Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

Know how long every operation takes on a computer. Once you do, identifying good vs bad vs acceptable performance becomes way easier. It all boils down to knowing your napkin math.

57,417 Aufrufe • vor 3 Monaten •via X (Twitter)

12 Kommentare

Profilbild von Ben Dicken
Ben Dickenvor 3 Monaten

And btw @Sirupsen is the expert here. Longer video on youtube:

Profilbild von Obaid
Obaidvor 3 Monaten

Ayy I was just thinking about this today! So cool.

Profilbild von Ben Dicken
Ben Dickenvor 3 Monaten

/napkin-math when?

Profilbild von Obaid
Obaidvor 3 Monaten

Already done: :)

Profilbild von Samian
Samianvor 3 Monaten

napkin math is underrated. how do you keep those numbers in your head without writing them down

Profilbild von Ben Dicken
Ben Dickenvor 3 Monaten

🧠

Profilbild von Philip
Philipvor 3 Monaten

good stuff! for GPUs the bottleneck is bandwidth because we need to feed all compute units in parallel. per GPU bandwidth: HBM (GPU ↔ GPU memory): TB/s (~5-8 for HBM3) NVLink / scale-up (GPU ↔ GPU): low TB/s InfiniBand / scale-out (rack ↔ rack): 100s GB/s

Profilbild von Ben Dicken
Ben Dickenvor 3 Monaten

We need the inference napkin math

Profilbild von K
Kvor 3 Monaten

if only there was a way to visualize this intuitively with balls

Profilbild von Sheikh
Sheikhvor 3 Monaten

Flame Graphs 🔥

Profilbild von Aadithya niranjan
Aadithya niranjanvor 3 Monaten

jus saw this yest while havin dinner....trus me i frgt what i had

Profilbild von Melon Usk
Melon Uskvor 3 Monaten

I like when math and number is used in the right way

Ähnliche Videos