes. BookPress renders classic themes on a newer path than block themes, and a few of the bundled classic themes still fall back to an error today. The pages above are block themes.

Check the numbers yourself

The demo is a real WordPress site with no page cache in front of it. Sign in, publish a post, then open the shop or a page no cache could have kept.

Open the live demo