Getty Images

Here's the situation: You've got a website to launch, and you don't want to spend forever figuring out the best place to host it. You just want something that works well, keeps your site running fast, and won't break the bank. Is that too much to ask? We don't think so.

Let's face it: Picking the best web hosting company for your needs can feel like you're lost in a maze of tech jargon and endless options. How much is too much per month? What kind of server load can your provider handle? And what's your site's uptime going to be like? Don't sweat it. We've got your back.

We've sorted through the noise to bring you a no-nonsense list of the best web hosting companies in 2023. Whether you're setting up a small blog or launching the next online empire, we've found options for everyone. So sit back, relax, and read on to find the perfect fit for your website.

And if you're thinking about launching a Taylor Swift fan site or something, a word to the wise: You're going to want to spring for something that can handle all that traffic.

If you need to host a website but don't want to get lost in a maze of plans and tech jargon, HostGator is your go-to. It's perfect for anyone looking for something solid but straightforward. Starting at just $12 a month, its Linux-based Hatchling plan hooks you up with unlimited disk space, data transfers, and email, Want more bang for your buck? They've got "Baby" and "Busines"s plans that pile on the perks like unlimited domains and nifty SEO tools.

More of a Windows person? They've included an option for that OS, too. Starting at $10 a month, you get all the space and data transfers you could want. And if you're big on domains, the Enterprise plan lets you juggle up to five of them.

In terms of building your website, if you need help with that, the Gator website builder is a simple drag-and-drop interface that makes it simple to put together your new online home. Need something fancier, or have a soft spot for WordPress? You can run pretty much any content management system you want. And its File Manager tool isn't just a glorified FTP client. You can edit files without jumping through hoops.

Plus, HostGator has a pretty good offer that not all of the competition matches. You can sign up for a year and snag a free domain name. It's not groundbreaking, but hey, it's a freebie that can save you some additional money by not having to go through a third-party to register.

Plus, if you need help after getting started, the host's 24/7 customer service is fantastic, based on our testing. You're talking to a human in under a minute -- not a robot -- and the reps are quick and savvy. They'll make sure you have what you need, and the know-how to tackle whatever issues come your way.

So, if you're after no-nonsense web hosting that doesn't skimp on features, HostGator is where you want to be, especially if you're a beginner to web design or hosting in general.

Key features of Hostgator:

Looking for web hosting that offers a lot of features without a big price tag? Hostinger might be for you. It's super affordable, especially considering what's on offer.

For example, its "Business" plan is only $4 a month currently (down from $15 and includes a variety of features such as NVMe storage, daily backups, and free SSL. It even supports up to 100 websites on a single plan, which is great news if you have multiple projects that also need hosting options.

It's also surprisingly easy to use, even though Hostinger uses a custom control panel called hPanel instead of the industry-standard cPanel. This change might be a bit frustrating for anyone who's used cPanel in the past, but hPanel is surprisingly intuitive and user-friendly. It's great for anyone new to website management while serving up advanced features for the people who already know their way around it.

Hostinger's shared hosting options start at $1.99 a month with basic features. They also offer a Premium plan for $2.99 a month that significantly expands the capabilities, including support for up to 100 websites and a free domain.

It's important to keep in mind, however, that while Hostinger is a solid choice for smaller sites or personal projects, it's not the best choice for any bigger, busier, enterprise-level needs. For everyone else, though? It's a great investment.

Key features of Hostinger:

Bluehost is known for its ease of use and affordability. So whether you're hosting your hundredth website or you're getting started on your very first page, it's a great choice to kick things off with.

Like many of its hosting competitors, Bluehost offers you a free domain name that comes with your subscription, which can save you a small sum upfront. The hosting plans start at a reasonable $3 per month, and you get plenty of great features to help set up your website without much hassle.

New to building websites and need a little more help getting started? Bluehost's cPanel and customer support are super user-friendly. The company also offers 24/7 customer service, so you're never without help. It may not seem like the greatest offering in terms of perks, but you never know when you're going to need more help.

Bluehost also offers shared hosting, which is a great option for small sites. That service comes with the added benefits of free SSL certification and a free domain. But if you plan on sticking with it and growing your website, you may find that shared hosting doesn't meet your performance requirements. In that case, Bluehost has tons of more robust options, like dedicated hosting and VPS hosting for more flexibility and bandwidth.

Bluehost offers a good balance of price and features for anyone just getting started with building websites or anyone who's satisfied with a smaller host. If you need something more robust you'll have to shop elsewhere, but for everything else, Bluehost has you covered.

Key features of Bluehost:

DreamHost has carved out a space in web hosting, thanks to its reliability and affordable plans. Unlike some hosts that scale back features in their monthly plans, DreamHost keeps most of its offerings available for both monthly and yearly subscribers.

When it comes to reliability, DreamHost is hard to beat. The company is so confident in its uptime that it offers a 100% guarantee. If you have any downtime at all, you'll likely get a refund. There's also a super generous, 97-day money-back guarantee that applies to certain hosting plans, but it gives you the peace of mind to try the service and see if you like it before committing.

On the security front, DreamHost doesn't cut corners. It offers free SSL certificates and additional security protocols. All of these come wrapped in a user-friendly, custom control panel that differs from the industry-standard cPanel, but it's still pretty intuitive.

However, DreamHost isn't without its drawbacks. It does have fewer global server locations than some competitors. And DreamHost's servers are only based in Virginia and Oregon. This can be a major concern for anyone needing a global audience, since having servers closer to your visitors generally means faster load times. Also, unlike other providers that offer free email accounts, DreamHost charges extra for this, at least in its basic plans.

Customer support is an area where DreamHost excels, despite charging a bit extra for phone callbacks. However, it does offer robust email and ticket support with a live chat and an active online forum.

