SKRIME - 25% OFF AMD RYZEN 5950X & 9950X KVM VPS from 3.74€/mo - Eygelshoven, NL

gameplayer2507

@SKRIME said:
Hey @gameplayer2507,

We don't have a standard trial, but as @zed said the XEON 1G at 0.99€/mo is about as low-risk as a test gets. If you want to trial a specific larger configuration before committing, open a ticket on skrime.eu/support and we can talk about what's possible.

If the bandwidth performance of 0.99€/mo is going to be same as for the RYZEN 4G , then ill take it and test it out!

gameplayer2507

If then> @SKRIME said:

Hey @gameplayer2507,

We don't have a standard trial, but as @zed said the XEON 1G at 0.99€/mo is about as low-risk as a test gets. If you want to trial a specific larger configuration before committing, open a ticket on skrime.eu/support and we can talk about what's possible.

If the bandwidth performance for the 0.99€/mo is going to be the same for RYZEN 4G , then ill take that and test it out!

atkl

Hello @SKRIME, that looks interesting which kind of iso/OS can be installed on those?

These are just some of our highlighted plans. We offer many more configurations on our website skrime.eu

The -25% offer is only on 4G & 8G?

xavcon

tested with the official Ookla binary (multi-threaded) to an Amsterdam server from your Amsterdam VM. that kills the single-threaded argument:

Server: PhoenixNAP Global IT Services - Amsterdam (id: 28922)
ISP: Tom Gewiese
Idle Latency: 12.45 ms (jitter: 0.63ms)
Download: 20.18 Mbps
Upload: 20.24 Mbps
Jitter: up to 279ms

20 Mbps on a '10 Gbit/s shared uplink', Amsterdam to Amsterdam.

Also re: 'single core is the CPU ceiling for IO' - fio uses direct IO with libaio, it bypasses the filesystem and CPU is not the bottleneck. the fio results showed 4.27% usr / 8.54% sys CPU usage, nowhere near saturated. the 13ms latency and 19K IOPS is the storage, not the CPU.

for reference, actual NVMe behind a proper stack does 200K+ IOPS at sub-1ms latency. this is doing 19K at 13ms. that's a 10x IOPS gap and 13x latency gap. not a single-core limitation.

EvoSteven

@xavcon said:
tested with the official Ookla binary (multi-threaded) to an Amsterdam server from your Amsterdam VM. that kills the single-threaded argument:

Server: PhoenixNAP Global IT Services - Amsterdam (id: 28922)
ISP: Tom Gewiese
Idle Latency: 12.45 ms (jitter: 0.63ms)
Download: 20.18 Mbps
Upload: 20.24 Mbps
Jitter: up to 279ms

20 Mbps on a '10 Gbit/s shared uplink', Amsterdam to Amsterdam.

Also re: 'single core is the CPU ceiling for IO' - fio uses direct IO with libaio, it bypasses the filesystem and CPU is not the bottleneck. the fio results showed 4.27% usr / 8.54% sys CPU usage, nowhere near saturated. the 13ms latency and 19K IOPS is the storage, not the CPU.

for reference, actual NVMe behind a proper stack does 200K+ IOPS at sub-1ms latency. this is doing 19K at 13ms. that's a 10x IOPS gap and 13x latency gap. not a single-core limitation.

Made me curious and there are worse ways to spend 1.5 EUR...

Also using speedtest here (not speedtest-cli as that one is outdated I think, gave me sub 1mbps results):

   Speedtest by Ookla

      Server: kyouma.net – AS203819 - Kassel (id: 15426)
         ISP: Tom Gewiese
Idle Latency:     8.91 ms   (jitter: 0.17ms, low: 8.75ms, high: 9.32ms)
    Download:  2158.48 Mbps (data used: 2.8 GB)
                 22.85 ms   (jitter: 15.89ms, low: 8.40ms, high: 160.60ms)
      Upload:  2358.41 Mbps (data used: 4.3 GB)
                  9.95 ms   (jitter: 4.49ms, low: 8.58ms, high: 98.22ms)
 Packet Loss:     0.0%

Unable to reproduce your network speeds.
No comment on the disk speed though, that part was reproducible.

fxdjclub

Has the offer expired?

