Earlier this year, Intel Xeon server processor (codenamed Skylake) became generally available on Google Compute Engine, providing you with the most powerful and technically advanced processors in the cloud. Paired with Compute Engine, Skylake benefits from finer-grained controls over your VMs, the ability to select your host CPU platform for any of our predefined and custom machine types, and new machine types that extend to 96 vCPUs and 1.4TB of memory per instance.
And now, we offer Skylake in our South Carolina, Mumbai and Singapore regions, joining Iowa, Oregon, Belgium and Taiwan, and bringing the total number of GCP regions with Skylake to seven globally. We’re also lowering the cost of Skylake VMs by 6-10 percent, depending on your specific machine configuration. With this price drop, we’re making it easier for you to choose the best platform for your applications. Just select the number of cores and amount of RAM you need and get all the computational power that Skylake on Compute Engine makes possible.
Already, what you’ve done with this additional computational power has been amazing! In the last six months, thousands of Compute Engine customers have used Skylake to run their applications faster, to achieve better performance, and to utilize new instruction sets like AVX512 to optimize their applications. Here’s what a few customers have to say about taking advantage of Compute Engine’s Skylake processors.
Alpha Vertex develops cognitive systems that provide advanced analytical capabilities to the financial community. Using compute engine 64-core machine types with Skylake for their ML systems allowed them to cut on training times for machine learning models.
“Using Google Cloud Platform, our Machine Learning (ML) training times have improved by 15 percent. We were able to build a Kubernetes cluster of 150 64-core Skylake processors in 15 minutes.”
— Michael Bishop, CTO, Alpha Vertex Inc
Milk VFX runs thousands of preemptible cores using one of the larger machine types (n1-highcpu-96) to create innovative and complex sequences for high-end television and feature films. At this scale, better performance means decreasing the runtime by days. With 96 vCPUs instances and preemptible machines, they were able to reduce the number of nodes they needed, and decrease their costs.
“By using Skylake with larger core machines we’ve been able to process more data faster, enabling our artists to be more productive, creative and cost effective. With preemptible machines we’ve cut the cost even more, so much so that we’re already seeing savings made in such a short timeframe. More importantly, for the past 12 weeks since we started rendering all 3D on the GCP, we have met our deadlines without any late nights or weekend work and everyone is really happy.”
— Dave Goodbourn, Head of Systems, Milk Visual Effects
QuantConnect breaks down barriers to algorithmic trading by providing market data and a cluster computer so any engineer can quickly design an algorithmic trading system. They’re constantly seeking the latest infrastructure innovations on Compute Engine.
“Our work at QuantConnect is constantly pushing the boundaries of cloud computing. When we learned of the addition of Skylake processors to the Google compute platform, we quickly joined as one of the early beta testers and converted infrastructure to harness it. The Skylake vCPUs improved our web-compiler speeds by 10 to 15 percent, making a measurable improvement to our user coding experience and increasing user satisfaction overall.”
— Jared Broad, Founder, QuantConnect
We’re committed to making all our infrastructure innovations accessible to all Compute Engine customers. To start using Skylake processors in Compute Engine today, sign up for a new GCP account and get $300 in free trial credits to use on Skylake-powered VMs.
Source: Skylake processors now available in seven regions
除非特别声明，此文章内容采用知识共享署名 3.0许可，代码示例采用Apache 2.0许可。更多细节请查看我们的服务条款。