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.
Understanding & Interpreting Benchmarks / YABS
bamboo4409
Member
in Help
Hope everyone had a nice New Years & Holiday. I was curious how to best understand the benchmarks that I often see here.
I tried running a few searches in the googles, but was only served ads and 10yo results.
For reference, I have two cloud providers. One is marketed as a premium provider that changes resources based on need, and the other is marketed as budget. When running YABS - I am surprised by how little difference there are in benchmarks. Or are they pretty different and I just don't understand?
What is acceptable disk speed? What should one look for between SSD/NVME/STD disks? What GeekBench values do you consider acceptable?
(also - I know everyone's time is important. don't expect a book or thorough response. but if anyone has a link or online resource that better explains this, happy to read)
"Premium" YABS
# ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## #
# Yet-Another-Bench-Script #
# v2022-12-04 #
# https://github.com/masonr/yet-another-bench-script #
# ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## #
Tue Jan 3 08:02:17 MST 2023
Basic System Information:
---------------------------------
Uptime : 15 days, 13 hours, 25 minutes
Processor : AMD EPYC Processor
CPU cores : 2 @ 3500.000 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ✔ Enabled
RAM : 15.6 GiB
Swap : 0.0 KiB
Disk : 256.0 TiB
Distro : Ubuntu 22.04.1 LTS
Kernel : 5.15.0-56-generic
fio Disk Speed Tests (Mixed R/W 50/50):
---------------------------------
Block Size | 4k (IOPS) | 64k (IOPS)
------ | --- ---- | ---- ----
Read | 240.62 MB/s (60.1k) | 1.51 GB/s (23.5k)
Write | 241.25 MB/s (60.3k) | 1.51 GB/s (23.7k)
Total | 481.88 MB/s (120.4k) | 3.02 GB/s (47.3k)
| |
Block Size | 512k (IOPS) | 1m (IOPS)
------ | --- ---- | ---- ----
Read | 1.52 GB/s (2.9k) | 1.47 GB/s (1.4k)
Write | 1.60 GB/s (3.1k) | 1.57 GB/s (1.5k)
Total | 3.12 GB/s (6.0k) | 3.04 GB/s (2.9k)
iperf3 Network Speed Tests (IPv4):
---------------------------------
Provider | Location (Link) | Send Speed | Recv Speed | Ping
----- | ----- | ---- | ---- | ----
Clouvider | London, UK (10G) | 480 Mbits/sec | 816 Mbits/sec | 132 ms
Scaleway | Paris, FR (10G) | 275 Mbits/sec | 868 Mbits/sec | 146 ms
NovoServe | North Holland, NL (40G) | 408 Mbits/sec | 1.19 Gbits/sec | 146 ms
Uztelecom | Tashkent, UZ (10G) | 471 Mbits/sec | 268 Mbits/sec | 241 ms
Clouvider | NYC, NY, US (10G) | 638 Mbits/sec | 1.50 Gbits/sec | 66.2 ms
Clouvider | Dallas, TX, US (10G) | 578 Mbits/sec | 574 Mbits/sec | 31.6 ms
Clouvider | Los Angeles, CA, US (10G) | 840 Mbits/sec | 915 Mbits/sec | 9.95 ms
iperf3 Network Speed Tests (IPv6):
---------------------------------
Provider | Location (Link) | Send Speed | Recv Speed | Ping
----- | ----- | ---- | ---- | ----
Clouvider | London, UK (10G) | 448 Mbits/sec | 232 Mbits/sec | 132 ms
Scaleway | Paris, FR (10G) | 503 Mbits/sec | 214 Mbits/sec | 149 ms
NovoServe | North Holland, NL (40G) | 580 Mbits/sec | 558 Mbits/sec | 146 ms
Uztelecom | Tashkent, UZ (10G) | 352 Mbits/sec | 203 Mbits/sec | 240 ms
Clouvider | NYC, NY, US (10G) | 651 Mbits/sec | 312 Mbits/sec | 65.6 ms
Clouvider | Dallas, TX, US (10G) | 605 Mbits/sec | 5.05 Gbits/sec | 28.4 ms
Clouvider | Los Angeles, CA, US (10G) | 842 Mbits/sec | 909 Mbits/sec | 9.49 ms
Geekbench 5 Benchmark Test:
---------------------------------
Test | Value
|
Single Core | 1223
Multi Core | 2339
Full Test | https://browser.geekbench.com/v5/cpu/19676930
YABS completed in 8 min 21 sec
"Budget" YABS
# ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## #
# Yet-Another-Bench-Script #
# v2022-12-04 #
# https://github.com/masonr/yet-another-bench-script #
# ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## #
Tue 03 Jan 2023 03:02:42 PM GMT
Basic System Information:
---------------------------------
Uptime : 0 days, 0 hours, 13 minutes
Processor : Intel(R) Xeon(R) CPU E5-2698 v4 @ 2.20GHz
CPU cores : 4 @ 2199.984 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ✔ Enabled
RAM : 7.4 GiB
Swap : 1024.0 MiB
Disk : 45.2 GiB
Distro : Ubuntu 20.04.5 LTS
Kernel : 5.4.0-135-generic
fio Disk Speed Tests (Mixed R/W 50/50):
---------------------------------
Block Size | 4k (IOPS) | 64k (IOPS)
------ | --- ---- | ---- ----
Read | 107.29 MB/s (26.8k) | 697.44 MB/s (10.8k)
Write | 107.57 MB/s (26.8k) | 701.11 MB/s (10.9k)
Total | 214.87 MB/s (53.7k) | 1.39 GB/s (21.8k)
| |
Block Size | 512k (IOPS) | 1m (IOPS)
------ | --- ---- | ---- ----
Read | 807.46 MB/s (1.5k) | 835.21 MB/s (815)
Write | 850.36 MB/s (1.6k) | 890.83 MB/s (869)
Total | 1.65 GB/s (3.2k) | 1.72 GB/s (1.6k)
iperf3 Network Speed Tests (IPv4):
---------------------------------
