Home Assistant, until recently, has been a wide-ranging and hard-to-define project.

The open smart home platform is an open source OS you can run anywhere that aims to connect all your devices together. But it’s also bespoke Raspberry Pi hardware, in Yellow and Green. It’s entirely free, but it also receives funding through a private cloud services company, Nabu Casa. It contains tiny board project ESPHome and other inter-connected bits. It has wide-ranging voice assistant ambitions, but it doesn’t want to be Alexa or Google Assistant. Home Assistant is a lot.

After an announcement this weekend, however, Home Assistant’s shape is a bit easier to draw out. All of the project’s ambitions now fall under the Open Home Foundation, a non-profit organization that now contains Home Assistant and more than 240 related bits. Its mission statement is refreshing, and refreshingly honest about the state of modern open source projects.

“We’ve done this to create a bulwark against surveillance capitalism, the risk of buyout, and open-source projects becoming abandonware,” the Open Home Foundation states in a press release. “To an extent, this protection extends even against our future selves—so that smart home users can continue to benefit for years, if not decades. No matter what comes.” Along with keeping Home Assistant funded and secure from buy-outs or mission creep, the foundation intends to help fund and collaborate with external projects crucial to Home Assistant, like Z-Wave JS and Zigbee2MQTT.

Home Assistant’s ambitions don’t stop with money and board seats, though. They aim to “be an active political advocate” in the smart home field, toward three primary principles:

• Data privacy, which means devices with local-only options, and cloud services with explicit permissions
• Choice in using devices with one another through open standards and local APIs
• Sustainability by repurposing old devices and appliances beyond company-defined lifetimes

Notably, individuals cannot contribute modest-size donations to the Open Home Foundation. Instead, the foundation asks supporters to purchase a Nabu Casa subscription or contribute code or other help to its open source projects.

From a few lines of Python to a foundation

Home Assistant founder Paulus Schoutsen wanted better control of his Philips Hue smart lights just before 2014 or so and wrote a Python script to do so. Thousands of volunteer contributions later, Home Assistant was becoming a real thing. Schoutsen and other volunteers inevitably started to feel overwhelmed by the “free time” coding and urgent bug fixes. So Schoutsen, Ben Bangert, and Pascal Vizeli founded Nabu Casa, a for-profit firm intended to stabilize funding and paid work on Home Assistant.

Through that stability, Home Assistant could direct full-time work to various projects, take ownership of things like ESPHome, and officially contribute to open standards like Zigbee, Z-Wave, and Matter. But Home Assistant was “floating in a kind of undefined space between a for-profit entity and an open-source repository on GitHub,” according to the foundation. The Open Home Foundation creates the formal home for everything that needs it and makes Nabu Casa a “special, rules-bound inaugural partner” to better delineate the business and non-profit sides.

Home Assistant as a Home Depot box?

In an interview with The Verge’s Jennifer Pattison Tuohy, and in a State of the Open Home stream over the weekend, Schoutsen also suggested that the Foundation gives Home Assistant a more stable footing by which to compete against the bigger names in smart homes, like Amazon, Google, Apple, and Samsung. The Home Assistant Green starter hardware will sell on Amazon this year, along with HA-badged extension dongles. A dedicated voice control hardware device that enables a local voice assistant is coming before year’s end. Home Assistant is partnering with Nvidia and its Jetson edge AI platform to help make local assistants better, faster, and more easily integrated into a locally controlled smart home.

That also means Home Assistant is growing as a brand, not just a product. Home Assistant’s “Works With” program is picking up new partners and has broad ambitions. “We want to be a consumer brand,” Schoutsen told Tuohy. “You should be able to walk into a Home Depot and be like, ‘I care about my privacy; this is the smart home hub I need.’”

