NVIDIA's GeForce RTX 50 "Blackwell" Gaming GPUs under the GB200 series are likely to retain the same memory bus configurations as Ada.
18.02.2024 - 10:37 / wccftech.com / Muhammad Zuhair / Ai / Ai Gpus
NVIDIA's supply chain of its AI accelerators has reportedly witnessed a massive improvement as the firm reduces delivery times drastically.
The news comes from UBS analysts, who have claimed that NVIDIA's efforts to diversify their supply chain have finally started to work out, as the firm has now reduced the wait time of its AI GPUs from 8-11 months to 3-4 months. The reason behind this significant cutback in delivery times hasn't been disclosed yet. Still, Team Green has been working for quite some time now when it comes to upscaling its partners in the supply chain, especially with the inclusion of Samsung Foundry to manage the HBM department.
Apart from this, existing partners like TSMC and SK hynix have also rapidly expanded their production capacities, with TSMC reportedly reaching 32,000 units of CoWoS output by the end of this year, which is huge because at the start of 2023, this number is much lower. Moreover, the same situation has been in the firm's wafer output, which means that the global semiconductor supply is in a much better position compared to what it was a year ago.
However, not everything is rainbows for NVIDIA since the firm has seen a decline in AI consumer interest, not because of its products but because of the political situation the company has been dragged into. With the ban on exporting the firm's cutting-edge H100 and A100 AI GPUs to China, the firm has already seen a decline in sales since Chinese clients are a huge part of NVIDIA's overall revenue from the data center segment. With "harsher" restrictions being imposed, Team Green has been cut off, and the new "workarounds" they devised didn't get the attention needed.
Moreover, competitors like AMD and Huawei are also moving towards closing the gap with their existing offerings, which has impacted NVIDIA, even though not much as of yet. NVIDIA still retains its dominance over 90% of the AI market so it will be interesting to see how the firm copes with the situation since the firm can't be caught slacking at this stage.
News Source: MyDrivers
NVIDIA's GeForce RTX 50 "Blackwell" Gaming GPUs under the GB200 series are likely to retain the same memory bus configurations as Ada.
Upcoming NVIDIA and AMD GPUs with GDDR7 memory could have 16 GB dies or greater, according to a reputable PC tech insider. If the specifications are true, the new GDDR7 NVIDIA and AMD graphics cards could offer incredible performance in demanding PC titles in the future.
The first generation of NVIDIA & AMD GPUs to utilize GDDR7 memory are most likely going to stick with 16 Gb dies for 2 GB VRAM capacities.
Following NVIDIA's Chat with RTX launch, AMD is now offering users their very own localized & GPT-based LLM-powered AI chatbot which can run on Ryzen AI CPUs & Radeon 7000 GPUs.
Dragon's Dogma 2's vast open world has a very surprising source of inspiration: Rockstar's GTA 5.
For many, the promise of a game lasting upwards of 100 hours is no longer the selling point that it used to be. People who grew up with games may now find themselves with families or other responsibilities — and so, at least from what we see in comments sections and the like, there's a clear demand for shorter titles that offer more focused experiences.
NVIDIA has banned the use of CUDA libraries on other platforms like AMD & Intel, as the firm adds a new warning with CUDA's EULA.
NVIDIA's qualification tests seem to be giving HBM manufacturers a hard time, as yield rates are significantly low compared to traditional memory products.
Ahead of launching, Alan Wake 2 made the news because it required graphics cards with hardware support for DirectX 12 Ultimate's Mesh Shaders feature. That left unsupported GPUs like NVIDIA's GTX 1000 series and AMD's RX 5000 Series in the dust.
NVIDIA's Blackwell AI GPU lineup will include two major accelerators, the B100 for 2024 and the B200 for 2025, as revealed by Dell.
Nvidia Chief Executive Jensen Huang on Friday said that artificial general intelligence could - by some definitions - arrive in as little as five years.
Sign up for the GI Daily here to get the biggest news straight to your inbox