Ran Yabs on my late 2012 Macbook Pro -
Clean install of Manjaro on OEM 128 GB disk. While I wait for the adapter to arrive that will help me upgrade to a 512 GB SSD.
Question: Why the Zeros for disk speed?
I used the command
curl -sL yabs.sh | bash -s -- -r9
bench.monster gave the following:
Disk Speed:
1st run : 2.9 GB/s
2nd run : 2.7 GB/s
3rd run : 2.9 GB/s
Thu Jul 29 17:26:55 IST 2021
Usage: ./yabs.sh [-fdighr49]
curl -sL yabs.sh | bash
curl -sL yabs.sh | bash -s -- -{fdighr49}
Flags:
-f/d : skips the fio disk benchmark test
-i : skips the iperf network test
-g : skips the geekbench performance test
-h : prints this lovely message, shows any flags you passed,
shows if fio/iperf3 local packages have been detected,
then exits
-r : reduce number of iperf3 network locations (to only three)
to lessen bandwidth usage
-4 : use geekbench 4 instead of geekbench 5
-9 : use both geekbench 4 AND geekbench 5
Detected Arch: x64
Detected Flags:
-i, skipping iperf network test
-g, skipping geekbench test
running geekbench 5
Local Binary Check:
fio not detected, will download precompiled binary
iperf3 not detected, will download precompiled binary
Detected Connectivity:
IPv4 connected
IPv6 not connected
Exiting...
Yikes... yeah don't think it's gonna fare well on something that old. Lots of bash-related syntax errors there. Maybe I could enforce some sort of minimum bash version or something rather than spewing out junk output.
@vyas not too sure about that one. I'd do what @LetMeBeSad suggested and install fio natively to see if it runs that way (the script will detect the locally installed version and use that instead). If that fails too, then I'd know it's not an issue with the binary that I compiled
@Not_Oles said: When running qemu on Hetzner AX101, how much difference does -cpu host and -enable-kvm make?
Nice trial and results! Very interesting
This is why I always request for CPU passthrough on any newly purchased VMs if it's not enabled by default... Those AES-NI instructions are really essential for some operations. Plus... the less time a client's VM spends processing instructions, the less impact to the load on a host node, so it's a win-win.
⭕ A simple uptime dashboard using UptimeRobot API https://upy.duo.ovh
⭕ Currently using VPS from BuyVM, GreenCloudVPS, Gullo's, Hetzner, HostHatch, InceptionHosting, LetBox, MaxKVM, MrVM, VirMach.
@Not_Oles said: When running qemu on Hetzner AX101, how much difference does -cpu host and -enable-kvm make?
Nice trial and results! Very interesting
This is why I always request for CPU passthrough on any newly purchased VMs if it's not enabled by default... Those AES-NI instructions are really essential for some operations. Plus... the less time a client's VM spends processing instructions, the less impact to the load on a host node, so it's a win-win.
would be interesting to also see some middle ground, aka qemu-kvm or similar but with AES-NI enabled (which should be possible with additional settings)... as you said I guess that option alone will make a very huge difference.
@Not_Oles any chance you can add this too, pretty please ;-)
⭕ A simple uptime dashboard using UptimeRobot API https://upy.duo.ovh
⭕ Currently using VPS from BuyVM, GreenCloudVPS, Gullo's, Hetzner, HostHatch, InceptionHosting, LetBox, MaxKVM, MrVM, VirMach.
@Falzo said: would be interesting to also see some middle ground, aka qemu-kvm or similar but with AES-NI enabled (which should be possible with additional settings)... as you said I guess that option alone will make a very huge difference.
@Not_Oles any chance you can add this too, pretty please ;-)
Here are results with qemu's default processor (as in the first test, above) but now flying the -enable-kvm flag. As might be expected, this third test gives an intermediate result, 1308:
KVM Status
AES Status
GB 5 Single Core Score
Off
Off
110
On
Off
1308
On
On
1524
For anybody trying to understand how KVM could be "off" in test 1, given that the processor model name -- "Common KVM processor" -- literally includes "KVM," please consider that possessing a capability is not the same as actually using the capability. It seems that calling qemu as qemu-kvm-x86_64 requires that root explicitly enable KVM with -enable-kvm in order for KVM to be used. However, when qemu is called simply as kvm then KVM is used automagically if the caller is root but not used if the caller is unprivileged.