Where does this leave existing Home Assistant enthusiasts, who are probably familiar with the feeling of a tech brand pivoting away from them? It’s hard to imagine Home Assistant dropping its advanced automation tools and YAML-editing offerings entirely. But Schoutsen suggested he could imagine a split between regular and “advanced” users down the line. But Home Assistant’s open nature, and now its foundation, should ensure that people will always be able to remix, reconfigure, or re-release the version of smart home choice they prefer.

https://arstechnica.com/?p=2019041

On Tuesday, Intel revealed a new AI accelerator chip called Gaudi 3 at its Vision 2024 event in Phoenix. With strong claimed performance while running large language models (like those that power ChatGPT), the company has positioned Gaudi 3 as an alternative to Nvidia’s H100, a popular data center GPU that has been subject to shortages, though apparently that is easing somewhat.

Compared to Nvidia’s H100 chip, Intel projects a 50 percent faster training time on Gaudi 3 for both OpenAI’s GPT-3 175B LLM and the 7-billion parameter version of Meta’s Llama 2. In terms of inference (running the trained model to get outputs), Intel claims that its new AI chip delivers 50 percent faster performance than H100 for Llama 2 and Falcon 180B, which are both relatively popular open-weights models.

Intel is targeting the H100 because of its high market share, but the chip isn’t Nvidia’s most powerful AI accelerator chip in the pipeline. Announcements of the H200 and the Blackwell B200 have since surpassed the H100 on paper, but neither of those chips is out yet (the H200 is expected in the second quarter of 2024—basically any day now).

Meanwhile, the aforementioned H100 supply issues have been a major headache for tech companies and AI researchers who have to fight for access to any chips that can train AI models. This has led several tech companies like Microsoft, Meta, and OpenAI (rumor has it) to seek their own AI-accelerator chip designs, although that custom silicon is typically manufactured by either Intel or TSMC. Google has its own line of tensor processing units (TPUs) that it has been using internally since 2015.

Given those issues, Intel’s Gaudi 3 may be a potentially attractive alternative to the H100 if Intel can hit an ideal price (which Intel has not provided, but an H100 reportedly costs around $30,000–$40,000) and maintain adequate production. AMD also manufactures a competitive range of AI chips, such as the AMD Instinct MI300 Series, that sell for around $10,000–$15,000.

Gaudi 3 performance

Intel says the new chip builds upon the architecture of its predecessor, Gaudi 2, by featuring two identical silicon dies connected by a high-bandwidth connection. Each die contains a central cache memory of 48 megabytes, surrounded by four matrix multiplication engines and 32 programmable tensor processor cores, bringing the total cores to 64.

The chipmaking giant claims that Gaudi 3 delivers double the AI compute performance of Gaudi 2 using 8-bit floating-point infrastructure, which has become crucial for training transformer models. The chip also offers a fourfold boost for computations using the BFloat 16-number format. Gaudi 3 also features 128GB of the less expensive HBMe2 memory capacity (which may contribute to price competitiveness) and features 3.7TB of memory bandwidth.

Since data centers are well-known to be power hungry, Intel emphasizes the power efficiency of Gaudi 3, claiming 40 percent greater inference power-efficiency across Llama 7B and 70B parameters, and Falcon 180B parameter models compared to Nvidia’s H100. Eitan Medina, chief operating officer of Intel’s Habana Labs, attributes this advantage to Gaudi’s large-matrix math engines, which he claims require significantly less memory bandwidth compared to other architectures.

Gaudi vs. Blackwell

Last month, we covered the splashy launch of Nvidia’s Blackwell architecture, including the B200 GPU, which Nvidia claims will be the world’s most powerful AI chip. It seems natural, then, to compare what we know about Nvidia’s highest-performing AI chip to the best of what Intel can currently produce.

For starters, Gaudi 3 is being manufactured using TSMC’s N5 process technology, according to IEEE Spectrum, narrowing the gap between Intel and Nvidia in terms of semiconductor fabrication technology. The upcoming Nvidia Blackwell chip will use a custom N4P process, which reportedly offers modest performance and efficiency improvements over N5.

Gaudi 3’s use of HBM2e memory (as we mentioned above) is notable compared to the more expensive HBM3 or HBM3e used in competing chips, offering a balance of performance and cost-efficiency. This choice seems to emphasize Intel’s strategy to compete not only on performance but also on price.

As far as raw performance comparisons between Gaudi 3 and the B200, that can’t be known until the chips have been released and benchmarked by a third party.

