New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
It was right at the end that I saw that it was from LowendBoxTV. Was that paid spot from Advin Servers?
The video is very good, but the plugin used to perform the test at the beginning of the video hasn't been shared. Does anyone know where I can download it? It doesn't seem to be public.
I haven’t watched it yet, but we had no idea that we would be mentioned in a video and haven’t paid for any advertising past the provider tag fee
Is there any article to read instead?
any summary
It's always YouTuber and their face lmao. I need wall of text.
I will call out, not all EPYCs are created equal.
We are running AMD EPYC 4564P processors from the EPYC 4004 series on the AM5 (Ryzen) socket platform.
These CPUs feature 16 cores and 32 threads, with a base clock of 4.5 GHz and boost speeds up to 5.7 GHz. Like Ryzen built on the Zen 4 architecture, they offer high single-thread performance, making them suitable for workloads like web hosting.
While they share performance similarities with Ryzen 7000 series CPUs, the EPYC 4564P includes server-grade features such as ECC memory support.
We love them!
Ryzen is better, more Ryzen cores will beat more EPYC cores, but single server can have more EPYC cores, so if website has huge amount of visitors at once, going for more EPYC cores would make sense (but more Ryzen cores will be even better). Caching can help significantly.
Ryzen also supports ECC memory (provided you have the right motherboard). The only exceptions are the APUs like 8600G which have better integrated graphics but come at some other costs like lack of ECC support.
in summary
Ryzen FTW
Ryzen with debian thx...
in essence a good combination
epyc or ryzen, another limiting factor is sometimes wordpress itself. see below:
are there any wordpress specialists here who can tell me what the “wordpress option manipulation” is all about and how to solve this problem? thank you!
I have seen where the backend database is actually the problem, leftover entries from removed plugins specifically can really slow thing down.
The wordpress optimization docs are pretty solid for the web / php side:
https://developer.wordpress.org/advanced-administration/performance/optimization/
Or you might need to run some basic mysql table optimizations such as:
But for wp database optimizations I usually just leverage the functions in the wp-cli tool: https://wp-cli.org
Hope this helps.
sadly not. i have still "Wordpress option manipulation 0"
I may be wrong but isn't option manipulation is just wp playing with data in options table with option(s) function?
no idea. but in the wp-benchmark the “option manipulation” has a negative effect on the score (in my installation).
maybe you need to enable some required php extensions?
Why not all providers choose this beast!!
what I read earlier is PHP is based on single-thread, so if the single CPU has high GHz it will outperform multi-core CPU. So, for a VPS, with 1CPU with 4.7GHz, the wordpress will load faster here, than on 2 or 3CPU with 2.0 GHz in laymen terms. But beyond that the overall accounts on a node, how busy it is, the i/o, mysql connections, database caching, object and opcode caching, etc - everything will affect wordpress load times.
For those interested, recently moved my site to 2 core 9950x plan, and the Wordpress benchmark certainly is singing. My site gets no traffic so me looking to optimise it is purely hobby/academic exercise.
Here’s the YABS for the server
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Yet-Another-Bench-Script
v2025-04-20
https://github.com/masonr/yet-another-bench-script
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Fri May 2 09:09:18 PM AEST 2025
Basic System Information:
Uptime : 14 days, 19 hours, 5 minutes
Processor : AMD Ryzen 9 9950X 16-Core Processor
CPU cores : 2 @ 4291.884 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ✔ Enabled
RAM : 3.8 GiB
Swap : 4.0 GiB
Disk : 99.9 GiB
Distro : Debian GNU/Linux 12 (bookworm)
Kernel : 6.1.0-33-amd64
VM Type : KVM
IPv4/IPv6 : ✔ Online / ✔ Online
IPv6 Network Information:
ISP : WebNX, Inc.
ASN : AS18450 WebNX, Inc.
Host : WebNX, Inc.
