@Neoon said:
A wild KS-LE in its natural habitat in RBX5
Lol looks like someone at OVH goofed up the bandwidth limits. Limited to 100 mbps outgoing on IPv4, but still get full gig over IPv6. Switch to v6 preferred and that's a sweet rig. The 100 mbit cap is what turned me off originally.
@Neoon said:
A wild KS-LE in its natural habitat in RBX5
Lol looks like someone at OVH goofed up the bandwidth limits. Limited to 100 mbps outgoing on IPv4, but still get full gig over IPv6. Switch to v6 preferred and that's a sweet rig. The 100 mbit cap is what turned me off originally.
Its all mixed up, one got 2x4TB all 100Mbit the other one 2x2TB with 1Gbit on IPv6.
If you are unlucky you stuck with 100Mbit on all links and 2x2TB.
Also notice the benchmark difference, no idea yet.
Likely cooling.
@Mason and others, can you please explain something to me? m_ _m
I am probably missing some crucial information on YABS and/or hardware, but on first sight, it feels like the math does not add up in the case of some dedicated servers... (like in the last 5 or so YABS results posted here) - namely, it feels like the given core count is either doubled, or the multi core GB5 result is halved...
Is my understanding correct in that
These 'cores' @ YABS are actually CPU threads
But this is the number of "CPUs" seen by the OS (using lscpu or in some other way)
Meanwhile, Geekbench uses CPU cores and not 'cores' (=threads)
@chimichurri said: @Mason and others, can you please explain something to me? m_ _m
I am probably missing some crucial information on YABS and/or hardware, but on first sight, it feels like the math does not add up in the case of some dedicated servers... (like in the last 5 or so YABS results posted here) - namely, it feels like the given core count is either doubled, or the multi core GB5 result is halved...
Is my understanding correct in that
These 'cores' @ YABS are actually CPU threads
But this is the number of "CPUs" seen by the OS (using lscpu or in some other way)
Meanwhile, Geekbench uses CPU cores and not 'cores' (=threads)
Or perhaps does the reason for this lie somewhere completely else? orz
There 'cores' @ YABS are the logical cores seen by the operating system.
Geekbench uses all those logical cores in multicore benchmark.
The point is that the number of logistic cores or threads doesn't determine the multicore performance.
The number of physical cores determines that.
Some CPUs have HT/SMT enabled, where makes the number of logical cores equal to two times number of physical cores.
(The HT/SMT in fact, can be disabled by motherboard settings)
So basically, considering a host node with 12 physical cores, and with HT/SMT enabled,
you will see 24 logical cores.
If you run a multithread program with 4 threads, you will see it is about 4 times faster because 12>4, i.e. you have enough physical cores,
but if you run a multithead program with 16 thread, you will see it is about 12 times fater, because 16>12, i.e. you only have 12 physical cores.
Then considering a 4-core (shared cores) VPS on such host node, which means such VPS can only have 4 threads equivalently.
Since 12>4, you will see the performance is 4 times of single core, as the host node have enough physical cores.
Geekbench 5 Benchmark Test:
---------------------------------
Test | Value
|
Single Core | 73
Multi Core | 123
Full Test | https://browser.geekbench.com/v5/cpu/11388337
.
Geekbench 5 Benchmark Test:
---------------------------------
Test | Value
|
Single Core | 84
Multi Core | 157
Full Test | https://browser.geekbench.com/v5/cpu/11388065
This is crazy, slower than the snail owned by cuciu shitserver.
⭕ 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.
Geekbench 5 Benchmark Test:
---------------------------------
Test | Value
|
Single Core | 73
Multi Core | 123
Full Test | https://browser.geekbench.com/v5/cpu/11388337
.
Geekbench 5 Benchmark Test:
---------------------------------
Test | Value
|
Single Core | 84
Multi Core | 157
Full Test | https://browser.geekbench.com/v5/cpu/11388065
This is crazy, slower than the snail owned by cuciu shitserver.
Geekbench 5 Benchmark Test:
---------------------------------
Test | Value
|
Single Core | 73
Multi Core | 123
Full Test | https://browser.geekbench.com/v5/cpu/11388337
.
Geekbench 5 Benchmark Test:
---------------------------------
Test | Value
|
Single Core | 84
Multi Core | 157
Full Test | https://browser.geekbench.com/v5/cpu/11388065
This is crazy, slower than the snail owned by cuciu shitserver.
Looks like it may be intentional, considering it's 2 digit in any location I choose. They might want you to use this for FTP/NFS/etc storage, not so much for seedboxes and such.
⭕ 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.
Comments
Spaceberg.cc - Your favorite Seedbox provider!
really wanted to and did so for 2 years, finally replacing it with Ryzen in HK haha.
wallets getting weaker..
I bench YABS 24/7/365 unless it's a leap year.
I have that too, I'm not going to renew it so I can transfer it
Due: 17.12.2021
17,65 EUR / 6m
NVMe Gen 4.0 goodness
@vyas
I bench YABS 24/7/365 unless it's a leap year.
@Mason my NVMe is 600K IOPS on paper, any insight why fio shows much less?
I bench YABS 24/7/365 unless it's a leap year.
@cybertech Try this
Spaceberg.cc - Your favorite Seedbox provider!
56Special Sing:
5 fair-share cores Xeon Gold @ 2.6GHz+
6GB RAM
56GB NVMe RAID-10
5600GB BW @ 10Gbps port
KVM @ SolusVM
Singapore DC2
$56/year
Here is yabs:
https://dttech.top - Personal Site.