SKRIME

Hey everyone, a few replies.

@xavcon, please open a ticket with your product ID and we will take a look at your VM and the storage directly, then see what we can do.

@gameplayer2507, bandwidth is the same across all plans, the port is not tiered by plan size. If you want to try a specific configuration first, you can request a test server through support.

@atkl, any OS from our template list installs straight from the panel and custom ISOs can be mounted via ticket. The discount applied to the whole RYZEN range, but it ended on the 19th so that pricing is no longer active.

@fxdjclub, yes the sale expired on the 19th, so the RYZEN range is back at regular pricing. The order pages show the current numbers.

alincupunct

@SKRIME said:
@alincupunct - fair spot. The GB6 numbers on our 9950X plans are lower than they should be right now - looking at the host it's boosting fine, so the gap is in the VM CPU model exposure. We're on it - once corrected, the 9950X plans will land where you'd actually expect them.

@JohnFilch123 - the Shared Hosting ToS section covers all shared-resource products including KVM on the shared cores side, but the clause is aimed at intentional sabotage or someone actually degrading service quality for others on the node, not "you moved a lot of GB this month". Traffic itself has no monthly cap, no counter, no per-GB overage. Two different things.

@fuqet - sorry to hear that. If you'd opened a ticket while it was happening, we'd have looked at the host and either moved you or fixed the neighbour. Genuinely open to hearing what went wrong if you want to DM me the details.

Got an update on the 9950x numbers? Tyvm!

SKRIME

@alincupunct said:

@SKRIME said:
@alincupunct - fair spot. The GB6 numbers on our 9950X plans are lower than they should be right now - looking at the host it's boosting fine, so the gap is in the VM CPU model exposure. We're on it - once corrected, the 9950X plans will land where you'd actually expect them.

@JohnFilch123 - the Shared Hosting ToS section covers all shared-resource products including KVM on the shared cores side, but the clause is aimed at intentional sabotage or someone actually degrading service quality for others on the node, not "you moved a lot of GB this month". Traffic itself has no monthly cap, no counter, no per-GB overage. Two different things.

@fuqet - sorry to hear that. If you'd opened a ticket while it was happening, we'd have looked at the host and either moved you or fixed the neighbour. Genuinely open to hearing what went wrong if you want to DM me the details.

Got an update on the 9950x numbers? Tyvm!

@alincupunct, following up on the 9950X numbers.

We went through it properly. The guest sees the real 9950X with the full AVX-512 set, the cores boost to 5.3 GHz under load, and steal time on the node is zero. Running the same single-thread test on the host and inside a VM on the same machine gave 3.19s against 3.35s, so virtualisation costs around 5%. We tried a few config changes on top of that and none of them moved the numbers.

EvoSteven

@SKRIME said:

@alincupunct said:

@SKRIME said:
@alincupunct - fair spot. The GB6 numbers on our 9950X plans are lower than they should be right now - looking at the host it's boosting fine, so the gap is in the VM CPU model exposure. We're on it - once corrected, the 9950X plans will land where you'd actually expect them.

@JohnFilch123 - the Shared Hosting ToS section covers all shared-resource products including KVM on the shared cores side, but the clause is aimed at intentional sabotage or someone actually degrading service quality for others on the node, not "you moved a lot of GB this month". Traffic itself has no monthly cap, no counter, no per-GB overage. Two different things.

@fuqet - sorry to hear that. If you'd opened a ticket while it was happening, we'd have looked at the host and either moved you or fixed the neighbour. Genuinely open to hearing what went wrong if you want to DM me the details.

Got an update on the 9950x numbers? Tyvm!

@alincupunct, following up on the 9950X numbers.

We went through it properly. The guest sees the real 9950X with the full AVX-512 set, the cores boost to 5.3 GHz under load, and steal time on the node is zero. Running the same single-thread test on the host and inside a VM on the same machine gave 3.19s against 3.35s, so virtualisation costs around 5%. We tried a few config changes on top of that and none of them moved the numbers.

Any chance something happened on the 5950x side as well? Some sort of configuration change? The performance on the 16th was a lot better than a couple days later and the deficiency persists till this day. I'm barely seeing any steal, so I'm not sure what happened. Maybe you could rerun geekbench there as well?

If the same "configuration" was already active on the 9950x but not on the 5950x during those first couple YABS runs that could explain the discrepancy.
Hope it's something as simple as that...

(I've got an open ticket already but it seems they're very busy these past couple of weeks. Didn't really want to unnecessarily drag it into the open, but it's the topic of discussion right now lol. Outside of that I'm quite satisfied with the service so far.)

SKRIME

@EvoSteven said:

@SKRIME said:

@alincupunct said:

@SKRIME said:
@alincupunct - fair spot. The GB6 numbers on our 9950X plans are lower than they should be right now - looking at the host it's boosting fine, so the gap is in the VM CPU model exposure. We're on it - once corrected, the 9950X plans will land where you'd actually expect them.

@JohnFilch123 - the Shared Hosting ToS section covers all shared-resource products including KVM on the shared cores side, but the clause is aimed at intentional sabotage or someone actually degrading service quality for others on the node, not "you moved a lot of GB this month". Traffic itself has no monthly cap, no counter, no per-GB overage. Two different things.

@fuqet - sorry to hear that. If you'd opened a ticket while it was happening, we'd have looked at the host and either moved you or fixed the neighbour. Genuinely open to hearing what went wrong if you want to DM me the details.

Got an update on the 9950x numbers? Tyvm!

@alincupunct, following up on the 9950X numbers.

We went through it properly. The guest sees the real 9950X with the full AVX-512 set, the cores boost to 5.3 GHz under load, and steal time on the node is zero. Running the same single-thread test on the host and inside a VM on the same machine gave 3.19s against 3.35s, so virtualisation costs around 5%. We tried a few config changes on top of that and none of them moved the numbers.

Any chance something happened on the 5950x side as well? Some sort of configuration change? The performance on the 16th was a lot better than a couple days later and the deficiency persists till this day. I'm barely seeing any steal, so I'm not sure what happened. Maybe you could rerun geekbench there as well?

If the same "configuration" was already active on the 9950x but not on the 5950x during those first couple YABS runs that could explain the discrepancy.
Hope it's something as simple as that...

(I've got an open ticket already but it seems they're very busy these past couple of weeks. Didn't really want to unnecessarily drag it into the open, but it's the topic of discussion right now lol. Outside of that I'm quite satisfied with the service so far.)

@EvoSteven, no problem bringing it up here.

Most of the team was on holiday last week, which is why the ticket queue moved slowly. We're catching up on that today, so yours will get an answer.

Nothing should have changed on the 5950X side configuration-wise, at least nothing we're aware of. But a regression that sticks around is worth looking at properly, so we'll go through that node and rerun the benchmarks there.

Thanks for the patience, and good to hear you're happy with the rest of it.

SKRIME

Update - the sale is back until 31.07.2026

We got a lot of requests asking about the discount after it ended, so we've decided to bring it back. It now runs until 31.07.2026, 23:59 CEST or while stocks last. DDR4 is 35% off, DDR5 is 25% off, and it applies across the whole RYZEN lineup rather than just the plans listed below.

