Learning About Gpus Through Measuring Memory Bandwidth

Posted5 months agoActive4 months ago

JasperBekkers

58 points

5 comments

evolvebenchmark.comTechstory

skepticalmixed

Debate

30/100

GPUMemory BandwidthBenchmarking

Key topics

GPU

Memory Bandwidth

Benchmarking

The article explores GPU architecture by measuring memory bandwidth, sparking discussion about the potential bias due to AMD sponsorship.

Snapshot generated from the HN discussion

Discussion Activity

Light discussion

First comment

Peak period

Day 1

Avg / period

2.5

Key moments

01Story posted
Aug 20, 2025 at 10:35 AM EDT
5 months ago
Step 01
02First comment
Aug 20, 2025 at 4:39 PM EDT
6h after posting
Step 02
03Peak activity
4 comments in Day 1
Hottest window of the conversation
Step 03
04Latest activity
Sep 1, 2025 at 3:27 AM EDT
4 months ago
Step 04

Generating AI Summary...

Analyzing up to 500 comments to identify key contributors and discussion patterns

Discussion (5 comments)

Showing 5 comments

jamesblonde

5 months ago

2 replies

Good article, but my spider sense tingles with AMD sponsored content.

JasperBekkersAuthor

5 months ago

2 replies

Thankfully the content is actually provided by us without any sponsorship! We ran into a memory bandwidth problem on one of these devices, wrote some small tests for it and ran it on the other devices as well. Indeed tho, the AMD gpus do seem to have extremely linear cache performance.

Full disclosure: I do run the company behind Evolve.

iamcreasy

4 months ago

Great article. What social account should I follow to keep upto date with that you guys are doing?

latchkey

5 months ago

Thanks for this article. Let me know if you want some CDNA / MI300x to play with. Happy to donate compute time to see what you can come up with.

alecco

5 months ago

Why do you think it's a good article?

View full discussion on Hacker News

ID: 44962361Type: storyLast synced: 11/20/2025, 7:45:36 PM