Physical cores pinned to your VM. No scheduler sharing, no noisy neighbors. AVX-512 passthrough via cpu: host and unlimited bandwidth on every plan.
Your VM is bound to specific physical CPU cores via cpuset. The Linux scheduler never migrates your workload to another core or another tenant's socket.
Full cpu: host mode exposes all Xeon Gold 6138 instruction sets including AVX-512. Useful for ML inference, cryptography, scientific computing, and financial workloads.
Every Dedicated Core plan uses the U.2 NVMe drive exclusively. No SSD fallback. Consistent low-latency I/O for databases, trading systems, and high-throughput workloads.
All Dedicated Core plans include unlimited inbound and outbound bandwidth. No monthly caps, no overage fees. Throttled to 10 Mbps after a very high usage threshold is reached.
1 dedicated physical core, pinned to NUMA-local CPU pair. AVX-512 exposed via cpu: host passthrough. Unlimited bandwidth.
Powered by Dual Intel® Xeon® Gold 6138s
2 dedicated physical cores, pinned to NUMA-local CPU pairs. AVX-512 exposed via cpu: host passthrough. Unlimited bandwidth.
Powered by Dual Intel® Xeon® Gold 6138s
4 dedicated physical cores, pinned to NUMA-local CPU pairs. AVX-512 exposed via cpu: host passthrough. Unlimited bandwidth.
Powered by Dual Intel® Xeon® Gold 6138s
Dedicated Core VMs use the same KVM OS templates. One-click reinstall from the portal, or mount your own ISO.
Standard KVM VMs share the CPU scheduler with other tenants. Dedicated Core pins your VM to specific physical cores that no other VM will ever touch. Consistent performance, every cycle.
Every Dedicated Core VPS runs from our Detroit, Michigan data center: the 123Net facility in Southfield, with Detroit-IX connected upstream routing for low-latency access across Michigan and the wider Midwest.
Compare with shared KVM VPS →Your VM is bound to dedicated physical core pairs. No other tenant runs on your cores. CPU-heavy workloads from neighboring VMs cannot affect your latency or throughput.
Intel Xeon Gold 6138 supports AVX-512F, AVX-512DQ, AVX-512BW, and AVX-512VL. With cpu: host passthrough enabled, your workloads can exploit 512-bit SIMD vectors directly.
Every Dedicated Core plan is provisioned on the U.2 NVMe drive. No mixed storage pools, no SSD fallback. Predictable I/O for latency-sensitive workloads like databases and caches.
No transfer caps on any Dedicated Core plan. Unlimited inbound and outbound. Sustained throughput for backups, replication, large model downloads, or high-traffic applications.
Cores are allocated from a single NUMA socket. Your VM's CPU and RAM access stay local to one socket, avoiding cross-NUMA latency penalties that occur when cores span both sockets.