DreamHost does simplify the installation process for some popular content management systems (CMSs) like WordPress. But if you're planning to use a less common CMS, be prepared for a steeper learning curve. If WordPress has what you're looking for and more, you should be good to go.

All things considered, DreamHost offers a lot of bang for your buck, especially for those who value flexibility and robust features.

Key features of DreamHost:

GoDaddy goes beyond its well-known role as a domain registrar to offer a variety of hosting services. And of course, it has its brand recognition to lean on, too. It also has plenty of user-friendly features, especially its one-click installations for popular apps like WordPress. This is a major time-saver and eliminates the need for manual setup, making it approachable for both beginners and experienced developers.

GoDaddy's website builder streamlines the process further. You can pick from a variety of templates to kick-start their website, although this convenience might be a trade-off for those who want more creative control.

But where GoDaddy really stands out is in its customer service. In an era where automated bots are becoming the norm, GoDaddy offers 24/7 phone and web chat support. The quick response times are a big plus, often less than two minutes for phone support.

On the performance front, GoDaddy offers quick load times, thanks to multiple data centers across different continents. While they are planning to extend their reach with additional data centers, their current setup already provides solid global performance. And as you might be expecting, the service offers some pretty awesome extras, like a free domain for the first year and complimentary Office 365 email.

You might not think it based on the somewhat cringeworthy ads we've seen from the service over the years, GoDaddy remains a strong choice for web hosting, even if it isn't the cheapest one you'll run across.

Key features of GoDaddy:

Web hosting just means finding a place online to store and display your website. Your content is uploaded and published so that others can see it. Using a web hosting service essentially gives you space on a third-party server where you can store all the elements that make up your website. Consider your hosting provider as your website's landlord, looking after all the server maintenance and security issues. Often, hosts even throw in some extras, like email. In short, if you want your website to be accessible on the internet, you're going to need some type of web hosting.

First, you choose a domain name, which is basically your website's address. Now what? You have to link that address to server space that your hosting company provides. So, when someone wants to visit your website, they'll type in your domain or click a link to your site. That action sends a request through the internet to your server. Your server responds by sending back the files that make up your website. The end user's browser takes those files and puts together the website for them to see. You must have somewhere for all of your website's materials to reside if you want to publish it for the world to see.

There are several ways to get web hosting. Shared hosting is the budget option of the hosting world. You're sharing a server (and all its resources like CPU, memory, etc.) with other websites. It's the cheapest option, but you're also getting what you pay for: lower speed, less security, the works. Next up is VPS, or Virtual Private Server hosting. It's like having your own apartment in a building; you still share some amenities (the server), but you've got a dedicated partition that's all yours. It offers a bit more oomph in terms of speed and reliability.

Dedicated hosting is the penthouse suite, essentially, and the most expensive. You get an entire server to yourself -- lots of space, lots of resources, and lots of control -- but it'll cost you. Lastly, cloud hosting is a newer option. It's a network of servers that work together to host your website. You can start small and grow your hosting resources as your website gains traction, without the hassle of moving everything to a larger server.

Your first order of business is figuring out what you actually need from a web host. The type of website you plan to launch will determine your hosting requirements. Performance is another important factor. You'll typically want to aim for fast server speeds and a minimum uptime of 99.9%. Good customer support should also be high on your list. Go with the provider that gives you the best support available: That's typically 24/7 support across multiple channels, so you can get help whenever you need it.

Of course, as your website grows, your hosting needs could change. Pick a provider that allows for easy scalability without hefty fees or downtime. It might be super tempting to opt for a cheap plan, but remember that you usually get what you pay for. Balance the cost with the features you genuinely need. Security is non-negotiable when it comes to web hosting, so make sure the provider you go with offers robust features like firewalls and SSL certificates. Lastly, don't forget to read reviews and seek recommendations to ensure you're making an informed choice.

Brittany Vincent has been covering gaming, tech, and all things entertainment for 16 years for a variety of online and print publications. She's been covering the commerce space for nearly a decade. Follow her on Twitter at @MolotovCupcake.

See the original post:
The best web hosting companies in 2023 - CBS News

Feature It has been a decade since Amazon RDS launched support for PostgreSQL. Since then, the relational system authored by Turing Award winner Michael Stonebraker in the 1980s has gone on to become the most popular database among professional developers, used by nearly half of them, according to Stack Overflow's 2023 Developer Survey.

I've seen people seduced by the cloud provider, and they fire their DBAs and everybody who knows about databases, but then they figure out when they need schema design, and query optimization well, Amazon's not going to help them

In parallel to PostgreSQL's rise in popularity comes a bewildering array of ways to deploy the database system in the cloud or exploit PostgreSQL-compatible database services. For example, as well as hosting standard versions of PostgreSQL as in RDS, the major three cloud providers, including Azure and Google, also provide PostgreSQL-compatiable enhanced database services such Aurora and AlloyDB.

Meanwhile, some vendors have created serverless systems with PostgreSQL-compatible front ends, such as CockroachDB and Yugabyte.

And that's just PostgreSQL. Similar options are available for other popular database systems, including MySQL, MongoDB and MariaDB. To navigate these choices, developers and database administrators need to understand the strengths and weaknesses of each approach.

As the author of the MySQL performance bible and founder and former CEO of opensource database consultancy Percona, Peter Zaitsev has witnessed the rise of the various ways of deploying database in the cloud and cautions about making choices lightly.

Whether the users might want to manage their deploy in a VM or adopt a serverless system managed by a vendor will depend on how much work they want to do, how much control and flexibility they want to have and how much they can tolerate being locked into a particular vendor.

Added into the mix, the cloud vendors offer proprietary databases for specific workloads, for Amazon offers DynamoDB, a fully managed proprietary value-key database, while Google offers BigQuery, a fully managed, serverless data warehouse.

