Research has evolved over the years. More and more, we are dealing with theoretical models and simulations that vastly change our understanding of the world around us. In fact, these models have spurred some of the innovations and breakthroughs that have been and continue to change the face of many industries. However, one of the biggest challenges in achieving these breakthroughs is access to high-performance computing. With limited stipends and funds, researchers have found themselves unable to afford dedicated high-performance computing machines. AMD has been addressing that with their High Performance Compute Fund.
Source: AMD
What initially started as a means to drive timely breakthroughs in the midst of the COVID-19 pandemic is now being expanded to more applications. Keeping with the social impact factor of research, the HPC Fund is being expanded to include more fields in medical research and social good. The expansion will, hopefully, drive more breakthroughs in various fields for the betterment of the global society.
In addition to the expansion of its scope, AMD’s recent acquisition of Xilinx is also expanding the technology offered in the program. In addition to access to AMD’s EPYC Processors and Instinct Accelerators, researchers will now be able to make use of the Xilinx Heterogeneous Accelerated Compute Clusters (HACC). This gives them access to Xilinx ALveo accelerators and Xilinx Versal ACAPs. The newly integrated technologies expand the potential applications and decrease the time needed to run complex models.
As of May 2022, AMD has contributed over 20 petaflops of computing capacity through the High Performance Compute Fund. That’s over USD$31 million in monetary value. Researchers in over 28 institutions in eight countries have benefitted from the HPC Fund. Access has been extended to nearly 6,000 researchers resulting in 55 published research papers covering key issues including the disparities in COVID-19 Vaccination rates by race and ethnicity and improvements in the classification of breast cancer imagery using deep vision techniques.
This includes the National University of Singapore (NUS) where AMD has donated AMD EPYC processors and AMD Radeon Instinct Accelerators to the School of Computing. According to Eric Han, a fourth year PhD student at NUS Computing, “We are studying highly complex Machine Learning techniques that require us to run multiple compute and memory-intensive experiments. The AMD cluster has a high core count with large RAM, allowing us to run multiple experiments across the cluster.”
AMD is granting cloud-based access to universities and research institutions around the world with support from leading system partners Supermicro and WEKA.io. Applications are now being accepted. Research institutions and universities can apply for the new round of computing power grants by submitting their application on their official HPC Fund Website.
AMD has been on a roll with a multitude of releases and powerful innovations that are changing the landscape of the semiconductor industry. AMD is a company that has made its bets 5 years ago in high-performance computing and is now starting to see them come to fruition. With the announcement of their RDNA 2 platform and Zen architecture, they are delivering more performance to consumers for less power. These innovations continue to create waves in the industry with the company seeing its technologies adopted in gaming consoles to cars.
AMD’s recent announcements have proven that the company is making strides; not only in the consumer and gaming space but also in high-performance computing. The company has even announced their take on graphic supersampling with the FidelityFX Super Resolution. So, what’s next for AMD?
We’re sitting down with Paige Shi, Consumer Lead for Asia Pacific over at AMD to find out what’s happening at AMD and how the company is adapting to the new normal, how it has affected consumer behaviour and also what we can expect from the company in the future!
Hot off the heels of AMD’s announcement of their new EPYC processors, Dell Technologies has revealed a series of new offerings that put the AMD EPYC front and centre. The new Dell EMC servers come optimised for multiple workloads to help businesses better cater to their needs. They up the ante with up to 64 cores with EPYC’s new Zen 3 architecture.
The new additions add to Dell Technologies’ already robust portfolio of Dell EMC PowerEdge offerings. They bring updated technologies with improved compute capabilities. The expanded portfolio allows for better handling of critical workloads and applications augmented by fast data performance thanks to the integration of PCIe Gen 4 technology. In addition, the new offerings in the Dell EMC PowerEdge portfolio come with up to six accelerators to help with large, challenging, data-intensive workloads. They continue to put data security and redundancy at the core with a well-established Root of Trust and what Dell Technologies refers to as a Cyber resilient architecture.
Of course, running on the Zen 3 augmented EPYC processors, the new PowerEdge servers are more power-efficient than before. Dell Technologies claims that the new racks offer up to 60% power efficiency compared to the previous generation. Adding to its power efficiency, Dell has equipped the new generation PowerEdge with multi-vector cooling. This technology helps direct airflow to the hottest portions of the server; helping maintain lower temperatures and better performance.
