Nvidia Benchmark Recipes Bring Deep Insights In AI Performance
As AI workloads and accelerated applications grow in sophistication and complexity, businesses and developers need better tools to assess their infrastructure's ability to handle the demands of both training and inference more efficiently. To that end, Nvidia has been working on a set of performance testing tools, called DGX Cloud Benchmark Recipes, that are designed to help organizations evaluate how their hardware and cloud infrastructure perform when running the most advanced AI models available today. Our team at HotTech had a chance to kick the tires on a few of these recipes recently, and found the data they can capture to be extremely insightful.
Nvidia's toolkit also offers a database and calculator of performance results for GPU-compute workloads on various configurations, including the number of Nvidia H100 GPUs and cloud service providers, while the recipes allow businesses to run realistic performance evaluations on their own infrastructure. The results can help guide decisions on whether to invest in more powerful hardware, cloud provider service levels, or tweak configurations to better meet machine learning demands. These tools also take a holistic approach that incorporates network technologies for optimal throughput.
Nvidia DGX Cloud Benchmarking Recipes are a set of pre-configured containers and scripts that users can download and run on their own infrastructure. These containers are optimized for testing the performance of various AI models under different configurations, making them very valuable for companies looking to benchmark systems, whether on prem or in the cloud, before committing to larger-scale AI workloads or infrastructure deployments.
Nvidia DGX Cloud Time To Train Performance Report
In addition to offering static performance data, time to train and efficiency calculated from its database, Nvidia has recipes readily available for download that let businesses run real-world tests on their own hardware or cloud infrastructure, helping them understand the performance impact of different configurations. The recipes include benchmarks for training models like Meta's Llama 3.1 and Nvidia's own Llama 3.1 branch, called Nemotron, across several cloud providers (AWS, Google Cloud, and Azure), with options for adjusting factors like model size, GPU usage, and precision. The database is broad enough to cover popular AI models, but it is primarily designed for testing large-scale pre-training tasks, rather than inference on smaller models. The benchmarking process also allows for flexibility. Users can tailor the tests to their specific infrastructure by adjusting parameters such as the number of GPUs and the size of the model being trained.
The default hardware configuration in Nvidia's database of results uses the company's high-end H100 80GB GPUs, but it is designed to be adaptable. Although currently, it does not include consumer or prosumer-grade GPUs (e.g., RTX A4000 or RTX 50) or the company's latest Blackwell GPU family, these options could be added in the future.
Running the DGX Cloud Benchmarking Recipes is straightforward, assuming a few prerequisites are met. The process is well-documented, with clear instructions on setting up, running the benchmarks, and interpreting the results. Once a benchmark is completed, users can review the performance data, which includes key metrics like training time, GPU usage, and throughput. This allows businesses to make data-driven decisions about which configurations deliver the best performance and efficiency for their AI workloads. This could also go a long way in helping companies maintain green initiatives in terms of meeting power consumption and efficiency budgets.
While the DGX Cloud Benchmarking Recipes offer valuable insights, there are a few areas where Nvidia's tools could be expanded. First, benchmarking recipes are currently focused primarily on pre-training large models, not on real-time inference performance. Inference tasks, such as token generation or running smaller AI models, are equally important in many business applications. Expanding the toolset to include more detailed inference benchmarks would provide a fuller picture of how different hardware configurations handle these real-time demands. Additionally, by expanding the recipe selection to include lower-end or even higher-end GPUs (like Blackwell or even competitive offerings), Nvidia could cater to a broader audience, particularly businesses that don't require the massive compute power of a Hopper H100 80GB cluster for every workload.
Regardless, Nvidia's new DGX Cloud Benchmarking Recipes look like a very helpful resource for evaluating the performance of AI compute infrastructure, before making major investment decisions. They offer a practical way to understand how different configurations—whether cloud-based or on-premises—handle complex AI workloads. This is especially valuable for organizations exploring which cloud provider best meets their needs, or if the company is looking for new ways to optimize existing infrastructure.
