github-starred/cs249r_book
2
0
Fork 0
mirror of https://github.com/harvard-edge/cs249r_book.git synced 2026-07-17 16:34:48 -05:00
Code Issues 276 Packages Projects Releases 26 Wiki Activity

[PR #1681] [MERGED] fix(book): numerical and clarity corrections across 11 chapter files #34489

New Issue
Closed
opened 2026-07-14 19:14:11 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-07-14 19:14:11 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/harvard-edge/cs249r_book/pull/1681
Author: @profvjreddi
Created: 5/5/2026
Status: ✅ Merged
Merged: 5/5/2026
Merged by: @profvjreddi

Base: dev ← Head: fix/numeric-corrections

📝 Commits (1)

  • ff9232a fix(book): numerical and clarity corrections across 11 chapter files

📊 Changes

11 files changed (+64 additions, -62 deletions)

View changed files

📝 book/quarto/contents/vol1/backmatter/glossary/glossary.qmd (+2 -2)
📝 book/quarto/contents/vol1/benchmarking/benchmarking.qmd (+2 -2)
📝 book/quarto/contents/vol1/hw_acceleration/hw_acceleration.qmd (+1 -1)
📝 book/quarto/contents/vol1/training/training.qmd (+1 -1)
📝 book/quarto/contents/vol2/conclusion/conclusion.qmd (+1 -1)
📝 book/quarto/contents/vol2/data_storage/data_storage.qmd (+4 -4)
📝 book/quarto/contents/vol2/edge_intelligence/edge_intelligence.qmd (+1 -1)
📝 book/quarto/contents/vol2/fault_tolerance/fault_tolerance.qmd (+5 -6)
📝 book/quarto/contents/vol2/inference/inference.qmd (+15 -8)
📝 book/quarto/contents/vol2/ops_scale/ops_scale.qmd (+4 -4)
📝 book/quarto/contents/vol2/performance_engineering/performance_engineering.qmd (+28 -32)

📄 Description

Summary

Author-initiated factual corrections that had been accumulated as uncommitted edits in the working tree. These tighten numbers, equations, and a few prose sentences. No structural rewrites.

Notable corrections:

  • training: GPT-2 total FLOPs 1e20 → 1e19
  • hw_acceleration: AmdahlH100 hw_speedup_factor 500 → 247 (H100 INT8 vs ~8 TOPS baseline CPU, more accurate)
  • data_storage: Adam optimizer state 700 GB → 1,400 GB (175B params × 2 moments × 4 bytes FP32)
  • edge_intelligence: A11→A17 Pro Neural Engine improvement 58× → 27×
  • fault_tolerance: λ 20 FITs → 20,000 FITs (×1000 correction); hourly_survival = availability_pct/100 (was 1 - 1/hours_per_year); LEGO dead-code cleanup (drop unused hours_str)
  • inference: GPT-3 batch-size table totals recalculated (b8 96.3, b16 row added at 74.4, b32 74.9); prose updated; check() updated to match new L target
  • ops_scale: cost_savings_pct 40% → 57%; capacity_threshold 4000 → 2500
  • performance_engineering: B200 dense TFLOPS 4500/5000 → 2250 (sparse vs dense throughput honesty); ridge points + figures recalculated consistently (139→281 FLOP/byte; 16× → 2× growth in 7 years)
  • glossary: GEMM and ReLU definitions wrap math in LaTeX delimiters
  • benchmarking: drop unused V100_FLOPS_FP32 import
  • conclusion (vol2): minor prose rewording

Verification:

Test plan

  • PDF + HTML render cleanly
  • Inline {python} references resolve correctly with new constant values
  • Roofline figures regenerate with B200 = 2250 TFLOPS

🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.