The new line up consists of six offerings catered for different workloads and computing environments. From powerful, performance-oriented racks to AI optimised racks to deceivingly powerful, slim racks, Dell’s new offerings have you covered. Their Dell EMC PowerEdge XE8545 marries the performance of AMD’s new EPYC cores with NVIDIA’s A100 GPUs for powerful, AI and machine learning optimised workloads. The PowerEdge R6515 packs a configurable, dual-socket setup in a 1U rack server for compact performance while the R7515 brings a scalable single-socket 2U rack with performance and affordability in mind. They’ve got the C6525 for high performance, dense computing environments while the R7525 and R6525 bring extended flexibility.
Just when we thought AMD is done with being the best at gaming, they proved that one title is never enough. They want to retain their title as the most powerful ever, with a full-stop right after. So, they went ahead and introduced what seems to be the most powerful processing chips in the world, the EPYC 7003 series server processors.
The new EPYC 7003 series is the third-generation server processors by AMD. That also means that it gets the Zen 3 architecture that the AMD Ryzen 5000 series line-up got. That same architecture also means that AMD’s processing core clusters gets to access all of the available cache memory module when necessary, instead of just half of it. Again, if you were doing some research with AMD’s latest Ryzen 5000 with Zen 3 architecture, it might sound like something simple and small. The results are quite significant though.
At least in terms of a server, Zen 3 architecture also allows the cores to process more data at any single time than before. In practical terms, that also means a much faster processor compared to before. Of course, AMD’s implementation also means more efficient power consumption.
The headline figures for EPYC’s flagship is up to 64 cores and 128 threads. With 64 cores, the processor features 256MB of cache, that is the first indication that this is not your regular consumer gaming PC stuff. All cores are clocked at 2.45GHz and can be boosted up to 3.675GHz when necessary.
Of course, these chips are also embedded with extra security measurements and algorithms. In a server, data center, cloud seservice, and virtualisation environment, protecting your data is of utmost importance. Security, in this case, has to start from the processors itself with AMD’s own Infinity Guard suite encryption technologies to keep everything it processes secure.
Of course, you are not going to see the AMD EPYC processors out on the street. These are highly specialised processors made for Artificial Intelligence implementation, data centers, and data backbones for larger organisations. These are the stuff that keeps thigs like Google Cloud, Microsoft Azure, and Amazon Web Services running for you.
The AMD EPYC processors range from 32 cores to 64 cores. The EPYC processors are usually the base for AMD’s Ryzen Threadripper platforms. There are obviously no announcements made yet on the availability of the prosumer processor platforms at this time. If they are launching a Threadripper soon though, the AMD EPYC processors could probably give you an insight to what you can expect.
The world is quickly changing and technology is moving forward at a pace which hasn’t been seen before. The main impetus behind the progress which we experience is the internet. The internet has changed the way we communicate and experience things. In fact, it’s made the world even smaller. However, the way we connect to it needs to continue evolving to keep up with the pace of innovations. We’re talking about low latency connectivity and processing.
Over the years, we’ve managed to push the current connectivity to the edge when it comes to connectivity. With the advent of 5G and more importantly, the Internet of Things (IoT), there is an increasing demand for greater compute power and quicker outcomes. This is where Amazon Web Services (AWS) has stepped up with their new AWS Outposts technology. While Amazon isn’t the first name you’d necessarily think of when it comes to internet technologies, the are one of the foremost companies in the world when it comes to providing cloud infrastructure.
Opening Up New Possibilities with Low Latency Hybrid Computing
AWS Outposts allows users and developers to have low latency or local data processing on premises. This enables companies to have near real time responses when using or interfacing with AWS’s many services. It also bridges the lag in communications when the service needs to rely on the cloud. This also opens up larger possibilities to automate and regulate processes in new places such as the factory floor or in production outfits that require low latency. AWS Wavelength further strengthens the low latency with single digit latencies and enables 5G connectivity.
The near instantaneous latency between data lakes and data points on premise allow companies to adapt and react to situations better. Take for instance in the case of healthcare and medicine, having AWS Outposts on site would allow real time diagnostic data to be communicated to physicians for them to be better informed when they make critical treatment decisions. It can also open avenues for real time robotics in surgery. Perhaps more relevant to us as consumers, AWS Outposts also allows more seamless setups when we look into things like streaming (Netflix and Amazon Prime Video) or event real-time cloud gaming. Even AR and VR will expand with the advent of such a powerful architecture. Effectively, the availability of these hybrid cloud solutions will effectively open new doors in every industry.
