tweetindex
FR

Vivek Galatage ✓

@vivekgalatage · Bangalore, India · joined 09 Feb 2010

browser baker • chromium contributor • enjoys compilers, systems, teaching • director of engineering: ai & systems @visteon • prev founding eng @browsercompany

21 383Followers
990Following
4 505Posts total
76.7KViews on collected posts

Derniers posts

If you ever need to know about HTTP - HyperText Transfer Protocol, reach out to https://t.co/C8wPe2TCme https://t.co/X2OwgF1XLK
881 views · 20 likes · 1 reposts · 0 replies Open on X →
Past 1k downloads!! 🎉 If you read technical docs or research papers in @obsdmd, and hate those bright white PDFs, give PDF Toggle Dark Mode a spin. https://t.co/4t0r8tW4Nq https://t.co/UHnhYP01fV
1.3K views · 10 likes · 1 reposts · 0 replies Open on X →
@vivekgalatage the highest-leverage page in any such handbook is the roofline one: decode is memory-bound, so bytes per token (weights + kv cache) over bandwidth predicts most of serving cost before you touch a kernel. everything else is second-order until that arithmetic closes. 48 views · 0 likes · 0 reposts · 0 replies Open on X →
Went back through NVIDIA’s CUDA Refresher series this week with some new folks. https://t.co/2aFSB542Ap Origins of GPU computing: why we ended up here. CPUs stopped getting faster the easy way, and parallel hardware picked up the workloads that could split. Getting started htt
8.4K views · 212 likes · 30 reposts · 2 replies Open on X →
@vivekgalatage Thank you for the shoutout! 238 views · 3 likes · 0 reposts · 0 replies Open on X →
Apparently, it’s an international launch everything day today!! 771 views · 3 likes · 0 reposts · 0 replies Open on X →
@vivekgalatage 这个手册很适合反过来提醒大家:先有能跑的模型,再谈 kernel 优化,别一上来就对着空 benchmark 调参。真实输入、显存占用和端到端延迟往往会改写结论。 164 views · 0 likes · 0 reposts · 0 replies Open on X →
GPU architecture | LLM Inference Handbook https://t.co/lPbF3l51T7 Add this to your LLM learning resource bundle. "Before writing or tuning GPU kernels, you need a working model of how a GPU runs code. Without it, suggestions like “increase occupancy” or "reduce shared memory h
26.8K views · 608 likes · 97 reposts · 4 replies Open on X →
@vivekgalatage just tried it - for someone who reads research papers every day, this is a nice quality of life improvement. 190 views · 1 likes · 0 reposts · 0 replies Open on X →
See it in action - https://t.co/1y1sIHTBx2
1.2K views · 6 likes · 0 reposts · 2 replies Open on X →
Find it here - https://t.co/4t0r8tW4Nq 592 views · 4 likes · 0 reposts · 0 replies Open on X →
I wrote an Obsidian plugin! "Adapt to theme" is nice, but it breaks badly with embedded images in PDFs - makes them harder to read. So I built better control: choose the darkness, color, and brightness for PDFs in dark mode. Introducing PDF Toggle Dark Mode - make PDF reading
16.4K views · 63 likes · 5 reposts · 3 replies Open on X →
Obsidian: Tried "Adapt to theme" for a PDF? https://t.co/Bt2uAdWzzb
6.3K views · 16 likes · 0 reposts · 0 replies Open on X →
Career update: I've joined Visteon as the Director of Engineering - AI and Systems. The automotive industry is evolving rapidly - AI, on-device inference, edge compute, and specialized hardware-software co-design are taking center stage. Keeping safety, performance, and 13.3K views · 214 likes · 4 reposts · 23 replies Open on X →

Face aux comptes de taille comparable

14 posts des 90 derniers jours, à côté de la tranche de 10K–100K abonnés. diffusé largement, mais peu de ces spectateurs réagissent.

Vues médianes1 042ce compte924médiane pour 10K–100K
Portée, %4.87%ce compte3.62%médiane pour 10K–100K
Engagement, %0.67%ce compte1.52%médiane pour 10K–100K
IndicateurCe compteMédiane pour 10K–100KRapport
Vues médianes par post1 0429241.13×
Portée (vues ÷ abonnés)4.87%3.62%1.35×
Taux d'engagement0.67%1.52%0.44×

Autres comptes de cette tranche →   Comparer avec un autre compte →   Comment ces repères sont établis →

Growth & engagement

How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.

Views per post

13.3K24 Jul
6.3K31 Jul
16.4K1 Aug
592
1.2K
190
26.8K22 Sep
164
771
238
8.4K23 Sep
48
1.3K
881

Last 14 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

1.82%24 Jul
0.27%31 Jul
0.44%1 Aug
0.68%
0.67%
0.53%
2.65%22 Sep
0.00%
0.39%
1.26%
2.90%23 Sep
0.00%
0.85%
2.38%

Reactions — likes, reposts, replies and quotes — divided by views. Median for 10K–100K accounts is 1.52%.

What the audience does

Likes46.7%1 160 in total
Reposts5.6%138 in total
Replies1.4%34 in total
Quotes0.2%4 in total
Bookmarks46.2%1 148 in total

Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.

The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.

Comptes similaires