ℹ️ This listing was auto-generated from Mastodon's public data. meowray hasn't claimed it, and it doesn't imply any partnership with or endorsement of SocialDB. Is this you? Claim it · Request removal (free).

meowray

Mastodon @meowray@hachyderm.io

⚔ Compare ☆ Watch Own this channel?

🔄 Data last refreshed 3 days ago

🤝

Make a deal with meowray

🔒 Paid proposals held safely in escrow — released only when the work's approved.

Followers

230

Account age

3 yrs

🧰 Free analysis for meowray

🕵️ Fake follower check 📊 Engagement rate 💰 What they charge

Known for

7000 commits to llvm-project — likely the last commit count worth announcing.
The Age of Agents has begun: they forge in an afternoon what took me years, and counting commits is counting leaves now. The forest matters, not the tally.
I'm not sailing west. The hands that forged 7000 rings are exactly the ones that know when to take one off.

3 views 7000 commits to llvm-project — likely the last commit count worth announcing. The Age of Agents has begun: they forge in an afternoon what took me years, and counting commits is counting leaves now. The forest matters, not the tally. I'm not sailing west. The hands that forged 7000 rings are exactly the ones that know when to take one off.

Google AI Overviews cited my blog post to say DT_GNU_HASH support started in glibc 2.4 - but my post didn't say that! No mention of glibc 2.4/2.5 at all.
After doing some archaeology in git log and mail archives, I can confirm it was actually introduced in glibc 2.5.

3 views Google AI Overviews cited my blog post to say DT_GNU_HASH support started in glibc 2.4 - but my post didn't say that! No mention of glibc 2.4/2.5 at all. After doing some archaeology in git log and mail archives, I can confirm it was actually introduced in glibc 2.5.

https://github.com/llvm/llvm-project/pull/206787 wants ~1900 lines in lld/ELF to rebuild the AMDGPU cross-TU device call graph, compute resource usage, and re-encode kernel descriptors.
My worry: perpetual AMDGPU-only changes to lld/ELF that ELF-layer maintainers can't review, and more leaky abstraction.
CUDA does this RDC mode handling in nvlink (a device linker sharing encoders with ptxas); the

2 views https://github.com/llvm/llvm-project/pull/206787 wants ~1900 lines in lld/ELF to rebuild the AMDGPU cross-TU device call graph, compute resource usage, and re-encode kernel descriptors. My worry: perpetual AMDGPU-only changes to lld/ELF that ELF-layer maintainers can't review, and more leaky abstraction. CUDA does this RDC mode handling in nvlink (a device linker sharing encoders with ptxas); the

📊 Post engagement

Avg engagement / post

2.6%

Engagement vs followers

Nov 2022

On Mastodon since

🔥 Top post: Since 2020 I've been the de facto reviewer for object file forma · 28 likes + reposts

📊 Activity & format

Posting cadence

3 / week

Posts several times a week.

Content mix

Mostly images

Recent: 6 text · 6 image · 0 video.

Follower / following

6.8×

Follows 34 back. A strong ratio — an audience that follows them, not a follow-for-follow network.

Recent posts

View on Mastodon ↗

🔥 Top post Since 2020 I've been the de facto reviewer for object file format stuff/MC/ELF linker in LLVM. Until I left XXX in 2024, it counted as a line item on my "community contribution". Since then it has come out of my own evenings. Four of the patches in my review queue right now come… ★ 28

Agents excel at the mechanical labor of rebasing. "We can't afford a fork, so we must upstream", the strongest argument for upstreaming, is losing force. What I expect isn't less upstreaming, though — it's adverse selection. The same tools… ★ 2 Every time I modify bolt/ (because I often refactor MC/ADT/Support and BOLT uses them in... weird ways), CODEOWNERS automatically adds 7 reviewers. You'd think: "Wow, BOLT actually has so many maintainers!" In reality, only one person ever… ★ 1

7000 commits to llvm-project — likely the last commit count worth announcing. The Age of Agents has begun: they forge in an afternoon what took me years, and counting commits is counting leaves now. The forest matters, not the tally. I'm n… ★ 3

