Show HN: detecting cache latency inside a Web browser

Marat_Dukhan 8 years ago |

Author here. I made this demo and a related matrix-matrix multiplication demo [1] back in 2015 for Robert van de Geijn's Linear Algebra: Foundations to Frontiers MOOC class [2]. In the light of Spectre attack and recent browsers' changes to reduce precision of timers, I remembered of this project, and decided to check if it still works now, 3 years later. Surprisingly, it still works well!

The source code is available on GitHub [3].

[1] https://maratyszcza.github.io/laff-demos/dgemm.html

[2] https://www.edx.org/course/linear-algebra-foundations-fronti...

[3] https://github.com/Maratyszcza/laff-demos

madez 8 years ago | |

Could you make the website viewable without JavaScript?

Edit: I think the downvotes are unjustified. For clarification, if that wasn't clear by context, I don't expect to get the JavaScript test results from my computer while viewing the website without JavaScript. Demanding that would obviously be nonsense. Rather than that I assume there is information on that website that is interesting to read even without using JavaScript personally. Or is using JavaScript now a requirement to learn about JavaScript?

detaro 8 years ago | | |

There's only a graph of the results, so you are not missing any other content.

madez 8 years ago | | |

Thanks for the info.

StavrosK 8 years ago | | |

I guess you'd have more luck if you had asked "can you make some examples accessible for those of us who don't run JS?".

madez 8 years ago | | |

I didn't know what to expect to see. I vaguely assumed to see some text and data. That's why I carelessly formulated my question like I did. Still, I think the difference is small and my question wasn't extraordinary.

Mister_Snuggles 8 years ago | | |

I use uMatrix, this site just shows up as a white page until I allow it to load some JS from a third party. Once I allow the JS, I see a graph that gets built without any explanation of what that graph means.

Cyph0n 8 years ago | | |

Is there a particular reason why you can't enable JS yourself? Metered connection? Low-end machine?

madez 8 years ago | | |

I consider it an unacceptable form of code deployment. It's unsafe in the computing sense, and leads to a ecosystem where users are less and less in control of the software they use.

Tehnix 8 years ago | | |

Metered wouldn't really matter would it? Scripts are still downloaded, right? (never had JS off, because the web)

kaoD 8 years ago |

On my pretty outdated Core 2 Duo Q9300: https://i.imgur.com/HQXm5FU.png

The CPU is advertised as 6 MB L2 cache[0] but it has "L1 = 4 x 32 KB 8-way set associative data caches and L2 = 2 x 3 MB 12-way set associative caches (each L2 cache is shared between 2 cores)"[1].

You can clearly see the constant time access to L1 cache up to 32KB. Then grows to (linear?) access time up to 2MB (where it still fits in one of the L2 caches) and then again a different (log-like?) function from there on.

This is really neat.

[0] https://ark.intel.com/products/33922/Intel-Core2-Quad-Proces...

[1] http://www.cpu-world.com/CPUs/Core_2/Intel-Core%202%20Quad%2...

esnard 8 years ago |

Somewhat related: using the cache latency to estimate the CPU cache size. https://fromwhenceitca.me/cache_size/cache_size.html

jacquesm 8 years ago |

So, who thought it was a great idea again to use a mechanism intended for document transfer to run unsigned, unaudited code on just about every client in the world?

Sephr 8 years ago |

This is similar to how I detected total CPU cores in 2013 to implement navigator.hardwareConcurrency: https://eligrey.com/blog/cpu-core-estimation-with-javascript...

laCour 8 years ago |

i5-4690K (L1 4x32KB; L2 4x256KB; L3 6MB): https://i.imgur.com/FUs1isW.png

LG v30 (L2 2MB): https://i.imgur.com/q5R3bLY.png

polyfemos 8 years ago |

And with the latest iOS 11.2.2, this (admittedly cool) piece of code no longer works.

Still works on the latest MacOS (10.13.2) though.

Marat_Dukhan 8 years ago | |

Hmm...I just tried on iPhone 7/iOS 11.2.2, and it still works, albeit takes very long to start.

mherrmann 8 years ago |

On my OnePlus X Android phone: https://imgur.com/a/C1hb7

chrisper 8 years ago | |

On my i7 8700k https://imgur.com/a/D71v1

Yetanfou 8 years ago |

Thinkpad T42p with Pentium M 1.8GHz, 2MB L2 cache

https://imgur.com/agxy8e7

Mister_Snuggles 8 years ago |

I'd like to see some text explaining what I'm looking at.

Based on the title, it's something to do with cache latency, but that doesn't really help me. What exactly is it measuring? How does it perform this measurement? What are the limitations of this technique? What are the wider implications of this? What uses does it have, both potentially nefarious ones and potentially beneficial ones?

yeukhon 8 years ago |

Is there plan to collect this data voluntarily? Would be interesting to see the different percentiles

Marat_Dukhan 8 years ago | |

No, it is a static web page, and all code runs only locally in your browser.

chrisper 8 years ago |

Is that Ram only or also L1 etc. caches?

TickleSteve 8 years ago | |

Its characterising memory accesses, so it will show up the result of using all the caches in the system.

You would typically see larger latencies when the caches are not being effective.

Multiple humps will indicate the presence of multi-level caches.

frik 8 years ago | |

Okay, so how can we deactivate ASM.js and WebAssembly? (in the light of Meltdown and Spectre)

The config in Chrome is broken, WebAssembly cannot be deactivated anymore with chrome://flags/#enable-webassembly , setting it to "deactivated" and it's still active.

Marat_Dukhan 8 years ago | | |

Asm.js is not necessary, simple JavaScript interpreter is enough. This demo used to work before most browsers implemented optimizers for Asm.js

based2 8 years ago | | |

Mozilla Foundation Security Advisory 2018-01 Speculative execution side-channel attack ("Spectre")

https://www.mozilla.org/en-US/security/advisories/mfsa2018-0...

rmrfrmrf 8 years ago | | |

You need to disable `SharedArrayBuffer` and `performance.now`, which Firefox already did.

wyldfire 8 years ago | | |

chrome://flags/#enable-site-per-process is a mitigation that at least raises the bar a little. :/