<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/graphics-programming-weekly---issue-460---october-25th-2026-xrvylagzv" -->

---
title: Graphics Programming weekly - Issue 460 - October 25th, 2026
description: This weekly roundup covers GPU-driven rendering patterns, shadow mapping, and shader tooling. One link explains the count/prefix-sum/write pattern used for...
canonical: https://daily.dev/posts/graphics-programming-weekly---issue-460---october-25th-2026-xrvylagzv
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: Graphics Programming weekly - Issue 460 - October 25th, 2026 | daily.dev
og:description: This weekly roundup covers GPU-driven rendering patterns, shadow mapping, and shader tooling. One link explains the count/prefix-sum/write pattern used for...
og:url: https://daily.dev/posts/graphics-programming-weekly---issue-460---october-25th-2026-xrvylagzv
og:image: https://api.daily.dev/og/posts/XRvYLAgZV.png
og:image:alt: Graphics Programming weekly - Issue 460 - October 25th, 2026
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Graphics Programming weekly - Issue 460 - October 25th, 2026

**[Jendrik Illner](https://daily.dev/sources/jendrikillner)** · 2 min read · 0 upvotes · 0 comments

## Summary

This weekly roundup covers GPU-driven rendering patterns, shadow mapping, and shader tooling. One link explains the count/prefix-sum/write pattern used for culling, LOD, and clustered lighting, plus push vs pull data models. Another breaks down the Metal compute shader paint simulation behind iOS app Fold & Squish, covering splats, thin-film flow, and folding. A third details a virtual shadow map system using cascaded virtual page tables, with performance timings on an RTX 3070. Other links cover a Vulkan Pong-to-3D tutorial, Godot 4 stencil/depth shader tricks, softmax numerical bugs across PyTorch, JAX, and TensorFlow, presenting Vulkan frames without libwayland, and a shader compiler survey benchmarking FXC, DXC, Metal, Tint, Naga, NIR, and Slang.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://www.jendrikillner.com/post/graphics-programming-weekly-issue-460>

## Questions this post answers

### What is the count, prefix-sum, write pattern used for in GPU-driven rendering?

It is a GPU-driven graphics pattern used for culling, level-of-detail selection, and clustered lighting, where work items are counted, their output positions computed via prefix sum, and results written in parallel. It covers expansion, compaction, and binning, and differs between push-based and pull-based data movement models, with implications for determinism versus using atomics.

_Engineers optimizing GPU culling pipelines can track pattern explainers like this via daily.dev._

### How does a virtual shadow map system using virtual page tables perform on an RTX 3070?

A virtual shadow map implementation using cascaded virtual page tables and physical page allocation includes bookkeeping, visible-page marking, 2D wraparound addressing, and hierarchical page buffer culling, with performance timings measured on an RTX 3070. Future work mentioned includes supporting translucency and local lights, along with caching and filtering improvements.

_Rendering engineers comparing shadow mapping techniques can follow benchmarks like this on daily.dev._

### Why do softmax and cross-entropy implementations in PyTorch, JAX, and TensorFlow produce numerical bugs?

Standard softmax, log-softmax, and cross-entropy implementations across PyTorch, JAX, and TensorFlow suffer from softmax collapse, where gradients vanish at high confidence, and softmax cancellation. A proposed fix uses log-space drop-in replacements built on softplus, log1p, and expm1 to eliminate these precision cliffs.

_ML engineers debugging vanishing gradients can keep up with numerical stability fixes via daily.dev._