As the race to power the tech industry’s thirst for AI computation heats up, IEEE Spectrum notes that the next generation of Intel’s Gaudi chip, code-named Falcon Shores, remains a point of interest. It also remains to be seen whether Intel will continue to rely on TSMC’s technology or leverage its own foundry business and upcoming nanosheet transistor technology to gain a competitive edge in the AI accelerator market.

https://arstechnica.com/?p=2016421

Il nuovo chip per l’AI di Meta

Presentato poche ore fa il nuovo modello di chip per l’intelligenza artificiale (AI) di Meta, che secondo i suoi ingegneri sarà tre volte più efficiente del primo.

Si tratta della nuova versione del Meta training and inference accelerator (Mtia) V1,dedicata specificatamente allo sviluppo e il potenziamento dell’AI generativa dell’azienda.

Un lavoro fatto in casa, sia a livello software, sia hardware, che consente a Meta di controllare l’intero flusso stack, che va dal frontend (l’interfaccia utente e le interazioni con l’applicazione) al backend (l’esecuzione dell’applicazione e la comunicazione con eventuali altri app, compresa l’elaborazione dati).

“Poiché controlliamo l’intero stack, possiamo ottenere una maggiore efficienza rispetto alle GPU disponibili in commercio“, hanno scritto Eran Tal, Nicolaas Viljoen, Joel Coburn e Roman Levenstein nell’articolo di lancio della nuova versione dell’Mtia sul blog di Meta.

Un lavoro a lungo termine

Il nuovo modello di chip però al momento non è ancora utilizzato per lo sviluppo diretto dell’AI proprietaria, più che altro è impiegato nel potenziamento degli algoritmi di classificazione e raccomandazione.

Il prossimo passaggio, secondo quanto affermato dalla società nel testo di presentazione, sarà dedicato proprio al rafforzamento della sua AI, come fa ad esempio Google, con il suo chip AI TPU v5p che addestra i modelli di grandi dimensioni e la famiglia di chatbot Gemini 1,5 Pro.

“Mtia sarà una parte importante della nostra tabella di marcia a lungo termine per costruire e scalare l’infrastruttura più potente ed efficiente possibile per i carichi di lavoro IA unici di Meta“, hanno affermato gli ingegneri di Meta.

La corsa all’AI di Meta, Google, Intel, Microsoft e Amazon

L’annuncio del gigante tecnologico di San Francisco segue di un giorno quello dei rivali Google e Intel.

Google ha rivelato i piani per un nuovo processore basato su tecnologia Arm, che punta su consumi energetici più bassi. Si chiama Axion e offre prestazioni migliori del 30% rispetto agli altri chip con architettura Arm. Sarà disponibile per i servizi cloud che le aziende possono noleggiare e utilizzare, dagli annunci su YouTube all’analisi dei big data.

Intel ha lanciato una nuova versione del suo chip acceleratore di intelligenza artificiale. Si chiama Gaudi 3 e promette prestazioni di calcolo doppie. L’azienda californiana punta a diventare un’alternativa a Nvidia che nel 2023 ha controllato l’83% del mercato dei chip per data center e che ha segnato una ultima trimestrale record.

Tutti questi annunci arrivano dopo che Microsoft nel novembre del 2023 fa ha rivelato i propri microprocessori personalizzati progettati per la sua infrastruttura cloud e per addestrare modelli linguistici di grandi dimensioni: l’Azure Maia AI Accelerator e l’Azure Cobalt CPU. Anche Amazon nello stesso mese ha presentato il nuovo processore su tecnologia Arm AWS Graviton4 e l’acceleratore AWS Trainium2.

L’obiettivo delle Big Tech è comunque sempre lo stesso, cioè ridurre la propria dipendenza da partner come Intel e Nvidia, competendo sui chip personalizzati che riescono a smaltire grandi carichi di lavoro sull’IA e il cloud.