## 📋 Pull Request Information **Original PR:** https://github.com/harvard-edge/cs249r_book/pull/1681 **Author:** [@profvjreddi](https://github.com/profvjreddi) **Created:** 5/5/2026 **Status:** ✅ Merged **Merged:** 5/5/2026 **Merged by:** [@profvjreddi](https://github.com/profvjreddi) **Base:** `dev` ← **Head:** `fix/numeric-corrections` --- ### 📝 Commits (1) - [`ff9232a`](https://github.com/harvard-edge/cs249r_book/commit/ff9232a2d232aff3021bfda0d062c365a5c59c1a) fix(book): numerical and clarity corrections across 11 chapter files ### 📊 Changes **11 files changed** (+64 additions, -62 deletions) <details> <summary>View changed files</summary> 📝 `book/quarto/contents/vol1/backmatter/glossary/glossary.qmd` (+2 -2) 📝 `book/quarto/contents/vol1/benchmarking/benchmarking.qmd` (+2 -2) 📝 `book/quarto/contents/vol1/hw_acceleration/hw_acceleration.qmd` (+1 -1) 📝 `book/quarto/contents/vol1/training/training.qmd` (+1 -1) 📝 `book/quarto/contents/vol2/conclusion/conclusion.qmd` (+1 -1) 📝 `book/quarto/contents/vol2/data_storage/data_storage.qmd` (+4 -4) 📝 `book/quarto/contents/vol2/edge_intelligence/edge_intelligence.qmd` (+1 -1) 📝 `book/quarto/contents/vol2/fault_tolerance/fault_tolerance.qmd` (+5 -6) 📝 `book/quarto/contents/vol2/inference/inference.qmd` (+15 -8) 📝 `book/quarto/contents/vol2/ops_scale/ops_scale.qmd` (+4 -4) 📝 `book/quarto/contents/vol2/performance_engineering/performance_engineering.qmd` (+28 -32) </details> ### 📄 Description ## Summary Author-initiated factual corrections that had been accumulated as uncommitted edits in the working tree. These tighten numbers, equations, and a few prose sentences. No structural rewrites. **Notable corrections:** - `training`: GPT-2 total FLOPs `1e20 → 1e19` - `hw_acceleration`: AmdahlH100 `hw_speedup_factor 500 → 247` (H100 INT8 vs ~8 TOPS baseline CPU, more accurate) - `data_storage`: Adam optimizer state `700 GB → 1,400 GB` (175B params × 2 moments × 4 bytes FP32) - `edge_intelligence`: A11→A17 Pro Neural Engine improvement `58× → 27×` - `fault_tolerance`: λ `20 FITs → 20,000 FITs` (×1000 correction); `hourly_survival = availability_pct/100` (was `1 - 1/hours_per_year`); LEGO dead-code cleanup (drop unused `hours_str`) - `inference`: GPT-3 batch-size table totals recalculated (b8 96.3, b16 row added at 74.4, b32 74.9); prose updated; `check()` updated to match new `L` target - `ops_scale`: cost_savings_pct `40% → 57%`; capacity_threshold `4000 → 2500` - `performance_engineering`: B200 dense TFLOPS `4500/5000 → 2250` (sparse vs dense throughput honesty); ridge points + figures recalculated consistently (139→281 FLOP/byte; 16× → 2× growth in 7 years) - `glossary`: GEMM and ReLU definitions wrap math in LaTeX delimiters - `benchmarking`: drop unused `V100_FLOPS_FP32` import - `conclusion (vol2)`: minor prose rewording **Verification:** - All pre-commit hooks pass (incl. LEGO dead-code, MLSys units, external image checks) - No conflicts with PRs #1677 / #1679 / #1680 ## Test plan - [ ] PDF + HTML render cleanly - [ ] Inline `{python}` references resolve correctly with new constant values - [ ] Roofline figures regenerate with B200 = 2250 TFLOPS --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
GiteaMirror added the pull-request label 2026-07-14 19:14:11 -05:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
area: book
area: collabs
area: kits
area: labs
area: socratiq
area: staffml
area: tinytorch
area: tools
area: website
bug
dependencies
format: epub
format: pdf
javascript
link-health
link-rot
priority-high
pull-request
Mirrored from GitHub Pull Request
 staffml
type: bug
type: citation
type: code
type: errata
type: improvement
type: new
type: question
vault-sli
No Label pull-request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/cs249r_book#34489
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?