xtacy

Curious - what’s this court filing?

> QUIC is not bad, but there are places where it either does not work at all or works too slow.

Curious: in your experience where does QUIC work bad/slow?

Are there good public examples of well designed APIs that have stood the test of time?

Is it GBps (gigabytes per second) or Gbps (giga bits per second)? I see mixed usage in this comment thread so I’m left wondering what it actually is.

The article is consistent and uses Gigabytes.

We do have a model. That’s statistical physics.

Any standard course goes over various derivations of classical physics laws (Newtonian dynamics) from quantum mechanics.

Yep, such ideas have been around. But congestion is a fundamental problem. Admission control is the only way to ensure there is no congestion collapse.

The technical issue is that you would need global arbitration to ensure that the _goodput_ (useful bytes delivered) is optimal. With training across 32k GPUs and more these days, global arbitration to ensure the correct packets are prioritised is going to be very difficult. If you are sending more traffic than the receiver's link capacity, packets _will_ get dropped, and it's suboptimal to transmit those dropped packets into the network as they waste link capacity elsewhere (upstream) within the network.

In principle, with perfect knowledge of flows at any given instant, you can assign credits/rate-of-transmission for each flow to prevent congestion. But, in practice this is somewhat nuanced to build, and there are various tradeoffs to consider: what happens if the flows are so short that coordinating with a centralised scheduler incurs a latency overhead that is comparable to the flow duration? There's been research to demonstrate that one can strike a sweet spot, but I don't think it's practical nor has it been really deployed in the wild. And of course, this scheduler has to be made reliable as it's a single point of failure.

Such ideas are, however, worth revisiting when the workload is unique enough (in this case, it is), and the performance gains are so big enough...

OP is referring to "Credit based flow control", which is a way to ensure a sender does not overwhelm a receiver with more data than it can handle.

Usually, this is line-rate, but if the other side is slow for whatever reason (say the consumer is not draining data), you wouldn't want the sender to continue sending data.

If you also have N hosts sending data to 1 host, you would need some way of distributing the bandwidth among the N hosts. That's another scenario where the credit system comes. Think of it as an admission control for packets so as to guarantee that no packets are lost. Congestion control is a looser form of admission control that tolerates lossy networks, by retransmitting packets should they be lost.

It's also a bit odd that they do not implement congestion control. Congestion control is fundamental unless you only have point-to-point data transfers, which is rarely the case. All-reduce operation during training requires N to 1 data transfer. In these scenarios the sender needs to control its data transfer rates so as to not overwhelm not just the receiver, but also the network... if this is not done, it will cause congestion collapse (https://en.wikipedia.org/wiki/Network_congestion#:~:text=ser...).

You can check the example described here: https://stats.stackexchange.com/questions/644280/stable-viol...

Judea Pearl’s book also goes into the above in some detail, as to why faithfulness might be a reasonable assumption.

Honest question: Are there successful cases where a PE takeover has transformed the company that has helped the company's mission?

You are not alone, me too, and several others have noted it and discussed it before. For e.g., Gmail made this change that I absolutely hated and instantly disabled when it was released: https://ux.stackexchange.com/questions/15291/are-grayscale-b... and https://news.ycombinator.com/item?id=17038130

The grumpy.website linked above is gold!

> to have a tool that is external to the VM (runs on the hypervisor host) that essentially has "read only" access to the kernel running on the VM to provide visibility into what's running on the machine without an agent running within the VM itself

Not quite what you are after, but comes close ... you could run gdb on the kernel in this fashion and inspect, pause, step through kernel code: https://stackoverflow.com/questions/11408041/how-to-debug-th....

While this is in part a good news, they are still not accepting MV2 extensions to the Chrome Web Store, even as unlisted/private extensions. There is no change to that policy: https://developer.chrome.com/docs/extensions/mv3/mv2-sunset/...

You will also like Arithmetic coding, then! https://en.wikipedia.org/wiki/Arithmetic_coding

I suspect that the web server's CPU usage will be pretty high (almost 100%), so C-state tuning may not matter as much?

