Introducing
Acceleration
_VIA ECCOE_
Introducing
Experiment with your favourite VFX software (Maya, Houdini, Nuke, Cinema4D) on Google Cloud high-performance GPU equipped workstations using Zero and Thin clients from Teradici
Observe a low-latency workstation design workflow using dual 4K monitors, full-colour accuracy & Wacom tablet
See how on-demand rendering works using Google Cloud VMs with your favourite scheduler (OpenCue, Deadline, Slurm etc) and renderer (Arnold, Renderman etc). Observe accelerated performance (30-40% faster) on Intel Xeon Scalable Processor hosted VMs.
Google Cloud Partner Interconnect for high speed, low latency direct connect to Google data centre
1 day workshop - to understand relevant Google Cloud services for your VFX workload and recommended next steps
7 day engagement - to perform a discovery of a customer’s requirements and build a working prototype that integrates Google Cloud services with a client’s existing studio / scheduler / rendering software
Solution Engineering - engage the ECCoE team to build production grade, Google Cloud hosted studio & rendering solutions integrated with a client’s on-premise infrastructure
Managed Services - 3rd line support for production solutions hosted on Google Cloud
Benchmarking - The ECCoE team can benchmark the performance of your VFX software on the latest CPU architectures on Google Cloud and the ECCoE servers (bare metal - latest Intel Xeon Scalable Processors)
During an ECCoE PoC, the team typically aims to demonstrate the technical solution shown below. Different filers and schedulers (including Deadline) can be substituted.
Google offers a variety of services that map well to VFX workloads. Cloud Storage offers an object store for VFX artefacts and different tiers for archival and cost savings. Compute Engine offers Virtual Machines (VMs) of any shape (core count and memory size) with a selectable CPU architecture (Skylake and Cascade Lake for performance). And all services are typically provisioned in a VPC (Virtual Private Cloud) linked to a customer’s on-premise network via Dedicated or Partner Interconnect.
Intel Xeon Scalable Processors (Skylake and Cascade Lake): The latest Xeon processors from Intel include the AVX512 instruction set for SIMD execution for a 30-40% performance boost for VFX workloads. Benchmarks run on Renderman and Cinebench R20 prove this out. Google Cloud is the first hyperscaler to offer Skylake and Cascade Lake to its customers.
Elastifile offers a scale-out NAS on Google Cloud for high-performance NFS. Aggregate performance of 30GB/s using Local SSDs and 17GB/s using PD SSDs is achievable with a multi-node Elastifile Clustered FileSystem (ECFS). Elastifile also offers ClearTier so that costs can be controlled by transparently splitting a Filesystem between high-performance Tier-1 storage (SSD backed) and a lower-cost Tier-2 Google Cloud Storage object store. The ECCoE team will demonstrate this in operation at any Virtual Studio engagement.
PCoIP (PC-over-IP) is a display protocol that encrypts and transports pixels to users devices with unrivalled quality. Teradici invented this protocol and it has become the industry standard for high-performance remote workstation visualisation. During an ECCoE engagement, the team will demonstrate using Google Cloud VMs with NVidia GPUs:
Google and Sony Pictures Imageworks collaborated to build OpenCue for Render management across existing on-premise VMs and dynamically provisioned Cloud VMs. OpenCue offers the following key benefits:
“Appsbroker has developed the ‘Extreme Cloud Centre of Excellence (ECCoE)’ to facilitate Google Cloud customers with the on-boarding of ‘Big Compute’ workloads. VFX customers can benefit from running their rendering and Virtual Studio assets in Google Cloud”