ShockHosting free dedi 3 months
https://dttech.top - Personal Site.
A wild KS-LE in its natural habitat in RBX5
Free NAT KVM | Free NAT LXC | Bobr
A wild KS-LE in its natural habitat in Gravelands
Free NAT KVM | Free NAT LXC | Bobr
Lol looks like someone at OVH goofed up the bandwidth limits. Limited to 100 mbps outgoing on IPv4, but still get full gig over IPv6. Switch to v6 preferred and that's a sweet rig. The 100 mbit cap is what turned me off originally.
Head Janitor @ LES • About • Rules • Support
Its all mixed up, one got 2x4TB all 100Mbit the other one 2x2TB with 1Gbit on IPv6.
If you are unlucky you stuck with 100Mbit on all links and 2x2TB.
Also notice the benchmark difference, no idea yet.
Likely cooling.
Free NAT KVM | Free NAT LXC | Bobr
@Mason and others, can you please explain something to me? m_ _m
I am probably missing some crucial information on YABS and/or hardware, but on first sight, it feels like the math does not add up in the case of some dedicated servers... (like in the last 5 or so YABS results posted here) - namely, it feels like the given core count is either doubled, or the multi core GB5 result is halved...
Is my understanding correct in that
...and if the above reasoning is correct, then why is the math OK [i.e. core count * GB5 ingle score = (more or less) GB5 multi score] in case of VPS results? ><"
Examples:
https://talk.lowendspirit.com/discussion/comment/77160/#Comment_77160
https://talk.lowendspirit.com/discussion/comment/76781/#Comment_76781
AFAIK both of the these are modern high end CPUs whose thread count is 2x their core count, contrary to let's say https://browser.geekbench.com/processors/1824 where core count = thread count
Or perhaps does the reason for this lie somewhere completely else? orz
Contribute your idling VPS/dedi (link), Android (link) or iOS (link) devices to medical research
There 'cores' @ YABS are the logical cores seen by the operating system.
Geekbench uses all those logical cores in multicore benchmark.
The point is that the number of logistic cores or threads doesn't determine the multicore performance.
The number of physical cores determines that.
Some CPUs have HT/SMT enabled, where makes the number of logical cores equal to two times number of physical cores.
(The HT/SMT in fact, can be disabled by motherboard settings)
So basically, considering a host node with 12 physical cores, and with HT/SMT enabled,
you will see 24 logical cores.
If you run a multithread program with 4 threads, you will see it is about 4 times faster because 12>4, i.e. you have enough physical cores,
but if you run a multithead program with 16 thread, you will see it is about 12 times fater, because 16>12, i.e. you only have 12 physical cores.
Then considering a 4-core (shared cores) VPS on such host node, which means such VPS can only have 4 threads equivalently.
Since 12>4, you will see the performance is 4 times of single core, as the host node have enough physical cores.
Contabo BF Promotion USA East (Top Plan)
Price was around 27 euro per month. Price is great though performance is a bit mediocre.
I am a representative of Advin Servers
OneProvider Lisbon 2TB HDD $12/month 1 Gbps Unmetered Bandwidth
OneProvider Israel 2TB HDD $12/month 1 Gbps Unmetered Bandwidth
Disk speeds are moderate, network speeds are meh and Geekbench5 results are terrible.
I am a representative of Advin Servers
Nothing as crazy as my $1/year sister from cociu that got involcurated in Romania, but still insane deals :
DataIdeas, Texas: $6/year
GreenCloud, Singapore: $68/2 years
Also managed to get advinserver's offer, awaiting provision
.
This is crazy, slower than the snail owned by cuciu shitserver.
⭕ 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.
I love 2 digit gb5 scores
———-
blog | exploring visually |
Looking for some outstanding disk I/O in Vienna?
Alwyzon - Virtual Servers in Austria starting at 4,49 €/month (excl. VAT)
Some updates pushed to the YABS repo. Give me a shout if anyone hits any issues.
Head Janitor @ LES • About • Rules • Support
Always interesting seeing the usage spikes on Black Friday. Seems like it'll continue to be an annual tradition
Head Janitor @ LES • About • Rules • Support
I blame @cybertech ..... :-)
p.S @Mason congratulations on the growth innuse and acceptance of YABS...
Also interesting is the spike around August /September where number of BMs exceed unique machines by a factor of 4x
———-
blog | exploring visually |
Thanks, I was waiting for this
Looks like it may be intentional, considering it's 2 digit in any location I choose. They might want you to use this for FTP/NFS/etc storage, not so much for seedboxes and such.
I am a representative of Advin Servers
Virmach
BF-SPECIAL-2021
$12.95/yr WOW
3172MB RAM
3 vCORE
35GB SSD (RAID 10)
5TB Traffic WOW
10Gbps Bandwidth WOW
BUFFALO, NY LOCATION
1 DEDICATED IPv4
NO IPv6 but its ok
I bench YABS 24/7/365 unless it's a leap year.
PHP Friends Schnupperspecial 2021
Low performance of CPU compared with Intel Xeon Gold 6240 that could get 1500-ish score
⭕ 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.
Virmach Special
Enjoy meditation without religion for one month.
Terrahost free LXC, seems improved IO
I bench YABS 24/7/365 unless it's a leap year.