EDIT: also, RSS happens on the NIC. RFS happens in the kernel, so it might not be as effective. For a uniform request workload like the one in the article, statically binding flows to a NIC queue should be sufficient. :)

They key takeaway really is that for high performance, the way you think about memory layout, data structures, memory management, etc. all start to really matter. It is possible to write programs with memory pools to reuse data structures as much as possible without having to defer the work to the GC, which can then introduce performance spikes.

That said, C# and JVM are both mature and have decades of GC improvements, which could explain why golang is falling behind. Time will tell.

Although most US markets carry just one type, the Cavendish, you can easily get a few more varieties at various ethnic markets. Pay a quick visit to the Indian/South Asian stores for a delight!

I have also heard about interesting varieties at Berkeley Bowl, but I haven't personally checked it out though.

Or, you could (re)schedule first and then generate the key? Assuming that key generation doesn't take an hour, of course. :)

A multi-buck is essentially an SDR / Special Drawing Rights: https://www.investopedia.com/terms/s/sdr.asp

I guess the main issue with BeyondCorp-style architecture is that while they work well with apps that mainly use HTTP, but it can be a pain to get them working for plain TCP or UDP-based apps.

> • The website is blank on Firefox.

Do you have Firefox FB containers enabled? I had the same issue, but when I opened the website inside a FB container it works.

The core problem seems to be:

   For instance, overwriting content on a page of the document is not allowed and thus leads to invalid signature verification. Nevertheless, some changes are considered harmless by the PDF applications and do not throw any warnings

I am trying to understand if there is any design consideration behind not signing the entire PDF, but only some sections?

Interesting link, thanks for sharing. I wonder what this means precisely:

  If you take out one of the characters or story elements of a conspiracy theory, the connections between the other elements of the story fall apart.

I guess I have to read the paper, but what are these "connections" and what does "fall apart" actually mean?

EDIT: I just skimmed the paper https://journals.plos.org/plosone/article?id=10.1371/journal...

The connections capture context-specific relationships, such as co-occurrences. The "fall apart" part comes from the fact that conspiracy theories rely on hidden, unsubstantiated, subjective interpretations of intent or actions whose validity can be questioned. If they are key pillars of the narrative, then their falsity can negate the truth of the narrative.

This reminds me of a philosophical discussion around what "truth" means. Coherent theory of truth: truth is defined as a property that's coherent among a set of beliefs. It can also be used as an epistemic justification -- that is, any set of internally consistent beliefs can be taken as true. Of course, in practice, certain truth statements have to correspond to reality, which is where the correspondence theory of truth comes in.

Didn't the Fed say it will buy up to $500B of municipal bonds as part of the new QE program?

I wonder if the "Lottery ticket hypothesis" work can be applied to this model to further shrink the number of parameters by 10x, to bring it closer to Google's T5 but with higher accuracy?

Slick! I too felt that many libraries were not slick/fast enough, but I eventually ended up using react-stockcharts (in canvas mode) for my use case and was quite satisfied with its performance.

Could you compare your performance against this one if possible?

https://github.com/rrag/react-stockcharts

As others have noted, Q2 would likely be worse hit than Q1 since the quarantine started later in the quarter.

FRB NY has a GDP Nowcast prediction of about -8% for Q2.

What is the recommended best practice for a stable API endpoint that supports things like filters, pagination, and some limited aggregation?

Would a REST-like endpoint backed by a database VIEW (to decouple schema changes from API changes) be a good approach?

Try the paper towel trick: wrap the seeds in a moist paper towel (not too wet, or it will encourage fungus growth). Then, keep them inside a ziplock bag in a warm place for a couple of days, checking on them every day. I keep them on top of the refridgerator where it's nice and warm.

Beans sprout super fast! For very dry seeds (lentils, beans, peas, edamamme) you could soak them in water overnight and just use the seeds that settle down after they've absorbed some moisture.