Daniel Lemire ✓

@lemire · Montreal, Quebec · joined 29 Nov 2007

Software performance expert. Ranked in the top 2% of scientists globally (Stanford/Elsevier 2025) and among GitHub's top 1000 developers. Father, husband.

37 439Followers

2 250Following

35 822Posts total

139.9KViews on collected posts

Latest posts

Daniel Lemire ✓@lemire4 Sep

@lemire There are multiple points to comment on, but this line of reasoning seems to misinterpret what the terms “time complexity” and “constant-time”/“O(1)” mean. Time complexity does not refer to actual time as in seconds or clock cycles. It refers to a more abstract notion of 196 views · 3 likes · 0 reposts · 2 replies Open on X →

Daniel Lemire ✓@lemire4 Sep

@lemire You made me look at my dict implementation and find a bug, add a benchmark 😅 https://t.co/ZL9jpDTfwS 833 views · 7 likes · 0 reposts · 1 replies Open on X →

Daniel Lemire ✓@lemire4 Sep

@lemire Great article! Another annoying thing about hash tables is that you need to pick a hash function based on a use case. Which is fine, but if a language doesn't expose a good way to customize it, you open yourself to stuff like this: https://t.co/N6E7UJ9bAN 983 views · 15 likes · 0 reposts · 2 replies Open on X →

Daniel Lemire ✓@lemire4 Sep

https://t.co/ZBlcOTLyOb 23.5K views · 119 likes · 9 reposts · 8 replies Open on X →

Daniel Lemire ✓@lemire4 Sep

@lemire What would happen if that mask is a zero whose value is only known at runtime? 50 views · 0 likes · 0 reposts · 0 replies Open on X →

Daniel Lemire ✓@lemire3 Sep

@lemire I mean it is a load unaligned instruction so it isn't very surprising it lets you use any address, I'd be more surprised if it did fault 530 views · 1 likes · 0 reposts · 1 replies Open on X →

Daniel Lemire ✓@lemire3 Sep

@lemire I guess they optimized the zero mask case in the microcode and they're not actually doing the load in that case. Smart. 1.2K views · 9 likes · 0 reposts · 2 replies Open on X →

Daniel Lemire ✓@lemire3 Sep

In C/C++, you can selectively load 64 bytes with the intrinsic function _mm512_maskz_loadu_epi8 on recent AMD/Intel processors. The first parameter is the MASK which tells you which bytes are loaded, it is effectively a 64-bit word where each bit corresponds to a byte. In the bit 14.9K views · 168 likes · 6 reposts · 6 replies Open on X →

Daniel Lemire ✓@lemire2 Sep

In C, you can define an enum type like so... enum status {OK = 0, ERROR = -1}; You would then use it as a type like so... enum status my_condition = OK; The underlying type was often chosen by the compiler to be an 'int'. Thus, the following would often compile... enum 14.6K views · 198 likes · 11 reposts · 20 replies Open on X →

Daniel Lemire ✓@lemire1 Sep

@lemire nice, and you get an actual size_t instead of whatever type the macro happens to expand to in context. only catch is C23 constexpr covers objects only, no functions, so the truly ugly macros survive anyway. still a clear win for stuff like this 1.8K views · 18 likes · 0 reposts · 1 replies Open on X →

Daniel Lemire ✓@lemire1 Sep

@lemire would still name consts all caps 4.6K views · 15 likes · 0 reposts · 1 replies Open on X →

Daniel Lemire ✓@lemire1 Sep

@lemire am i too old school for thinking the new way is the ugly one? 6.2K views · 60 likes · 1 reposts · 7 replies Open on X →

Daniel Lemire ✓@lemire1 Sep

By introducing constexpr, C23 allows you to clean up your C code from a lot of ugly macros. https://t.co/S0feRMx4QW 70.7K views · 466 likes · 12 reposts · 47 replies Open on X →

Against accounts of the same size

13 posts from the last 90 days, next to the 10K–100K follower range. shown widely, but few of those viewers react.

Median views1 785this account924median for 10K–100K

Reach, %4.77%this account3.62%median for 10K–100K

Engagement, %0.96%this account1.52%median for 10K–100K

Metric	This account	Median for 10K–100K	Ratio
Median views per post	1 785	924	1.93×
Reach (views ÷ followers)	4.77%	3.62%	1.32×
Engagement rate	0.96%	1.52%	0.63×

Others in this range → Compare with another account → How these benchmarks are built →

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

70.7K1 Sep

6.2K

4.6K

1.8K

14.6K2 Sep

14.9K3 Sep

1.2K

530

504 Sep

23.5K

983

833

196

Last 13 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

0.75%1 Sep

1.10%

0.35%

1.06%

1.58%2 Sep

1.23%3 Sep

0.91%

0.38%

0.00%4 Sep

0.59%

1.73%

0.96%

2.55%

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

What the audience does

Likes72.0%1 079 in total

Reposts2.6%39 in total

Replies6.5%98 in total

Quotes0.8%12 in total

Bookmarks18.1%271 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.

Latest posts

Against accounts of the same size

Growth & engagement

Views per post

Engagement rate per post

What the audience does

Similar accounts