Skip to main content
  1. Home
  2. Computing
  3. News

Valve already has a higher-performance, cheaper Lighthouse chip

Add as a preferred source on Google

Although the tracking solution for the HTC Vive headset is certainly our preferred method of VR tracking right now, Valve already has access to a better version. The new sensor chip is one the headset co-creator is recommending anyone use if they build a new SteamVR headset and peripherals.

The HTC Vive’s tracking system works by sensors on the headset and controllers detecting laser pulses and sweeps from the Lighthouse base stations mounted on the wall. Combined with a computational chip built into each sensor, they let the system generate an accurate interpretation of where the user is in the real world and translate that into the game.

Recommended Videos

It’s accurate and efficient, but Valve and its partners can do better. The new version of the chip that helps translate all of that data is far simpler, cheaper, and more efficient, too. The TS3633 integrated circuit – created by Triad Semiconductors in partnership with Valve – has just nine components, versus the 41 in the original chip’s design.

These chips are used as part of every sensor on the Vive and its controllers, and reducing the components in the chip by that much equates to around 2,560 less components across the entire headset, according to RoadtoVR. That’s a big savings in cost and complexity, which should make the headset more reliable in some cases.

While that’s all well and good for those who don’t have a VR headset yet, those who do have headsets and are willing and able to pay for another in the future, want to know about performance. You’ll be pleased to know that the TS3633 improves that, too.

Longer range detection sensitivity has been improved, along with an improvement in the detection of pulse width — which should improve accuracy. We’re also told by Triad’s vice president of marketing & sales, Reid Wender, that these new sensors will reduce that part of the headset’s power draw by 50-70 percent. That’s great for those trying to build low-power VR systems, but is even more important if we consider that a second-generation Vive is likely to be completely wireless.

Jon Martindale
Jon Martindale covers how to guides, best-of lists, and explainers to help everyone understand the hottest new hardware and…
AI chatbots will agree and misinform if you just put a little pressure, warns research
ChatGPT 3.5 proved the most vulnerable when false claims were repeated over and over
Claude AI on an iPhone.

AI chatbots have a well-documented habit of hallucinating information and sometimes agreeing with users even when they are wrong. A new study suggests that simply refusing to take no for an answer can make the problem worse.

Researchers from the University of Arizona tested seven AI models, including GPT-3.5, GPT-4o, GPT-4o-mini, Claude 3.5 Sonnet, Gemini 1.5 Pro, Llama 3 70B, and DeepSeek-R1. Instead of judging them from a single response, researchers kept conversations going while repeatedly feeding the models information they knew was false.

Read more
Google brings its best AI music model Lyria 3.5 to the Gemini app
Developers get full API access through Google AI Studio.
Google-gemini-lyria-3.5

Google has added Lyria 3.5, its most advanced music generation model yet, to the Gemini app. Previously available through Google's AI filmmaking tool Flow, the model is now rolling out to all Gemini users, making it easier to generate polished songs, instrumentals, and soundtracks from simple text prompts or even photos.

Lyria 3.5 makes AI-generated music sound more natural

Read more
AMD’s new workstation packs 576GB of GPU memory and enough horsepower for trillion-parameter AI models
AMD just built a desktop workstation with more raw memory than most small data centers had a decade ago.
People, Person, Performer

AMD just built a desktop that really has no business sitting on an actual desk. At IFA 2026, the company unveiled the Threadripper Halo Station.

As the name probably already suggests, it is a liquid-cooled machine that can run AI models with more than a trillion parameters without touching the cloud even once. If you don't already know, those numbers are reserved for server shelves, not someone's personal desk.

Read more