Urgent.News

What's breaking now, across thousands of outlets.

Tech

My 12GB RTX 4070 Ti runs multiple DLSS 5 layers thanks to one OptiScaler setting

Nothing like an open-source tool to squeeze more power

My 12GB RTX 4070 Ti runs multiple DLSS 5 layers thanks to one OptiScaler setting

A user reported that their 12GB RTX 4070 Ti graphics card can handle multiple DLSS 5 layers with the help of one OptiScaler setting. The neural rendering files released by Nvidia for NBA 2K27 have been shared by the open-source and modding community, allowing DLSS 5 to be injected into many games. However, this adds a significant performance cost, especially when multiple neural rendering passes are added.

OptiScaler offers a simple feature that changes what the user can achieve with their 12GB GPU. It reports the RTX 4070 Ti as an RTX 50-series GPU, unlocking multi-frame generation and allowing the use of DLSS 5 and neural rendering in games. By manipulating the order of the neural rendering pass in the pipeline, the user can achieve better performance while still experiencing the visual benefits of DLSS 5.

This technique is particularly useful when running games at high-to-ultra graphics presets with ray tracing at 1440p or 4K resolution, as it prevents the RTX 4070 Ti from reaching its VRAM limit. OptiScaler grants access to DLSS 5 features that even RTX 50-series owners do not have officially, such as adjusting the position of neural rendering in the rendering pipeline and stacking multiple neural rendering passes.

Nvidia's implementation lacks the same level of control over the number of neural rendering passes. The user's experiment demonstrates the power of software-defined DLSS 5 and how open-source tools can push the boundaries of hardware capabilities, even for last-gen GPUs.

Written by urgent.news from XDA Developers's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at xda-developers.com →

More in Tech

The first hour after a token launch: a Telegram lockdown checklist

The first hour after a launch is when a Telegram group is most exposed: new members flood in, nobody knows who the real admins are yet, and scammers know it. This is the checklist I'd run, in order.

  • Pin real contract address in pinned message, restrict DMs to admins
  • Set captcha, block links/media for new members during launch
  • Assign monitor for new joins, watch for usernames resembling admins

Claude Haiku 5.5 400 Error? 5 Fixes That Work (2026)

Claude Haiku 5.5 400 Error? 5 Fixes That Work (2026) Last updated: October 31, 2026 Short answer: Changing only the model ID to Claude Haiku 5.5 can turn a working integration red, because requests…

  • Claude Haiku 5.5 generates 400 errors when model ID changes
  • Anthropic's migration guide provides five fix categories
  • October 2026 pricing confirms $0.10/1M input token tier for Haiku 5.5

The 2017 S3 Outage Was One Wrong Input. AWS Fixed the Tool, Not the Engineer

Book: System Design Pocket Guide: Fundamentals Also by me: Thinking in Go (2-book series) — Complete Guide to Go Programming + Hexagonal Architecture in Go My project: Hermes IDE | GitHub — an IDE for…

  • S3 team executed command with incorrect input, removing more servers than intended
  • Incident caused four-hour outage from 9:37 AM to 1:54 PM Pacific
  • AWS fixed the tool, not the engineer, implementing safety changes to prevent recurrence

More from Saturday 7 November →