Hacker News Reader

tntnon Cray, AMD to Extend DOE’s Exascale Frontier

It's probably not a great comparison to compare the theoretical numbers of Frontier to the achieved numbers of the green 500. The achieved flops is pretty much always considerably lower than the theoretical flops. Titan is a 27 PFLOPS machine that achieves 17.6 PFLOPS, sequoia is a 20 PFLOPS machine that achieves 17, summit is a 200 PFLOPS machine that achieves 143, ...

~ 37 GFLOPS/W is probably a better projection if we assume (out of nowhere) that the theoretical/achieved ratio of Frontier is comparable to Summit (75%). Still very impressive.

tntnon Estimating branch probability using Intel LBR feature

i7-6950X: 256 KiB L2/core, 2.5 MiB L3/core

i9-7960X: 1024 KiB L2/core, 1.375 MiB L3/core

Seems to be exactly what I described? L2/core went up by 768 KiB, L3/core went down by 1.125 MiB.

tntnon Cray, AMD to Extend DOE’s Exascale Frontier

IMO the coolest thing about Summit/Sierra is that the GPUs and CPUs have a fully coherent single address space with all memory available to the GPUs by default, meaning that your stack- and malloc-allocated variables can be used directly from the GPUs.

I wonder if that will be the case on Frontier.

tntnon Estimating branch probability using Intel LBR feature

But skylake also has much less L3 per core, so the increased L2 is less like cache growth and more like a rebalancing.

The change to a non-inclusive L3 is more likely to contribute to greater capacity.

tntnon Ray-Strips: A Compact Mesh Representation for Interactive Ray Tracing (2007) [pdf]

(2007)

tntnon Biomarker for chronic fatigue syndrome identified

This is less a criticism of the paper than it is a statement of misaligned expectations.

> but the mechanism behind this is entirely hand-waving.

The objective of this paper was not to specify, or even attempt to specify, the mechanism.

> The paper's use of machine learning seems entirely pointless, basically drawing the obvious line between the two sets of points.

Yeah, that is pretty much the entire objective of SVMs.

tntnon Janus cosmology: what is negative mass?

Dr. Petit has said that his theories are based on his personal interactions with aliens and that the U.S. Army has tested antimatter bombs on Jupiter, so there's that.

tntnon GCC 9.1 Released

Anyone know what happened circa GCC 5 that changed the version number scheme? It felt like GCC 4 was around forever and now they do whole number increments every year or so.

tntnon Use mmap with care

> This simply won't die, will it?

Please clarify how you think that statement is incorrect. As far as I'm concerned, it won't die because that's how virtual memory works, and you have something going on in your head that is either wrong or massive hairsplitting.

I would have expected a better understanding from someone bloviating about how the try of the commenters are too thick to understand mmap and virtual memory.

tntnon Use mmap with care

> here is using mmap for general file IO

> nobody has ever suggested avoiding virtual memory except you

Do you know what the "anonymous" in "anonymous mapping" means? You are the one that started asserting that anonymous mapping from mmap have the same difficulties as mapping of normal files and therefore too dangerous to use.

https://news.ycombinator.com/item?id=19807322

tntnon Use mmap with care

As if "avoid virtual memory" is substantially more of a "sweeping generalization" than "avoid mmap."

If you apply the same reasoning that you've used to conclude that everyone should avoid mmap, than you are led directly to the conclusion that everyone should avoid virtual memory.

The "same problems" that you are pointing out are possible with anonymous memory aren't unique to memory you get directly from mmap, they also apply to all memory, period. mmap'ed anonymous memory might have th same problems as file-backed memory, but those are the same problems that .text and .bss have.

The "powerful (and consequently hazardous) OS feature" here isn't mmap, it's literally virtual memory. At the moment you concede that memory may be backed by something besides physical memory at any point in time, you get the possibility of all those "exotic error paths."

tntnon Use mmap with care

So should we extend your conclusion above to the following?

