HNHacker News
TopNewBestAskShowJobs

zbjornson

1,062 karma · joined October 30, 2015

Microbiologist (virology) & immunologist by training. High-performance bioinformatics software dev by trade.
submissionscomments
zbjornson··on Variolation
Well, for one, orthopoxviruses cause skin lesions, replicate in keratinocytes, dermal fibroblasts and other skin cells and evoke strong immune responses there. Coronaviruses don't.
zbjornson··on AVX-512 Mask Registers, Again
> Who is going to be making heavy use of x87 or MMX (both obsolete) along with AVX-512

Edge case and still not likely to cause contention: x87 fadd has 3-cycle latency while AVX has 4-c lat through ICL. If you need to prepare some constant with a chain of high-precision additions to broadcast into an AVX reg, it's faster. (Especially if the alternative is compensated addition.)

zbjornson··on Jepsen: MongoDB 4.2.6
I'm not sure I understand your point.

> the master is the one doing all writes. The replicas are read only. By default replicas are asynchronous

The same is true with MongoDB's defaults in an unsharded cluster.

zbjornson··on Jepsen: MongoDB 4.2.6
It'd be especially interesting given that MongoDB claims this:

> Postgres has both asynchronous (the default) and synchronous replication options, neither of which offers automatic failure detection and failover [12]. The synchronous replication only waits for durability on one additional node, regardless of how many nodes exist [13]. Additionally, Postgres allows one to tune these durability behaviors at the user level. When reading from a node, there is no way to specify the durability or recency of the data read. A query may return data that is subsequently lost. Additionally, Postgres does not guarantee clients can read their own writes across nodes.

From http://www.vldb.org/pvldb/vol12/p2071-schultz.pdf

zbjornson··on Lomuto's Comeback
For one: https://arxiv.org/abs/1704.08579
zbjornson··on Can You Write a Vectorized Reduction Operation? (2016)
It's also slow (5-6 cycle latency and 2 cycle throughput on Intel processors, compared to 4/0.5 for multiplication).
zbjornson··on There is one, and only one strain of SARS-CoV-2
The authors of the D614G paper do attempt to rule out a founder effect. I'm not saying their analysis is bullet proof, but the OP doesn't address its merits.

To differentiate between founder effects and a selective advantage driving the increasing frequency of the G clade in the GISAID data, we applied the suite of tools that we had been developing for the SARS-CoV-2 analysis pipeline (Fig. 2, Fig. 3, Fig. S2 and Fig. S3). A clear and consistent pattern was observed in almost every place where adequate sampling was available. In most countries and states where the COVID-19 epidemic was initiated and where sequences were sampled prior to March 1, the D614 form was the dominant local form early in the epidemic (orange in Figs. 2 and 3). Wherever G614 entered a population, a rapid rise in its frequency followed, and in many cases G614 became the dominant local form in a matter of only a few weeks (Fig. 3 and S3).

zbjornson··on Opt out of Clearview.ai
If you both opt-out and request data deletion, how do they keep you opt'ed-out? That seems impossible for them to comply with.
zbjornson··on The proximal origin of SARS-CoV-2
Pathogens can definitely evolve to improve fitness and settle into an ecological balance with their hosts. I more meant to contest that most start out very harmful. I don't think we have data to support that, or to be able to separate virus vs. host adaptations that would cause the appearance of lessened severity, given that most viruses circulating today have ancient origins.
zbjornson··on The proximal origin of SARS-CoV-2
Source on the first claim?

The S/L strain study has been heavily challenged, see commentary at https://nextstrain.org/help/coronavirus/FAQ#is-one-strain-of...

zbjornson··on Using SIMD to aggregate billions of values per second
There are several accurate summation algorithms; Kahan is one of them. Looks like Postgresql uses an accurate summation method[0] (and not Kahan), so that's probably a big factor in why theirs is so much slower and also makes for an unfair comparison. FWIW I posted AVX2 and AVX512 Kahan summation implementations here[1].

[0] https://github.com/postgres/postgres/blob/3ed2005ff595d34927...