"These systems are only available if you buy them from a specific cloud vendor: you cannot run it on your own," Zaitsev said.

Alternatively, users can get a standard system based on a popular open source database like PostgreSQL or MySQL, but significantly enhanced and presented as a fully managed service like Amazon Aurora, and Google's AlloyDB.

Lastly, there are fully managed "shrinkwrapped" services based on MySQL or PostgreSQL, such as Googles Cloud SQL or Amazon's RDS.

"This is some standard database technology just with some GUI and interface on top of it and some automatic backups and stuff like that," Zaitsev said.

Going from first to last, users face the most lock-in to the least lock-in with each of these choices. But they should also question what cloud vendors mean by a "fully managed service."

"That is what the cloud vendors recommend to users and what they push them towards, and it also typically comes with the highest cost, because they charge more for that compared to just the basic infrastructure to run a database," he said.

"But when they talk about a fully managed services, you can ask, 'OK, who's responsible for performance or security?' And they come back and say 'This is a shared responsibility.' They expect you to do your part while they keep the environment up and running. That is often misunderstood. I've seen people seduced by the cloud provider, and they fire their DBAs and everybody who knows about databases, but then they figure out when they need schema design, and query optimization well, Amazon's not going to help them. Any cloud provider would turn around and say, 'Hey, guys, we are keeping the database up and running, but all that stuff, which is specific to application and database usage, is on you'," Zaitsev said.

Another challenge to using shrink-wrapped or enhanced database services from the cloud vendors arrives when users want to use systems across cloud infrastructure from different cloud providers, according to corporate policy or geographic limitations.

"Amazon RDS, for example, sounds simple until you have to run it in different clouds. Then you have to deal with the nuances of RDS and the cloud infrastructure as well, and then it becomes very complicated," Zaitsev said.

Users can manage database deployment in the cloud themselves using virtual machines, but the fastest growing approach to cloud deployment of database is via Kubernetes, the open-source container orchestration which originated with Google.

"It gives us a programmable infrastructure, which is much more flexible and advanced than you get just dealing with VMs. At the same time, can you run it on-prem and on all the clouds. Kubernetes has become much more mature and much more capable to run a database compared to the early stages when it was designed to be as solutions for stateless applications," Zaitsev said.

Into the throng of database options in the cloud, a group of vendors have begun offering serverless systems, which is their own back end, but a front end compatible with a common database. For examples, CockroachDB and Yugabyte both offer serverless database with PostgreSQL-compatible front end.

In June, Cockroach CEO and co-founder Spencer Kimball told The Register it took five years to port the serverless system to Azure, a "non-trivial amount of work" that involved understanding the tolerances and failure of a different cloud architecture.

While Yugabyte claims 100 percent compatibility with PostgreSQL, and MariaDB recently launched a PostgreSQL-compatible front end to its distributed MariaDB back end, Kimball admitted CockroachDB does not have full PostgreSQL compatibility, but it is getting there.

Users, however, should question what lies behind serverless databases, Zaitsev said. "There are really servers in the end, right? It is just you are not charged for them and you may or may not be aware about what is going on with the servers."

One approach to serverless was to scale the instance size up and down according to the load. Another was to offer a multi-tenant approach in Google Spanner or CockroachDB.

"They have a different idea. You have a distributed database which is shared by multiple tenants. The benefit of that approach is, you have more ready to use capacity, which can be dynamically shared. If you need more resources, you don't need to reallocate and spin up the larger instance size," Zaitsev said.

Serverless is convenient if the load is very irregular. Users do not pay for keeping a system up and running when it is not in use. On the other hand, if the system is well used, and the operator understands and can predict demands on the system, then it can become less valuable from a pricing perspective, he said.

Earlier this year, Gartner said the DBMS market grew by 14.4 percent in 2022, reaching $91 billion, with the cloud platform-as-a-service model capturing nearly all the gain, with cloud spend at 55 percent exceeding on-premises at 45 percent.

The progress to the cloud is slower than Gartner predicted in 2019, when it said by 2022 75 percent of all databases would be deployed or migrated to a cloud platform. Users seem to be taking their time to navigate the many options available to them in deciding their future database strategy.

Original post:
There are lots of ways to put a database in the cloud here's what to consider - The Register

Adobe is announcing the next set of enhancements to Creative Cloud for video, adding features to Premiere Pro, After Effects and Frame.io in this update cycle.

In May 2023, Adobe released text-based editing in Premiere Pro after a public beta cycle. It is important to note, as our Adobe press briefers reiterated to us, that text-based editing does not require an internet connection for text-to-speech transcription, which does set it apart from other applications that send audio to cloud servers. Adobe is adding Filler Word Detection to this feature in response to user requests. This means that Premiere Pro can detect pauses and so-called filler words like all the uhs and ums and, with one click, delete them. It is also possible to set a duration of pause to be detected and deleted. This also works across multi-track audio.

Enhance Speech removes noise from voice with just one click. AI technology can make voices sound like studio rather than field recordings. This feature too comes from user requests, in this case from the podcast community.

A frequent request from all users is timeline performance. Adobe claims up to a 5x faster performance in timeline thanks in good measure to clearing out considerable legacy code. I guess it is always good to clean out your closets occasionally.

Color sees improvements in automatic tone mapping with options for three new tone mapping methods. Settings in the Lumetri Color panel are also consolidated, and there are improvements to LUT management. But the most significant improvement in color is a fix to the dreaded Quick Time gamma differences. The sequence gamma can be set to match QT, so no more of that gamma 2.2 or 2.4 manual gyrations! Writers notethank you, Adobe.

Ever had a third-party plug-in crash the system? Effects Manager can detect a third-party crash, isolate that plug-in, and, on relaunching the application, recover your place automatically.

