Netcup RS 1000 G12 In-Depth Review: 4-Core EPYC 9645 Benchmarks & Performance
Netcup recently upgraded its popular Root Server lineup to the Generation 12 (G12). Less than a year after the launch of the G11 series, this update comes as a pleasant surprise. The primary change is the CPU upgrade from the AMD EPYC�?9634 to the newer, Zen 5-based AMD EPYC�?9645, while the rest of the specifications and pricing remain identical.
In this review, we run comprehensive benchmark tests on the RS 1000 G12—Netcup’s most mainstream Root Server—and provide the full, unmodified logs from our sysbench, YABS, and Unixbench runs.
📋 Hardware Specifications
- CPU: AMD EPYC�?9645 (Zen 5 Architecture)
- Cores: 4 Dedicated Cores
- RAM: 8 GB DDR5 ECC RAM
- Storage: 256 GB NVMe SSD
- Network Port: 2.5 Gbps (with a fair-use limit: if traffic exceeds 3 TB in 24 hours, speed is temporarily throttled to 300 Mbps until the traffic usage drops below the threshold)
- Virtualization: KVM (Root Server class, meaning dedicated resources)
🚀 Detailed Benchmark Reports
Below are the full raw test logs from our server runs.
1. Basic System & Disk I/O (Bench.sh)
Here is the full output of bench.sh verifying system characteristics, disk I/O, and speedtest bandwidth limits:
-------------------- A Bench.sh Script By Teddysun -------------------
Version : v2025-05-08
Usage : wget -qO- bench.sh | bash
----------------------------------------------------------------------
CPU Model : AMD EPYC 9645 96-Core Processor
CPU Cores : 4 @ 2296.548 MHz
CPU Cache : 512 KB
AES-NI : �?Enabled
VM-x/AMD-V : �?Disabled
Total Disk : 251.9 GB (1.1 GB Used)
Total Mem : 7.8 GB (288.5 MB Used)
System uptime : 0 days, 0 hour 42 min
Load average : 0.02, 0.01, 0.06
OS : Debian GNU/Linux 12
Arch : x86_64 (64 Bit)
Kernel : 6.1.0-39-amd64
TCP CC : cubic
Virtualization : KVM
IPv4/IPv6 : �?Online / �?Online
Organization : AS197540 netcup GmbH
Location : Nürnberg / DE
Region : Bavaria
----------------------------------------------------------------------
I/O Speed(1st run) : 1.2 GB/s
I/O Speed(2nd run) : 2.1 GB/s
I/O Speed(3rd run) : 2.1 GB/s
I/O Speed(average) : 1843.2 MB/s
----------------------------------------------------------------------
Node Name Upload Speed Download Speed Latency
Speedtest.net 934.00 Mbps 992.61 Mbps 0.64 ms
Paris, FR 2489.04 Mbps 2348.30 Mbps 22.25 ms
Amsterdam, NL 2518.30 Mbps 2386.96 Mbps 9.76 ms
Shanghai, CN 1029.41 Mbps 1398.91 Mbps 245.10 ms
Hong Kong, CN 475.64 Mbps 1571.77 Mbps 188.40 ms
Singapore, SG 44.40 Mbps 285.99 Mbps 284.76 ms
Tokyo, JP 411.72 Mbps 2465.19 Mbps 274.70 ms
----------------------------------------------------------------------
Finished in : 3 min 26 sec
Timestamp : 2025-09-11 13:26:38 CEST2. Full YABS (Yet Another Bench Script) Logs
The YABS test covers detailed disk I/O performance via fio, network throughput via iperf3 (both IPv4 and IPv6), and standard Geekbench CPU benchmarking scores.