"There is probably enough evidence in this thread to use it as a reference for why typical apps should avoid virtual memory whenever possible -- it's clear almost nobody fully understands it"

I'd suggest that is ludicrous, and for the same reason your original conclusion is also excessive.

tntnon L.A.’s Elite on Edge as Prosecutors Pursue More Parents in Admissions Scandal

Are you dying soon? If not, there's no deadline. If the people you know did a cool thing at 20 or 21, and you do a cool thing at 24 or 25 (or 45 or 55 or 65 or ...), it is still cool, is it not?

It seems clear that you aren't actually interested primarily in sound the cool things, but doing more cool things earlier than peers.

tntnon Tesla sued in wrongful death lawsuit that alleges Autopilot caused crash

Humans can be 30+% better than "humans" by 1. not using a phone and 2. not being drunk or stoned.

A few percent better than "humans" is quite bad relative to a semi-responsible human driver.

tntnon Perchance to Dream

"Aye, that's the rub" thought the audience while contemplating escaping their life to escape the boredom.

tntnon Teen Suicide Spiked After Debut Of Netflix's '13 Reasons Why,' Study Says

> what about the Bible (Jesus)?

Erm, what?

> what about Socrates?

Erm, what?

tntnon Show HN: Applying the Unix philosophy to neural networks

Unless you share it between processes.

tntnon V8: A Year with Spectre

I don't see how this would really work. As soon as one core pulls in a cache line, all the other cores know about it, so somehow you would have to roll back all the coherency messaging as well.

tntnon V8: A Year with Spectre

... yeah.

And that is a different discussion. The article, and the discussion here, is regarding side channels attacks within a single process. I'm pretty sure everyone agrees that the hardware (or some conspiracy of the hardware and kernel) must provide process isolation.

tntnon The black hole picture is both an astonishing achievement and a deceptive image

... there's no criticism in this article.

tntnon Ford CEO says the company 'overestimated' self-driving cars

either that or NYC is just really inefficient in terms of cost.

Keep in mind you've cited probably the first or second most expensive mile of any subway ever built, and subways usually cost much more than other trains.

China regularly builds high speed rail for $30M per mile. Europe is usually closer to $50-60M. Tokyo has built subways for $400M per mile.

https://www.nytimes.com/2017/12/28/nyregion/new-york-subway-...

tntnon Tesla Shakes Up Model 3 Lineup

I guess they are just counting on testing via the irresponsible tesla drivers who tape water bottles to the steering wheel and such.

tntnon Ford CEO says the company 'overestimated' self-driving cars

A third of fatal accidents involve drunk drivers.

I find it interesting that the apparent objective is to make driving safer, but apparently the powers that be have decided that it is easier to make cars that drive themselves than to convince people not to drive while drunk.

tntnon Ford CEO says the company 'overestimated' self-driving cars

> Trains involve massive capital outlays

And self-driving cars don't? How many trains could be built with the billions invested so far in self-driving cars?

tntnon San Francisco’s Slow-Motion Suicide

You would prefer we segregate people into cities based on some measure of worth? No thanks.

What do you have in mind? The stanford elites get to live in san francisco, and the less-elite get shipped elsewhere?

tntnon San Francisco’s Slow-Motion Suicide

Perfect for a response to 'nipponese's comment.

tntnon San Francisco’s Slow-Motion Suicide

Because people tend to distinguish between "cities" and "towns." There are a lot of ghost towns, but few ghost cities. Once a settlement becomes large enough to be called a city, it is rare for it to die.

tntnon San Francisco’s Slow-Motion Suicide

> die all the time

> ur died, Babylon died.

Your two examples are from 2-3 millennia ago. I would think if cities "die all the time," you would have more recent examples.

Maybe it proves the author correct that the first cities that came to mind are from 2500 years ago?

tntnon San Francisco’s Slow-Motion Suicide

http://www.bayareacensus.ca.gov/counties/SanFranciscoCounty....

I don't think there are other major cities where the population is < 14% children. NYC is around 24%.

Your view of the "pretty good community" is irrelevant. Compared to other cities, San Francisco has "very few kids/families."

tntnon NNCP: Lossless Data Compression with Neural Networks

10x the speed of what? lstm-compress achieves a ratio of 1.64 bpb with a speed of 4.4 kB/s, Bellard's LSTM (small) achieves the same ratio with a speed of 41.7 kB/s.