According to Adobe, here are a few more features based on community feedback. Metadata and TC can be burned in. New project templates, custom export across projects and export to Media Encoder are added. There is guidance in installing Blackmagic RAW, and perhaps the most significant feature for many of us, batch selection of markers is implemented.

While Premiere Pro is a mature product to which Adobe adds new functionalities, After Effects continues to add features.

Advancing After Effects 3D capabilities is True 3D Workspace for motion graphics. It is now possible to import, animate, light, shade and render 3D models regardless of the source of those 3D models. Users can combine 2D and 3D elements in this new workspace. A new GPU-accelerated rendering engine can deliver photo-realistic results. And through Creative Cloud libraries, Substance 3D assets are available for free.

Rotoscoping gets AI! Simply using the rotoscope tool, just draw a line along the object or subject and AI will figure out the rest. It is particularly effective for overlapping hair or limbs as well as transparent elements. Of course, manual adjustments can be made, but it does seem remarkably accurate in picking up details that could take hours of manual refinement pre-AI rotoscoping.

Adobe continues the development of Frame.io at a rapid pace in response to individual and enterprise-level users.

Frame.io now recognises audio, video, images or PDF assets, and these can be compared side by side, matched and annotated.

ProRes RAW and 10bit 4K workflows are now possible with Atomos Ninja and Ninja Ultra.

The newly announced 102MP Fujifilm GFX-100 II, also with 8K video, is now supported for camera to cloud at the high end, and at the other end, the Accsoon SeeMo and SeeMo Pro are supported for c2c.

Enterprise users constitute a significant part of the Frame.io base, and in direct response to the storage requirements of this segment, AWS S3 bucket can be connected directly to Frame.io. This represents both a cost-saving as well as workflow efficiency for enterprise users in c2c workflows.

All of these features will be released in public beta on September 13, 2023, with release versions coming at some point in Fall 2023.

Read this article:
New features for Premiere Pro, After Effects, and Frame.io from Adobe - RedShark News

French Infrastructure-as-a-Service (IaaS) provider OVHcloud is rolling out an exhaustive carbon calculator to its customers to help make it easier for them to track the Scope 1, Scope 2 and Scope 3 emissions generated by their cloud usage.

Eight months in the making, the tool is accessible via the OVHcloud customer panel, and according to the company will provide users with a granular level of detail about the carbon footprint of their cloud infrastructure.

It has also been co-developed with Sopra Steria, and its results are location-based, meaning it factors in the different energy mix that OVHcloud draws on to power its datacentres, which can differ depending on where they are situated.

The tool takes into account the estimated electrical consumption of servers from OVHcloud datacentre monitoring and maps them to their carbon equivalent, taking into account the cooling and networking equipment, as well as freight, manufacturing, end of life and waste management, to provide a complete picture of the actual carbon footprint, the company said, in a statement.

The company first went public with details of its planned carbon calculator tool in Spring 2023, with OVHclouds pledge that it would provide Scope 1, Scope 2 and Scope 3 emissions data from launch, resulting in it garnering comparisons with a similar tool from Amazon Web Services (AWS).

The latters take on the same technology launched in March 2022 has seen the public cloud giant come in for criticism for failing to provide a comprehensive enough view of the carbon footprint of its customers on account of the fact it does not track Scope 3 emissions.

In response, AWS confirmed to Computer Weekly in May 2023 that it was working to provide its customers with Scope 3 data from early 2024.

OVHCloud CEO Michel Paulin said that, coupled with the work the company has done to improve the energy and water efficiency of its datacentres, it has sustainability rooted in its DNA, and constantly challenges itself to improve the carbon footprint of its entire operations.

We are more than ever aware of the importance for our customers of calculating their carbon footprint as accurately as possible, he said.

We are therefore extremely happy to give them a precise reading and understanding of it, all with a single click of the mouse.

Fabienne Mathey-Girbig, executive director of corporate responsibility and sustainable development at Sopra Steria, said the calculator will enable businesses to easily understand the environmental impact of their cloud activities.

As a major tech player in Europe, we have a key role to play in supporting a more sustainable digital landscape across the entire value chain, including employees, clients, suppliers and partners, she said.

We are proud of the trust placed in us to play an active role in this decarbonisation initiative.

Link:
OVHcloud debuts comprehensive carbon calculator for customers - ComputerWeekly.com

MLCommons this week issued the results of its latest MLPerf Inference (v3.1) benchmark exercise. Nvidia was again the top performing accelerator, but Intel (Xeon CPU) and Habana (Gaudi1 and 2) performed well. Google provided a peak at its new TPU (v5e) performance. MLCommons also debuted a new MLPerf Storage (v0.5) benchmark intended to measure storage performance under ML training workloads. Submitters in the first Storage run included: Argonne National Laboratory (ANL), DDN, Micron, Nutanix, and Weka.

Digging through the latest Inference results more than 12,000 performance and 5,000 power inferencing results from 120 systems is a challenge. There were a more modest 28 results in the storage category. From a usefulness perspective, MLCommons provides direct access to results spreadsheets that permit potential system users/buyers to drill down onto specific system configurations and benchmark tests for comparison. (Links to Inference Datacenter and Edge v3.1 results and Storage v0.5 results)

In the past, HPCwire has tended to try to cover the full exercise in a single article. The rising number of results and introduction of a new category make this less tenable. Instead, well present a broad overview in this article and drill deeper into some vendor-specific results in separate articles (Nvidia and Intel/Habana). By now, you may be familiar with the MLPerf release cadence which is twice yearly for training and inference, with each released on alternate quarters. - so, inference results are released in spring and (early) fall; training results are released in winter and summer. The HPC Training benchmark is released just once yearly, close to the annual SC conference.