Note that there are additional changes beyond just AES status when switching from the default "Common KVM processor" to -cpu host which is AMD Ryzen 9 5950X 16-Core Processor.
root@fsn1 ~ # qemu-system-x86_64 -enable-kvm -m 8192m -nographic debian-sid-nocloud-amd64-daily.qcow2
root@debian:~# curl -sL yabs.sh | bash
# ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## #
# Yet-Another-Bench-Script #
# v2021-06-05 #
# https://github.com/masonr/yet-another-bench-script #
# ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## #
Sat Jul 31 21:39:44 UTC 2021
qemu-system-x86_64: Slirp: external icmpv6 not supported yet
Basic System Information:
---------------------------------
Processor : Common KVM processor
CPU cores : 1 @ 3393.622 MHz
AES-NI : ❌ Disabled
VM-x/AMD-V : ❌ Disabled
RAM : 7.8 GiB
Swap : 0.0 KiB
Disk : 1.9 GiB
Less than 2GB of space available. Skipping disk test...
iperf3 Network Speed Tests (IPv4):
---------------------------------
Provider | Location (Link) | Send Speed | Recv Speed
| | |
Clouvider | London, UK (10G) | 931 Mbits/sec | 923 Mbits/sec
Online.net | Paris, FR (10G) | 936 Mbits/sec | 937 Mbits/sec
WorldStream | The Netherlands (10G) | 935 Mbits/sec | 936 Mbits/sec
Biznet | Jakarta, Indonesia (1G) | 761 Mbits/sec | 269 Mbits/sec
Clouvider | NYC, NY, US (10G) | 885 Mbits/sec | 395 Mbits/sec
Velocity Online | Tallahassee, FL, US (10G) | 728 Mbits/sec | 476 Mbits/sec
Clouvider | Los Angeles, CA, US (10G) | 843 Mbits/sec | 293 Mbits/sec
Iveloz Telecom | Sao Paulo, BR (2G) | 767 Mbits/sec | 292 Mbits/sec
iperf3 Network Speed Tests (IPv6):
---------------------------------
Provider | Location (Link) | Send Speed | Recv Speed
| | |
Clouvider | London, UK (10G) | 919 Mbits/sec | 910 Mbits/sec
Online.net | Paris, FR (10G) | 924 Mbits/sec | 924 Mbits/sec
WorldStream | The Netherlands (10G) | 921 Mbits/sec | 918 Mbits/sec
Clouvider | NYC, NY, US (10G) | 873 Mbits/sec | 336 Mbits/sec
Clouvider | Los Angeles, CA, US (10G) | 829 Mbits/sec | 332 Mbits/sec
Running GB5 benchmark test... *cue elevator music*
Geekbench 5 Benchmark Test:
---------------------------------
Test | Value
|
Single Core | 1308
Multi Core | 1308
Full Test | https://browser.geekbench.com/v5/cpu/9101589
root@debian:~#
@Not_Oles thanks, interesting indeed, esp. because the gap is much smaller with that setting and AES still is disabled. I'd guess KVM+AES then comes really close to full host passthru
Just a mini-update towards the Alwyzon Performance Server line: instead of Xeon® E-2174G (no longer sold), you might now get a Xeon® E-2244G (newer model) delivered instead. But no worries, those are at least as beefy as the E-2174G. 🥳
Comments
Tuned a little bit 4k test
Spaceberg.cc - Your favorite Seedbox provider!
10/2 Gbit/s uplink. Network is awesome all around the world.
Spaceberg.cc - Your favorite Seedbox provider!
wow which provider?!
I bench YABS 24/7/365 unless it's a leap year.
@cybertech OVH.
Spaceberg.cc - Your favorite Seedbox provider!
DO droplet, got it free a little while ago from this thread
juicy network
YABS doesn't run on a Cobalt RaQ4r
I tried using
bash2and got a bit further. Commented out a section in yabs and got:I think basically it's all just too old lol.
Ran Yabs on my late 2012 Macbook Pro -
Clean install of Manjaro on OEM 128 GB disk. While I wait for the adapter to arrive that will help me upgrade to a 512 GB SSD.
