Digital Design Automation (EDA) workloads require excessive computing efficiency and a big reminiscence footprint. These workloads are delicate to quicker CPU efficiency and better clock speeds because the quicker efficiency permits extra jobs to be accomplished on the decrease variety of cores. At AWS re:Invent 2020, we launched Amazon EC2 M5zn cases which use second-generation Intel Xeon Scalable (Cascade Lake) processors with an all-core turbo clock frequency of as much as 4.5 GHz, which is the quickest of any cloud occasion.
Our clients have loved the excessive single-threaded efficiency, high-speed networking, and balanced memory-to-vCPU ratio of EC2 M5zn cases. They’ve requested for cases that can leverage these options whereas additionally offering them a larger reminiscence footprint per vCPU.
In the present day, we’re launching Amazon EC2 X2iezn cases, which use the identical Intel Xeon Scalable processors as M5zn cases, with an all-core turbo clock frequency of 4.5 GHz and as much as 1.5 TiB of reminiscence, which is the quickest of any cloud occasion for EDA workloads. These cases are able to delivering as much as 55 p.c higher price-performance per vCPU in comparison with X1e cases.
X2iezn cases provide 32 GiB of reminiscence per vCPU and can help as much as 48 vCPUs and 1536 GiB of reminiscence. Constructed on the AWS Nitro, they ship as much as 100 Gbps of networking bandwidth and 19 Gbps of devoted Amazon EBS bandwidth to enhance efficiency for EDA functions.
You may need seen that we’re now utilizing the “i” suffix within the occasion sort to specify that the cases are utilizing an Intel processor, “e” within the memory-optimized occasion household to point prolonged reminiscence, “z” which signifies high-frequency processors, and “n” to help increased community bandwidth as much as 100 Gbps.
X2iezn cases are VPC solely, HVM-only, and EBS-Optimized, with help for Optimize vCPU. As you possibly can see, the memory-to-vCPU ratio on these cases is identical as that of previous-generation X1e cases:
|Occasion Identify||vCPUs||RAM (GiB)||Community Bandwidth (Gbps)||EBS-Optimized Bandwidth (Gbps)|
|x2iezn.2xlarge||8||256||As much as 25||3.170|
|x2iezn.4xlarge||16||512||As much as 25||4.750|
Many purchasers will be capable to profit from utilizing X2iezn cases to enhance efficiency and effectivity for his or her EDA workloads. Listed below are some examples:
- Annapurna Labs examined the X2iezn cases with Calibre’s Design Rule Checking, which has proven a 40 p.c quicker runtime in comparison with X1e cases, and a 25 p.c quicker runtime over R5d cases.
- Astera Labs is a fabless, cloud-based semiconductor firm growing purpose-built CXL, PCIe, and Ethernet connectivity options for data-centric methods. They have been in a position to see efficiency features of as much as 25 p.c in comparison with related EDA workloads operating on R5 cases.
- Cadence examined the X2iezn cases utilizing their Pegasus True Cloud characteristic, which permits designers to run bodily verification jobs on the cloud and noticed a 50 p.c efficiency enchancment over R5 cases. They see X2iezn cases as a wonderful atmosphere for testing EDA workloads.
- NXP Semiconductors labored with AWS to run their Calibre and Spectre workloads on Amazon EC2 X2iezn cases, which measured 10-15 p.c increased efficiency utilizing X2iezn cases in comparison with their on-premises, Xeon Gold 6254 with max turbo frequency of 4.0GHz.
- Siemens EDA labored with AWS to check the brand new Amazon EC2 X2iezn HPC/EDA targeted cases with the trade efficiency and sign-off chief Calibre evaluating superior node DRC workloads. They have been happy to exhibit efficiency enhancements of as much as 14% utilizing the 4.5 GHz all core turbo frequency of X2iezn cases for all VMs within the run. Moreover, they efficiently demonstrated using a heterogeneous server configuration utilizing the X2iezn as the first node and different decrease reminiscence VMs for distant compute – offering an 11% pace up and enticing worth. These outcomes confirmed the X2iezn is an efficient match for main server EDA workloads for Calibre Bodily and Circuit verification functions.
- Synopsys IC Validator supplies extremely scalable high-performance bodily verification signoff. They achieved 15 p.c efficiency enchancment, scalability to 1000s of cores, and 30 p.c higher effectivity utilizing IC Validator’s distinctive elastic CPU administration know-how versus R5d cases.
Issues to Know
Listed below are some enjoyable details in regards to the X2iezn cases:
Optimizing CPU—You’ll be able to disable Intel Hyper-Threading Know-how for workloads that carry out properly with single-threaded CPUs, like some HPC functions.
NUMA—You may make use of non-uniform reminiscence entry (NUMA) on x2iezn.12xlarge cases. This superior characteristic is value exploring when you’ve got a deep understanding of your software’s reminiscence entry patterns.
Amazon EC2 X2iezn cases are actually out there within the US East (N. Virginia), US West (Oregon), Asia Pacific (Tokyo), and Europe (Eire) Areas. You should utilize On-Demand Cases, Reserved Cases, Financial savings Plan, and Spot Cases. Devoted Cases and Devoted Hosts are additionally out there.