Broadly, inferencing and training are the foundational pieces of ML applications, with training deemed the more computational-intense of the two (i.e. think of training LLMs with trillions of parameters). Inferencing, though, is the volume workhorse, sitting behind every chatbot and similar applications.

MLPerf Inference v3.1 introduced two new benchmarks to the suite. The first is a large language model (LLM) using the GPT-J reference model to summarize CNN news articles; it garnered results from 15 different submitters, reflecting the rapid adoption of generative AI. The second change is an updated recommender, meant be more representative of industry practices, using the DLRM-DCNv2 reference model and larger datasets; it had 9 submissions. These new tests, say MLCommons, help advance AI by ensuring that industry-standard benchmarks represent the latest trends in AI adoption to help guide customers, vendors, and researchers, says MLCommons.

In a pre-briefing, David Kanter, MLCommons executive director, said, We added our first generation recommender a couple of years ago and are now updating it. The LLM (inference) benchmark is brand new and reflects the explosion of interest in what people are calling generative AI, large language models. An LLM had been added to the MLPerf Training benchmark in the spring (see HPCwire coverage, MLPerf Training 3.0 Showcases LLM; Nvidia Dominates, Intel/Habana Also Impress)

No ML benchmarking effort today would be complete without LLM coverage and MLCommon (parent organization for MLPerf) now has that.

Its important to understand large language models operate on tokens. A token is typically a piece of a word. An LLM simply takes a set of tokens as input and predicts the next token. Now, you can chain this together to actually build a predicted sentence. In practice, LLM s are used in a wide variety of applications. You can use them in search and in generating content, like essays or summaries. Summarization is what we do here, said Kanter.

The MLPerf LLM inference benchmark is quite different from the training benchmark, he emphasized.

One of the critical differences is the inference LLM is fundamentally performing a generative task. It's writing fairly lengthy sentences, multiple sentences, [but] its also actually a different and smaller model, he said. Many folks simply don't have the compute or the data to really support a really large model. The actual task we're performing with our inference benchmark is text summarization. So we feed in an article and then tell the language model to summarize the article.

As is MLCommons practice, submitting organizations are invited to submit brief statements on their submissions. These range in quality from pure marketing to providing more granular technical descriptions of a submission's distinguishing features. Given the high number of results, a fast review of the vendor statements can be informative in conjunction with consulting the spreadsheet.

Both Inference and storage submitter statements are appended to the end of this article. As examples, here are a few snippets from a few vendor statements in MLPerf Inference v3.1 exercise:

Azure promoted its online versus on premise showing access to H100 instances. Azure was the only submitter to publish results for virtual machines in the cloud, while matching the performance of on premises and bare metal offerings. This has been possible thanks to innovative technologies including: AI supercomputing GPUs: Equipped with eight NVIDIA H100 Tensor Core GPUs, these VMs promise significantly faster AI model performance than previous generations, empowering businesses with unmatched computational power; Next-generation computer processing unit (CPU): Understanding the criticality of CPU performance for AI training and inference, we have chosen the 4th Gen Intel Xeon Scalable processors as the foundation of these VMs, ensuring optimal processing speed."

CTuning Foundation, the non-profit ML tool developer, noted that it [delivered the new version of the open-source MLCommons CM automation language, CK playground and modular inference library (MIL) that became the 1st and only workflow automation enabling mass submission of more than 12000 performance results in a single MLPerf inference submission round with more than 1900 power results across more than 120 different system configurations.

Google touted its new TPU v5e. TPU v5e systems use multiple accelerators linked together by a high-speed interconnect and can be configured with a topology ranging from 1x1 to 16x16 (256 chips), giving the user the flexibility to choose the system that best meets their needs. This wide range of topology options offered by TPU systems allows users to run and scale AI inference workloads cost-effectively, without compromising on performance."

In this submission, Google Cloud used a TPU v5e system with a 2x2 topology (4 TPU chips) to run the 6-billion-parameter GPTJ benchmark. This benchmark demonstrates both the ease of scaling and the cost-efficiency offered by the TPU v5e systems for inference of large language models. Users can easily add more TPU v5e instances to achieve higher total queries per second (QPS), while maintaining the same performance per dollar advantage.

HPE reported, In the datacenter category, HPE Cray systems with eight (8) NVIDIA GPUs led our portfolio in performance, delivering more than 340,000 samples per second throughput for ResNet-50 Computer Vision, and more than 28,000 samples per second throughput for Bert 99.0 NLP. HPE also submitted for the first time the newly available HPE ProLiant DL380a Gen11 and HPE ProLiant DL320 Gen11 servers with NVIDIA H100 and L4 GPUs. The HPE ProLiant DL380a Gen11 with four (4) NVIDIA H100 GPUs is ideal for NLP and LLM inference. The HPE ProLiant DL320 Gen11 with four (4) NVIDIA L4 GPUs is a 1U server positioned for computer vision inference.

Intel discussed Gaudi2 accelerators, 4th Gen Intel Xeon Scalable processors and Intel Xeon CPU Max Series. Gaudi2 performance on both GPT-J-99 and GPT-J-99.9 for server queries and offline samples are 78.58/second and 84.08/second, respectively. These outstanding inference performance results complement our June training results and show continued validation of Gaudi2 performance on large language models. Performance and model coverage will continue to advance in the coming benchmarks as Gaudi2 software is updated continually with releases every six to eight weeks.

Intel remains the only server CPU vendor to submit MLPerf results. Our submission for 4th Gen Intel Xeon Scalable processors with Intel AMX validates that CPUs have great performance for general purpose AI workloads, as demonstrated with MLPerf models, and the new and larger DLRM v2 recommendation and GPT-J models.

You get the general flavor. Its necessary to dig into the spreadsheet for meaningful comparisons.