Location : Los Angeles, California (CA)
Country : United States
fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/vda3):
iperf3 Network Speed Tests (IPv4):
iperf3 Network Speed Tests (IPv6):
Geekbench 6 Benchmark Test:
Test | Value
|
Single Core | 3194
Multi Core | 5596
Full Test | https://browser.geekbench.com/v6/cpu/11780823
YABS completed in 11 min 58 sec
Moved from this plan which itself is good. More Epyc cores and stacks of RAM, but again my traffic and site can’t take advantage of those resources. I don’t have a screenshot of the same WP benchmarks, but I had this at about 8.4 overall.
May I ask which supplier provides the YABS for the AMD Ryzen 9 9950X server?
Greencloud VPS. Picked it up in the provider sale they ran around Feb.
Looks interesting. Not accurate as Not physical cores
I need this but 8vcpu
Hello, creator of video here!
Summary of the video
I've researched what is the most commonly used EPYC among all providers (smaller and bigger) and that is EPYC Milan.
I've decided to compare cores from Ryzen 9 9950X (very common at LET) to EPYC Milan. My Wordpress benchmark showed almost 3x performance difference between these two, so that is massive, but that's just CPU performance, what about real-world browsing experience? What about core count scaling?
For that I've benchmarked these VPSes that are located in Germany from NY by using LoaderIO, to not overhype the differences by testing locally.
That 3x performance difference in benchmark shrunk to ~514ms vs ~345ms response time with one visitor per second, so now it's not even 2x faster on Ryzen, because how much time is spent on network and IO.
When I added more visitors per second the CPU limitation became an issue and it became clear that one core from Ryzen is as fast as two cores from EPYC.
I've added more and more requests per second, on 50rq/s even 4core Ryzen couldn't achieve acceptable result (1760ms, when 1000ms/one second is the limit of not having any timeouts after longer period of time).
Then I activated Varnish page cache, still on 50rq/s
Performance skyrocketed.
Now EPYC sends back response in ~114ms, Ryzen in ~94ms.
Yup, on 50x higher load (50rq vs 1rq) we get ~4x faster response, just because of caching.
Then I tested 500rq/s and that was the limit of 1vCPU EPYC, but 1vCPU Ryzen still did well and didn't slow down.
So page caching in Wordpress on 500x more load gives you 4x faster response on 1vCPU Ryzen. 2000x difference.
In conclusion
These things make it clear that if you need high performance VPS for Wordpress you want to get some Ryzen and use page cache, with just 1vCPU it can work with 5000rq/s (300k rq/minute!), but in the conclusion segment of the video I also added a table which compares performance of different generations of Zen architecture (EPYC/Ryzen generations), so you can better apply my results to offers that you see. Ryzen 9 9950X is still king, but Genoa and Turin are not that far away as Milan.
On my thumbnails you see my face and topic, in this example you have Wordpress logo.
This way you can recognize my video - I don't steal your time if you watched my video in the past and didn't like it and if you liked my video you know its worth your time.
In both cases you get better experience. There is no clickbait, no extreme emotions in my face, no oversaturated background, no red arrows.
I've created that benchmark plugin. Initially I wanted to share it, but I've decided against it as I'm not sure I can properly communicate how to use it. If user wants to compare two servers he needs to use same panels, same PHP/DB settings, very similar WP installation with same theme, same amount of data in database etc. and I'm worried that it will cause misunderstandings or faked results. So basically I spent couple of hours for couple of seconds of the single video haha
No. In my videos I frequently show offers of LET providers (look at earlier videos) and that's the main purpose of this channel - to show broader audience offers from smaller providers and opensource apps (earlier video was about Coolify).
This way viewer gets knowledge and better deals, while LEB/LET providers and opensource project get more recognition and customers. Absolute win-win situation.
Advin Servers was an example of Ryzen hosting, because I know he provides reliable service and is available in multiple regions.
Bero-host, the VPS hosting that I've used in this benchmark video is only in Germany and I wanted to emphasise that you can get Ryzens in any region, thus I'm showing offer from Advin.
When I'm testing some VPS or service then it's 60% that I already paid for it, 30% I'll pay for it just for video, 10% I'll request an trial for testing from the provider.
In case of this video:
in summary
Ryzen ftw
Mad respect for this reply!
Sorry for being direct, it is not about you or your content.
And yes, RYZEN FTW !!