Every day, AI continues to accelerate the pace of business. For organizations that require on-device AI processing rather than cloud-based models—whether for increased speed, security, or personalization—HP recently released its line of Next Gen AI PCs, designed specifically to “tackle complex AI tasks running on your device on in the cloud while boosting performance for your daily work.”¹

To put this claim to the test, we assessed the performance of an HP EliteBook 8 G1a 14 inch Notebook Next Gen AI PC powered by an AMD Ryzen^™ AI 7 PRO 350 processor, a Dell^™ Pro 14 Plus laptop powered by an Intel^® Core^™ Ultra 7 268V processor with Intel vPro^®, and aLenovo ThinkPad T14s Gen 6 laptop powered by an Intel Core Ultra 7 268Vprocessor with Intel vPro.

We found that the HP EliteBook 8 G1a 14 inch Notebook Next Gen AI PC achieved higher performance across a variety of productivity and AI workloads. Plus, with a workday’s worth of battery life, serviceable parts, a built-in AI assistant, and intelligent privacy features, this AMD Ryzen^™ AI PRO processor-powered AI PC can help propel your workforce forward in the age of AI.

What and how we tested

We equipped each PC with Windows 11 Pro, 32 GB of memory, and 1 TB of SSD storage:^*

*The results we report reflect the specific configurations we tested. Any difference in the configurations—as well as screen brightness, network traffic, and software additions—can affect these results. For a deeper dive into our testing parameters and procedures, see the science behind the report.

To measure productivity and on-device AI performance, we ran these benchmark tests:

• Cinebench 2024
• Geekbench AI
• Geekbench 6
• LM Studio
• Procyon^® AI Computer Vision Benchmark
• Procyon Office Productivity Benchmark

We also ran real-world tests to measure system performance while multitasking and to examine battery life under punishing workloads. Additionally, as the systems ran a sustained Cinebench 2024 workload, we measured their heat output. We also assessed each system’s privacy features, serviceability, and built-in AI assistants. Read on to learn more about these results and what they might mean to you.

Note: The graphs in this report use different scales to keep a consistent size. Please be mindful of each graph’s data range as you compare.

Productivity testing

Punch through your checklist

For multiple angles on potential productivity benefits, we ran several tests. To examine CPU performance, the Cinebench 2024 benchmark stresses the processor by rendering a 3D scene using CPU-intensive Redshift for Cinema 4D software.² To evaluate system performance, the Geekbench 6 benchmark executes multiple CPU-intensive tasks at the same time.³ Finally, Procyon Office Productivity Benchmark measures CPU performance around common office productivity tasks, reflecting a typical day at the office. It even leaves Microsoft 365 apps “running in the background as the focus moves from one task to another.”⁴

We also ran a real-world multitasking test with the Procyon Office Productivity Benchmark. In the first test, we ran Microsoft Teams and Windows Studio Effects in the background during benchmark testing, utilizing the systems’ NPUs. Then, in the second test, we ran only Microsoft Teams in the background, which did not utilize NPUs.

AI testing

Accelerate next-gen workloads

Geekbench AI measures on-device AI performance at different precision levels using real-world machine learning (ML) apps.⁷ The precision levels reflect different requirements for AI models: Full Precision (FP32) is more accurate but requires more resources, Half Precision (FP16) is less accurate but more efficient, and the Quantized (INT8) precision level is the least accurate but most resource-efficient.⁸ In our testing, we used the Open Neural Network Exchange (ONNX) AI framework and DirectML AI backend for machine learning on Windows.

In AI testing on CPU, the HP EliteBook 8 G1a 14 inch Notebook Next Gen AI PC outperformed the competition (see Figure 6). While its integrated NPU handles AI work with up to 50 TOPS, these results indicate its processor offers high performance for AI apps that don’t yet utilize the NPU.

LM Studio uses large language models (LLMs) to evaluate the AI chat capabilities that many of us use every day.⁹ In our testing, we ran the test on CPU with the Meta-Llama-3.1-8B-Instruct-Q4_K_M model to measure the time to first token. Tokens are “the basic units of input and output in a language,” typically words, subwords, or characters.¹⁰ The LLM predicts the most likely token to follow a sequence of input tokens and generates valuable output. The less time users must wait for the first token—or the first part of their answer—the smoother their experience will feel.

For more on using LLMs locally on the devices, see Built-in AI assistant assessment: Accurate answers on device and from the cloud.

UL Procyon AI Computer Vision Benchmark measures AI inference with various engines.¹³ In our testing, we used the API optimized for each system’s CPU, GPU, and NPU backends: the AMD Ryzen^™ AI API on the AMD processor-based system and the Intel OpenVINO^™ inference API on the Intel processor-based systems. The AI inference engines and their use cases include:

• MobileNetV3 and Inception-v4: Research institutions, tech companies, and individuals use these models for image recognition, object detection, and image classification tasks.^{14, 15}
• YOLOv3: Video surveillance companies, healthcare providers, and manufacturers use these Deep Neural Network (DNN) architectures to distinguish between different objects and features within images and videos.¹⁶

In integer-optimized testing, we found the wait times were lowest on the MobileNetV3 model, which is designed for real-time applications on mobile devices.¹⁷

Battery life

Collaborate from anywhere

For our real-world battery life assessment, we set up a Microsoft Teams meeting for nine participants and measured how long the laptops’ batteries lasted. We used the Windows 11 Best power efficiency mode. In our tests, the HP EliteBook 8 G1a 14 inch Notebook Next Gen AI PC delivered more than a full workday’s worth of battery life and a strong power efficiency ratio.

Thermal tests

Stay cool without compromise

High-performing device and processor combinations can produce excessive heat during intensive workloads. If this sensation creates a distraction for users completing business-critical work, productivity and creativity could slow down.

The HP EliteBook 8 G1a 14 inch Notebook Next Gen AI PC includes HP Smart Sense to “optimize PC performance. Work without distraction with HP Smart Sense which makes automatic adjustments keeping your system cool and quiet and lets you enable performance mode when you need it most,” according to HP.¹⁹ The other PCs we tested also include built-in intelligent features: Dell Optimizer for the Dell Pro 14 Plus, and Lenovo Intelligent Cooling for the Lenovo ThinkPad T14s Gen 6. For our physical user experience testing, we turned on each device’s intelligent feature.

When we ran a Cinebench 2024 workload for 30 minutes, the HP EliteBook 8 G1a 14 inch Notebook Next Gen AI PC scored significantly higher under this intensive workload (see Figure 11). And, it achieved this score while staying 13.3°F cooler than the Dell Pro 14 Plus (see Figure 12). While the Lenovo ThinkPad T14s Gen 6 also ran cool, its performance score lagged behind. The results speak for themselves: The HP EliteBook 8 G1a 14 inch Notebook Next Gen AI remained cool under load without compromising on performance.

Serviceability tests

Sustainably save time, fleet wide

If your team can easily replace device components when the unexpected happens, your organization could cut down on repair costs and downtime. Serviceable devices are also a more sustainable option, enabling your team to avoid replacing entire systems if just one component needs servicing.

When we tested the three devices’ serviceability, we found that the HP EliteBook 8 G1a 14 inch Notebook Next Gen AI PC enabled us to service three components in less time than the competitors. It was the only device with two fans—and despite our replacing them both, we still did so in less time than the other PCs. It was also the only device that allowed us to replace the memory modules. While the minutes saved may seem minor in isolation, they can make a significant difference to budget and uptime across a fleet of devices.