Meta, Microsoft, Alphabet, Apple, Nvidia e Amazon hanno effettuato l’acquisto di 88 tra startup e imprese di AI dal 2010 ad oggi.

https://www.key4biz.it/meta-risponde-a-google-e-intel-e-presenta-il-suo-nuovo-chip-ai/486395/

La corsa USA ai chip

Gli Stati Uniti hanno da tempo avviato un programma di investimenti, pubblici e privati, per sostenere e potenziare la produzione nazionale di semiconduttori e chip/microchip. Secondo quanto riportato da un articolo pubblicato sul Wall Street Journal, la SK Hynix, società sudcoreana di semiconduttori, avrebbe intenzione di investire 4 miliardi di dollari nella costruzione di un impianto a West Lafayette, in Indiana.

La struttura, dedicata all’imballaggio di chip, dovrebbe entrare in funzione nel 2028, per dare lavoro a circa 1000 dipendenti. L’operazione rientra nel Chips and Science Act voluto da Washington per supportare questa industria negli Stati Uniti e aumentare il potenziale competitivo su scala globale, soprattutto con i cinesi, e vedrà il concorso di incentivi statali e federali, oltre che degli investitori privati.

L’azienda è uno dei principali partner del mega produttore di chip Nvidia, a cui fornisce chip di memoria ad alta larghezza di banda (HBM), che, combinati con le GPU Nvidia, provvedono ad alimentare i modelli di intelligenza artificiale generativa più avanzati, tra cui ChatGPT.

Grazie all’IA, vola la capitalizzazione di mercato della SK Hynix

Sempre sulle pagine del Wsj, si legge che all’inizio di quest’anno, l’amministratore delegato di SK Hynix, Kwak Noh-jung, ha dichiarato che il boom dell’intelligenza artificiale potrebbe portare la valutazione della società di chip sui mercati globali a 200 trilioni di won, ovvero 148,8 miliardi di dollari. Già nel 2023 la capitalizzazione di mercato di questa azienda è più che raddoppiata, raggiungendo i 96 miliardi di dollari.

Per lo stesso motivo, il gigante Tsmc aumenterà del 30% la propria forza lavoro globale

Tra i clienti di Nvidia c’è anche il grande produttore di chip taiwanese Tsmc (Taiwan Semiconductor Manufacturing Company), una delle prime dieci grandi aziende di maggior valore al mondo.

Grazie al boom dell’intelligenza artificiale generativa la Tsmc ha deciso di aumentare la propria forza lavoro, secondo le dichiarazioni rese alla Cnn dal vicepresidente Lora Ho, con l’obiettivo di arrivare a 100 mila dipendenti in tutto il mondo entro i prossimi dieci anni.

Oggi, la più grande fonderia di semiconduttori su scala planetaria cota circa 77 mila dipendenti (nel 2020 non superavano le 56 mila unità).

Il gigante Taiwanese ha aperto a febbraio di quest’anno la sua prima fabbrica giapponese, la Japan Advanced Semiconductor Manufacturing (JASM), che dovrebbe avviare la produzione a fine 2024, mentre una seconda è attualmente in fase di progettazione. Per entrambe Tsmc ha pianificato un investimento complessivo di 20 miliardi di dollari.

I problemi di Tsmc in Arizona

Più difficile, a quanto pare, è l’apertura di un nuovo impianto in Arizona, negli Stati Uniti, perché la Tsmc continua a lamentare una certa difficoltà a reperire personale specializzato, nonostante un investimento da 40 milioni di dollari e la promessa di migliaia di posti di lavoro “ben retribuiti”, secondo quanto riportato dal quotidiano The Guardian.

Stando ad un rapporto diffuso dalla Casa Bianca, negli ultimi tre decenni la quota statunitense della produzione globale di semiconduttori è scesa dal 37% ad appena il 12%.

Il Presidente della società taiwanese, Mark Liu, ha anche offerto l’aiuto di 500 tecnici specializzati che l’azienda potrebbe inviare negli Stati Uniti per formare i nuovi dipendenti, ma è più probabile che il vero ostacolo sia solo e sempre finanziario.

La Tsmc probabilmente attende un’offerta “adeguata” da parte Washington, mentre i sindacati locali denunciano anche una mossa aziendale per favorire l’assunzione di manodopera a basso costo.

https://www.key4biz.it/semiconduttori-la-coreana-sk-hynix-investe-4-miliardi-di-dollari-negli-usa/485130/

Nvidia ha superato ancora una volta la capitalizzazione di mercato di 2mila miliardi di dollari e ha chiuso a 2,06 miliardi di dollari venerdì, per la prima volta oltre questa soglia e superando così Aramco, la la compagnia nazionale di idrocarburi dell’Arabia Saudita per diventare così la terza azienda che vale più al mondo.

Il produttore di chip è diventato il primo nel suo settore a raggiungere una valutazione di 2mila miliardi di dollari a febbraio, grazie a utili che hanno superato le aspettative degli analisti e a un settore dell’intelligenza artificiale in forte espansione. Nvidia aveva precedentemente superato Amazon e la società madre di Google, Alphabet, diventando la terza azienda di maggior valore negli Stati Uniti per capitalizzazione di mercato.

Aramco in flessione

Nel frattempo, secondo Bloomberg, Aramco ha visto le sue azioni crollare del 5% quest’anno a causa della minore produzione di petrolio derivante dai tagli dell’OPEC+ e di una possibile offerta successiva da parte del governo questo mese.

Il produttore di chip ha visto crollare le sue azioni prima di pubblicare gli utili per il quarto trimestre, poiché gli investitori hanno risposto alle voci secondo cui i suoi utili sarebbero stati inferiori. Ma i prezzi delle azioni sono rimbalzati alla grande dopo che Nvidia ha registrato un fatturato di 22 miliardi di dollari, un enorme aumento del 270% rispetto all’anno precedente.

“Il calcolo accelerato e l’intelligenza artificiale generativa hanno raggiunto il punto di svolta”, ha detto in una nota il fondatore e CEO di Nvidia Jensen Huang. “La domanda è in aumento in tutto il mondo tra aziende, industrie e nazioni”.

Microsoft e Meta diventano autonomi

Secondo quanto riferito, gli analisti erano preoccupati per le entrate di Nvidia a causa di alcuni dei suoi maggiori clienti, tra cui Microsoft e Meta, che sviluppavano i propri chip AI. Nvidia considera entrambe le società come i maggiori clienti del suo chip H100 da 30mila dollari che alimenta i modelli di intelligenza artificiale delle società. L’anno scorso Microsoft e Meta hanno speso 9 miliardi di dollari per i chip. Altri grandi investitori includono Alphabet, Amazon e Oracle.

Ma la performance di Nvidia preoccupa ancora gli scettici su quanto a lungo riuscirà a mantenere il suo dominio e su cosa potrebbe significare per la bolla tecnologica.

“Un altro trimestre di successo di Nvidia solleva la questione di quanto durerà la sua performance in ascesa”, ha scritto Jacob Bourne, analista senior di Insider Intelligence, in una nota dopo gli utili. “La forza di mercato di Nvidia a breve termine è duratura, anche se non invincibile”.

https://www.key4biz.it/nvidia-supera-aramco-e-diventa-la-terza-azienda-che-vale-di-piu-al-mondo/482227/

Capitalizzazione di mercato, Nvidia supera Amazon

Nvidia ha chiuso a 721,28 dollari per azione, con un valore di mercato di 1,78 trilioni di dollari, rispetto alla capitalizzazione di mercato di Amazon, che si è attestata a 1,75 trilioni di dollari. È la prima volta in 11 anni.

È questo il valore delle capitalizzazioni di mercato che hanno raggiunto le due Big Tech ieri. un segno, secondo cnbc.com, di quanto sia forte la domanda globale di chip per l’intelligenza artificiale (IA) e della volontà di investire ancora nei grandi chip maker.

Il prossimo 21 febbraio Nvidia pubblicherà i nuovi dati sugli utili trimestrali e gli esperti si attendono una crescita delle vendite del +118% su base annua, a oltre 59 miliardi di dollari.