Seamless Integration with the Cloud
AWS Outposts doesn’t necessarily work alone. In fact, the service will allow companies to quicken their eventual move to the cloud. AWS believes that even with the hybrid computing approach now, “in the fullness of time, the vast majority of companies will run almost all of their IT workloads in the cloud“. With that in mind, they’ve designed AWS Outposts to work seamlessly with their existing cloud services. Companies are able to adapt and experiment with workloads on the cloud and over Outposts for the best outcomes. What’s more, the AWS Outpost rack can be customised to meet the unique needs of the customer – this includes the recently announced AMD EPYC powered instances.
Even when it comes to the user interface, AWS Outposts share the same interface with AWS cloud services. Users won’t need to relearn a new interface to take advantage of AWS Outposts. In fact, they can use the exact same interface for both on-premises and cloud. This also applies to services that are running atop AWS such as VMWare’s on premise solutions. The only difference would be that they will still be able to access their on-premises AWS Outposts server even when internet connectivity goes down. This also serves as a redundancy that will keep local workloads running. Once internet is available again, the servers will seamlessly synchronise with the cloud where necessary. Prolonged outage, though, may cause some data loss.
That said, Outposts doesn’t require the fastest internet to be able to function. In fact, AWS states that the minimum required is a 10Mbps connection. However, they advise to have 1Gbps for better performance.
Innovating Securely & With Peace of Mind
Bringing AWS Outposts to your premises can be daunting when you think of it; you’ll be having a server rack in house. So what happens if something goes awry? Well, first off, you can bring down your blood pressure a little bit as AWS Outposts is a fully managed service which means that AWS will come and service the hardware on a regular basis just like they would any of their own data centers. It also means that if anything goes awry – you can simply call AWS and they’ll get it sorted for you.
Servicing and maintenance aside, with the Outposts possibly handling a large amount of sensitive data, it wouldn’t be surprising to worry about data security. AWS reassures us that the AWS Outposts is safe. In fact, their approach to security starts with the rack which comes equipped with built-in tamper detection and a lockable door. The door is outfitted with a special key that, when tampered with, automatically encrypts the data within the server; making life a miserable mess for anyone who tries to take it.
On top of that, AWS Outposts builds on AWS Nitro – a security system for AWS which continually monitors, protects and verifies the hardware and firmware of the setup. It also secures and locks down instances running on the Outpost to ensure that human error is reduced. However, AWS does stress that the security of AWS Outposts as a whole is a shared effort where it can only secure the hardware and instances being run; but when it comes to the applications and software being developed on the platform, it needs to be secured by the developers.
That said, AWS Outposts is an impetus that could lead to further developments that will spur more advancement. It brings the power of the AWS cloud on-premise for corporations needing low latency connectivity to their compute and also bridges the transition for companies which are still struggling to adopt the cloud fully.
This article was written based on an interview with Paul Chen, Head of Solutions Architect for ASEAN at Amazon Web Services (AWS).
Amazon Web Services (AWS) is one of the most prolific web service platform in the world. In fact, it’s estimated that over half of the world’s small and medium businesses have adopted the technology platform as their platform of choice when it comes to dealing with cloud services for their needs. AWS provides one of the most diverse platforms supporting Artificial Intelligence, Machine Learning and even rudimentary data storage. They provide their different services as deployable modules which allow companies to deploy and terminate instances as they need to.
AMD is one of the latest additions to the platforms array of instance which can be deployed. The new AMD EPYC instances will be powering the Amazon Elastic Compute Cloud (EC2) C5a instances. These instances will be powered by AMD’s 2nd generation server grade EPYC processors which also power one of the largest supercomputers in the world. The AMD EPYC processors will be able to run at frequencies of up to 3.3GHz and will be able to provide users with high performance x86 processing for large compute workloads. This includes batch processing, distributed analytics, data transformations, log analytics and web applications.
The new EPYC powered C5a instance joins the increasing number of AMD powered instances available on AWS. It will be available in eight configurations with up to 96 virtual CPUs (vCPUs). The new AMD EPYC powered instance also delivers on AMD’s promise of being able to deliver high performance compute at affordable prices. On AWS, the new C5a instance is the one of the lowest cost per x86 vCPU in Amazon’s portfolio.
The C5a instance is already available in AWS U.S. East, AWS U.S. West, AWS Europe and AWS Asia Pacific regions. AMD also has five other instances already available on AWS under the EC2 catalogue: M5a, M5ad, R5a, R5ad and T3a. These instances also provide users with compute capabilities that are catered to their needs and price points.
