Wordpress WPPerformanceTester Results

Something strange happens with MySQL 8 (on Ubuntu LEMP stack, non-Centminmod):



So I guess this test is not reliable with MySQL8?

Btw, CHI1 works now, but
->
I get the same CPU on CHI1:
1 Intel(R) Xeon(R) CPU E5-2687W v3 @ 3.10GHz
1 3099.998

 

@eva2000

Was a simple Ubuntu LEMP stack, non-Centminmod. I have not yet figured out how to utilize MySQL8 on Centminmod. I will try to read through the forums and mysql-control.

The original purpose for this odysseia was to optimize a huge daily WP-All Import job on Wordpress/Woocommerce setup. Now it is mainly a curiosity.

Thanks for noticing the above change conserning the ENCODE(str,pass_str) function.

 

So I've been continuing to benchmark with this tool, since it's convenient and closely correlates with eva2000 trademark benchmark's results.

Today I stumbled upon Unesty: Unesty - Mein Warenkorb

4 vCores @ 4,0 GHz / 1 GB RAM / 25 GB SSD
3.49 EUR / month
3.5 / 2400

This is a clear winner at the moment, almost good true to be true. What I am missing here? It is "optimized for gaming", but has unlimited bandwith and 1Gbps connection.

w/
tinof/custom_config.inc
tinof/kernel5.sh

# install centmin mod latest beta with php-fpm 7.3 default
yum -y update; curl -O https://centminmod.com/betainstaller73.sh && chmod 0700 betainstaller73.sh && bash betainstaller73.sh
---
Other results. Best result from ~10 runs, except Hetzner, which has the highest variance and I follow it daily. Upcloud is the most stable, Vultr is nearly there. These are just my notes.

CENTMINMOD
Hezner EX 62-NME (Intel® Core™ i9-9900K: @ 3.60GHz)
4.6 / 3500

Vultr 1CPU1GB Amsterdam (Intel @ 3.8GHz)
6.1 / 2400

Upcloud 2CPU4GB (Intel Xeon Gold 6136 @ 3.00Ghz)
6.5 / 2300

Upcloud 1CPU1GB (Intel Xeon Gold 6136 @ 3.00Ghz)
6.5 / 2100

Upcloud 1CPU1GB (Intel(R) Xeon(R) CPU E5-2687W v4@ 3.00GHz)
7.5 / 1850

Upcloud 1CPU1GB (Intel(R) Xeon(R) CPU E5-2687W v3 @ 3.10GHz)
8.3 / 1850

Hezner 1CPU2GB
8.9-12 / 600-1529

Aruba 1CPU1GB
+12 / 1090

Scaleway DEV1-S
+10/400

 

NexusBytes GER-NVMe-SSD-Pro (8GB / 4-CPU)
model name : AMD Ryzen 7 3700X 8-Core Processor
3.4 / 2700

 

Nope, but I received few abuse notices for just playing around with Centminmod and GCC..

 

Just to clarify; I meant abuse notice for excessive CPU usage

 

@Jimmy I agree, you get what you pay for here. A cheap way to get access to Zen2 (AMD Ryzen 3xxx).

Currently messing around with Hetzner AX-41 (not NVMe). This seems very unstable, I mean the queries per second start from around 2300. The second run is <2000 and within five runs its down to 500. And this with or without GCC9 & other tryouts.

Wierd. I have not seen this kind of behavior before.

 

Can they do that, on a dedicated server?

PS. Please publish your latest server results @eva2000, so this kind of amateur testing is not needed

 

Oddities with Hetzner AX41 (Ryzen 3600) and AX51-NVME (Ryzen 3700X):

So with 5.4 Kernel and betainstaller73.sh and modified custom_config

AX41 HDD (Ryzen 3600): PHPBench 3.4 / 500 - 2300 req / s (not stable)
AX51 NVMe (Ryzen 3700X): PHPBench 3.2 / 1800 - 1900 req / s (not very stable)

I find it strange, that the request per second vary so much. This much variance can be seen only with Hetzner cloud, and there it is affecting the PHP Bench as well. Here only req / s is affected.

Also, the result for req / s is not very good compared to i9-9900K.

Hetzner EX 62-NME (Intel® Core™ i9-9900K: @ 3.60GHz): PHPBench 4.7 / 3600 req / s (stable)


Building with GCC9 does not change anything.