AMD RYZEN 5950X (DDR4) - 35% off

Plan	Cores	RAM	Storage	Sale Price	Regular
RYZEN 4G	2	4 GB	50 GB	3.24€/mo	4.99€/mo
RYZEN 8G	2	8 GB	100 GB	5.84€/mo	8.99€/mo

Order: skri.me/ryzen5950X

AMD RYZEN 9950X (DDR5) - 25% off

Plan	Cores	RAM	Storage	Sale Price	Regular
RYZEN 4G	2	4 GB	50 GB	5.24€/mo	6.99€/mo
RYZEN 8G	2	8 GB	100 GB	8.24€/mo	10.99€/mo

Order: skri.me/ryzen9950X

Longer billing terms stack on top of the sale price with 6 months at -5%, 12 months at -10% and 24 months at -15%. At 24 months the DDR4 RYZEN 4G effectively lands at 2.75€/mo and the DDR5 RYZEN 4G at 4.45€/mo.

mp11

how can i add 5tb san disk to this offer?

SKRIME

@mp11 said:
how can i add 5tb san disk to this offer?

Hey @mp11,

The RYZEN plans in this offer are fixed configurations, so extra storage can't be added on top of them. We also don't offer SAN or network block storage at all - everything is local NVMe in RAID 1.

DK87

I can confirm that something strange with disk io happened last days on 5950x vps series.
Here is output from ryzen 5950 plan and epyc 7543 plan.

## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##

Yet-Another-Bench-Script

v2026-07-24

https://github.com/masonr/yet-another-bench-script

## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##

Wed Jul 29 15:20:31 UTC 2026

Basic System Information:

Uptime : 0 days, 0 hours, 4 minutes
Processor : AMD Ryzen 9 5950X 16-Core Processor
CPU cores : 2 @ 3393.622 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ✔ Enabled
RAM : 3.8 GiB
Swap : 0.0 KiB
Disk : 48.4 GiB
Distro : Ubuntu 24.04.4 LTS
Kernel : 6.8.0-136-generic
VM Type : KVM
IPv4/IPv6 : ✔ Online / ✔ Online

IPv6 Network Information:

ISP : Tom Gewiese
ASN : AS215365 Tom Gewiese
Host : Tom Gewiese
Location : Frankfurt am Main, Hesse (HE)
Country : Germany

fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/sda1):

Block Size	4k (IOPS)	64k (IOPS)
Read	170.00 MB/s (41.5k)	563.05 MB/s (8.5k)
Write	170.45 MB/s (41.6k)	566.01 MB/s (8.6k)
Total	340.46 MB/s (83.1k)	1.12 GB/s (17.2k)
Block Size	512k (IOPS)	1m (IOPS)
------	--- ----	---- ----
Read	704.58 MB/s (1.3k)	767.17 MB/s (731)
Write	742.02 MB/s (1.4k)	818.26 MB/s (780)
Total	1.44 GB/s (2.7k)	1.58 GB/s (1.5k)

Geekbench 6 Benchmark Test:

Test | Value
|
Single Core |
Multi Core |
Full Test | https://browser.geekbench.com/v6/cpu/18861301

## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##

Yet-Another-Bench-Script

v2026-07-24

https://github.com/masonr/yet-another-bench-script

## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##

Wed Jul 29 18:24:23 +03 2026

Basic System Information:

Uptime : 19 days, 11 hours, 21 minutes
Processor : AMD EPYC 7543 32-Core Processor
CPU cores : 2 @ 2800.000 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ✔ Enabled
RAM : 961.3 MiB
Swap : 1024.0 MiB
Disk : 9.6 GiB
Distro : Ubuntu 24.04.4 LTS
Kernel : 6.17.0-35-generic
VM Type : KVM
IPv4/IPv6 : ✔ Online / ✔ Online

IPv6 Network Information:

ISP : Tom Gewiese
ASN : AS215365 Tom Gewiese
Host : Tom Gewiese
Location : Frankfurt am Main, Hesse (HE)
Country : Germany

fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/sda1):