La crescita delle Big Tech

Le azioni di Nvidia sono aumentate in valore di oltre il 246% negli ultimi 12 mesi, proprio a causa della forte domanda per i suoi chip IA per server, che possono costare più di 20.000 dollari ciascuno. Ad aziende come Microsoft, OpenAI e Meta ne servono decine di migliaia per far funzionare ChatGPT e Copilot.

Le azioni di Amazon sono aumentate di circa il 78% negli ultimi 12 mesi, grazie ad utili trimestrali migliori del previsto e come risultato di un buon controllo della spesa generale e del licenziamento di più di 27 mila dipendenti.

All’inizio dell’anno Microsoft ha superato Apple diventando l’azienda statunitense con il maggior valore in termini di capitalizzazione di mercato, in gran parte frutto della sua partnership cloud con OpenAI e alle nuove funzionalità AI in Windows e Office.

https://www.key4biz.it/nvidia-vale-piu-di-amazon-grazie-ai-chip-ia/479961/

In many ways, 2023 was a long-awaited return to normalcy for people who build their own gaming and/or workstation PCs. For the entire year, most mainstream components have been available at or a little under their official retail prices, making it possible to build all kinds of PCs at relatively reasonable prices without worrying about restocks or waiting for discounts. It was a welcome continuation of some GPU trends that started in 2022. Nvidia, AMD, and Intel could release a new GPU, and you could consistently buy that GPU for roughly what it was supposed to cost.

That’s where we get into how frustrating 2023 was for GPU buyers, though. Cards like the GeForce RTX 4090 and Radeon RX 7900 series launched in late 2022 and boosted performance beyond what any last-generation cards could achieve. But 2023’s midrange GPU launches were less ambitious. Not only did they offer the performance of a last-generation GPU, but most of them did it for around the same price as the last-gen GPUs whose performance they matched.

The midrange runs in place

Not every midrange GPU launch will get us a GTX 1060—a card roughly 50 percent faster than its immediate predecessor and beat the previous-generation GTX 980 despite costing just a bit over half as much money. But even if your expectations were low, this year’s midrange GPU launches have been underwhelming.

The worst was probably the GeForce RTX 4060 Ti, which sometimes struggled to beat the card it replaced at around the same price. The 16GB version of the card was particularly maligned since it was $100 more expensive but was only faster than the 8GB version in a handful of games.

The regular RTX 4060 was slightly better news, thanks partly to a $30 price drop from where the RTX 3060 started. The performance gains were small, and a drop from 12GB to 8GB of RAM isn’t the direction we prefer to see things move, but it was still a slightly faster and more efficient card at around the same price. AMD’s Radeon RX 7600, RX 7700 XT, and RX 7800 XT all belong in this same broad category—some improvements, but generally similar performance to previous-generation parts at similar or slightly lower prices. Not an exciting leap for people with aging GPUs who waited out the GPU shortage to get an upgrade.

The best midrange card of the generation—and at $600, we’re definitely stretching the definition of “midrange”—might be the GeForce RTX 4070, which can generally match or slightly beat the RTX 3080 while using much less power and costing $100 less than the RTX 3080’s suggested retail price. That seems like a solid deal once you consider that the RTX 3080 was essentially unavailable at its suggested retail price for most of its life span. But $600 is still a $100 increase from the 2070 and a $220 increase from the 1070, making it tougher to swallow.

In all, 2023 wasn’t the worst time to buy a $300 GPU; that dubious honor belongs to the depths of 2021, when you’d be lucky to snag a GTX 1650 for that price. But “consistently available, basically competent GPUs” are harder to be thankful for the further we get from the GPU shortage.

Marketing gets more misleading

If you just looked at Nvidia’s early performance claims for each of these GPUs, you might think that the RTX 40-series was an exciting jump forward.

But these numbers were only possible in games that supported these GPUs’ newest software gimmick, DLSS Frame Generation (FG). The original DLSS and DLSS 2 improve performance by upsampling the images generated by your GPU, generating interpolated pixels that make lower-res image into higher-res ones without the blurriness and loss of image quality you’d get from simple upscaling. DLSS FG generates entire frames in between the ones being rendered by your GPU, theoretically providing big frame rate boosts without requiring a powerful GPU.

