marcosalvi Profile Banner
Marco Salvi Profile
Marco Salvi

@marcosalvi

Followers
5K
Following
12K
Statuses
15K

Computationally responsible deep learning and real-time graphics research @ NVIDIA. Personal opinions only.

Seattle
Joined April 2009
Don't wanna be here? Send us removal request.
@marcosalvi
Marco Salvi
8 years
My SIGGRAPH presentation with notes is online. Enjoy and feel free to ask questions. Thanks @mirror2mask and @aaronlefohn for inviting me!
@mirror2mask
Natalya Tatarchuk 🇺🇦
8 years
Slides for deep learning application for real-time rendering from @marcosalvi posted on #rtropenproblems
Tweet media one
4
51
194
@marcosalvi
Marco Salvi
5 hours
RT @PeteButtigieg: Huh I thought he was elected to bring prices down and make our lives better. Even my Trumpiest neighbors weren’t lookin…
0
9K
0
@marcosalvi
Marco Salvi
8 hours
RT @PeteButtigieg: You are not powerless, and he is not unstoppable. Look at, and learn from, the funding freeze and how quickly he was for…
0
14K
0
@marcosalvi
Marco Salvi
9 hours
RT @SenAdamSchiff: This morning, I sent a letter to the Office of Government Ethics (OGE) regarding @elonmusk's potential conflicts of inte…
0
8K
0
@marcosalvi
Marco Salvi
11 hours
@funkkiy @SebAaltonen It supports mip mapping and anisotropic filtering is achieved via STF and temporal reconstruction, often yielding better results than HW aniso due to filtering after shading:
Tweet media one
0
0
0
@marcosalvi
Marco Salvi
11 hours
@SebAaltonen NTC supports mip mapping. The desired LOD is provided as input to the decompressor/decoder and one can use multiple (pairs of) latent grids at different resolutions. Details are provided in the paper:
0
0
14
@marcosalvi
Marco Salvi
2 days
RT @DJSnM: NVIDIA are pushing a graphics future where the GPU renders at low resolution and uses AI to interpolate more detail. This is ba…
0
635
0
@marcosalvi
Marco Salvi
2 days
RT @davidaxelrod: A year ago, Biden refused the traditional Super Bowl interview and the unparalleled audience it would have yielded. That…
0
627
0
@marcosalvi
Marco Salvi
4 days
RT @jonfavs: Elon now bragging about destroying an $800 million/year agency that has returned $21 billion to consumers who've been screwed…
0
6K
0
@marcosalvi
Marco Salvi
4 days
RT @NVIDIAGameDev: 👀Now available: RTX Mega Geometry with Vulkan Samples. Render high quality assets with higher instance counts and grea…
0
37
0
@marcosalvi
Marco Salvi
4 days
RT @NVIDIAGameDev: Render path-traced hair in real time.⚡ Available today, developers can use linear swept sphere (LSS) to ray trace hair…
0
48
0
@marcosalvi
Marco Salvi
4 days
RT @more_fps: We've just released NVIDIA RTX Kit - a collection of SDKs for developing next-gen rendering applications, including Neural Sh…
0
23
0
@marcosalvi
Marco Salvi
4 days
RT @NVIDIAGameDev: Explore the latest SDKs available now through NVIDIA RTX Kit, a suite of neural rendering technologies that’s powering t…
0
8
0
@marcosalvi
Marco Salvi
5 days
RT @milesmacklin: NVIDIA Warp 1.6.0 is out! Including native Blackwell support, up to 2x faster geometric queries, and tile-based Cholesky…
0
36
0
@marcosalvi
Marco Salvi
8 days
RT @VulkanAPI: NVIDIA has just released Vulkan samples for RTX Mega Geometry. They should run on all RTX GPUs using the new drivers: https…
0
36
0
@marcosalvi
Marco Salvi
11 days
RT @Jukanlosreve: The greatest analysis in history. Within a few days, SemiAnalysis’s analysis will dominate headlines worldwide. https://t…
0
66
0
@marcosalvi
Marco Salvi
11 days
RT @sama: first full 8-rack GB200 NVL72 now running in azure for openai—thank you @satyanadella and jensen!
0
624
0
@marcosalvi
Marco Salvi
13 days
RT @docmilanfar: amazing he also invented etc
Tweet media one
0
10
0
@marcosalvi
Marco Salvi
15 days
RT @fchollet: The key is really this: AI usefulness scales logarithmically with inference time compute. Right now for many use cases the am…
0
310
0
@marcosalvi
Marco Salvi
15 days
RT @satyanadella: Jevons paradox strikes again! As AI gets more efficient and accessible, we will see its use skyrocket, turning it into a…
0
2K
0
@marcosalvi
Marco Salvi
16 days
RT @dylan522p: Deepseek V3 and R1 discourse boils down to this. Shifting the curve means you build more and scale more dummies https://t.co…
0
494
0