## Similar posts on daily.dev

- [Graphics Programming weekly - Issue 439 - May 4th, 2026](https://daily.dev/posts/graphics-programming-weekly---issue-439---may-4th-2026-b3yniwzj3) · Jendrik Illner · 0 upvotes · 0 comments
- [Graphics Programming weekly - Issue 418 - November 30th, 2025](https://daily.dev/posts/graphics-programming-weekly---issue-418---november-30th-2025-zckkikomd) · Jendrik Illner · 0 upvotes · 0 comments
- [Graphics Programming weekly - Issue 457 - September 27th, 2026](https://daily.dev/posts/graphics-programming-weekly---issue-457---september-27th-2026-ttollowih) · Jendrik Illner · 0 upvotes · 0 comments
- [Graphics Programming weekly - Issue 433 - March 22nd, 2026](https://daily.dev/posts/graphics-programming-weekly---issue-433---march-22nd-2026-eufnfym8u) · Jendrik Illner · 1 upvotes · 0 comments
- [Graphics Programming weekly - Issue 428 - February 15th, 2026](https://daily.dev/posts/graphics-programming-weekly---issue-428---february-15th-2026-gbrezp4ts) · Jendrik Illner · 1 upvotes · 0 comments

---

Tags: [#vulkan](https://daily.dev/tags/vulkan), [#graphics-programming](https://daily.dev/tags/graphics-programming), [#godot](https://daily.dev/tags/godot)

[View this post on daily.dev](https://daily.dev/posts/graphics-programming-weekly---issue-460---october-25th-2026-xrvylagzv)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"TechArticle","headline":"Graphics Programming weekly - Issue 460 - October 25th, 2026","url":"https://daily.dev/posts/graphics-programming-weekly---issue-460---october-25th-2026-xrvylagzv","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/graphics-programming-weekly---issue-460---october-25th-2026-xrvylagzv"},"datePublished":"2026-10-29T18:35:15.919Z","dateModified":"2026-10-29T18:41:55.699Z","description":"This weekly roundup covers GPU-driven rendering patterns, shadow mapping, and shader tooling. One link explains the count/prefix-sum/write pattern used for...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/6b7cd64b6a2051fe0dcfbe5859b92616?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/6b7cd64b6a2051fe0dcfbe5859b92616?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Jendrik Illner","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Jendrik Illner","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/c55bfe232d344991adfeb9d15c202904","url":"https://daily.dev/sources/jendrikillner"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/graphics-programming-weekly---issue-460---october-25th-2026-xrvylagzv","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"vulkan,graphics-programming,godot","timeRequired":"PT2M"}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Jendrik Illner","item":"https://daily.dev/sources/jendrikillner"},{"@type":"ListItem","position":3,"name":"Graphics Programming weekly - Issue 460 - October 25th, 2026"}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/graphics-programming-weekly---issue-460---october-25th-2026-xrvylagzv#faq","mainEntity":[{"@type":"Question","name":"What is the count, prefix-sum, write pattern used for in GPU-driven rendering?","acceptedAnswer":{"@type":"Answer","text":"It is a GPU-driven graphics pattern used for culling, level-of-detail selection, and clustered lighting, where work items are counted, their output positions computed via prefix sum, and results written in parallel. It covers expansion, compaction, and binning, and differs between push-based and pull-based data movement models, with implications for determinism versus using atomics. Engineers optimizing GPU culling pipelines can track pattern explainers like this via daily.dev."}},{"@type":"Question","name":"How does a virtual shadow map system using virtual page tables perform on an RTX 3070?","acceptedAnswer":{"@type":"Answer","text":"A virtual shadow map implementation using cascaded virtual page tables and physical page allocation includes bookkeeping, visible-page marking, 2D wraparound addressing, and hierarchical page buffer culling, with performance timings measured on an RTX 3070. Future work mentioned includes supporting translucency and local lights, along with caching and filtering improvements. Rendering engineers comparing shadow mapping techniques can follow benchmarks like this on daily.dev."}},{"@type":"Question","name":"Why do softmax and cross-entropy implementations in PyTorch, JAX, and TensorFlow produce numerical bugs?","acceptedAnswer":{"@type":"Answer","text":"Standard softmax, log-softmax, and cross-entropy implementations across PyTorch, JAX, and TensorFlow suffer from softmax collapse, where gradients vanish at high confidence, and softmax cancellation. A proposed fix uses log-space drop-in replacements built on softplus, log1p, and expm1 to eliminate these precision cliffs. ML engineers debugging vanishing gradients can keep up with numerical stability fixes via daily.dev."}}]}
```

