Is the CMP 170HX good for training?

It is better suited to inference and memory-bound jobs. With 12.1 TFLOPS FP32 it is slower at compute-heavy training than cards like the A40 or RTX 3090, so test your workload before planning a long run.

Why test my workload first?

The CMP 170HX is a compute-only card and some software paths behave differently on it than on mainstream GPUs. A short hourly session tells you whether your framework, drivers and model run as expected.

What is the CMP 170HX's strongest spec?

Memory bandwidth: about 1516 GB/s, higher than the A100 PCIE's listed 1313 GB/s, which helps any workload that waits on memory more than on math.

How long do I have to keep a CMP 170HX server on Cloudj?

As long as the machine stays listed: each machine shows the date it is available until, after which the instance and its disk are removed. Usage is billed per minute while it runs, and you can destroy the instance as soon as the test or job finishes.

GPU Cloud · NVIDIA · Ampere

Rent an NVIDIA CMP 170HX GPU with 64 GB VRAM

Name: NVIDIA CMP 170HX cloud GPU
Brand: Cloudj
Availability: InStock

Rent CMP 170HX cloud GPU servers by the hour on Cloudj: 64 GB VRAM listed, Ampere architecture and about 1516 GB/s memory bandwidth for memory-bound compute. It is a headless compute card for developers who care about bandwidth and capacity more than FP32 speed (12.1 TFLOPS, 250 W, compute capability 8.0). Pick a live configuration, start from the Ubuntu 22.04 + CUDA or PyTorch + JupyterLab template and connect over SSH with your key.

Rent CMP 170HX All GPUs

CMP 170HX specifications

VRAM per GPU

64 GB

Architecture

NVIDIA Ampere

Compute capability

FP32 compute

≈ 12.1 TFLOPS

Memory bandwidth

≈ 1516 GB/s

Power limit

250 W

Values reported by the infrastructure provider for machines currently listed; real performance varies by machine.

Live CMP 170HX configurations and prices

From $0.43/h

incl. 32 GB disk · Live price, updated every 15 minutes

Deploy

What you can run on the CMP 170HX cloud GPU

Bandwidth-bound inference

LLM token generation is limited mainly by memory bandwidth, and 1516 GB/s is in the same range as A100-class cards.

Quantized LLM hosting

Load quantized models in Ollama or llama.cpp and serve them over an API for testing or internal tools.

CUDA kernel experiments

Profile memory-heavy kernels on an Ampere card with compute capability 8.0 before moving them to larger hardware.

Batch data processing

Stream large arrays through GPU memory for transforms, hashing or analytics jobs that are bound by memory access.

Is the CMP 170HX the right GPU for your workload?

Choose the CMP 170HX when your workload is bottlenecked by memory capacity and bandwidth rather than raw single-precision compute. Opt for a GPU with higher compute throughput if you require faster training iterations, or select a smaller consumer card if your models do not need 64 GB of VRAM.

CMP 170HX vs similar cloud GPUs

GPU	VRAM per GPU	FP32 compute	Power limit
NVIDIA CMP 170HX	64 GB	≈ 12.1 TFLOPS	250 W
NVIDIA A800 PCIE	80 GB	≈ 15.6 TFLOPS	300 W	Details →
NVIDIA A40	48 GB	≈ 29.9 TFLOPS	300 W	Details →
NVIDIA H100 PCIE	80 GB	≈ 41 TFLOPS	350 W	Details →
NVIDIA H100 SXM	80 GB	≈ 53.5 TFLOPS	700 W	Details →

Guides for CMP 170HX

GPU cloud Asia guide: Japan, Singapore, Korea and nearbyA practical GPU cloud Asia guide: which country codes to search for East and Southeast Asian users, a fallback order, and how to pair a GPU with a VPS.GPU cloud by location: how to pick a country for your GPUHow GPU cloud by location works on Cloudj: filter a live list of machines by country code, decide between near users, lowest price or data rules.GPU cloud Europe: choosing a country for your AI workloadsA practical guide to GPU cloud Europe searches: filter by several country codes at once, weigh Western, Nordic and Eastern hosts, and plan where data lives.GPU cloud North America: renting GPUs in the US and CanadaA GPU cloud North America guide: how to search US and CA hosts, read the results, pick a VPS front end and decide when a US GPU suits users elsewhere.

All articles →

CMP 170HX cloud GPU: frequently asked questions

Rent the CMP 170HX cloud GPU by the hour

Pick a live configuration, choose a template and connect over SSH with your key. Billing stops when you destroy the instance.

Rent CMP 170HX All GPUs