正在加载视频...

视频加载失败

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.

57,442 次观看 • 3 个月前 •via X (Twitter)

12 条评论

Ben Dicken 的头像
Ben Dicken3 个月前

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

Obaid 的头像
Obaid3 个月前

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

Ben Dicken 的头像
Ben Dicken3 个月前

/napkin-math when?

Obaid 的头像
Obaid3 个月前

Already done: :)

Samian 的头像
Samian3 个月前

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

Ben Dicken 的头像
Ben Dicken3 个月前

🧠

Philip 的头像
Philip3 个月前

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

Ben Dicken 的头像
Ben Dicken3 个月前

We need the inference napkin math

K 的头像
K3 个月前

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

Sheikh 的头像
Sheikh3 个月前

Flame Graphs 🔥

Aadithya niranjan 的头像
Aadithya niranjan3 个月前

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

Melon Usk 的头像
Melon Usk3 个月前

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

相关视频