estimated reading effort: 3 minutes
In fact, it’s been over two and a half months since I video chatted with Google’s Charlie Sheridan and Intel’s Ricky Watts. The main topic at that time was: How can one imagine the cooperation between the two companies and what can be expected in this regard in the future. This blog post is about both.
Of course, the Intel and Google collaboration has a lot to do with edge and cloud computing. Both are increasingly used, especially in the construction environment, as the vast mass of data is used in a targeted way, analyzed and used for new insights and resulting decisions. Under these aspects, a powerful cloud platform that combines both is necessary: the shortest possible processing cycles, combined with powerful cloud technology that also deserves the name.
Google Manufacturing Data Engine and Connect for unified data analytics
For these and other purposes, Google has developed a Manufacturing Data Engine, including Manufacturing Connect. This is primarily intended to allow production developers to understand the generated data faster and better. This will lead to production systems that are as trouble-free and optimized as possible and offer the highest level of quality. Here artificial intelligence algorithms are used that process data in real-time and make it available directly to edge endpoints. At the same time, more than 250 machine protocols are available with Manufacturing Connect, with which edge and cloud instances can be connected to the respective manufacturing plant with little effort.
But Google’s solution is also used to ensure quality by evaluating engine sensor data in real time. This allows faster and more reliable detection of possible machine faults. The underlying methods can also be used for planned and predictive maintenance. And it goes without saying that Intel plays a key role behind the scenes. Finally, in many Google Cloud cases, the scalable Intel Xeon processor provides the required computing power, both when evaluating data and when running artificial intelligence algorithms.
Intel and Google enable cloud-based HPC
But it continues. In addition to joint efforts in the industrial environment, a partnership announced in early July targets the “high-performance computing” industry. Because here too there is great potential in terms of analyzing and delivering HPC workloads. For this reason, Google has developed the Cloud HPC Toolkit, which features various Intel software components. This includes parts of the Intel oneAPI toolkit and Intel Select Solutions for Simulations & Modeling. This allows necessary and required HPC use cases to be adapted to the appropriate Google Cloud instance with relatively little effort.
The Intel oneAPI toolkit provides standard programming tools such as Intel MPI Library, Intel oneAPI Math Kernel, C++. SYCL, Fortran, OpenMP, MPI and Python available. This allows the required HPC algorithms to be adapted uniformly, regardless of the underlying hardware. Because regardless of whether HPC workloads are to run on CPUs, GPUs or other silicon components, with Intel oneAPI uniform approaches are available for all components.
The Google HPC Toolkit offers modular HPC environments
What’s special about the Google HPC Toolkit is its modular structure. This allows the required HPC workloads to be built using a single plan and migrated to the cloud instance of choice on that basis. This is exactly where Intel oneAPI comes into play, since ultimately it doesn’t matter which CPU or GPU your cloud presence is based on.
Denial of responsibility: I was commissioned by Intel to write and publish this blog post. I had almost a free hand in designing the content.