Laurie does an amazing job of reimagining Google's strange job optimisation technique (for jobs running on hard disk storage) that uses 2 CPUs to do the same job. The technique simply takes the result of the machine that finishes it first, discarding the slower job's results... It seems expensive in resources, but it works and allows high priority tasks to run optimally.
Laurie re-imagines this process but for RAM!! In doing this she needs to deal with Cores, RAM channels and other relatively undocumented CPU memory management features.
She was even able to work out various undocumented CPU/RAM settings by using her tool to find where timing differences exposed various CPU settings.
She's turned "Tailslayer" into a lib now, available on Github, https://github.com/LaurieWired/tailslayer
You can see her having so much fun, doing cool victory dances as she works out ways of getting around each of the issues that she finds.
The experimentation, explanation and graphing of results is fantastic. Amazing stuff. Perhaps someone will use this somewhere?
As mentioned in the YT comments, the work done here is probably a Master's degrees worth of work, experimentation and documentation.
Go Laurie!
This is the sort of thing which was done before in a world where there was NUMA, but that is easy. Just task-set and mbind your way around it to keep your copies in both places.
The crazy part of what she's done is how to determine that the two copies don't get get hit by refresh cycles at the same time.
Particularly by experimenting on something proprietary like Graviton.
Tis just probabilities and unlikelihood of hitting a refresh cycle across that many memory channels all at once.
The hedging technique is a cool demo too, but I’m not sure it’s practical.
At a high level it’s a bit contradictory; trying to reduce the tail latency of cold reads by doubling the cache footprint makes every other read even colder.
I understand the premise is “data larger than cache” given the clflush, but even then you’re spending 2x the memory bandwidth and cache pressure to shave ~250ns off spikes that only happen once every 15us. There’s just not a realistic scenario where that helps.
Especially HFT is significantly more complex than a huge lookup table in DRAM. In the time you spend doing a handful of 70ns DRAM reads, your competitor has done hundreds of reads from cache and a bunch of math. It’s just far better to work with what you can fit in cache. And to shrink what doesn’t as much as possible.
1) Can we take this library and turn it into a a generic driver or something that applies the technique to all software (kernel and userspace) running on the system? i.e. If I want to halve my effective memory in order to completely eliminate the tail latency problem, without having to rewrite legacy software to implement this invention.
2) What model miniature smoke machine is that? I instruct volunteer firefighters and occasionally do scale model demos to teach ventilation concepts. Some research years back led me to the "Tiny FX" fogger which works great, but it's expensive and this thing looks even more convenient.
what I wished I had during this project is a hypothetical hedged_load ISA instruction. Issue two requests to two memory controllers and drop the loser. That would let the strategy work on a single thread! Or, even better, integrating the behavior into the memory controller itself, which would be transparent to all software without recompilation. But, you’d have to convince Intel/AMD/someone else :)
2. It’s called a “smokeninja”. Fairly popular in product photography circles, it’s quite fun!
Wouldn't you have a tail latency problem on the write side though if you just blindly apply it every where? As in unless all the replicas are done writing you can't proceed.
Her videos have really high production value, but man I just really struggle to watch them.
But all the accounts are old/legit so I think that you and me have just become paranoid...
In all seriousness, agreed. The top comment at time of this writing seems like a poor summarizing LLM treating everything as the best thing since sliced bread. The end result is interesting, but neither this nor Google invented the technique of trying multiple things at once as the comment implies.