The new storage benchmark (v0.5) has been in the works for two years. MLCommons says, Its the first open-source AI/ML benchmark suite that measures the performance of storage for ML training workloads. The benchmark was created through a collaboration spanning more than a dozen leading industry and academic organizations and includes a variety of storage setups including: parallel file systems, local storage, and software defined storage. The MLPerf Storage Benchmark will be an effective tool for purchasing, configuring, and optimizing storage for machine learning applications, as well as for designing next-generation systems and technologies.

Although its being introduced along with the latest inference results, storage performance in ML is typically a more sensitive system element in training. MLCommons notes, Training neural networks is both a compute and data-intensive workload that demands high-performance storage to sustain good overall system performance and availability. For many customers developing the next generation of ML models, it is a challenge to find the right balance between storage and compute resources while making sure that both are efficiently utilized.

MLPerf Storage is intended to help overcome this problem by accurately modeling the I/O patterns posed by ML workloads, providing the flexibility to mix and match different storage systems with different accelerator types. The new benchmark reports results in sample/s and MB/s. Of course, the choice of storage hardware, protocol/filesystem, and network all influence performance.

The MLPerf Storage benchmark suite is built on the codebase of DLIO, a benchmark designed for I/O measurement in high performance computing, adapted to meet current storage needs.

Talking about the motivation and goals for the new benchmark, Kanter said Id heard about pretty large hyperscalers, who deployed really large training clusters, that could not hit their peak utilization because they didn't have enough storage. That [suggested] there's fundamentally a hard problem in storage and one that's under appreciated. Most hyperscalers that are buying 1000s, or tens of 1000s of accelerators also have engineers on staff to design proper storage subsystems."

The key accomplishment is we created a tool that represents ML training IO patterns, that doesn't require having any compute or accelerators, said Kanter. That's important, because if you want to size a storage subsystem for 1000 accelerators, you don't want to have to buy 1000 accelerators. Another interesting thing is its a dynamic tool that is coupled to compute. The metric for MLPerf storage is how many samples per second can be streamed out, for a given compute utilization; so we model a compute subsystem. If your storage falls behind too much, the compute subsystem will be idle, and we only allow 10% idle due to storage.

If the storage system us too slow, you cant run the benchmark, said Kanter. Obviously, these are early days for MLPerf Storage and it will take some time for the community take its full measure. There are already plans for additions. Given its newness, its best look through MLCommons documentation. (Link to MLPerf Storage Benchmark Rules)

Link to MLCommons, https://mlcommons.org/en/

ASUSTeK

ASUStek recently benchmarked its new AI servers using the MLPerf Inference v3.1 suite, aiming to highlight its performance across varied deep learning tasks. Our results exhibit our system's competency in inferencing some of the most demanding models with remarkable efficiency.

In the modern era of AI, speed and efficiency in deploying machine learning models to production are paramount. Enter ASUS GPU Server portfolios - designed to redefine the standards of inference, as validated by our recent MLPerf Inference benchmarks. Harness the power of AI frameworks like TensorFlow, PyTorch, and more. ASUS servers are not just about raw power; they're about smart power. Optimized software-hardware integrations ensure that you get the most out of every tensor operation. Power doesnt have to come at the cost of the planet. ASUS GPU servers not only boast top-tier performance metrics but do so with impressive energy efficiency ratings, as highlighted in the MLPerf power efficiency results. Seamlessly scale your AI workloads. With our multi-GPU configurations and optimized in hardware and software, ASUS GPU servers are built to handle increasing data demands, ensuring youre always ahead of the curve.

System Configuration:

Hardware: ASUS flagship AI Server ESC8000A-E12 with Dual AMD Genoa CPU up to 8 NVIDIA H100 GPUs, and ESC4000A-E12 with Dual AMD Genoa CPU up to 8 L4 GPUs

The results signify the DL system's enhanced performance and capability to address contemporary deep learning challenges, making it an apt choice for researchers and industries requiring accelerated inferencing workloads.

Azure

Microsoft Azure announced the general availability of the ND H100 v5-series for Generative AI at scale. These series of virtual machines vary in sizes ranging from eight to thousands of NVIDIA H100 GPUs interconnected by NVIDIA Quantum-2 InfiniBand networking. Azure was the only submitter to publish results for virtual machines in the cloud, while matching the performance of on premises and bare metal offerings. This has been possible thanks to innovative technologies including:

The ND H100 v5 is now available in the East United States and South Central United States Azure regions. Enterprises can register their interest in access to the new VMs or review technical details on the ND H100 v5 VM series at Microsoft Learn.

CTuning

As a founding member of MLCommons, cTuning.org is committed to democratizing MLPerf benchmarks and making them accessible to everyone to deliver the most efficient AI solutions while reducing all development, benchmarking and optimization costs.

We are proud to deliver the new version of the open-source MLCommons CM automation language, CK playground and modular inference library (MIL) that became the 1st and only workflow automation enabling mass submission of more than 12000 performance results in a single MLPerf inference submission round with more than 1900 power results across more than 120 different system configurations from different vendors (different implementations, all reference models and support for DeepSparse Zoo, Hugging Face Hub and BERT pruners from the NeurIPS paper, main frameworks and diverse software/hardware stacks) in both open and closed divisions!

This remarkable achievement became possible thanks to open and transparent development of this technology as an official MLCommons project with public Discord discussions, important feedback from Neural Magic, TTA, One Stop Systems, Nutanix, Collabora, Deelvin, AMD and NVIDIA, and contributions from students, researchers and even school children from all over the world via our public MLPerf challenges. Special thanks to cKnowledge for sponsoring our developments and submissions, to One Stop Systems for showcasing the 1st MLPerf results on Rigel Edge Supercomputer, and to TTA for sharing their platforms with us to add CM automation for DLRMv2 available to everyone.