Provider | Location (Link) | Send Speed | Recv Speed | Ping
----- | ----- | ---- | ---- | ----
Clouvider | London, UK (10G) | 1.47 Gbits/sec | 1.08 Gbits/sec |
Scaleway | Paris, FR (10G) | 1.34 Gbits/sec | 1.27 Gbits/sec | 103 ms
NovoServe | North Holland, NL (40G) | 894 Mbits/sec | 1.65 Gbits/sec |
Uztelecom | Tashkent, UZ (10G) | 1.03 Gbits/sec | 594 Mbits/sec |
Clouvider | NYC, NY, US (10G) | 2.99 Gbits/sec | 2.76 Gbits/sec |
Clouvider | Dallas, TX, US (10G) | 2.99 Gbits/sec | 4.85 Gbits/sec |
Clouvider | Los Angeles, CA, US (10G) | 1.68 Gbits/sec | 1.76 Gbits/sec |
Geekbench 5 Benchmark Test:
---------------------------------
Test | Value
|
Single Core | 389
Multi Core | 1336
Full Test | https://browser.geekbench.com/v5/cpu/19676949
YABS completed in 8 min 46 sec
Comments
H u h?
YABS is just a tool showing
current(and temporary!) state of machine and everyone need different things - you seems to be happy with 389 score on single core on CPU (sloooooooow) - I guess you don't have things needing CPU power running in single thread - maybe your things are multicore so you benefit more from having more slower cores rather than one fast. In theory your Premium machine have single core faster than all other cores together on the Budget machine (simplification), but if you don't utilize those resources you won't notice a thing. Other people could need faster single thread operations rather than multicore (single threaded games, shitty PHP websites etc). Depends on usage.Disk speeds depend from usage too - if all you do for example is proxying stuff on network level and you never hit the disk - you won't notice difference except booting time / updating packages (bah, you can even use slow HDD for that!). Most people won't notice difference in disk speed between your benchmarks - it's far enough for any "average" use - but if you need heavy disk-write related database operation you could see it.
Golden rule is: if you are happy with current state then you are happy and you don't pay attention to those pesky numbers :-)
the network is better on the budget YABS .. on how fast the info is transferet to the end user doesn`t count? Can we consider that the budget VPS has a premium network and the premium VPS has a budget network?
256TiB disk? Seriously?
It's probably some sort of shared environment/instance
@bamboo4409
Are you running into any service degradation on either of your servers?
If yes, it will be nice to understand what's the challenge and we can assist better
If no, and you are simply looking at yabs to prove you have a shiny medal...
It all boils down to what you plan to do with the vps at the end of the day.
If you had searched the let forum you would have seen other threads asking the same and not getting anywhere because the poster was clueless and probably just liked getting higher numbers because they paid for it..
Plus most of the services come under shared resources and the providers allow some burst usage. So at the time your Benchmark is running and if another user is running some intensive tasks then your numbers are not going to be indicative of the true meaning.
Hope that helps
I appreciate everyone's response and time, thank you!
Yeah. This Premium Hosting provider allocates space on how much you use. For each 10 GB of storage, they charge $0.10/day. So they see it as you pay only for the resources you use.
I did not intend for this thread to be a pissing contest. To be honest, the main reason of asking is based off a poor experience I had in the past.
Back in the day, I had an account with CloudAtCost.com - and regardless of how many "vCPU" I purchased, the service was so slow. Just running apt update/upgrade took ages. So now I am trying to avoid that same mistake.
I am indifferent at this point. Ideally I'd like to host a few sites, maybe run a matrix server. Or set up Plex.
I'm assuming a single thread benchmark should be 1k+? Many seem to prefer Xeon since it is my understanding that is more a server-grade processor. But I wonder if it really makes that large of a difference, or maybe its just over-sold?
They charge how much? That's $3 per month for 10GB, you can easily get 1TB for that.
It's probably NVMe, not HDD (so you can't really find $3/mo for 1TB NVMe)
@Advin - it is advertised as NVMe. But do you think it is safe to assume that is true based off the benchmarks?
Definitely oversold on cpu performance. But for the price, it’s worth it. Epyc is also an enterprise-grade processor series that belongs to AMD, I’d like to say for these years, AMD got lots of improvements especially on server-side. Unless it’s Xeon 3rd gen scalable processors, I’d like to choose Epyc instead. If it’s dedicated core, single thread performance of 2698v4 should be around 750 or higher.