Nvidia’s next GPU may have 32GB of VRAM and 17 billion transistors

By Brad Bourque Published July 24, 2015

nvidia pascal gpu hbm confirmed nvidiahq — Shutterstock

Big changes may be on the horizon for Nvidia’s new GPU, Pascal, if the rumors Fudzilla is reporting are accurate. Information has leaked about the upcoming revision that points to more than twice the transistors of the current Maxwell chip found in the Titan X, with support for up to 32GB of RAM.

Nvidia has been stuck on the 28nm die size for several years now, but it looks like the Pascal chip will break that cycle, moving the process to FINFet 16nm. The sources also reported that it’s actually a very similar process to the 14nm used by Intel and Samsung, with identically sized logic gates. The Pascal GPU supposedly packs more than 17 billion transistors into its smaller size.

Recommended Videos

One of the targets for Pascal that Nvidia shared late last year was to bring HBM into the manufacturing process, and it looks like they’ve hit that mark too. AMD introduced the new RAM stacking technology on its Fury line of cards at E3, which lowers power needed to access memory, and speeds up bandwidth drastically. Nvidia’s first card with the tech will actually feature HBM 2.0, which gives it support for up to 32GB of RAM, although that’s likely to only end up on the highest-end card.

NvidiaRoadmap — Image used with permission by copyright holder

It’s no surprise that Nvidia is showing up late to the HBM party, especially when you consider AMD is co-developing the technology. There isn’t a lot of detail here for use in confirming if these rumors are true or not, but they’re certainly in line with what we can expect from Pascal, based on previous statements and roadmaps from Nvidia. Chances are we won’t see, or even hear more about, the Pascal GPU in 2015, and if we do it will just be more rumors.

Topics

Former Digital Trends Contributor

Brad Bourque is a native Portlander, devout nerd, and craft beer enthusiast. He studied creative writing at Willamette…

Computing

Intel’s next laptop chips may have a secret weapon

Intel Core Ultra Series 2 Lunar Lake chipset.

An upcoming Intel graphics solution, namely the Intel Arc 140T, has recently been spotted on GFXBench. The most interesting bit is that it offers a noticeable performance advantage over the Xe2-based Arc 140V iGPU that recently made its appearance on Intel’s latest Lunar Lake mobile CPUs.

A post by X (formerly Twitter) user Michael (@miktdt) compares GFXbench scores of the Arc 140T with two Arc 140V SKUs—one with 16GB memory and the other with 8GB. The 8GB Arc 140V reached 6,613 frames with an average of 106.7 framers per second (fps), while the 16GB version achieved 6,839 frames at 110.3 fps. However, the Intel Arc 140T, equipped with 16GB of memory, excelled with 11,056 frames at an average of 178.3 fps, surpassing the 16GB Arc 140V by 62%, despite both GPUs sharing eight Xe cores.

Computing

Nvidia’s next-gen GPU may be a hit for laptop gamers

Forza Horizon 5 running on the Alienware x14 R2.

Nvidia's RTX 50-series is still shrouded in mystery, but today, we got an exciting report about the upcoming RTX 5060 laptop GPU. According to the chairman of Hasee, a Chinese laptop and desktop manufacturer, the RTX 5060 will feature a sizable upgrade, all the while cutting back on the power consumption. Is the RTX 5060 going to be one of the best graphics cards for laptop gamers, including the ones who prefer thin notebooks?

The juicy scoop, first shared via Weibo, comes from Hasee's chairman Wu Haijun, who talked about the RTX 5060 during a meeting with the media. The chairman claimed that the next-gen xx60 graphics card will only consume 115 watts of power, as opposed to the 140W total graphics power (TGP) we've seen in the RTX 4060.

Computing

Nvidia’s next-gen GPUs may be delayed due to ‘design flaws’

Nvidia introducing its Blackwell GPU architecture at GTC 2024.

Dark clouds are looming over the future of Nvidia's best graphics cards. According to a new report, Nvidia told some of its partners that it will be delaying its upcoming Blackwell GPUs, and is now aiming for an early 2025 release instead. Delays are one thing, but the cause is perhaps the most worrying part of it all -- design flaws. What does this mean for Nvidia's RTX 50-series?

This worrying report originates from The Information, which cites two sources who helped produce the Blackwell chip, as well as its server hardware. Bloomberg recounts that the chips may be delayed by over three months at this point. Nvidia is preparing B100 and B200 chips for some of the world's most prominent tech companies, including Meta, Google, and Microsoft, so these delays could hit pretty hard.