Heres a couple:*Researchers have noted challenges in accurately...

Heres a couple:

*Researchers have noted challenges in accurately assessing the hardware performance of Intel's Loihi neuromorphic processors when accessing them via cloud-based platforms. This is primarily due to the shared nature of cloud resources, which can introduce variability and limit direct control over hardware-specific parameters.

For instance, a study titled "Performance Evaluation of Neuromorphic Hardware for Onboard Satellite Applications" highlights that, during their research, primary access to Loihi 2 was through Intel’s Neuromorphic Research Cloud. The authors mention that remote access introduced limitations, as it was not possible to control for concurrent usage by other users, potentially affecting performance evaluations. arXiv

Similarly, the "Access Intel Loihi Hardware" documentation from Intel's Neuromorphic Computing Lab notes that most members access Loihi hardware via the Neuromorphic Research Cloud, a shared pool of virtual machines and Loihi systems. While this setup facilitates broad accessibility, it may not provide the same level of performance consistency as dedicated hardware, potentially impacting precise performance measurements. Intel NCL

These observations suggest that while cloud-based access to Loihi hardware offers convenience and broad accessibility, it may pose challenges for researchers aiming to obtain consistent and accurate hardware performance metrics due to factors like shared resource contention and limited control over hardware-specific configurations.

*GPT4o