Hacker Newsnew | past | comments | ask | show | jobs | submit | jsolson's commentslogin

I realize this isn't the meat of your post, but you ate a lobster sandwich at Walmart?

I cannot help but think of https://youtu.be/Pj-D0jc17D0?si=BiEGWr9aacGdAkGW


The walmart experience is so wildly different depending on who they're catering to.

When I lived in poor areas, I've literally been shaken down by gigantic bouncers in the parking lot for "stealing" (but not) a bag of $5 cat litter. Why on earth this is even worth the liability of sending bouncers is beyond me. And everything even remotely valuable and easy to steal is behind locked showcases, but no one gives a fuck to get it out to help you buy it.

In rich areas, it's lobster sandwiches and no one even bothering you for your receipt on the way out, and maybe someone might even help you find something in the store. Nothing you want is behind glass, except maybe something unusually valuable by walmart standards and miniscule like an SD card.


yeah it was pretty central walmart in new rochelle. it looked like a high end store in europe.


“Gas Station Sushi: Fresh one day a week, but no one knows which one…”

I read the post as involving seeing the other family at Walmart, but the buying of things not needed and consumption of lobster as happening elsewhere.


nah it was walmart. just a fancy one


You might have even better precision if you stay away from CPU0 and also set idle=poll in your kernel command line. Lots of things (including other interrupts) often land on CPU0. It would not be my first choice for something where I wanted high timing precision.


I came here to post this. We make a lot of the same sorts of optimizations for our OS distro (debian based) -- disabling frequency scaling, core pinning, etc. Critically, CPU0 has a bunch of stuff you cannot push, and you're better off with using one of the other cores as an isolated island.

This is what the scheduler latency looks like on our isolated core:

# Total: 000300000 # Min Latencies: 00001 # Avg Latencies: 00005 # Max Latencies: 00059 # Histogram Overflows: 00000

(those are uS!)


Very cool. What are you running on it? What's your use case?


Worth a shot! I'll give it a go later today.




I'd say certain countries in Europe give us a run for our money: https://caw.ceu.edu/other-activities/academic-blog/politics/...


This is completely different.

The is the politicians not being able to form a coalition. Imagine your congress not agreeing to fill some commissions or confirm some positions in government.

It is a 'shutdown' of the legislative branch, not the executive.


In the US, the legislative branch controls the budget, so when their budget expires, if the legislative branch is shutdown, so is the money.


Not really comparable because the bureaucratic system kept running. It's just a completely different political system and "there is no government" means a different thing. US-style "government shutdowns" don't really happen in Belgium.


I agree with nearly everything except your point (1).

Periodic polling is awkward on both sides: you add arbitrary latency _and_ increase database load proportional to the number of interested clients.

Events, and ideally coalesced events, serve the same purpose as interrupts in a uniprocess (versus distributed) system, even if you don't want a proper queue. This at least lets you know _when_ to poll and lets you set and adjust policy on when / how much your software should give a shit at any given time.


From a database load perspective, Postgres can get you pretty far. The reads triggered by each poll should be trivial index-only scans served right out of RAM. Even a modest Postgres instance should be able to handle thousands per second.

The limiting factor for most workloads will probably be the number of connections, and the read/write mix. When you get into hundreds or thousands of pollers and writing many things to the queue per second Postgres is going to lose its luster for sure.

But in my experience with small/medium companies, a lot of workloads fit very very comfortably into what Postgres can handle easily.


+1 on connections being the limiting factor. Curious how further a connection pooler can get you


That would make sense.

My first thought was "Arc'teryx will probably adopt this immediately." They (and similar brands) are already pushing as hard as they can on seamlessness or very very tight seams.


Doesn't Arc'teryx make outdoor gear? That's something where I absolutely would not buy a product that I couldn't repair in the field with a needle and thread.


Yes, Arcteryx makes outdoor gear, but more along the line of technical climbing and backcountry skiing, not backpacking. Not likely to carry a needle and thread to repair your 3L Goretex shell.

Of course the brand has been diluted to cater to a more mainstream buyer.


This actually makes sense from a spec perspective if you want to give enough to allow hardware to catch up with the specs and to support true interop.

Contrast this with the wild west that is "Ethernet" where it's extremely common for speeds to track well ahead of specs and where interop is, at best, "exciting."


Datacenter Blackwell is Gen6, which is critical when pairing it with CX8 (2x400G) as otherwise you'd be stranding NIC BW.


please link me a B200 (not a B300) system with CX8 that uses gen6 x16 instead of gen5 x32.

to the best of my knowledge this does not exist, but I'd be happy to stand corrected.

(the official NVIDIA DGX B200 is gen5).


> Those are already unstoppable because they now use Direct Memory Access over the PCI-E bus, so the cheats don't even run on the same computer anymore.

Working on mostly server platforms, I had forgotten that IOMMU enablement (and, where relevant, enforcement) was not the default.

Consumer hardware and software is terrifying.


Not sure how that's relevant, unless you find it terrifying that owners of hardware have control over their hardware.


It's your IOMMU, you can do what you want with it. Maybe you need to write heaps of stuff to take advantage of it, but what's new there?

The only thing you're getting by saying "no IOMMU" is "I want any devices in my machine to be able to do anything, not just what I want them restricted to".


Okay, but he's specifically brought it up in the context of a computer's owner doing something that the software vendor (and also myself as another gamer harmed by cheating) would prefer he did not.


And if they want complete control, they can choose not to use a vendor and do it themself, for all the control they could want.

Hooray, freedom!


I have no idea what you're trying to say


> unless you find it terrifying that owners of hardware have control over their hardware

I mean that the presence or absence of an IOMMU doesn't impact whether owners of hardware have control over their hardware.

It just means that the owner of the machine is able to limit what memory the devices in their system are able to access, in the same way that MMUs limited what memory every process on your system could access.


In my world, we won't let a system boot with production credentials unless the IOMMU is enabled.

This is enforced by a greatly enriched TPM (and it's willingness to unwrap credentials). We have trust several layers of firmware and OS software, but the same mechanism allows us to ensure that known-bad versions of those aren't part of the stack that booted.

If I wanted secure games (and the market would tolerate it), I'd push for enforcement of something similar in the consumer space.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: