PCIe 6.0 connectivity demoed at NVIDIA GTC 2024: ready for Blackwell B100, B200, GB200 AI GPUs

Anthony Garreffa | Video Cards & GPUs | Mar 27, 2024 11:59 PM CDT

Astera Labs was demoing its working PCIe 6.0 connectivity with next-gen Aries PCIe 6.0 retimers, spotted on the NVIDIA GTC 2024 showfloor by our friends at Tom's Hardware. Check it out:

PCIe 6.0 connectivity demoed at NVIDIA GTC 2024: ready for Blackwell B100, B200, GB200 AI GPUs

We don't even have PCIe 5.0 across the board with all PC users and gamers, but PCIe 6.0 is coming with NVIDIA's new Blackwell B100, B200, and GB200 AI GPUs all ready for Gen6. AI workloads require ridiculous amount of throughput through the GPUs to process everything at full speed, with NVIDIA using its in-house custom NVLink and NVSwitch to offer massive GPU-to-GPU communication, offering up to 1.8TB/sec of bandwidth on GB200-powered systems.

Not every customer wants the entire NVIDIA solution, which is where PCIe 5.0 x16 connections step in for the current fleet of AI GPUs, but Blackwell introduces PCIe 6.0, doubling the 128GB/sec of bandwidth of an x16 connection on PCIe 5.0 to a huge 256GB/sec bandwidth on PCIe 6.0 x16.

Continue reading: PCIe 6.0 connectivity demoed at NVIDIA GTC 2024: ready for Blackwell B100, B200, GB200 AI GPUs (full post)

Digital Xbox Series X spotted, no disc drive and all white like the Series S

Kosta Andreadis | Gaming | Mar 27, 2024 11:31 PM CDT

Images of a yet-to-be-announced digital-only Xbox Series X have appeared online, showcasing an all-white version of the current console sans optical drive. The dimensions and size look the same in the surprisingly low-resolution leaked images presented by eXtas1s over on exputer.

Digital Xbox Series X spotted, no disc drive and all white like the Series S

Many have been expected to see the all-digital Xbox Series X sometime this year because it was included in that massive Microsoft document leak last year. The version of the console showcased there presented a cylindrical-shaped console with notes that the all-digital console would be released sometime in June or July of this year.

The leaked images, if real, show that Microsoft has moved on from creating a new look for the digital Xbox Series X console. Instead, it will be the same box in a white form factor to match the drive-free baseline Xbox Series S console. It's also reported to include an improved heatsink.

Continue reading: Digital Xbox Series X spotted, no disc drive and all white like the Series S (full post)

China president: we don't need ASML, 'no force can stop the pace of China' tech development

Anthony Garreffa | Business, Financial & Legal | Mar 27, 2024 11:00 PM CDT

ASML is the only company in the world with advanced High-NA EUV lithography machines, with Intel as its latest $380 million customer. Now, Chinese President Xi Jinping has told visiting Dutch Prime Minister Mark Rutte-ASML, who is in the Netherlands, that attempts to restrict China's success to technology will not stop his country's advancement.

China president: we don't need ASML, 'no force can stop the pace of China' tech development

The Netherlands slapped down export licensing requirements in 2023, stopping the sale of high-end machinery that is capable of making the world's most advanced processors. The US started by blocking access to advanced chips and the equipment required to make them, citing security concerns with China, seeing the Netherlands impose similar restrictions.

Chinese President Xi Jinping said: "The Chinese people also have the right to legitimate development, and no force can stop the pace of China's scientific and technological development and progress."

Continue reading: China president: we don't need ASML, 'no force can stop the pace of China' tech development (full post)

MAINGEAR PRO AI workstation PC with Phison aiDaptiv+ platform: uses SSDs, DRAM to boost AI perf

Anthony Garreffa | Storage | Mar 27, 2024 10:28 PM CDT

Phison had something very cool to show off at NVIDIA GTC 2024 last week, with a single MAINGEAR PRO AI workstation PC packing 4 x NVIDIA RTX 6000 Ada workstation GPUs using its in-house aiDaptiv+ software to use SSDs and DRAM to make up for the lack of VRAM, using it to super-speed AI workloads on the cheap.

MAINGEAR PRO AI workstation PC with Phison aiDaptiv+ platform: uses SSDs, DRAM to boost AI perf

By cheap, we mean cheaper than using full-fledged NVIDIA Hopper H100 or new H200 AI GPUs, let alone the next-gen Blackwell B100 and B200 AI GPUs that will be very hard to come by this year, and a huge chunk into 2025. Phison's aim with its aiDaptiv+ platform is to lower the (cost) barrier of AI LLM training by using systems' already-there DRAM and SSDs to augment the amount of GPU VRAM available for AI training.

Phison says that achieving intense generative AI training workloads at a far lower cost than using standard GPUs will come at a cost: reduced AI training performance and longer training times. If you want the best, you'll spend magnitudes more money on it, but more off-the-shelf AI systems from the likes of MAINGEAR and its PRO AI system, putting Phison's new aiDaptiv+ platform in between is a big win.

Continue reading: MAINGEAR PRO AI workstation PC with Phison aiDaptiv+ platform: uses SSDs, DRAM to boost AI perf (full post)

Marvel Rivals is a new team-based hero PVP game from Netease that looks a lot like Overwatch

Kosta Andreadis | Gaming | Mar 27, 2024 10:04 PM CDT

Described as a "Super Hero team-based PVP shooter," Marvel Rivals from NetEase and Marvel Games gives off serious Overwatch and Valorant vibes. The free-to-play PC game was announced earlier today, and the announcement trailer showcases a stacked lineup of Marvel characters battling it out, including Black Panther, Spider-Man, Magneto, Rocket Raccoon, Iron Man, and more.

Marvel Rivals is a new team-based hero PVP game from Netease that looks a lot like Overwatch

According to the press release, the global team developing the game comprises talent who has worked on franchises like Call of Duty and Battlefield. The game will, of course, fall into the live service online competitive space currently populated by Overwatch, Valorant, and even Counter-Strike, with the game set to feature an "ever-evolving all-star squad of Super Heroes and Super Villains." So expect new maps, characters, Season Passes, and all that.

Marvel is calling it one of the most ambitious game projects it has worked on to date. It is set to feature characters from the worlds of the Avengers, Guardians of the Galaxy, X-Men, and more. An Alpha Test for Marvel Rivals begins in May, so a full or Early Access launch should happen this year.

Continue reading: Marvel Rivals is a new team-based hero PVP game from Netease that looks a lot like Overwatch (full post)

Introspect ships the world's first GDDR7 memory test system, with up to 40Gbps GDDR7

Anthony Garreffa | Video Cards & GPUs | Mar 27, 2024 9:33 PM CDT

Introspect Technology is a member of JEDEC, the body that covers memory -- including the next-gen GDDR7 memory standard -- with Introspect shipping the world's first GDDR7 memory test system.

Introspect ships the world's first GDDR7 memory test system, with up to 40Gbps GDDR7

The just-launched Introspect M5512 GDDR7 Memory Test System is the world's first commercial solution for testing JEDEC's fresh new JESD239 Graphics Double Data Rate (GDDR7) SGRAM specification. GDDR7 is a category-creating memory solution that enables graphics memory engineers, GPU design engineers, product engineers in both memory and GPU spaces, and system integrators to quickly design new GDDR7 memory devices, debug protocol errors, characterize signal integrity, and perform detailed memory read/write functional stress testing without requiring any other tool.

The new M5512 GDDR7 Memory Test System is the perfect solution for these engineers, AMD and NVIDIA, and AIB partners to get ready to launch GDDR7-powered graphics cards, the first of which will be NVIDIA's next-gen GeForce RTX 50 series GPUs.

Continue reading: Introspect ships the world's first GDDR7 memory test system, with up to 40Gbps GDDR7 (full post)

Microsoft's Copilot AI will run locally on AI PCs that have at least 40 TOPS of NPU performance

Kosta Andreadis | Artificial Intelligence | Mar 27, 2024 9:02 PM CDT

We already know that in the age of the Windows AI PC, devices are set to arrive with dedicated Copilot keys. Microsoft's AI is on track to be integrated with all parts of the operating system. Today, Intel executives at Intel's AI Summit in Taipei confirmed that Copilot AI services will soon run locally on PCs - as long as they meet a certain performance threshold.

Microsoft's Copilot AI will run locally on AI PCs that have at least 40 TOPS of NPU performance

That threshold is 40 TOPS of performance on the Neural Processing Unit (NPU) found in an AI PC. Based on this information, which Tom's Hardware confirmed, the definition of an AI PC has been given a specific hardware or performance baseline. Anything below 40 TOPS will still be able to run Copilot AI tasks and processes in the cloud, but it will serve as a way to differentiate hardware-based 'AI PCs' versus could-only devices.

Though it might take a generation or two to get there, Intel says it has next-gen products lined up that will fall into this 40 TOPS of performance category. Currently, The Meteor Lake NPU in Core Ultra chips offers only 10 TOPS performance, with AMD's Ryzen Hawk Point offering 16 TOPS.

Continue reading: Microsoft's Copilot AI will run locally on AI PCs that have at least 40 TOPS of NPU performance (full post)

NVIDIA's new Hopper H200 AI GPU tested: 3x faster GenAI with TensorRT-LLM in MLPerf 4.0 results

Anthony Garreffa | Artificial Intelligence | Mar 27, 2024 8:32 PM CDT

NVIDIA might have just announced its next-generation Blackwell B200 AI GPU, but the beefed-up Hopper H200 AI GPU is smashing performance records in the very latest MLPerf 4.0 results.

NVIDIA's new Hopper H200 AI GPU tested: 3x faster GenAI with TensorRT-LLM in MLPerf 4.0 results

NVIDIA's optimizations on TensorRT-LLM have been a non-stop chain of progression since the company released its AI Software suite last year. There were major performance increases from MLPerf 3.1 results to MLPerf 4.0, with NVIDIA amplifying Hopper's AI performance.

Using these new TensorRT-LLM optimizations, NVIDIA has pulled out a huge 2.4x performance leap with its current H100 AI GPU in MLPerf Inference 3.1 to 4.0 with GPT-J tests using an offline scenario. With server-based scenarios using GPT-J, NVIDIA's current H100 AI GPU had a huge 2.9x increase in MLPerf 3.1 to 4.0 performance.

Continue reading: NVIDIA's new Hopper H200 AI GPU tested: 3x faster GenAI with TensorRT-LLM in MLPerf 4.0 results (full post)

Cocoa bean price hits record high: $10,000 per metric ton, biggest shortfall in over 60 years

Anthony Garreffa | Business, Financial & Legal | Mar 27, 2024 7:52 PM CDT

If you thought the insatiable AI GPU demand was (and is, and will continue) nuts, well... the cocoa beans market is in trouble, with the price of cocoa beans skyrocketing. The price of cocoa beans has smashed through $10,000 a ton for the first time in the futures market.

Cocoa bean price hits record high: $10,000 per metric ton, biggest shortfall in over 60 years

The last major cocoa tree planting spree happened throughout the early 2000s in West Africa, which, if you didn't think was important to the global cocoa business, you're wrong: West Africa counts for around 75% of the global supply of cocoa beans. After already bad weather amplified by the ongoing El Nino phase of the Pacific Ocean, cocoa trees throughout Ghana and the Ivory Coast have been decimated.

Fast-forward to today, and the world is heading to an annual supply deficit of between 300,000 and 500,000 tons of cocoa beans, the largest shortfall in over 60 years. Right now, the price of cocoa is more expensive than copper on the commodities market, and we're sprinting directly into Easter, which is when chocolate makers have now increased pricing for consumers after multiple years of subpar cocoa harvests in West Africa with another on the way.

Continue reading: Cocoa bean price hits record high: $10,000 per metric ton, biggest shortfall in over 60 years (full post)

Intel's next-gen Lunar Lake-MX spotted: new CPU cores, Xe2 GPU, and on-package LPDDR5X memory

Anthony Garreffa | CPU, APU & Chipsets | Mar 27, 2024 7:08 PM CDT

Intel's next-gen Lunar Lake-MX mobile processor for future-gen thin and light laptops has been leaked by our good friend Igor at Igor's Lab. Check it out:

Intel's next-gen Lunar Lake-MX spotted: new CPU cores, Xe2 GPU, and on-package LPDDR5X memory

The next-gen Intel Lunar Lake-MX processor will have up to 8 cores, with 4 powerful "Lion Cove" P-Cores and 4 efficient "Skymont" E-Cores. There's no need for massive amounts of cores in thin and light laptops, as Intel is focusing on balancing performance and efficiency, something Lunar Lake-MX is expected to deliver.

On the integrated GPU side of things, Lunar Lake will feature an Intel Arc Battlemage GPU with a maximum of 64 Xe2-based Execution Units (EUs) and 8 x Xe2 GPU cores. Igor's report says that this will provide a "significant leap in graphics performance that could set new benchmarks for integrated graphics solutions".

Continue reading: Intel's next-gen Lunar Lake-MX spotted: new CPU cores, Xe2 GPU, and on-package LPDDR5X memory (full post)