Block Size	4k (IOPS)	64k (IOPS)
Read	137.52 MB/s (33.5k)	904.11 MB/s (13.7k)
Write	137.88 MB/s (33.6k)	908.87 MB/s (13.8k)
Total	275.40 MB/s (67.2k)	1.81 GB/s (27.6k)
Block Size	512k (IOPS)	1m (IOPS)
------	--- ----	---- ----
Read	1.20 GB/s (2.3k)	1.23 GB/s (1.1k)
Write	1.27 GB/s (2.4k)	1.31 GB/s (1.2k)
Total	2.47 GB/s (4.7k)	2.55 GB/s (2.4k)

Geekbench test failed and low memory was detected. Add at least 1GB of SWAP or use GB4 instead (higher compatibility with low memory systems).

YABS completed in 13 min 22 sec

ignore cpu benchmark test on epyc vps - its fails at gb6 version test cause of ram lack.

difference between disk speed performace is clearly strange. epyc node just faster in cheaper price.

SKRIME

@DK87 said:
I can confirm that something strange with disk io happened last days on 5950x vps series.
Here is output from ryzen 5950 plan and epyc 7543 plan.

## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##

Yet-Another-Bench-Script

v2026-07-24

https://github.com/masonr/yet-another-bench-script

## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##

Wed Jul 29 15:20:31 UTC 2026

Basic System Information:

Uptime : 0 days, 0 hours, 4 minutes
Processor : AMD Ryzen 9 5950X 16-Core Processor
CPU cores : 2 @ 3393.622 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ✔ Enabled
RAM : 3.8 GiB
Swap : 0.0 KiB
Disk : 48.4 GiB
Distro : Ubuntu 24.04.4 LTS
Kernel : 6.8.0-136-generic
VM Type : KVM
IPv4/IPv6 : ✔ Online / ✔ Online

IPv6 Network Information:

ISP : Tom Gewiese
ASN : AS215365 Tom Gewiese
Host : Tom Gewiese
Location : Frankfurt am Main, Hesse (HE)
Country : Germany

fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/sda1):

Block Size	4k (IOPS)	64k (IOPS)
Read	170.00 MB/s (41.5k)	563.05 MB/s (8.5k)
Write	170.45 MB/s (41.6k)	566.01 MB/s (8.6k)
Total	340.46 MB/s (83.1k)	1.12 GB/s (17.2k)
Block Size	512k (IOPS)	1m (IOPS)
------	--- ----	---- ----
Read	704.58 MB/s (1.3k)	767.17 MB/s (731)
Write	742.02 MB/s (1.4k)	818.26 MB/s (780)
Total	1.44 GB/s (2.7k)	1.58 GB/s (1.5k)

Geekbench 6 Benchmark Test:

Test | Value
|
Single Core |
Multi Core |
Full Test | https://browser.geekbench.com/v6/cpu/18861301

## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##

Yet-Another-Bench-Script

v2026-07-24

https://github.com/masonr/yet-another-bench-script

## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##

Wed Jul 29 18:24:23 +03 2026

Basic System Information:

Uptime : 19 days, 11 hours, 21 minutes
Processor : AMD EPYC 7543 32-Core Processor
CPU cores : 2 @ 2800.000 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ✔ Enabled
RAM : 961.3 MiB
Swap : 1024.0 MiB
Disk : 9.6 GiB
Distro : Ubuntu 24.04.4 LTS
Kernel : 6.17.0-35-generic
VM Type : KVM
IPv4/IPv6 : ✔ Online / ✔ Online

IPv6 Network Information:

ISP : Tom Gewiese
ASN : AS215365 Tom Gewiese
Host : Tom Gewiese
Location : Frankfurt am Main, Hesse (HE)
Country : Germany

fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/sda1):

Block Size	4k (IOPS)	64k (IOPS)
Read	137.52 MB/s (33.5k)	904.11 MB/s (13.7k)
Write	137.88 MB/s (33.6k)	908.87 MB/s (13.8k)
Total	275.40 MB/s (67.2k)	1.81 GB/s (27.6k)
Block Size	512k (IOPS)	1m (IOPS)
------	--- ----	---- ----
Read	1.20 GB/s (2.3k)	1.23 GB/s (1.1k)
Write	1.27 GB/s (2.4k)	1.31 GB/s (1.2k)
Total	2.47 GB/s (4.7k)	2.55 GB/s (2.4k)

Geekbench test failed and low memory was detected. Add at least 1GB of SWAP or use GB4 instead (higher compatibility with low memory systems).