Since its impossible to describe all the compelling performance and power-efficient results achieved by our collaborators in a short press-release, we will make them available with various derived metrics (power efficiency, cost, etc) and reproducibility reports at the MLCommons CK playground (x.cKnowledge.org), github.com/mlcommons/ck_mlperf_results and github.com/mlcommons/ck/blob/master/docs/news-mlperf-v3.1.md shortly after official release.

We continue enhancing the MLCommons CM/CK technology to help everyone

automatically co-design the most efficient end-to-end AI solutions

based on their requirements and constraints. We welcome all submitters to join our public MLCommons Task Force on Automation and Reproducibility if you want to automate your future MLPerf submissions at scale.

Connect Tech Inc

As a new member of MLCommons, Connect Tech ran performance and accuracy benchmarks in the Inference: Edge category in its recent MLPerf submission. Using Connect Techs feature-rich Hadron carrier board with the NVIDIA Jetson Orin NX, a high-performance, energy-efficient platform, showcased remarkable levels of performance across various AI workloads.

Connect Tech additionally supports NVIDIA Jetson Orin NX with Photon and Boson carrier boards, and system devices like Polaris and Rudi-NX. By deploying on Connect Techs production-ready hardware, customers can take immediate advantage of Jetson Orin NX for performance improvements and enhanced user experience with robotics and other edge AI applications.

Connect Tech's involvement in MLCommons signifies more than just technical achievement. It reflects the company's commitment to pushing the envelope of what's possible in the world of AI at the edge. The seamless integration of Connect Tech's hardware with NVIDIA's cutting-edge technology presents engineers and scientists with the tools to drive AI and machine learning innovations across diverse industries, including robotics, industrial automation, and healthcare.

Connect Tech is a hardware design and manufacturing company, specializing in rugged, small form factor solutions. As an Elite NVIDIA Jetson ecosystem partner, Connect Tech designs carrier boards, enclosures, and embedded systems for each Jetson generation. With a rich history of innovation, Connect Tech integrates edge AI solutions within various industries, empowering engineers and scientists to harness the potential of machine learning.

Connect Tech remains at the forefront as the world delves deeper into AI and machine learning. Navigating the complex landscape of embedded AI computing is made easier by using NVIDIA and Connect Techs innovative products.

Dell

Enterprise IT is bracing for the most transformative technology trend in decades: generative AI. Dell Technologies is ready to meet this demand with the worlds broadest Generative AI solutions portfolio from desktop to edge to data center to cloud, all in one place.

For the MLPerf inferencing v3.1 benchmark testing, Dell submitted 230 results, including the new GPT-J and DLRMv2 benchmark results, across 20 system configurations. Dell Technologies works with customers and collaborators, including NVIDIA, Intel, and Qualcomm, to optimize performance and efficiency, boosting inferencing workloads, including generative AI.

The Dell PowerEdge XE accelerated server family continues to deliver tremendous performance gains across several benchmarks. Here are some of the latest highlights:

Generate higher quality, faster time-to-value predictions and outputs while accelerating decision-making with powerful solutions from Dell Technologies. Take a test drive in one of our worldwide Customer Solution Centers. Collaborate with our Innovation Lab and tap into one of our Centers of Excellence.

Fujitsu

Fujitsu offers a fantastic blend of systems, solutions, and expertise to guarantee maximum productivity, efficiency, and flexibility delivering confidence and reliability. Since 2020, we have been actively participating in and submitting to inference and training rounds for both data center and edge divisions.

In this round, Fujitsu demonstrated the performance of PRIMERGY CDI with four A100-PCIe-80GB GPUs installed in an external PCIe BOX and measured the benchmark program only for the data center closed division. Fujitsu Server PRIMERGY CDI is expertly engineered to deploy the necessary resources according to each customer's unique workload, releasing them when no longer needed. CDI stands for Composable Disaggregated Infrastructure, a next-generation technology that supports the diversification of data processing. This results in an efficient operation that maximizes resource utilization, while providing user-friendly services that eliminate the drawbacks of traditional physical servers.

As demonstrated by the impressive results of this round, the PRIMERGY CDI confirms that even with GPUs mounted in an external PCIe BOX, it delivers outstanding performance and remarkable scalability for PCIe components.

Our purpose is to make the world more sustainable by building trust in society through innovation. With a rich heritage of driving innovation and expertise, we are dedicated to contributing to the growth of society and our valued customers. Therefore, we will continue to meet the demands of our customers and strive to provide attractive server systems through the activities of MLCommons.

Giga Computing Technology, a subsidiary wholly owned by GIGABYTE, is the enterprise unit that split off from GIGABYTE that designs, manufactures, and sells servers, server motherboards, immersion solutions, and workstations. As the GIGABYTE brand is widely recognized, Giga Computing will continue to use and promote it, and that includes at expos where we will join as GIGABYTE. Although the company name has changed, our customers can still expect the same quality and services as before. Giga Computing strives to do better and that includes greater push for efficiency and cooling with immersion and DLC technology. As well as providing public AI benchmarks.

As one of the founding members of MLCommons, GIGABYTE has continued to support the communitys efforts in benchmarking server solutions for various AI training & inference workloads. In the latest round of MLPerf Inference v3.1, Giga Computing submitted a powerful GIGABYTE system for platforms: Intel Xeon & NVIDIA H100 SXM5, and the results speak for themselves while showing great efficiency as measured in performance/watt. We did find that our system achieved excellent performance in some tests such as rnnt-Server and bert99-offline. We would have liked to have more benchmarks, but due to resource limitations we are not able; however, we found that our partners NVIDIA, Qualcomm, and Krai chose our GIGABYTE servers to do their own testing.

