HBM seemingly confirmed for Nvidia’s Pascal GPU

By Brad Jones Published October 5, 2015

nvidia pascal gpu hbm confirmed nvidiahq — Shutterstock

Despite near radio silence from Nvidia, we’ve heard quite a lot about the company’s Pascal project in recent months. It’s long been rumored that the GPU will release in 2016, and the increasing frequency of information leaks seems to suggest that we might be about to start seeing some official word on specifics.

The latest leak of information comes from GTC Taiwan 2015, according to a report from Tech Spot. An image of the GPU taken from the conference apparently confirms that it will be making use of High Bandwidth Memory, as has been expected for some months.

Recommended Videos

High Bandwidth Memory — commonly referred to as HBM — is a relatively recent innovation in the GPU manufacturing process, and has recently been a major focus for AMD. The process stacks memory on the chip itself, offering improvements in terms of power, efficiency, and form factor.

Get your weekly teardown of the tech behind PC gaming

Pascal is set to take advantage of HBM 2.0, a refined version of the process which allows for memory bandwidth close to 1 TB/s — roughly double that of HBM 1.0. Had there been any doubt before, this should establish that Nvidia is aiming high with the Pascal project.

Pascal will be going head to head with AMD’s impending Arctic Islands line, which also uses HBM 2.0. However, given that AMD has had a major hand in developing HBM, the fact that Nvidia is comparatively inexperienced with the tech might suggest that the company will face an uphill battle against its longtime rival.

However, we’ll have to wait until 2016 to see how that contest between Nvidia and AMD shakes out. Until then, it’ll be interesting to see how long Nvidia waits before sharing more information on Pascal. Given that we’re nearing the end of 2015, it would follow that official details can’t be too far off.

Topics

Former Digital Trends Contributor

Brad is an English-born writer currently splitting his time between Edinburgh and Pennsylvania. You can find him on Twitter…

Computing

This GPU just beat the RTX 4090 — and Nvidia didn’t make it

The board of the RTX 4090 Super graphics card.

Modders are doing what Nvidia won't. The team at Teclab put together a Frankenstein graphics card, which it calls the RTX 4090 Super, that was able to beat the RTX 4090 by 13% in benchmarks.

You can't buy this graphics card, of course, but it's an interesting look into how splicing together the best components can deliver big performance gains. The heart of the RTX 4090 Super is the RTX 4090 GPU, which is Nvidia's AD102 die. Teclab changed everything else about the graphics card, though.

Computing

Report: High RAM prices will continue to skyrocket

Product shot of the Crucial DDR5 memory modules on a blue background.

DDR5 RAM has become the new standard for most consumer-grade PCs and laptops. While DDR4 is still around, the first DDR5 chips came out in 2018. However, the standard was officially released in 2020, and actual commercial kits made their first appearance by mid-2021. Back then, a kit of twin 16GB sticks at DDR5-4800 speeds launched at a price tag of $311. That’s more than three times the price of what you'll pay today.

But that may change soon as DDR5 RAM prices are expected to continuing increasing in the coming months. Analysts from market research firm TrendForce suggest that the primary reason for this surge is being attributed to the manufacturers who are reallocating resources toward the production of High Bandwidth Memory (HBM).

Computing

Meet Blackwell, Nvidia’s next-generation GPU architecture

Nvidia introducing its Blackwell GPU architecture at GTC 2024.

We finally have our first taste of Nvidia's next generation of GPUs, named Blackwell. Sure, they're built for enterprises, and no, they won't run Cyberpunk 2077 (at least not officially). But this is the first look we've have at what Nvidia is cooking up for its RTX 50-series GPUs, which are rumored to launched sometime in the next year.

The GPU we have today is the B200 -- Blackwell 200, if you can spot it -- that comes packed with 208 billion transistors. The architecture is built on TSMC's 4NP node, which is an enhanced version of the 5nm node. It's a little surprising given that Nvidia's Ada Lovelace GPUs are built with TSMC's 4N node -- one refinement step away from 4NP. Nvidia notes that it's using a custom version of this process, however.