The technology is impressive when it works, and it’s been successful enough to spawn hardware-agnostic imitators like the AMD-backed FSR 3 and an alternate implementation from Intel that’s still in early stages. But it has notable limitations—mainly, it needs a reasonably high base frame rate to have enough data to generate convincing extra frames, something that these midrange cards may struggle to do. Even when performance is good, it can introduce weird visual artifacts or lose fine detail. The technology isn’t available in all games. And DLSS FG also adds a bit of latency, though this can be offset with latency-reducing technologies like Nvidia Reflex.

As another tool in the performance-enhancing toolbox, DLSS FG is nice to have. But to put it front-and-center in comparisons with previous-generation graphics cards is, at best, painting an overly rosy picture of what upgraders can actually expect.

https://arstechnica.com/?p=1991746

Most Ars readers still probably know Nvidia best for its decades-old GeForce graphics cards for gaming PCs, but these days Nvidia’s server GPU business makes GeForce look like a hobby project.

That’s the takeaway from Nvidia’s Q3 earnings report, which shows Nvidia’s revenue up 206 percent from the same quarter last year and 34 percent from an already-very-good Q2. Of the company’s $18.12 billion in revenue, $14.51 billion was generated by its data center division, which includes AI-accelerating chips like the H200 Tensor Core GPU as well as other cloud and server offerings.

And though GeForce revenue was a much smaller $2.86 billion, this was still a solid recovery from the same quarter of Nvidia’s fiscal 2023, when GeForce GPUs earned just $1.51 billion and were down 51 percent compared to fiscal 2022. Nvidia has released several new mainstream GeForce RTX 40-series GPUs this year, including the $299 RTX 4060. And while these more affordable GPUs aren’t staggering upgrades from previous-generation cards, Steam Hardware Survey data shows the RTX 4060 and 4060 Ti are being adopted pretty quickly, more than can be said of competing GPUs like AMD’s RX 7600 or Intel’s Arc series.

The company’s overall revenue numbers weren’t looking nearly this good a year ago, either—in Q3 of fiscal 2023, the company’s revenue had fallen 17 percent year over year. The quarter before that, the company missed its own projections by $1.4 billion due to an oversupply of GPUs and a crypto-mining crash that reduced sales.

Demand for Nvidia’s AI-accelerating GPUs probably won’t be as volatile as demand for cryptomining GPUs was. For starters, there are big companies with big money buying up just about every HGX GPU that Nvidia can make, and companies like Microsoft and Amazon continue to make major AI announcements and investments at a steady clip—Nvidia also said it’s partnering with Dropbox, Foxconn, Lenovo, and multiple other companies on various AI initiatives. And, just as in PC and workstation graphics cards, Nvidia’s dominance can beget more dominance as software tools are designed for and optimized for Nvidia’s chips first.

Still, the crypto-mining example is instructive. If this bubble bursts, or if competing products from AMD or Intel begin making a dent in Nvidia’s sales, Nvidia could be in for a rough few quarters as these stratospheric revenue numbers come back to earth. Nvidia has also had trouble selling its AI chips in China, where US export restrictions on some high-performance chips have caused Nvidia to modify or stop offering some of its products to meet requirements.

Nvidia’s workstation GPUs and automotive divisions also grew year over year, though, at $416 million and $261 million, respectively, both divisions contribute a lot less to Nvidia’s bottom line than the data center or GeForce products.

https://arstechnica.com/?p=1986147

On Monday, Nvidia announced the HGX H200 Tensor Core GPU, which utilizes the Hopper architecture to accelerate AI applications. It’s a follow-up of the H100 GPU, released last year and previously Nvidia’s most powerful AI GPU chip. If widely deployed, it could lead to far more powerful AI models—and faster response times for existing ones like ChatGPT—in the near future.

According to experts, lack of computing power (often called “compute”) has been a major bottleneck of AI progress this past year, hindering deployments of existing AI models and slowing the development of new ones. Shortages of powerful GPUs that accelerate AI models are largely to blame. One way to alleviate the compute bottleneck is to make more chips, but you can also make AI chips more powerful. That second approach may make the H200 an attractive product for cloud providers.

