BenoitP

43 ms·

481.
by BenoitP 6y ago
I think it is spelled Dunning-Kruger.
482.
by BenoitP 6y ago
Not so much a unification, as the new constructs sort of enhance and retrofit current primitives types. These new primitives are "class instances that lack object identity and can be stored and passed directly, without object headers o
483.
by BenoitP 6y ago
"- A component of the main electric trim system became inoperative. Our pilots ran the appropriate checklist, which included manually trimming the aircraft. They returned to MIA and landed uneventfully. The issue was not related to MCA
484.
by BenoitP 6y ago
Turkey is part of NATO at the moment, so it would not trigger an automatic NATO-wide engagement. But for a Baltics invasion, it would.
485.
by BenoitP 6y ago
> jumping through AWS hoops I found S3 static hosting to be fairly straightforward. Once set up, a single `aws s3 sync . s3://my-bucket/path` updates the content.
486.
by BenoitP 6y ago
> for how long the GC is going to kicks in. 1ms (max, average at 50us) And for the 'when' I'll add that the very concept of having a concurrent GC means you don't need to do a (potentially pausing) malloc right in the
487.
by BenoitP 6y ago
There is an overhead: They use higher bits in the address space to indicate various stages in the object's collection (Shenandoah has a forwarding pointer IRCC). This means you may not activate the compressed pointers optimization.
488.
by BenoitP 6y ago
> you can expect to see average GC pause times of around 0.05ms (50 µs) This is nuts (and very well below OS jittering)
489.
by BenoitP 6y ago
> I assume it's a structural thing? Maybe repairing the first floor will generate too much vibrations and additional dust on the external environment.
490.
by BenoitP 6y ago
That's actually par for the course for a lot of Java content, and that's ok. I'm not the youngest developer, and I like to see content on my steady bread-winner so I upvote it. Java is in a mature phase, so there are not a lo
491.
by BenoitP 6y ago
The Frontpage rank I'd like to see is a PageRank. Quality is a transitive property; I want to see what is upvoted by those who are upvoted (comments as well as posts)!
492.
by BenoitP 6y ago
Well, making a blocking call -say writing to a log for auditing purposes- in a tight loop can dramatically decrease performance. There will be no issue in terms of program correctness, but there will be one in terms of performance assumptio
493.
by BenoitP 6y ago
This makes me think, with the arrival of continuous querying: * kSQL' EMIT CHANGES * Spark has stream-stream JOINs * Flink's default behavior ('time-varying relations') is to emit the changes * Materialize's value p
494.
by BenoitP 6y ago
So are they using Parquet, ORC or Arrow under the hood; or do they have a custom format?
495.
by BenoitP 6y ago
On the other hand AWS and GCP services almost have a 1-to-1 mapping; so one only need to know one platform to be able to start working with the other.
496.
by BenoitP 6y ago
> that he would "short" it if he could This takes on a new meaning in light of GameStop's massive naked short selling, with buyers never really possessing hard shares [1]. I'm not saying a short selling infrastructure
497.
by BenoitP 6y ago
GPU mostly does SIMD. This would enable to do independent workloads. If your simple single threaded webserver only requires 20MB of RAM, you could just run it in its tiny chunk of RAM; and it would sometimes ask the CPU to coordinate with t
498.
by BenoitP 6y ago
For predicate pushdown in S3, I'd see that operation in the SSD controller. You wouldn't even have to load the date in RAM before filtering it.
499.
by BenoitP 6y ago
Well, in a balanced article that'd be something relevant to mention. I think too that transaction fees will dominate the block subsidy, but the effect of an eventually successful level 2 solution remain to be evaluated. Part of the fee
500.
by BenoitP 6y ago
Why is it that these articles never mention halvings?
501.
by BenoitP 6y ago
Yep, this is indeed very little detail. To me this PR is the PR you send when you just discovered you're late, but try to retain the clients' attention. They felt the wind turning, saw the market ask for it, saw companies like UPM
502.
by BenoitP 6y ago
> All that stuff is not just relevant to tensorflow / pytorch stuff but also databases. Yes! and that's the beauty of it. It is not an accelerator, these are fully generic cores. Not equivalent to 'smart' Intel cores
503.
by BenoitP 6y ago
Transport uncached 32 bits from RAM: 650 pJ 32 bit multiplication : 3 pJ [1] The energy savings come from not transporting data. [1] http://www.sigmod2014.org/damon/slides/picojoule.kozyrakis.p...
504.
by BenoitP 6y ago
> They named it AI because buzzword and marketing bullshit... They named it AI because it massively boosts embarrassingly parallel workloads. You can think of Processing In Memory as rendering the mapPartitions() operation free in Spark&
505.
by BenoitP 6y ago
This will take a specification: https://kirkmcdonald.github.io/calc.html#tab=graph&data=1-1-... But 98% of the problem remains: generating a layout.
506.
by BenoitP 6y ago
Nice! I did search for something like this, but I guess my Google-fu is not the best!
507.
by BenoitP 6y ago
I feel like APL would really benefit from a shadertoy-like site. It is an array language, an image is worth a thousand words, and an array[3][2][500] is a square image. It'd be fun to tinker with visual effects of an APL function, and
508.
by BenoitP 6y ago
> most of the compute time spent is in serializing and deserializing data. This is to be viewed in light how hardware evolves now. CPU compute power is no longer growing as much (at least for individual cores). But one thing that's
509.
by BenoitP 6y ago
But it can optimize heuristics to death.
510.
by BenoitP 6y ago
Nice! I guess the next target is to do RISC-V. I think it only works with the interpreter at the moment.

More ›