Basic System Information:
---------------------------------
Uptime : 0 days, 0 hours, 4 minutes
Processor : AMD EPYC 9645 96-Core Processor
CPU cores : 4 @ 2296.548 MHz
AES-NI : �?Enabled
VM-x/AMD-V : �?Disabled
RAM : 7.8 GiB
Swap : 0.0 KiB
Disk : 251.9 GiB
Distro : Debian GNU/Linux 12 (bookworm)
Kernel : 6.1.0-39-amd64
VM Type : KVM
IPv4/IPv6 : �?Online / �?Online
IPv6 Network Information:
---------------------------------
ISP : ANEXIA
ASN : AS47147 ANEXIA Internetdienstleistungs GmbH
Location : Vienna, Vienna (9)
Country : Austria
fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/vda4):
---------------------------------
Block Size | 4k (IOPS) | 64k (IOPS)
------ | --- ---- | ---- ----
Read | 374.13 MB/s (93.5k) | 2.06 GB/s (32.2k)
Write | 375.11 MB/s (93.7k) | 2.07 GB/s (32.4k)
Total | 749.24 MB/s (187.3k) | 4.13 GB/s (64.6k)
| |
Block Size | 512k (IOPS) | 1m (IOPS)
------ | --- ---- | ---- ----
Read | 1.92 GB/s (3.7k) | 2.10 GB/s (2.0k)
Write | 2.02 GB/s (3.9k) | 2.24 GB/s (2.1k)
Total | 3.94 GB/s (7.7k) | 4.34 GB/s (4.2k)
iperf3 Network Speed Tests (IPv4):
---------------------------------
Provider | Location (Link) | Send Speed | Recv Speed | Ping
----- | ----- | ---- | ---- | ----
Clouvider | London, UK (10G) | 2.68 Gbits/sec | 2.37 Gbits/sec | 26.2 ms
Eranium | Amsterdam, NL (100G) | 2.73 Gbits/sec | 2.37 Gbits/sec | 20.8 ms
Uztelecom | Tashkent, UZ (10G) | 520 Mbits/sec | 1.36 Gbits/sec | 91.7 ms
Leaseweb | Singapore, SG (10G) | 1.48 Gbits/sec | 1.16 Gbits/sec | --
Clouvider | Los Angeles, CA, US (10G) | 1.14 Gbits/sec | 1.06 Gbits/sec | 161 ms
Leaseweb | NYC, NY, US (10G) | 2.36 Gbits/sec | 2.20 Gbits/sec | 96.4 ms
Edgoo | Sao Paulo, BR (1G) | 2.14 Gbits/sec | 1.02 Gbits/sec | 196 ms
iperf3 Network Speed Tests (IPv6):
---------------------------------
Provider | Location (Link) | Send Speed | Recv Speed | Ping
----- | ----- | ---- | ---- | ----
Clouvider | London, UK (10G) | 1.98 Gbits/sec | 2.26 Gbits/sec | 26.7 ms
Eranium | Amsterdam, NL (100G) | 2.67 Gbits/sec | 2.32 Gbits/sec | 20.7 ms
Uztelecom | Tashkent, UZ (10G) | 2.40 Gbits/sec | 1.81 Gbits/sec | 91.7 ms
Leaseweb | Singapore, SG (10G) | 1.16 Gbits/sec | 1.70 Gbits/sec | 151 ms
Clouvider | Los Angeles, CA, US (10G) | 932 Mbits/sec | 701 Mbits/sec | 161 ms
Leaseweb | NYC, NY, US (10G) | 2.16 Gbits/sec | 2.06 Gbits/sec | 96.5 ms
Edgoo | Sao Paulo, BR (1G) | 1.86 Gbits/sec | 1.18 Gbits/sec | 199 ms
Geekbench 5 Benchmark Test:
---------------------------------
Test | Value
|
Single Core | 1827
Multi Core | 7027
Full Test | https://browser.geekbench.com/v5/cpu/23777546
Geekbench 6 Benchmark Test:
---------------------------------
Test | Value
|
Single Core | 2290
Multi Core | 7559
Full Test | https://browser.geekbench.com/v6/cpu/13758688
YABS completed in 13 min 24 sec3. Unixbench Scores
Unixbench measures the performance of a Unix-like system. We ran both single-core (1 parallel copy) and multi-core (4 parallel copies) tests:
Benchmark Run: Thu Sep 11 2025 13:30:55 - 13:58:51
4 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 70546709.3 lps (10.0 s, 7 samples)
Double-Precision Whetstone 8892.4 MWIPS (10.0 s, 7 samples)
Execl Throughput 5790.0 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 2398531.4 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 619612.9 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 7688487.3 KBps (30.0 s, 2 samples)
Pipe Throughput 3448692.0 lps (10.0 s, 7 samples)
Pipe-based Context Switching 89135.3 lps (10.0 s, 7 samples)
Process Creation 9478.8 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 15850.9 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 4610.5 lpm (60.0 s, 2 samples)
System Call Overhead 3110247.7 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 70546709.3 6045.1