As AI's role in business and our everyday lives continues to grow, tools like this will become essential for guiding infrastructure decisions, balancing performance versus cost and power consumption, and optimizing AI applications to meet real-world demands. As Nvidia expands these recipes to include more inference-focused benchmarks and potentially expands its reference data with a wider range of GPU options, these tools could become even more indispensable to businesses and developers of all sizes.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Tom's Guide
6 minutes ago
- Tom's Guide
Meta just revealed a new XR display prototype — and it may be the biggest leap in smart glasses since Google Glass
Smart glasses are evolving at a rapid pace — incredible display tech coming to AR specs and AI breakthroughs making the likes of the Oakley Meta HSTN's incredibly intelligent. But we all know the end goal is a pair of glasses that brings these two worlds together. And researchers at Meta Reality Labs and Stanford University may have just given us the clearest glimpse yet of this future. Yes, I know the researcher pictured above kind of looks like a Cyberpunk pirate, but that eyepatch is a prototype holographic XR display — creating full 3D holograms on a screen thin enough to be used in a standard-sized pair of glasses. This is the holy grail that companies have been chasing down, and Meta just took one giant step closer to it. So let's get into the details of this mixed reality tech. Basically, it's a combination of custom glass and silicon along with AI-driven algorithms to render 'perceptually realistic 3D images,' as the researchers say in their paper. To make this happen, there's a custom ultra-thin waveguide display driving the visuals — using a laser to project onto a uniquely-textured part of the lens glass for picture clarity. After this, it goes through a polarizer so we can see it, and a custom-designed Spatial Light Modulator that will (you guessed it) modulate light — the special thing being that it will do so on an individual pixel-by-pixel basis to render a 'full-resolution holographic' image we can see. Get instant access to breaking news, the hottest reviews, great deals and helpful tips. That's right. Holograms. Unlike your standard VR headsets and AR glasses that use eye-confusion techniques to simulate depth, this system can produce true holograms by reconstructing them entirely through light. There's both a wide field of view and a large eyebox to accommodate all possible pupil positions for accessibility. 'The world has never seen a display like this with a large field of view, a large eyebox, and such image quality in a holographic display,' Gordon Wetzstein, Stanford electrical engineering professor told the Stanford Report. 'It's the best 3D display created so far and a great step forward – but there are lots of open challenges yet to solve.' And even better? All of this is crammed into a panel just 3mm thin! That is significant for the future progress of stuffing displays into glasses without needing bird bath optics. One challenge to note, though, is this is mixed reality and not augmented. The wording is critical here, as the mixed reality reference means the optics are not transparent. But that being said, with this being the second of three projects to bring this to life in a commercial project, there's no doubt that what we're looking at here is a breakthrough. Don't expect this tech to come to glasses you can buy for another few years, but if this mixed reality display can become augmented, that's the dream I've been having over the past four years of testing and writing about smart glasses! My apologies to the phones team, but the clock has just started ticking on a new frontier of personal devices. Follow Tom's Guide on Google News to get our up-to-date news, how-tos, and reviews in your feeds. Make sure to click the Follow button.
CNN
18 minutes ago
- CNN
Microsoft has become the next $4 trillion company
Microsoft became the second company to reach a $4 trillion market valuation on Thursday, after its blockbuster earnings report on Wednesday night boosted the tech giant's shares. Microsoft's shares (MSFT) jumped nearly 4.5% after the market opened on Thursday, pushing its intraday valuation to $4.01 trillion. The company's shares have risen roughly 28% since the start of this year. The milestone comes just a year and a half after Microsoft reached a $3 trillion valuation. The company first cracked the $1 trillion mark in April 2019. It follows Nvidia into the $4 trillion valuation club, which hit the mark earlier this month. Microsoft forecast a record $30 billion in capital spending for the current fiscal first quarter to fuel its AI ambitions, and reported booming sales in its Azure cloud computing business on Wednesday. Wall Street analysts also noted that Microsoft's Copilot AI chatbot appears to have driven meaningful growth in its Microsoft 365 enterprise software business. Microsoft's move to $3 trillion was more measured than other tech giants, Nvidia (NVDA) and Apple (AAPL), with AI bellwether Nvidia tripling its value in just about a year and clinching the $4 trillion milestone before any other company on July 9. Apple was last valued at $3.12 trillion. Lately, breakthroughs in trade talks between the US and its trading partners ahead of President Donald Trump's August 1 tariff deadline have buoyed stocks, propelling the S&P 500 and the Nasdaq to record highs. Microsoft, the second largest US company, has rebounded nearly 50% from its April 2025 lows, when global markets were rattled by Trump's tariff offensive. Microsoft's multibillion-dollar bet on OpenAI is proving to be a game-changer, powering its Office Suite and Azure offerings with cutting-edge AI and fueling the stock to more than double its value since ChatGPT's late-2022 debut. Armed with exclusive access to OpenAI's models, Microsoft has raced to the front of the generative AI pack — supercharging its Azure cloud business, now the company's top revenue driver, and solidifying its dominance in the tech landscape, compared to Google's cloud and Amazon's web services. Wall Street's surging confidence in the company comes on the heels of back-to-back record revenues for the tech giant since September 2022. The stock's rally had also received an extra boost as the tech giant trimmed its workforce and doubled down on AI investments — determined to cement its lead as businesses everywhere race to harness the technology. While sweeping US tariffs had investors bracing for tighter business spending, Microsoft's strong earnings have shown that the company's books are yet to take a hit from the levies. But despite its meteoric rise, Microsoft has been streamlining its workforce in recent months. The company said earlier this month that it would cut around 9,000 employees, approximately 4% of its staff, its largest reduction since 2023. Those layoffs came after the company cut 6,000 workers in May. A company spokesperson said the July cuts were, in part, a reflection of new technologies that have made employees more productive. While AI wasn't mentioned specifically, the cuts came as Silicon Valley giants deploy the technology they're building to make their workers more efficient. Microsoft CEO Satya Nadella said earlier this year that 20% to 30% of the company's code was being generated by AI, and it's is pouring billions into AI infrastructure investments.
