HPC

Who needs a supercomputer when you can get a couple of petaflops on AWS?

Or you could run it on-premises three times cheaper says University of Cambridge bod


Descartes Labs, an outfit that analyses big data, has managed to nab the 136th spot on the top 500 list of the world's fastest publicly known supercomputers – with $5,000 and an Amazon Web Services (AWS) account.

The AWS "supercomputer" has 41,472 cores, 157,824 GB RAM, and achieved 1,926.4 TFlop/s using the LINPACK benchmark, which solves a dense system of linear equations. That is a long way short of the 148,600 TFlop/s posted by the number one supercomputer, Summit at the Oak Ridge National Laboratory in the US, but remarkable for pay-as-go computing on the public cloud.

"Our team merely followed the standard steps to request a 'placement group', or high-network throughput instance block, which is sort of like reserving a mini-Oakridge inside the AWS infrastructure," Descartes Labs commented. "We were granted access to a group of nodes in the AWS US-East 1 region for approximately $5,000 charged to the company credit card."

The system ran Amazon Linux 2 on normal EC2 instances and was managed by Descartes Labs CTO Mike Warren, using HashiCorp Packer to build automated machine images and MPI (Message Passing Interface) for parallel computing.

Fleet

AWS has featured in the Top500 before, its best performance being in 2013 when a submission from AWS itself achieved a rank of 64 with 484.2 TFlop/s. It could no doubt do better today especially as "we provisioned them before the instances were launched publicly, so no one else was on the fleet at the time," said Amazon's Deepak Singh on Twitter. However, Descartes Labs believes it has achieved the highest Top500 performance to date for a system on public cloud.

"We believe that true HPC applications will eventually migrate over to the cloud en masse," the biz said, though since the company specialises in a "data-refinery on a cloud-based supercomputer" such a declaration is hardly surprising. Cloud HPC does have many attractions, though, including relatively low cost of entry and the ability to burst capacity as needed.

Aurora supercomputer

Hey, US taxpayers. Filed your taxes? Good, good. $500m of it is going on an Intel-Cray exascale boffinry supercomputer

READ MORE

There is a cloud premium to pay, which is the margin AWS (in this instance) makes on the compute resources. You also have to factor the cost of getting your data into the cloud, which can be an issue.

Dr Paul Calleja, director of the University of Cambridge Research Computing Services, which operates Cumulus, 107 on the Top500, spoke to The Reg from the International Supercomputing Conference under way in Frankfurt. "The cost of running off-prem is significantly higher than the cost of running on-prem," he said. "With our cost models it's roughly 3x which is a big number when you are talking petascale."

Another issue is performance. "AWS may run LINPACK well, which is notoriously easy. I/O performance sucks on AWS. A more HPC-like infrastructure has RDMA (Remote Direct Memory Access) networks. So there's many issues with this. A big LINPACK number doesn't mean anything."

OpenStack

The Cambridge team has some bragging rights when it comes to I/O, occupying the world number 1 spot on the IO-500 list in the latest ranking.

Calleja is an advocate of hybrid cloud using OpenStack, an open source cloud computing API which his team is customising for this purpose. "It allows us to instantiate a on-prem cloud configuration within an off-prem commercial cloud. So you would use off-prem if you really needed to boost must larger than your in-house infrastructure, with time to solution as a commercial driver. If time to solution does not have a commercial driver you might as well just wait and run it on-prem three times cheaper." ®

Bootnote

Another approach to democratising HPC is this up-to-3,000 core Raspberry Pi system which you can set up without a data centre. Performance is not that great, but it costs a lot less to test your code on this before renting space on public cloud.

600 juicy cores of Raspberry Pi

Juicy cores of Raspberry Pi in a box

Similar topics

Narrower topics


Other stories you might like

  • Elasticsearch server with no password or encryption leaks a million records
    POS and online ordering vendor StoreHub offered free Asian info takeaways

    Researchers at security product recommendation service Safety Detectives claim they’ve found almost a million customer records wide open on an Elasticsearch server run by Malaysian point-of-sale software vendor StoreHub.

    Safety Detectives’ report states it found a StoreHub sever that stored unencrypted data and was not password protected. The security company’s researchers were therefore able to waltz in and access 1.7 billion records describing the affairs of nearly a million people, in a trove totalling over a terabyte.

    StoreHub’s wares offer point of sale and online ordering, and the vendor therefore stores data about businesses that run its product and individual buyers’ activities.

    Continue reading
  • D-Wave opens up access to small-scale Advantage2 quantum computer
    Via a cloud subscription, natch – this is the 2020s

    D-Wave Systems has put its next-generation Advantage2 quantum computer into the cloud, or at least some form of it.

    This experimental machine will be accessible from D-Wave's Leap online service, we're told. We first learned of the experimental system last year when the biz revealed its Clarity Roadmap, which includes plans for a gate-model quantum system. Advantage2 sports D-Wave's latest topology and qubit design that apparently increases connectivity and aims to deliver greater performance by reducing noise.

    "By making the Advantage2 prototype available in the Leap quantum cloud service today, the company is providing an early snapshot for exploration and learning by developers and researchers," D-Wave said in a canned statement.

    Continue reading
  • Price hikes, cloud expansion drive record datacenter spending
    High unit costs and fixed capex budgets propelling enterprises cloudwards

    The major hyperscalers and cloud providers are forecast to spend 25 percent more on datacenter infrastructure this year to $18 billion following record investments in the opening three months of 2022.

    This is according to Dell’Oro Group research, which found new cloud deployments and higher per-unit infrastructure costs underpinned capex spending in Q1, which grew at its fastest pace in nearly three years, the report found.

    Datacenter spending is expected to receive an additional boost later this year as the top four cloud providers expand their services to as many as 30 new regions and memory prices trend upward ahead of Intel and AMD’s next-gen processor families, Dell’Oro analyst Baron Fung told The Register

    Continue reading
  • Germany to host Europe's first exascale supercomputer
    Jupiter added to HPC solar system

    Germany will be the host of the first publicly known European exascale supercomputer, along with four other EU sites getting smaller but still powerful systems, the European High Performance Computing Joint Undertaking (EuroHPC JU) announced this week.

    Germany will be the home of Jupiter, the "Joint Undertaking Pioneer for Innovative and Transformative Exascale Research." It should be switched on next year in a specially designed building on the campus of the Forschungszentrum Jülich research centre and operated by the Jülich Supercomputing Centre (JSC), alongside the existing Juwels and Jureca supercomputers.

    The four mid-range systems are: Daedalus, hosted by the National Infrastructures for Research and Technology in Greece; Levente at the Governmental Agency for IT Development in Hungary; Caspir at the National University of Ireland Galway in Ireland; and EHPCPL at the Academic Computer Centre CYFRONET in Poland.

    Continue reading

Biting the hand that feeds IT © 1998–2022