Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: ChartGPU – WebGPU-powered charting library (1M points at 60fps)

github.com

Post reply on HN
Print view View original

111–120 of 226 posts

Re: Show HN: ChartGPU – WebGPU-powered charting library (1M points at 60fps)

#111

Post by leeoniya »

zokier wrote: post #86

If you have tons of datapoints, one cool trick is to do intensity modulation of the graph instead of simple "binary" display. Basically for each pixel you'd count how many datapoints it covers and map that value to color/brightness of that pixel. That way you can visually make out much more detail about the data. In electronics world this is what "digital phosphor" etc does in oscilloscopes, which started out as just…

agreed, heatmaps with logarithmic cell intensity are the way to go for massive datasets in things like 10,000-series line charts and scatter plots. you can generally drill downward from these, as needed.

Re: Show HN: ChartGPU – WebGPU-powered charting library (1M points at 60fps)

#114

Post by altern8 »

kettlecorn wrote: post #94
altern8 wrote: post #93

All charts in the demo failed for me. Error message: "WebGPU Error: Failed to request WebGPU adapter. No compatible adapter found. This may occur if no GPU is available or WebGPU is disabled.".

Does your browser support WebGPU yet? It's likely it does not. WebGPU is supported on Chrome and on the latest version of Safari. On Linux with all browsers WebGPU is only supported via an experimental flag.

I'm not sure. I'm using the latest version of Chrome.

Maybe I messed with the settings at some point and disabled something.

Re: Show HN: ChartGPU – WebGPU-powered charting library (1M points at 60fps)

#115

Post by vlovich123 »

leeoniya wrote: post #104

uPlot maintainer here. this looks interesting, i'll do a deeper dive soon :) some notes from a very brief look at the 1M demo: - sampling has a risk of eliminating important peaks, uPlot does not do it, so for apples-to-apples perf comparison you have to turn that off. see https://github.com/leeoniya/uPlot/pull/1025 for more details on the drawbacks of LTTB - when doing nothing / idle, there is significant cpu being…

Is there any techniques using wavelet decomposition to decimate the high frequency component while retaining peaks? I feel like that's a more principled approach than sampling but I haven't seen any literature on it describing the specific techniques (unless the idea is fundamentally unsound which is not obvious to me).

Re: Show HN: ChartGPU – WebGPU-powered charting library (1M points at 60fps)

#117

Post by vanderZwan »

akomtu wrote: post #108
huntergemmer wrote: post #92

Earlier quoted context omitted.

Great suggestion - density mapping is a really effective technique for overplotted data. Instead of drawing 1M points where most overlap, you're essentially rendering a heatmap of point concentration. WebGPU compute shaders would be perfect for this - bin the points into a grid, count per cell, then render intensity. Could even do it in a single pass. I've been thinking about this for scatter plots especially, where…

You don't need webgpu for that. It's a standard vertex shader -> fragment shader pass with the blending mode set to addition.

That works if more overdraw = more intensity is all you care about, and may very well be good enough for many kinds of charts. But with heat map plots one usually wants a proper mapping of some intensity domain to a color map and a legend with a color gradient that tells you which color represents which value. Which requires binning, counting per bin, and determining the min and max values.

Re: Show HN: ChartGPU – WebGPU-powered charting library (1M points at 60fps)

#118

Post by MindSpunk »

akomtu wrote: post #108
huntergemmer wrote: post #92

Earlier quoted context omitted.

Great suggestion - density mapping is a really effective technique for overplotted data. Instead of drawing 1M points where most overlap, you're essentially rendering a heatmap of point concentration. WebGPU compute shaders would be perfect for this - bin the points into a grid, count per cell, then render intensity. Could even do it in a single pass. I've been thinking about this for scatter plots especially, where…

You don't need webgpu for that. It's a standard vertex shader -> fragment shader pass with the blending mode set to addition.

Drawing lots of single pixels with alpha blending is probably one of the least efficient ways to use the rasterizer though. A good compute shader implementation would be substantially faster.

Re: Show HN: ChartGPU – WebGPU-powered charting library (1M points at 60fps)

#119

Post by akdor1154 »

The rendering is very cool, but what i really want is this as a renderer i can plug into Vega.

Vega/VGlite have amazing charting expressivity in their spec language, most other charting libs don't come close. It would be very cool to be able to take advantage of that.

Re: Show HN: ChartGPU – WebGPU-powered charting library (1M points at 60fps)

#120

Post by huntergemmer »

vlovich123 wrote: post #115
leeoniya wrote: post #104

uPlot maintainer here. this looks interesting, i'll do a deeper dive soon :) some notes from a very brief look at the 1M demo: - sampling has a risk of eliminating important peaks, uPlot does not do it, so for apples-to-apples perf comparison you have to turn that off. see https://github.com/leeoniya/uPlot/pull/1025 for more details on the drawbacks of LTTB - when doing nothing / idle, there is significant cpu being…

Is there any techniques using wavelet decomposition to decimate the high frequency component while retaining peaks? I feel like that's a more principled approach than sampling but I haven't seen any literature on it describing the specific techniques (unless the idea is fundamentally unsound which is not obvious to me).

Interesting idea - I haven't explored wavelet-based approaches but the intuition makes sense: decompose into frequency bands, keep the low-frequency trend, and selectively preserve high-frequency peaks that exceed some threshold.

My concern would be computational cost for real-time/streaming use cases. LTTB is O(n) and pretty cache-friendly. Wavelet transforms are more expensive, though maybe a GPU compute shader could make it viable.

The other question is whether it's "visually correct" for charting specifically. LTTB optimizes for preserving the visual shape of the line at a given resolution. Wavelet decomposition optimizes for signal reconstruction - not quite the same goal.

That said, I'd be curious to experiment. Do you have any papers or implementations in mind? Would make for an interesting alternative sampling mode.

Post reply on HN