Question: Why the Zeros for disk speed?
I used the command
bench.monster gave the following:
Disk Speed:
1st run : 2.9 GB/s
2nd run : 2.7 GB/s
3rd run : 2.9 GB/s
Average : 2901.3 MB/s
Thu Jul 29 17:08:32 IST 2021 Basic System Information: --------------------------------- Processor : Intel(R) Core(TM) i5-3210M CPU @ 2.50GHz CPU cores : 4 @ 1759.019 MHz AES-NI : ✔ Enabled VM-x/AMD-V : ✔ Enabled RAM : 7.7 GiB Swap : 8.8 GiB Disk : 101.8 GiB fio Disk Speed Tests (Mixed R/W 50/50): --------------------------------- Block Size | 4k (IOPS) | 64k (IOPS) ------ | --- ---- | ---- ---- Read | 0.00 KB/s (0) | 0.00 KB/s (0) Write | 0.00 KB/s (0) | 0.00 KB/s (0) Total | 0.00 KB/s (0) | 0.00 KB/s (0) | | Block Size | 512k (IOPS) | 1m (IOPS) ------ | --- ---- | ---- ---- Read | 0.00 KB/s (0) | 0.00 KB/s (0) Write | 0.00 KB/s (0) | 0.00 KB/s (0) Total | 0.00 KB/s (0) | 0.00 KB/s (0) iperf3 Network Speed Tests (IPv4): --------------------------------- Provider | Location (Link) | Send Speed | Recv Speed | | | Clouvider | London, UK (10G) | 7.96 Mbits/sec | 9.81 Mbits/sec Online.net | Paris, FR (10G) | busy | 8.53 Mbits/sec Clouvider | NYC, NY, US (10G) | 7.25 Mbits/sec | 5.83 Mbits/sec Geekbench 4 Benchmark Test: --------------------------------- Test | Value | Single Core | 3359 Multi Core | 6253 Full Test | https://browser.geekbench.com/v4/cpu/16283053 Geekbench 5 Benchmark Test: --------------------------------- Test | Value | Single Core | 646 Multi Core | 1201 Full Test | https://browser.geekbench.com/v5/cpu/9060975Output from the 'h' flag below
Thu Jul 29 17:26:55 IST 2021 Usage: ./yabs.sh [-fdighr49] curl -sL yabs.sh | bash curl -sL yabs.sh | bash -s -- -{fdighr49} Flags: -f/d : skips the fio disk benchmark test -i : skips the iperf network test -g : skips the geekbench performance test -h : prints this lovely message, shows any flags you passed, shows if fio/iperf3 local packages have been detected, then exits -r : reduce number of iperf3 network locations (to only three) to lessen bandwidth usage -4 : use geekbench 4 instead of geekbench 5 -9 : use both geekbench 4 AND geekbench 5 Detected Arch: x64 Detected Flags: -i, skipping iperf network test -g, skipping geekbench test running geekbench 5 Local Binary Check: fio not detected, will download precompiled binary iperf3 not detected, will download precompiled binary Detected Connectivity: IPv4 connected IPv6 not connected Exiting...Also,
gave
fio Disk Speed Tests (Mixed R/W 50/50): --------------------------------- Block Size | 4k (IOPS) | 64k (IOPS) ------ | --- ---- | ---- ---- Read | 0.00 KB/s (0) | 0.00 KB/s (0) Write | 0.00 KB/s (0) | 0.00 KB/s (0) Total | 0.00 KB/s (0) | 0.00 KB/s (0) | | Block Size | 512k (IOPS) | 1m (IOPS) ------ | --- ---- | ---- ---- Read | 0.00 KB/s (0) | 0.00 KB/s (0) Write | 0.00 KB/s (0) | 0.00 KB/s (0) Total | 0.00 KB/s (0) | 0.00 KB/s (0)———-
blog | exploring visually |
@vyas Install fio manually via brew.
Spaceberg.cc - Your favorite Seedbox provider!
Thanks- forgot to mention I run Manjaro Linux. YABS was on Manjaro
———-
blog | exploring visually |
Then it should work, at least as i know Manjaro is based on Arch. Maybe there is some misconfiguration for Arch(Manjaro) tho?