Google

Google Cloud recently launched an expansion to its AI infrastructure portfolio - Cloud TPU v5e - and is proud to announce its performance results in this round of MLPerf Inference (data center category). TPU v5e systems use multiple accelerators linked together by a high-speed interconnect and can be configured with a topology ranging from 1x1 to 16x16 (256 chips), giving the user the flexibility to choose the system that best meets their needs. This wide range of topology options offered by TPU systems allows users to run and scale AI inference workloads cost-effectively, without compromising on performance.

In this submission, Google Cloud used a TPU v5e system with a 2x2 topology (4 TPU chips) to run the 6-billion-parameter GPTJ benchmark. This benchmark demonstrates both the ease of scaling and the cost-efficiency offered by the TPU v5e systems for inference of large language models. Users can easily add more TPU v5e instances to achieve higher total queries per second (QPS), while maintaining the same performance per dollar advantage.

We are looking forward to seeing what Google Cloud customers achieve with the new TPU v5e systems.

HPE

HPE successfully submitted results in partnership with Intel, NVIDIA, Qualcomm, and Krai. HPE demonstrated a range of high-performing inference systems for both the datacenter and edge in Computer Vision, natural language processing (NLP), and large language models (LLM).

In the datacenter category, HPE Cray systems with eight (8) NVIDIA GPUs led our portfolio in performance, delivering more than 340,000 samples per second throughput for ResNet-50 Computer Vision, and more than 28,000 samples per second throughput for Bert 99.0 NLP.

HPE also submitted for the first time the newly available HPE ProLiant DL380a Gen11 and HPE ProLiant DL320 Gen11 servers with NVIDIA H100 and L4 GPUs. The HPE ProLiant DL380a Gen11 with four (4) NVIDIA H100 GPUs is ideal for NLP and LLM inference. The HPE ProLiant DL320 Gen11 with four (4) NVIDIA L4 GPUs is a 1U server positioned for computer vision inference. The HPE ProLiant DL380a Gen11 showed strong inference performance using 4th Gen. Intel Xeon Scalable Processors in CPU-only inference scenarios. The HPE ProLiant DL385 Gen10 Plus v2 with eight (8) Qualcomm Cloud AI 100 Standard accelerators remained well balanced for over-network inference compared to offline datacenter performance. Qualcomm Cloud AI 100 Standard is ideal for both computer vision and NLP inference.

In the Edge category, HPE Edgeline e920d powered by four (4) Qualcomm Cloud AI 100 Standard accelerators remains one of the lowest latency systems in the Edge category for SingleStream and MultiStream inference scenarios. The HPE Edgeline e920d also achieved strong performance improvements in throughput and energy efficiency.

Many thanks to Krais collaboration in achieving high-performance and energy efficiency for Qualcomm Cloud AI 100 accelerators.

IEI

IEI Industry Co., LTD is a leading provider of data center infrastructure, cloud computing, and AI solutions, ranking among the worlds top 3 server manufacturers. Through engineering and innovation, IEI delivers cutting-edge computing hardware design and extensive product offerings to address important technology arenas like open computing, cloud data center, AI, and deep learning.

In MLCommons Inference v3.1, IEI submitted the NF5468M6 system.

NF5468M6 is a highly versatile 4U AI server supporting between 4 and 16 NVIDIA single and double-width GPUs, making it ideal for a wide range of AI applications including AI cloud, IVA, video processing and much more. NF5468M6 offers ultra-high storage capacity and the unique function of switching topologies between Balance, Common and Cascade in one click, which helps to flexibly adapt to various needs for AI application performance optimization.

Intel

Intel is pleased to report MLPerf Inference v3.1 performance results for our Gaudi2 accelerators, 4th Gen Intel Xeon Scalable processors and Intel Xeon CPU Max Series. These results reinforce Intels commitment to delivering the full spectrum of products to address wide-ranging customer AI requirements.

Gaudi2 performance on both GPT-J-99 and GPT-J-99.9 for server queries and offline samples are 78.58/second and 84.08/second, respectively. These outstanding inference performance results complement our June training results and show continued validation of Gaudi2 performance on large language models. Performance and model coverage will continue to advance in the coming benchmarks as Gaudi2 software is updated continually with releases every six to eight weeks.

Intel remains the only server CPU vendor to submit MLPerf results. Our submission for 4th Gen Intel Xeon Scalable processors with Intel AMX validates that CPUs have great performance for general purpose AI workloads, as demonstrated with MLPerf models, and the new and larger DLRM v2 recommendation and GPT-J models.

The results confirm that 4th Gen Intel Xeon Scalable processor with optimized data pre-processing, modeling and deployment tools and optimizations, is an ideal solution to build and deploy general purpose AI workloads with the most popular open source AI frameworks and libraries.

For the GPT-J 100-word summarization task of a news article of approximately 1,000 to 1,500 words, 4th Gen Intel Xeon processors summarized two paragraphs per second in offline mode and one paragraph per second in real-time server mode.

This is the first time weve submitted MLPerf results for our Intel Xeon CPU Max Series, which provides up to 64GB of high-bandwidth memory. For GPT-J, it was the only CPU able to achieve 99.9% accuracy, which is critical for usages for which the highest accuracy is of paramount importance.

With our ongoing software updates, we expect continued advances in performance and productivity, and reporting new training metrics with the November training cycle.

For more details, please see MLCommons.org.

Notices & Disclaimers

Performance varies by use, configuration and other factors. Learn more at http://www.Intel.com/PerformanceIndex .

Performance results are based on testing as of dates shown in configurations and may not reflect all publicly available updates. See

backup for configuration details. No product or component can be absolutely secure. Your costs and results may vary.

Follow this link:
MLPerf Releases Latest Inference Results and New Storage ... - EnterpriseAI