What’s the H200 good for? Despite the “G” in the “GPU” name, data center GPUs like this typically aren’t for graphics. GPUs are ideal for AI applications because they perform vast numbers of parallel matrix multiplications, which are necessary for neural networks to function. They are essential in the training portion of building an AI model and the “inference” portion, where people feed inputs into an AI model and it returns results.

“To create intelligence with generative AI and HPC applications, vast amounts of data must be efficiently processed at high speed using large, fast GPU memory,” said Ian Buck, vice president of hyperscale and HPC at Nvidia in a news release. “With Nvidia H200, the industry’s leading end-to-end AI supercomputing platform just got faster to solve some of the world’s most important challenges.”

For example, OpenAI has repeatedly said it’s low on GPU resources, and that causes slowdowns with ChatGPT. The company must rely on rate limiting to provide any service at all. Hypothetically, using the H200 might give the existing AI language models that run ChatGPT more breathing room to serve more customers.

4.8 terabytes/second of bandwidth

According to Nvidia, the H200 is the first GPU to offer HBM3e memory. Thanks to HBM3e, the H200 offers 141GB of memory and 4.8 terabytes per second bandwidth, which Nvidia says is 2.4 times the memory bandwidth of the Nvidia A100 released in 2020. (Despite the A100’s age, it’s still in high demand due to shortages of more powerful chips.)

Nvidia will make the H200 available in several form factors. This includes Nvidia HGX H200 server boards in four- and eight-way configurations, compatible with both hardware and software of HGX H100 systems. It will also be available in the Nvidia GH200 Grace Hopper Superchip, which combines a CPU and GPU into one package for even more AI oomph (that’s a technical term).

Amazon Web Services, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure will be the first cloud service providers to deploy H200-based instances starting next year, and Nvidia says the H200 will be available “from global system manufacturers and cloud service providers” starting in Q2 2024.

Meanwhile, Nvidia has been playing a cat-and-mouse game with the US government over export restrictions for its powerful GPUs that limit sales to China. Last year, the US Department of Commerce announced restrictions intended to “keep advanced technologies out of the wrong hands” like China and Russia. Nvidia responded by creating new chips to get around those barriers, but the US recently banned those, too.

Last week, Reuters reported that Nvidia is at it again, introducing three new scaled-back AI chips (the HGX H20, L20 PCIe, and L2 PCIe) for the Chinese market, which represents a quarter of Nvidia’s data center chip revenue. Two of the chips fall below US restrictions, and a third is in a “gray zone” that might be permissible with a license. Expect to see more back-and-forth moves between the US and Nvidia in the months ahead.

https://arstechnica.com/?p=1983396

Enlarge / A look at the Nvidia T234 that could be the basis for a scaled-down custom chip in the next Nintendo console. (credit: Nvidia / Imgur)

In recent months, the long-running speculation surrounding Nintendo’s inevitable follow-up to the Switch has become more frequent and more specific, pointing to a release sometime in late 2024. Now, the pixel-counting boffins over at Digital Foundry have gone deep with some informed speculation on the system, dissecting leaked details on what they’re convinced is the Nvidia chip Nintendo will be putting in their Switch follow-up.

That chip is the Nvidia T239, a scaled-down, custom variant of the Nvidia Orin T234 that is popular in the automotive and robotics markets. While Digital Foundry can’t say definitively that this is the next Switch chip with “absolute 100 percent certainty,” the website points to circumstantial links and references to the chip in a number of leaks, a recent Nvidia hack, LinkedIn posts from Nvidia employees, and Nvidia’s own Linux distribution.

“From my perspective, the bottom line is that by a process of elimination, T239 is the best candidate for the processor at the heart of the new Nintendo machine,” Digital Foundry’s Richard Leadbetter writes. “With a mooted 2024 release date, there have been no convincing leaks whatsoever for any other processor that could find its way into the new Switch.”

Read 9 remaining paragraphs | Comments

https://arstechnica.com/?p=1981123