Spaceberg.cc - Your favorite Seedbox provider!
Yikes... yeah don't think it's gonna fare well on something that old. Lots of bash-related syntax errors there. Maybe I could enforce some sort of minimum bash version or something rather than spewing out junk output.
@vyas not too sure about that one. I'd do what @LetMeBeSad suggested and install fio natively to see if it runs that way (the script will detect the locally installed version and use that instead). If that fails too, then I'd know it's not an issue with the binary that I compiled
Head Janitor @ LES • About • Rules • Support
Thanks @LetMeBeSad and @Mason
installing fio worked.
Not bad for a 10 year old OEM disk. (Samsung SSD for MBP12)*
fio Disk Speed Tests (Mixed R/W 50/50): --------------------------------- Block Size | 4k (IOPS) | 64k (IOPS) ------ | --- ---- | ---- ---- Read | 62.51 MB/s (15.6k) | 64.78 MB/s (1.0k) Write | 62.62 MB/s (15.6k) | 65.18 MB/s (1.0k) Total | 125.13 MB/s (31.2k) | 129.96 MB/s (2.0k) | | Block Size | 512k (IOPS) | 1m (IOPS) ------ | --- ---- | ---- ---- Read | 108.74 MB/s (212) | 116.51 MB/s (113) Write | 114.52 MB/s (223) | 124.27 MB/s (121) Total | 223.26 MB/s (435) | 240.78 MB/s (234)*But allows me to build a case to upgrade the drive. Not just for storage but also speed.
———-
blog | exploring visually |
Interesting. So my prebuilt fio has some issues with Manjaro (and possibly Arch?) that I'll have to investigate in the future.
Glad the fio install worked
Head Janitor @ LES • About • Rules • Support
HostHatch 250GB big plate chicken in Chicago
SSH connection interrupted, after re-login I see
geekbench_claim.urlwith link to result:https://browser.geekbench.com/v5/cpu/9086575
HostBrr aff best VPS; VirmAche aff worst VPS.
Unable to push-up due to shoulder injury 😣
hosthatch
hongkong 香港
1 vCPU 2690 V2
2GB ram 内存
20GB 硬盘 Hard Disk NVMe RAID-10
1gbps bandwidth
1TB monthly transfer
Above only applies if paid for 2 years (USD $15 / year)
I bench YABS 24/7/365 unless it's a leap year.
When running qemu on Hetzner AX101, how much difference does -cpu host and -enable-kvm make?
Note -- using -cpu host without -enable-kvm gives an error:
First, without -cpu host and without -enable-kvm
Second, with -cpu host and -enable-kvm
I hope everyone gets the servers they want!
Nice trial and results! Very interesting
This is why I always request for CPU passthrough on any newly purchased VMs if it's not enabled by default... Those AES-NI instructions are really essential for some operations. Plus... the less time a client's VM spends processing instructions, the less impact to the load on a host node, so it's a win-win.
Head Janitor @ LES • About • Rules • Support
I run yabs inside a screen, so I can leave it anytime without worry..
⭕ A simple uptime dashboard using UptimeRobot API https://upy.duo.ovh
⭕ Currently using VPS from BuyVM, GreenCloudVPS, Gullo's, Hetzner, HostHatch, InceptionHosting, LetBox, MaxKVM, MrVM, VirMach.
would be interesting to also see some middle ground, aka qemu-kvm or similar but with AES-NI enabled (which should be possible with additional settings)... as you said I guess that option alone will make a very huge difference.
@Not_Oles any chance you can add this too, pretty please ;-)
Huge Poland KVM (WebHorizon)
Free NAT KVM | Free NAT LXC | Bobr
ITS WEDNESDAY MY DUDES
That is amazing VPSes you have, guys..
⭕ A simple uptime dashboard using UptimeRobot API https://upy.duo.ovh
⭕ Currently using VPS from BuyVM, GreenCloudVPS, Gullo's, Hetzner, HostHatch, InceptionHosting, LetBox, MaxKVM, MrVM, VirMach.
terrahost dedi
Hi @Falzo!