[1] http://blog.zachbjornson.com/2019/08/11/fast-float-summation...

zbjornson··on Using SIMD to aggregate billions of values per second
Cool. I see you're doing a regular (not compensated) horizontal sum in a loop. Horizontal sums are slow, but I'm guessing you wanted to have exactly the same result as if the sum was calculated sequentially (for doubles)? Do you know if any databases use more accurate summation methods (compensated summation)?
zbjornson··on New Intel Instructions for Alder Lake, Also BF16 for Sapphire Rapids
That's correct. Info on LBR in general: https://lwn.net/Articles/680985/
zbjornson··on Simdjson 0.3: Faster JSON parser
The interchange into v8 is indeed an issue, see another comment: https://news.ycombinator.com/item?id=22745941.

> JSON parse is CPU-blocking, so if you get a large object, your server cannot handle any other web request until it finishes parsing

Well, your CPU core is busy on one request or another, so I don't understand why this is an issue as long as you're guarding against maliciously large bodies. Blocking I/O is different because your core is partially idle while other hardware is doing async work. Using Node.js' cluster module lets you keep more cores busy. Chunking CPU-limited work increases total CPU time and memory required. (This is a pet peeve of mine and a hill I'm willing to die on :-) .)

zbjornson··on Ask HN: Freelancer? Seeking freelancer? (April 2020)
SEEKING FEEELANCER

Primity Bio | Remote | QA/Test Engineer

We're seeking a QA/test engineer to help build out and improve our testing for a data analysis Web application. Must have strong knowledge of Cypress.

We're a small, all-remote team with membees in the US and Europe.

Email: my username at primitybio.com.

zbjornson··on Ask HN: Who is hiring? (April 2020)
Primity Bio | Remote; part-time or consulting/contract okay | QA/Test Engineer

We're seeking a QA/test engineer to help build out and improve our testing for a data analysis Web application. Must have strong knowledge of Cypress.

We're a small, all-remote team with membees in the US and Europe.

Email: my username at primitybio.com.

zbjornson··on Groceries and Delivered Goods Decontamination Station Instructions
Hydrogen peroxide is excellent for many viruses, including coronaviruses. There's a "new" generation of disinfectants based on hydrogen peroxide (search for Clorox or Diversey brand) and they list a 30 second contact time for human coronavirus inactivation. It's an oxidizer like bleach, but like you said degrades to water and oxygen, making it safe and residue-free. I use it regularly to disinfect lab equipment that would get destroyed by bleach.
zbjornson··on AMD processors from 2011 to 2019 vulnerable to two new attacks
Was submitted as https://news.ycombinator.com/item?id=22505466
zbjornson··on New coronavirus can spread between humans, but started in a wildlife market
I'm amazed that there's not a movement to stop eating and being around animals as much world-wide, given that so much disease (and environmental damage) comes from consumption of and proximity to animals:

- MERS-CoV: bats via dromedary camels

- nCoV-2019: bats or snakes?

- Influenza: pigs, birds

- E. coli O157:H7: cows and other farm animals.

...

(Edit: I mean along the lines of how safer sex is encouraged to prevent STDs, hand washing is encouraged to prevent food-borne illnesses, ... I haven't seen a call for the above.)

zbjornson··on CCPA goes into effect January 1, but nobody’s sure how the new rules work
The regulation itself is very short: https://leginfo.legislature.ca.gov/faces/codes_displaySectio...
zbjornson··on Ask HN: Who is hiring? (November 2019)
Primity Bio, a Caprion Company | San Francisco Bay Area and Montreal | Senior Frontend Dev | Remote

We're building a high-performance bioinformatics data analysis application and are seeking a front-end developer to join our team. Users can analyze millions of single cells on-the-fly and use interactive tools to create visualizations and statistics for clinical trials and drug development, for example.