Double-Precision Whetstone 55.0 8892.4 1616.8
Execl Throughput 43.0 5790.0 1346.5
File Copy 1024 bufsize 2000 maxblocks 3960.0 2398531.4 6056.9
File Copy 256 bufsize 500 maxblocks 1655.0 619612.9 3743.9
File Copy 4096 bufsize 8000 maxblocks 5800.0 7688487.3 13256.0
Pipe Throughput 12440.0 3448692.0 2772.3
Pipe-based Context Switching 4000.0 89135.3 222.8
Process Creation 126.0 9478.8 752.3
Shell Scripts (1 concurrent) 42.4 15850.9 3738.4
Shell Scripts (8 concurrent) 6.0 4610.5 7684.1
System Call Overhead 15000.0 3110247.7 2073.5
========
System Benchmarks Index Score 2630.0
------------------------------------------------------------------------
Benchmark Run: Thu Sep 11 2025 13:58:51 - 14:26:48
4 CPUs in system; running 4 parallel copies of tests
Dhrystone 2 using register variables 273492319.3 lps (10.0 s, 7 samples)
Double-Precision Whetstone 35597.6 MWIPS (10.0 s, 7 samples)
Execl Throughput 14169.5 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 1086793.0 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 271925.1 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 3573753.1 KBps (30.0 s, 2 samples)
Pipe Throughput 13577272.5 lps (10.0 s, 7 samples)
Pipe-based Context Switching 807728.4 lps (10.0 s, 7 samples)
Process Creation 43410.7 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 37835.1 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 5236.8 lpm (60.0 s, 2 samples)
System Call Overhead 3423066.3 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 273492319.3 23435.5
Double-Precision Whetstone 55.0 35597.6 6472.3
Execl Throughput 43.0 14169.5 3295.2
File Copy 1024 bufsize 2000 maxblocks 3960.0 1086793.0 2744.4
File Copy 256 bufsize 500 maxblocks 1655.0 271925.1 1643.1
File Copy 4096 bufsize 8000 maxblocks 5800.0 3573753.1 6161.6
Pipe Throughput 12440.0 13577272.5 10914.2
Pipe-based Context Switching 4000.0 807728.4 2019.3
Process Creation 126.0 43410.7 3445.3
Shell Scripts (1 concurrent) 42.4 37835.1 8923.4
Shell Scripts (8 concurrent) 6.0 5236.8 8728.0
System Call Overhead 15000.0 3423066.3 2282.0
========
System Benchmarks Index Score 4891.0📺 Streaming Unlocking & IP Quality
Using standard detection scripts, we verified the streaming media geo-locking status for the server IP:
- Netflix: Yes (Fully Unlocked, Germany Region)
- YouTube Premium: Yes (Germany Region, routed via Anexia Vienna cache)
- Disney+: Yes (Germany Region)
- ChatGPT / Claude: Yes (Supported)
- Google Gemini: No
The IP address reputation is clean, with very low abuse scores across standard databases, meaning it is unlikely to trigger verification captchas on search engines.
🌐 Network Routing (Back to Asia)
The server is physically located in Netcup’s Nuremberg datacenter. The return routing to China is as follows:
- China Telecom: 163 (Ordinary direct route)
- China Unicom: AS4837 (Ordinary direct route)
- China Mobile: CMI (Ordinary direct route)
The latency to East Asia averages between 180ms to 240ms. The routing is standard for European servers—not optimized for low-latency gaming in Asia, but perfectly stable for web hosting, storage, backups, and staging environments.
💬 Conclusion
The Netcup RS 1000 G12 is an absolute powerhouse. By upgrading the CPU to the AMD EPYC 9645 (Zen 5) architecture while keeping the price unchanged, Netcup has cemented its place as the unbeatable value king in the Root Server space. You get 4 fully dedicated EPYC cores, 8 GB DDR5 ECC memory, and NVMe SSD speed for under �?/month.
How to Get a �? Discount
If you are ordering a server for the first time, you can apply our �? new customer voucher code (available on our homepage) during the checkout process to get a discount on your first hosting bill.