Modifying LLVM GenericCycleInfo to a single-pass DFS (SAS 2007) instead of Havlak-Tarjan. Claude Fable 5 created a patch after I fed it my OI-style C++ code, but its version had many inefficiencies: [screenshot] Still impressive though. Fi…

Google AI Overviews cited my blog post to say DT_GNU_HASH support started in glibc 2.4 - but my post didn't say that! No mention of glibc 2.4/2.5 at all. After doing some archaeology in git log and mail archives, I can confirm it was actua… ★ 3 https://maskray.me/blog/2026-07-12-irreducible-loops Irreducible loops ★ 5

LLVM CycleInfo其实很有潜力替代 LoopInfo(基于domtree的reducible loop)，可惜其初始实现相当糟糕，导致现在需要做大量的增量改进。这也算是一个经典的“OI 教训”：少用指针/链状结构，多用 4 字节整数/阵列。 Optimizing clang... Dominator tree achieves best performance when most array indices/values are in the pre-order number space (https://github.com/llvm/llvm-project/pull/207474; bottom green row) Storing predecessors using Linked Forward St… ★ 1

小学时我曾用五子棋棋盘、黑白棋子，以及几个不同颜色的桌游角色标识来模拟回合制策略游戏。绿色代表主角，蓝色和红色代表我方英雄；白棋正反面朝上分别代表我、敌方小兵，黑棋为障碍物，用二或八进制记录角色血量。较为简单的提示符即让Fable 5搭出了主要… lld/ELF实现了section和local symbol的并行初始化，而mold和wild则采用完全并行的文件和符号表解析。在Opus 4.5 to 4.8我都尝试过引入完全的文件解析。4.7/4.8能以牺牲--threads=1性能为代价提升多核性能，但代码复杂度太高了。 Fable 5刚发布时和前两天，…

https://github.com/llvm/llvm-project/pull/206787 wants ~1900 lines in lld/ELF to rebuild the AMDGPU cross-TU device call graph, compute resource usage, and re-encode kernel descriptors. My worry: perpetual AMDGPU-only changes to lld/ELF th… ★ 2

🐘 Community & instance

Home server

hachyderm.io

A tech & engineering server — its members and audience skew toward that niche, so expect a community-aligned, engaged following.

On Mastodon since

Nov 2022

Joined in the Twitter-exodus wave of late 2022 — part of the migration that made Mastodon a real destination.

💡 Facts

🗓️Joined Mastodon in 2022 — 3 years ago.

👁️Averages 6 views per post.

📤Posts about 3× per week.

🕵️ Fake follower check

Estimated

80/100

Excellent Credibility score

97%

Real Real audience

Low Fake-follower risk

High Data confidence

Est. 97% real, active audience · Low fake-follower risk.
Strong engagement (~2.6% of followers engage each post) — an active, real audience.
Established account (3+ years old).

Heuristic estimate from engagement, follower ratios, account age & growth — a screening signal, not a guarantee.

📸 Gallery

🔀 Audience overlap

Estimated

Estimated shared audience with similar creators — useful for avoiding overlap (or doubling down) when planning a campaign.

Avani 18%

Sunny 17%

Ron Starman 14%

Oscar Bazaldua 14%

Chris Thomas 12%

élytres - François Dumeaux 11%

More like this

Find more →

Mastodon

Sunny

@sunny@universeodon.com

230 Followers

Mastodon

élytres - François Dumeaux

@rstarman@infosec.exchange

230 Followers

Mastodon

Chris Thomas

@killfile@newsie.social

230 Followers

Mastodon

Avani

@avani@wandering.shop

meowray

Make a deal with meowray

Known for

📊 Post engagement

📊 Activity & format

Recent posts

🐘 Community & instance

💡 Facts

🕵️ Fake follower check

📸 Gallery

🔀 Audience overlap

More like this

✉ Message meowray

🎤 Event / appearance with meowray