A new AI gaming feature introduced by NVIDIA for its RTX hardware is off to a terrible start.
This week, NVIDIA unveiled their Project G-Assist during a broader update to the NVIDIA App. Alongside this release, they introduced some fresh Deep Learning Super Sampling (DLSS) adjustments, custom control panel settings, and Manufacturer Friendly Game (MFG) compatibility for several titles.
Initially, when Project G-Assist debuted on personal computers, I found myself taken aback by the swiftly negative responses it received. However, considering some people’s apprehension towards AI and NVIDIA’s less than stellar performance in the last few months following the release of RTX 5000 GPUs, I wasn’t entirely unprepared for this reaction.
A substantial number of unhappy gamers swiftly populated several Reddit discussions, examining both the necessary system requirements and the AI tool’s capabilities.
With absolute assurance, I’d tell you that the enjoyment level doesn’t escalate any further when you begin utilizing Project G-Assist on a gaming computer.
First off, let’s clarify what Project G-Assist from NVIDIA is intended to accomplish.
What is NVIDIA Project G-Assist?
In my analysis, I’ve found that Project G-Assist, as detailed by NVIDIA, is a compact language model designed to operate directly on your personal computer’s RTX hardware.
Theoretically, NVIDIA aims for G-Assist to be accessible whenever you require it, offering “instant diagnostics and suggestions to relieve performance issues, enhance energy savings, fine-tune game settings, boost your GPU’s performance, and a host of other functions.
G-Assist, theoretically speaking, can understand and execute “ordinary language commands,” which means you can provide it with common instructions (for now, only in English) and it will carry them out for you.
NVIDIA views it as an all-encompassing gaming companion that can control lighting and cooling fans on compatible devices, along with providing essential computer components details to you.
It sounds neat, but like most things AI, it’s not as impressive once you start using it.
This brings me to the first major sticky spot: system requirements.
Yes, NVIDIA’s Project G-Assist requires at least 12GB of VRAM to run.
The statement that NVIDIA made at their CES 2025 keynote, which suggested that VRAM (Video Random Access Memory) is less significant due to AI advancements, may still hold true when it comes to gaming. However, it’s worth noting that NVIDIA’s tools no longer support 8GB and 10GB RTX GPUs now.
Given whispers about the potential release of RTX 5060 and RTX 5060 Ti models featuring 8GB VRAM, it appears that NVIDIA’s AI-focused future may become increasingly limited to a select few.
Here’s another tidbit from NVIDIA explaining how G-Assist works behind the scenes:
As an analyst, I can share that beneath its surface, G-Assist has been upgraded to utilize a Llama-based Instruct model, boasting 8 billion parameters. This means it’s able to cram a considerable amount of language understanding into a minuscule fraction of the size typically seen in large-scale AI models. This improvement enables G-Assist to operate locally on GeForce RTX hardware. Furthermore, given the rapid advancements in Small Language Models (SLM) research, these compact models are consistently becoming more capable and efficient at regular intervals.
When you press Alt+G to request assistance from G-Assist, such as when adjusting graphics settings or monitoring GPU temperatures, your GeForce RTX GPU temporarily dedicates a segment of its processing power for AI inference. If you’re actively gaming or running another resource-intensive application at the same time, there might be a momentary decrease in rendering speed or AI task completion during this brief period. Once G-Assist completes its assigned task, the GPU resumes delivering optimal performance back to the game or application.
The project named G-Assist seems promising, but unfortunately, it’s hindered by certain hardware limitations and, to my disappointment, it was launched in an unfortunate condition.
NVIDIA Project G-Assist tested at home — it’s completely unusable
To start off, let me clarify that G-Assist has shown signs of functioning correctly for certain users.
I discovered a Reddit user who shared an experience where they were able to successfully undervolt their GPU, thanks to recommendations made by an AI assistant.
In a test conducted by my coworker Richard Devine on his gaming computer equipped with an RTX 5080 Founders Edition graphics card, it was found that the performance of G-Assist was unworkable during gameplay, despite the GPU’s impressive 16GB VRAM and high-end capabilities.
What’s the point of a gaming assistant if you can’t use it while you’re running a game?
Upon activation, G-Assist seemed to only reduce the frame rates drastically, sometimes dropping them from high triple figures to mere singles. In extreme instances, it even caused the entire system to freeze solid.
It’s possible that one part of the issue could originate because G-Assist needs the NVIDIA overlay to operate, and this overlay may potentially impact performance.
Divine tried running Project G-Assist alongside Call of Duty: Black Ops 6, Atomfall, and Avowed, but in each instance, the frame rate significantly dropped when activated.
It’s more problematic, whether you ask G-Assist a query or issue a command, to find that Devine’s system got stuck during the AI’s prolonged processing time.
Even asking a simple question like “Is G-Sync enabled?” nearly crashed the games.
From where I stand, it appears that Project G-Assist isn’t quite at its public debut yet, and it seems it might encounter challenges when running on lower-end RTX graphics cards, despite them boasting 12GB or more of VRAM.
As a researcher, I find myself optimistic about the potential growth and improvement of NVIDIA’s G-Assist, despite its initial bumps in the road. The more users engage with it, the more refined and effective it is likely to become, offering a promising outlook for its future.
As a researcher, I acknowledge that it will be challenging to garner much constructive feedback when the product proves to be incompatible with some of the leading video games.
There is some good news.
You can easily remove Project G-Assist from the NVIDIA App, and it appears that no other functions will be affected in the process, based on my current understanding.
To uninstall the NVIDIA App, simply open it, find the three vertical dots situated at the top-right corner within the G-Assist interface, then select ‘Uninstall’.
I’ve been wondering, have you given G-Assist a spin on your computer yet? If so, what kind of hardware setup and games are you using it with? How did it go for you? Feel free to share your experience in the comments!
Read More
- Solo Leveling Season 3: What You NEED to Know!
- ETH/USD
- Disney’s Snow White Dwarfs Controversy: THR’s Shocking Edit Exposed!
- Solo Leveling Arise Amamiya Mirei Guide
- Kanye West’s Wife Stuns Completely Naked at 2025 Grammys
- Shocking Truth Revealed: Is Cassandra Really Dead in Netflix’s Terrifying Mini-Series?
- AUCTION/USD
- See Channing Tatum’s Amazing Weight Loss Transformation
- Kim Kardashian’s Hilarious NSFW Confession Gets Silenced by Animal Noises!
- Jelly Roll’s 120-Lb. Weight Loss Leads to Unexpected Body Changes
2025-03-28 17:10