Yahoo
19 minutes ago
- Yahoo
Is Figma a Buy After Tripling on Its IPO?
Key Points Figma seemed likely to pop on its debut after it priced its IPO for less than Adobe had offered it three years ago. The company's financials look great, including strong revenue growth and double-digit GAAP operating margins. Following the opening-day surge, Figma's price-to-sales ratio appeared to approach 70. 10 stocks we like better than Figma › Figma (NYSE: FIG) stock soared on its opening day on Thursday. After pricing its initial public offering (IPO) at $33, the stock opened at $85 around 2:00 p.m. ET, 158% higher than the listing price, and then continued to climb from there. The surge came as little surprise. Figma went public with the stock market at all-time highs, and with risk-on tech stocks back in favor, stocks like Nvidia have soared as tariff fears have cooled. Figma also has just about everything investors could want from an IPO, which also seemed to set the stock up to skyrocket in its debut. A model start-up IPOs are inherently risky, but there are a number of features that investors look for to persuade them that it's worth taking the risk. Those include strong revenue growth, profitability, a large addressable market, and scalability of the underlying product or service. Figma is the rare IPO that has all those qualities, and even better, it has a stamp of approval from the biggest company in its industry. Adobe (NASDAQ: ADBE), of course, tried to buy the company for $20 billion back in 2022, but regulators blocked the deal. That event may be the biggest endorsement for Figma's IPO, as not only did the leader in design software want to acquire it, but it was also willing to pay a steep premium of $20 billion at the time. Additionally, the combination of the two companies was so feared by regulators that they blocked the deal on account of it having a negative impact on competition in the design software industry. You can't ask for an IPO to prove itself much more than that, and the fact that Figma's IPO price valued the company for less than what Adobe offered nearly three years ago seemed like a gift to investors able to get their hands on the offering. Figma by the numbers It's not just the Adobe deal that gives Figma street cred. Its quarterly numbers are rock-solid. Over its last four quarters, revenue has grown 46% to $821 million. Even better, the company is already generating wide operating margins on a generally accepted accounting principles (GAAP) basis at 17%. There are plenty of software companies that have been public for years that remain unprofitable, as software investors tend to reward growth over profitability. Finally, it has a 91% gross margin, showing its core product is highly cost-efficient. That high gross margin makes it easier for the overall business to turn a profit, giving the company more money to spend on product development, sales, and overhead. The operating margin of 17% also indicates that Figma's product and competitive position are strong enough that it doesn't need an outsize sales force to sell the product, a good sign for the long-term health of the business. Already, 78% of the Forbes 2000 uses Figma, showing big business has embraced the product. However, there is plenty of room for it to ramp up spending, as Figma believes it has a total addressable market (TAM) of $33 billion. That TAM is likely to grow. Is Figma a buy? IPOs are risky in part because it generally takes time for a stock to reach its equilibrium price, and investors may have to see a few more quarters of earnings reports before the post-IPO volatility runs its course. As a business, Figma appears to have a bright future ahead of it, but after tripling on its IPO day, the stock now has a market cap of roughly $60 billion, giving it a price-to-sales ratio (P/S) of around 70. That's a sky-high valuation for even the brightest of IPOs. At that price, I'm more inclined to wait on the sidelines for now to see whether some of the IPO euphoria fades. I'd prefer to see the P/S ratio fall below 50, which is still pricey, before considering it a buy. That may not happen, but its current valuation seems hard to sustain, especially considering the much larger Adobe has a market cap of only $150 billion. Investors shouldn't ignore valuation, but there's a lot to admire about Figma. It deserves a healthy premium from the market. Should you buy stock in Figma right now? Before you buy stock in Figma, consider this: The Motley Fool Stock Advisor analyst team just identified what they believe are the for investors to buy now… and Figma wasn't one of them. The 10 stocks that made the cut could produce monster returns in the coming years. Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you'd have $638,629!* Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you'd have $1,098,838!* Now, it's worth noting Stock Advisor's total average return is 1,049% — a market-crushing outperformance compared to 182% for the S&P 500. Don't miss out on the latest top 10 list, available when you join Stock Advisor. See the 10 stocks » *Stock Advisor returns as of July 29, 2025 Jeremy Bowman has positions in Nvidia. The Motley Fool has positions in and recommends Adobe and Nvidia. The Motley Fool has a disclosure policy. Is Figma a Buy After Tripling on Its IPO? was originally published by The Motley Fool
