rerx

At least they integrated more automation so that many titles from Amazon's catalogs appear on Goodreads quite soon without manual work. So a lot of the recent stuff in the main languages at least often doesn't need to be added manually anymore.

It's crazy how much Goodreads relies on unpaid "librarian" contributions to keep their data in check. One problem with allowing the general population to add titles to the database is that the chance of getting malformed entries that will require manual fix ups (from these volunteers) later on is pretty high.

> I just checked back - that was a month ago, and I even made a forum post about the book that it be added, but nope, book's not there...it seems nobody has even looked at the post.

Volunteer librarians + zero dev support means that, if you really want to see the addition, you should bump your post daily or so until by chance it catches the eye of somebody...

It's great for following people who read interesting books and may leave reviews that are worth reading. Of course this is an an extension of the database of books you've read, because you can compare your ratings quite easily with those of other people and find things they liked that you didn't know about yet. That's a great source of recommendations, much better than the automated systems.

V100 GPUs are from 2017, so it's more than two years. A100 already appeared there years ago, btw.

An eight GPU DGX-1 server cost ~149k$ back then (googled news postings). A current gen DGX H100 is 520k$ with 5 years of support. Of course it holds 5x the memory, plus GPUs and interconnect are much faster. But when comparing costs, take price hikes into account.

That must be a regional difference again. Maybe because fast shipping across North America is more expensive than inside a densely populated country like Germany? (There is monthly Prime here, including video and shipping, and it's slightly more expensive than the yearly package.)

I typically sign up for Prime for a month or two when I book vacation stays on booking.com. They have a very nice promotion where prime members receive 10% in credit. Then I watch a couple of films or shows on Prime Video, because they actually have some good stuff there usually. But free shipping below 39€ is really not worth the subscription to me (Germany).

I disagree. As a tourist I would usually prefer to use bike share, similarly to the systems I'm used to at home.

Getting and returning bikes at dedicated rental shops is quite a hassel, typically much more expensive, and unnecessary if I just want a simple city bike.

In fact I suspect that it's often intentional that tourists have a hard times registering for bike share systems. They would greatly impact the business of preexisting bike rental shops.

What?

The original IBM PC came with BASICA: https://en.wikipedia.org/wiki/IBM_BASIC#IBM_Advanced_BASIC

From 1983, MS DOS came with GW-BASIC: https://en.wikipedia.org/wiki/GW-BASIC

From 1991, GW-BASIC was replaced by QBasic: https://en.wikipedia.org/wiki/QBasic

I would say, DOS always shipped with Basic. QBasic remained included up to Windows ME (the last Windows that came with DOS).

If you add multiple GPUs to the same PCIe switch, their bandwidth to main memory of the SOC will be shared. For some use cases that bandwidth was already the bottle neck.

Iglberger's "C++ Software Design: Design Principles and Patterns for High-Quality Software" and Viafore's "Robust Python: Write Clean and Maintainable Code" sound most useful to me personally, but there are several interesting books.

The thing with AMD is that the selection of GPUs that can actually run your GPGPU code is typically much more limited then with Nvidia, so things work on far fewer consumer GPUd. Here it's limited to "CDNA2 (MI-210/250) GPUs". Those are priced comparatively to Nvidia's A100, 10k+ per card.

Speed running MI2 takes a bit over 30 minutes if you were wondering: https://youtu.be/5_I7hlaWVCE

Very much how he rolls, see Thimbleweed Park.

There are screenshots on the author's web site: https://www.dgray.com/hwpage.htm

Now I'm contemplating buying the set for 6$.

Could it be Hugo's House of Horrors? Somehow I remember the name quite distinctly, although I couldn't play it much back then (did not know English at that age). https://en.m.wikipedia.org/wiki/Hugo%27s_House_of_Horrors

That's because France like many other countries, but not Canada, signed an international treaty that standardizes road signs: https://en.wikipedia.org/wiki/Vienna_Convention_on_Road_Sign... That makes it a bit easier to drive safely across countries without reading all local languages.

Edit: Huh, apparently "stop" or the equivalent in the local language are allowed by the treaty. Didn't know that. https://en.wikipedia.org/wiki/Stop_sign

Rather than 'lactose' being derived from 'lactuca', I'm pretty sure that both words just have a common root 'lac', the Latin word for 'milk'

That's not trivial, in particular with C++. However, the talk actually talks about setting up tooling as early as you can as a best practice, like multiple different compilers or regression test suits on all targeted platforms.

Youtube Premium is €11.99 per month here, including YouTube Music. Spotify is €9.99. The pricing seems very fair to me as long as you make use of the music offering.

I literally can't tell how much you're joking. There's nothing French about the spelling of these contemporary English words, even though they have norman roots.

There's the Fulbright program with bilateral exchanges with many countries. According to wikipedia.de, on average each German university or college sends 55 students in a year to the US with a Fullbright scholarship.

US export controls certainly must slow down China somewhat. They can't import Nvidia's fastest GPUs, but are stuck with buying hobbled variations. Getting a domestic industry on par is hard.

Have you tried https://www.phind.com ?

_This_ comment was dead until I vouched for it.

Not sure what's going on here.

> There's no such thing as "contaminated" marijuana, that's an urban myth.

Is it a myth? In Germany black market cannabis has been found to be lined with synthetic cannabinoids in many cases. Those substances are not researched well and hundreds of their users (who typically didn't expect any synthetic drugs) required emergency medical care according to this German (sorry) source from two years ago: https://www.vice.com/de/article/m7a533/chemisch-gestrecktes-...

The German minister of health cites this situation as his main motivation for working towards legalization, so people wouldn't have to buy from questionable sources.

Lyrics by Chat-GPT?

> LSD

Is there actually evidence of LSD being more harmful than beneficial to its consumers? I haven't read Michael Pollan's pop science book yet, but from watching the Netflix rendition of "How to change your mind" I got the impression that it was mainly criminalized because its users opposed conscription during the Vietnam war.

Reportedly, the German government scaled down its original legalization plans as they would be in violation of current EU law. Private is in the scope of member state legislation, but virtually everything related to trade is subject to EU regulations, and they currently demand dealing with cannabis to be criminal. (Not a lawyer, that's how I understood the news.) Germany's current plans could be a first step before pushing for a change of EU law. That may not be trivial though, as some other member state governments appear to be opposed to the idea.

> Blindsight, by Peter Clark

* Peter Watts

Can't find documentation online yet for this year's H100 systems, but here's a schematic for an A100 server: https://docs.nvidia.com/dgx/dgxa100-user-guide/introduction-... There 2 GPUs, 2 NICs and 1 NVME are connected via one PCIe switch.

The torus topology makes sense for ring algorithms: Allreduce, Allgather, Reducescatter. For purely data parallel training you could put all model replicas into the same ring (although Nvidia also uses hierarchical algorithms that benefit from lower lately). With added model parallelism one will need smaller rings running concurrently. I guess the TPU cluster layout will then put constraints on the most efficient model architectures (as does the network topology of a GPU cluster).