No experience in biology needed. Ideally you have a passion for and experience with building highly interactive, creative interfaces (e.g. buisness intelligence or dashboarding apps, products like Google Docs) and keen attention to detail. Our front-end is built with Vue. Opportunities to contribute to open-source projects. Our team is close-knit, but scattered around the US and Europe.

zbjornson at primitybio . com

zbjornson··on Linux Kernel Fastboot [pdf]
> compiled with Intel compilers

This is a common misconception. It's not built with ICC, it's built with GCC (and some parts maybe with Clang).

When did you try Clear? They have about 6200 packages now, but I'm not sure if that's still "limited" in relative terms.

zbjornson··on Continuous Compliance
This is a major problem and one of the artifacts of the reg not having been updated since 1997. The industry expectation for the validation process is for it to take four to six months, and it's specific to one version. We have 99.99% of our software requirements specification testing covered by several thousand automated tests that run in several minutes. Nonetheless, auditors have been wary of frequent updates, even though by their own admission our test strategy is far superior to any manual testing strategy and, critically, is the only realistic way to monitor for regressions specific to our usage of software that the vendor might not catch.
zbjornson··on Chicken Hypnotism
By far the best part of the article:

> The United States military when trying to avoid divulging information gives reporters briefings with 25 minutes of intentionally dull PowerPoint presentations and 5 minutes left at the end for questions from anyone who is still awake. The presentations are called hypnotizing chickens.

zbjornson··on Ask HN: What can I do to accelerate scientific research?
Shameless plug: come work for Primity Bio. My team builds a bioinformatics web application that (among other things) is several orders of magnitude faster than the other tools, which fundamentally changes the scope of analysis that researchers can feasibly do. Email in bio.
zbjornson··on Performance Speed Limits
The Intel Optimization Manual also lists bypass delays for Skylake (table 2-3), Broadwell (2-8) and Nehalem (2-29). It's only ever a single cycle per uop for Skylake.
zbjornson··on Ask HN: Who is hiring? (June 2019)
Primity Bio | San Francisco Bay Area | Frontend Dev | Onsite or Remote

We're building a high-performance bioinformatics data analysis application and are seeking front-end developers to join our team. Users can analyze millions of single cells on-the-fly and use interactive tools to create visualizations and statistics for clinical trials and drug development, for example.

No experience in biology needed. Ideally you have a passion for and experience with building highly interactive, creative interfaces (e.g. buisness intelligence or dashboarding apps, products like Google Docs) and keen attention to detail. Our front-end is built on Vue. Opportunities to contribute to open-source projects. Our team is close-knit, but scattered around the globe; remote or on-site welcome.

zbjornson at primitybio . com

zbjornson··on Efficient IO with io_uring [pdf]
I made a prototype of a replacement for Node.js' fs.read/write using io_uring in hopes of it ultimately replacing libuv's current threadpool implementation. Definitely promising! Many props to Jens for his hard work on this and for his helpful discussions on the API.

Benchmarks at [0] and prototype repo at [1]

[0] https://github.com/libuv/libuv/issues/1947#issuecomment-4852... [1] https://github.com/zbjornson/node-io-uring

zbjornson··on Instrumenting HTTP requests in Node
Timeouts in Node.js HTTP requests I think are one of the most confusing aspects of all of Node.js, in terms of technical details that are critical to successful production operations. The core documentation is poorly worded in some places, and some of the helper libraries make it very difficult to figure out exactly what timeouts are being set. For example, node-fetch (mentioned in the article) had/has a pretty non-sense timeout on the entire request duration [0], and I've seen that blindly propagated into other libraries [1]. When timeouts only happen under error conditions (flaky servers), it's a nightmare to debug.

Glad to see this article explain the various timeouts (although I don't see the idle socket timeout mentioned). Huge +1 from me for using node's core http library only and adding explicit, readable timeouts in your own code.

[0] https://github.com/bitinn/node-fetch/issues/446

[1] https://github.com/googleapis/nodejs-storage/issues/27 (two different timeout bugs)

zbjornson··on REST and long-running jobs (2014)
206 is only valid to respond to range requests, which can only be GET requests.
← PreviousPage 2 of 9Next →