Here are results with qemu's default processor (as in the first test, above) but now flying the -enable-kvm flag. As might be expected, this third test gives an intermediate result, 1308:
For anybody trying to understand how KVM could be "off" in test 1, given that the processor model name -- "Common KVM processor" -- literally includes "KVM," please consider that possessing a capability is not the same as actually using the capability. It seems that calling qemu as
qemu-kvm-x86_64requires that root explicitly enable KVM with-enable-kvmin order for KVM to be used. However, when qemu is called simply askvmthen KVM is used automagically if the caller is root but not used if the caller is unprivileged.Note that there are additional changes beyond just AES status when switching from the default "Common KVM processor" to
-cpu hostwhich is AMD Ryzen 9 5950X 16-Core Processor.I hope everyone gets the servers they want!
Hey @Mason! Thanks for the kind words! Best wishes from the desert! 👨💻🏜️
I hope everyone gets the servers they want!
@Not_Oles thanks, interesting indeed, esp. because the gap is much smaller with that setting and AES still is disabled. I'd guess KVM+AES then comes really close to full host passthru
Just a mini-update towards the Alwyzon Performance Server line: instead of Xeon® E-2174G (no longer sold), you might now get a Xeon® E-2244G (newer model) delivered instead. But no worries, those are at least as beefy as the E-2174G. 🥳
— Michael
Alwyzon - Virtual Servers in Austria starting at 4,49 €/month (excl. VAT)
Picked up a few UK VMs to benchmark as I'm looking for a new home for a few smaller projects.
VM Specialist
OVHcloud
Nexus Bytes
Nice performance throughout, but will probably hold onto the VM Specialist one - kudos @Mr_Tom
NexusBytes VPS-6G Ryzen
Basic System Information: --------------------------------- Processor : AMD Ryzen 9 3900X 12-Core Processor CPU cores : 6 @ 3792.872 MHz AES-NI : ✔ Enabled VM-x/AMD-V : ✔ Enabled RAM : 5.8 GiB Swap : 0.0 KiB Disk : 88.5 GiB fio Disk Speed Tests (Mixed R/W 50/50): --------------------------------- Block Size | 4k (IOPS) | 64k (IOPS) ------ | --- ---- | ---- ---- Read | 352.17 MB/s (88.0k) | 3.71 GB/s (58.0k) Write | 353.10 MB/s (88.2k) | 3.73 GB/s (58.3k) Total | 705.28 MB/s (176.3k) | 7.44 GB/s (116.4k) | | Block Size | 512k (IOPS) | 1m (IOPS) ------ | --- ---- | ---- ---- Read | 4.33 GB/s (8.4k) | 4.21 GB/s (4.1k) Write | 4.56 GB/s (8.9k) | 4.50 GB/s (4.3k) Total | 8.90 GB/s (17.3k) | 8.71 GB/s (8.5k) iperf3 Network Speed Tests (IPv4): --------------------------------- Provider | Location (Link) | Send Speed | Recv Speed | | | Clouvider | London, UK (10G) | 318 Mbits/sec | 249 Mbits/sec Online.net | Paris, FR (10G) | 235 Mbits/sec | 755 Mbits/sec WorldStream | The Netherlands (10G) | 139 Mbits/sec | 419 Mbits/sec Biznet | Jakarta, Indonesia (1G) | 635 Mbits/sec | 452 Mbits/sec Clouvider | NYC, NY, US (10G) | 431 Mbits/sec | 301 Mbits/sec Velocity Online | Tallahassee, FL, US (10G) | 16.4 Mbits/sec | 57.8 Mbits/sec Clouvider | Los Angeles, CA, US (10G) | 624 Mbits/sec | 411 Mbits/sec Iveloz Telecom | Sao Paulo, BR (2G) | busy | 614 Kbits/sec Geekbench 5 Benchmark Test: --------------------------------- Test | Value | Single Core | 1333 Multi Core | 6131 Full Test | https://browser.geekbench.com/v5/cpu/9134126Entrybytes Entry-8G
UK node
Basic System Information: --------------------------------- Processor : Intel(R) Xeon(R) CPU E5-2630L v2 @ 2.40GHz CPU cores : 4 @ 2399.996 MHz AES-NI : ✔ Enabled VM-x/AMD-V : ✔ Enabled RAM : 7.8 GiB Swap : 0.0 KiB Disk : 78.7 GiB fio Disk Speed Tests (Mixed R/W 50/50): --------------------------------- Block Size | 4k (IOPS) | 64k (IOPS) ------ | --- ---- | ---- ---- Read | 171.05 MB/s (42.7k) | 1.87 GB/s (29.2k) Write | 171.50 MB/s (42.8k) | 1.88 GB/s (29.4k) Total | 342.55 MB/s (85.6k) | 3.75 GB/s (58.7k) | | Block Size | 512k (IOPS) | 1m (IOPS) ------ | --- ---- | ---- ---- Read | 4.16 GB/s (8.1k) | 5.54 GB/s (5.4k) Write | 4.38 GB/s (8.5k) | 5.91 GB/s (5.7k) Total | 8.54 GB/s (16.6k) | 11.45 GB/s (11.1k) iperf3 Network Speed Tests (IPv4): --------------------------------- Provider | Location (Link) | Send Speed | Recv Speed | | | Clouvider | London, UK (10G) | 947 Mbits/sec | 933 Mbits/sec Online.net | Paris, FR (10G) | 939 Mbits/sec | 930 Mbits/sec WorldStream | The Netherlands (10G) | busy | busy Biznet | Jakarta, Indonesia (1G) | 401 Mbits/sec | 78.4 Mbits/sec Clouvider | NYC, NY, US (10G) | 759 Mbits/sec | 404 Mbits/sec Velocity Online | Tallahassee, FL, US (10G) | 25.0 Mbits/sec | 100 Mbits/sec Clouvider | Los Angeles, CA, US (10G) | 349 Mbits/sec | 180 Mbits/sec Iveloz Telecom | Sao Paulo, BR (2G) | 180 Mbits/sec | 205 Mbits/sec Geekbench 5 Benchmark Test: --------------------------------- Test | Value | Single Core | 484 Multi Core | 1825 Full Test | https://browser.geekbench.com/v5/cpu/9134169NL node
Basic System Information: --------------------------------- Processor : Intel(R) Xeon(R) CPU E5-2620 v4 @ 2.10GHz CPU cores : 4 @ 2095.162 MHz AES-NI : ✔ Enabled VM-x/AMD-V : ✔ Enabled RAM : 7.8 GiB Swap : 0.0 KiB Disk : 78.7 GiB fio Disk Speed Tests (Mixed R/W 50/50): --------------------------------- Block Size | 4k (IOPS) | 64k (IOPS) ------ | --- ---- | ---- ---- Read | 141.89 MB/s (35.4k) | 1.85 GB/s (29.0k) Write | 142.27 MB/s (35.5k) | 1.86 GB/s (29.1k) Total | 284.16 MB/s (71.0k) | 3.72 GB/s (58.2k) | | Block Size | 512k (IOPS) | 1m (IOPS) ------ | --- ---- | ---- ---- Read | 4.17 GB/s (8.1k) | 4.30 GB/s (4.2k) Write | 4.39 GB/s (8.5k) | 4.59 GB/s (4.4k) Total | 8.57 GB/s (16.7k) | 8.90 GB/s (8.6k) iperf3 Network Speed Tests (IPv4): --------------------------------- Provider | Location (Link) | Send Speed | Recv Speed | | | Clouvider | London, UK (10G) | 971 Mbits/sec | 949 Mbits/sec Online.net | Paris, FR (10G) | 981 Mbits/sec | 948 Mbits/sec WorldStream | The Netherlands (10G) | 979 Mbits/sec | 952 Mbits/sec Biznet | Jakarta, Indonesia (1G) | 272 Mbits/sec | 63.0 Mbits/sec Clouvider | NYC, NY, US (10G) | 429 Mbits/sec | 885 Mbits/sec Velocity Online | Tallahassee, FL, US (10G) | 24.0 Mbits/sec | 234 Mbits/sec Clouvider | Los Angeles, CA, US (10G) | 308 Mbits/sec | 505 Mbits/sec Iveloz Telecom | Sao Paulo, BR (2G) | 229 Mbits/sec | 246 Mbits/sec Geekbench 5 Benchmark Test: --------------------------------- Test | Value | Single Core | 551 Multi Core | 2016 Full Test | https://browser.geekbench.com/v5/cpu/9134170