YABS completed in 13 min 22 sec

ignore cpu benchmark test on epyc vps - its fails at gb6 version test cause of ram lack.

difference between disk speed performace is clearly strange. epyc node just faster in cheaper price.

@DK87 Thanks for the numbers, we'll take a look at the node.

Worth noting the two node types don't run the same NVMes, so a difference in the disk profile is expected.

If you're seeing a drop compared to before, open a ticket with the product ID and we'll check it properly.

DK87

@SKRIME said: @DK87 Thanks for the numbers, we'll take a look at the node.

Worth noting the two node types don't run the same NVMes, so a difference in the disk profile is expected.

If you're seeing a drop compared to before, open a ticket with the product ID and we'll check it properly.

Yeah, thank you for the reply. I'll open it and send you compares. In fact that CPU performance and network speed is remain quite good, only disk performance affected - just curious what can be happened and as i noticed not only in my case.

EvoSteven

@SKRIME said:

@EvoSteven said:

@SKRIME said:

@alincupunct said:

@SKRIME said:
@alincupunct - fair spot. The GB6 numbers on our 9950X plans are lower than they should be right now - looking at the host it's boosting fine, so the gap is in the VM CPU model exposure. We're on it - once corrected, the 9950X plans will land where you'd actually expect them.

@JohnFilch123 - the Shared Hosting ToS section covers all shared-resource products including KVM on the shared cores side, but the clause is aimed at intentional sabotage or someone actually degrading service quality for others on the node, not "you moved a lot of GB this month". Traffic itself has no monthly cap, no counter, no per-GB overage. Two different things.

@fuqet - sorry to hear that. If you'd opened a ticket while it was happening, we'd have looked at the host and either moved you or fixed the neighbour. Genuinely open to hearing what went wrong if you want to DM me the details.

Got an update on the 9950x numbers? Tyvm!

@alincupunct, following up on the 9950X numbers.

We went through it properly. The guest sees the real 9950X with the full AVX-512 set, the cores boost to 5.3 GHz under load, and steal time on the node is zero. Running the same single-thread test on the host and inside a VM on the same machine gave 3.19s against 3.35s, so virtualisation costs around 5%. We tried a few config changes on top of that and none of them moved the numbers.

Any chance something happened on the 5950x side as well? Some sort of configuration change? The performance on the 16th was a lot better than a couple days later and the deficiency persists till this day. I'm barely seeing any steal, so I'm not sure what happened. Maybe you could rerun geekbench there as well?

If the same "configuration" was already active on the 9950x but not on the 5950x during those first couple YABS runs that could explain the discrepancy.
Hope it's something as simple as that...

(I've got an open ticket already but it seems they're very busy these past couple of weeks. Didn't really want to unnecessarily drag it into the open, but it's the topic of discussion right now lol. Outside of that I'm quite satisfied with the service so far.)

@EvoSteven, no problem bringing it up here.

Most of the team was on holiday last week, which is why the ticket queue moved slowly. We're catching up on that today, so yours will get an answer.

Nothing should have changed on the 5950X side configuration-wise, at least nothing we're aware of. But a regression that sticks around is worth looking at properly, so we'll go through that node and rerun the benchmarks there.

Thanks for the patience, and good to hear you're happy with the rest of it.

Quick follow up on their claim: I can confirm their support is moving a lot quicker over the past few days and my ticket ended up getting resolved, so credit where credit is due.