Benefits of HP Z Boost
Accelerated design cycles: Shorten design cycles and unblock teams by dynamically assigning shared GPUs to the users and tasks that need them. On-demand, on-prem resources accelerate workflows from idea to output no waiting for upgrades, reconfigurations, or render queues.
Low-latency shared GPU access: On prem GPU compute, optimized for low latency LAN grade network performance,
Maximized GPU utilization: Deliver value by turning underutilized workstation GPUs into shared, centrally managed resources. Dynamically allocate individual GPU resources avoiding overprovisioning, increasing utilization, and extending the workstation value.
Local data. Secure workflows: Keep sensitive project files and data on-prem, on-device and under your control. HP Z Boost lives within your network, minimizing cloud related risks and preventing unpredictable cloud-computer cost.
Architecture overview
Keep your application on your device. Connect supported GPU work to compatible HP Z workstations over your network.
Your device & application
Compatible clients from HP or other vendors
Work locally in supported design, visualization and AI applications. Suitable server workloads can also access shared GPU resources.
HP Z Boost
Shared GPU access across your network
GPU processing runs on the HP Z host, and results return to your workflow. An on-premises deployment can keep shared compute within your network.
HP Z GPU hosts
Compatible HP Z workstations
Use compatible GPUs in an existing HP Z fleet or add HP Z GPU hosts to complement the clients and suitable servers you already have.
One GPU fabric. Two ways to work.
One GPU fabric · Interactive work
1. Keep working on your device
Run a supported application on a compatible client from HP or another vendor.
2. Access shared HP Z GPU power
Connect the supported GPU work to compatible HP Z workstation hosts over your network.
3. Review the result in your workflow
Evaluate the design or AI task from your application, with GPU processing on the host.
The HP Z Boost advantage
See how shared GPU access addresses real workflow challenges, from design reviews to smaller AI jobs alongside HPC.
AI / ML alongside HPC
The challenge
Smaller AI experiments waited while long-running jobs occupied cluster GPUs.
The approach
Connect available workstation GPUs to suitable smaller jobs using available server resources and tailored orchestration.
Why it matters
Make distributed capacity useful alongside core training.
Explore the customer experience
A customer solution connected available workstation GPUs to overnight AI experiments. GPU availability changed with workstation activity. The deployment required workload orchestration and availability controls.
Product design
The challenge
A rendering workflow involved application switching and waiting for resources.
The approach
Use shared GPU access to support the desired visualization workflow.
Reported experience
Rendering became more accessible, allowing designers to iterate more often.
Explore the customer experience
The deployment reduced waiting for rendering resources. Test a representative design task to measure the outcome in your own environment.
Design teams
The challenge
Uneven local GPU capability limited real-time visualization for some users.
The approach
Give supported workflows access to HP Z GPU power while people keep their compatible devices.
Reported experience
A direct path from designing to reviewing on the same device.
Explore the customer experience
A design customer demonstrated real-time sessions from existing devices, using remote HP Z GPU resources. Application versions and configuration must be validated for each deployment.
Industrial AI
The challenge
Heat and dust made a production line unsuitable for powerful GPU hardware.
The approach
A rugged device collects images and accesses workstation GPU power elsewhere in the facility.
Reported experience
Separate the point of work from the location of the GPU hardware.
Explore the customer experience
The customer solution combines image inspection during the day with model refinement using collected data in the evening. Test workload support, network conditions and latency in the PoC.
Explore your fleet's unused GPU capacity
728,000 potential unused GPU-hours per year
Highlighted example: 500 workstations, with GPUs in use 30% of working hours (70% unused).
Assumptions behind the estimate: Estimate based on 2,080 annual reference hours per GPU (8 hours/day × 5 days/week × 52 weeks) and the GPU utilization entered by the user. The estimate represents potential unused GPU capacity and does not guarantee available or usable compute hours, time savings, or equivalent HPC-server performance. Actual usable capacity varies based on workload fit, system availability, network conditions, orchestration, and other factors.
Estimated unused GPU-hours per year, across the fleet
HP Z
Workstations
200
300
400
500
600
Average GPU time in use (% of working hours)
20%
332,800
499,200
665,600
832,000
998,400
30%
291,200
436,800
582,400
728,000
873,600
40%
249,600
374,400
499,200
624,000
748,800
50%
208,000
312,000
416,000
520,000
624,000
60%
166,400
249,600
332,800
416,000
499,200
70%
124,800
187,200
249,600
312,000
374,400
80%
83,200
124,800
166,400
208,000
249,600
Resources
Frequently asked questions
What is HP Z Boost?
How does HP Z Boost work?
What problem does HP Z Boost solve?
Who is HP Z Boost designed for?
What applications can benefit from HP Z Boost?
Does HP Z Boost replace VDI or remote desktop?
Does HP Z Boost require public cloud infrastructure?
Is my data sent to the cloud?
How secure is HP Z Boost?
What networking requirements should I consider?
Can multiple users share the same workstation GPUs?
How do I know if HP Z Boost is a good fit for my organization?
How can I request a demo or speak with HP?
Disclaimers
- The information contained herein is subject to change without notice. The only warranties for HP products and services are set forth in the express warranty statements accompanying such products and services. Nothing herein should be construed as constituting an additional warranty. HP shall not be liable for technical or editorial errors or omissions contained herein.