Log in

izard's Journal

Recent Entries

You are viewing the most recent 25 entries.

31st May 2019

10:08am: This blog
I'm writing the blog in English because I type twice as fast this way than in Russian (yes I am improving...), and 90% of my friends and friends can read English any way. I answer comments in the same language they are posted.

I am trying not to post anything related to my employer, but if I do so, this expresses only my own views and does not represent an official position of my employer.

When I post about some technical topic which seems non trivial and is related to my employer's product, don't expect it to be inside information. If I post about it, it means that this info is already public. Usually I do not post any personal information or anything that is related to my family.

All photos are mine, and I allow anyone to copy, change, do anything you please with them. I don't post "friends only".

Useful tags are: software , idea , Deutschland, freediving and trip report

9th October 2015

10:58am: Linuxcon is over
I attended all performance related presentations, and I am glad to admit that mine was the most down-to-u-arch details. (Not if that level is really required for most of real world performance analysis projects)

7th October 2015

10:44am: Met b0rg in Dublin
LJ is still a great tool, and russian engineers are everywhere. b0rg shown me a nice place to have a beer in Dublin, and explained me what is the fuss with nosql (as a system developer I am overlooking stuff happening higher in the stack).

5th October 2015

10:20pm: туда, блин x7
I am in Dublin again.. Just had a beer sitting in a bar next to Linux Torvalds, now back in a hotel. By they way on this weekend I spoke with an experienced but young s/w developer who does not know who is Mr Torvalds :). No plans for next two evenings. Thoughts? (I am not really into pubs here)

28th September 2015

5:42pm: New words
Recently I hear a word "baremetal" in a new meaning: application running as a usermode Linux process, where Linux runs on real hardware. I keep hearing it from cloud folks who use OpenStack a lot. Each time I ask: "are you sure your app is baremetal?"

Am I too old-fashioned? It is very misleading, even in virtualization context (and especially there) there are real baremetal images. I think even openstack defines baremetal stacks, or does it? Never seen in the wild..

16th September 2015

2:55pm: First ever memory chip and Intel Edison
Here it is on my desk, a first ever memory chip [which can store a whopping 8 bytes!],
operating as an external storage for Edison:

This chip has "intel" marking not "i" and no Malaysia/Philippines on its back so I assume it was manufactured and packaged in 1969 or early 1970 in a Santa Clara fab...

The code is in my UPM branch. Should I push it to master, I wonder :) ?

9th September 2015

5:22pm: Bias
One of a more difficult parts of performance engineering is confirmation bias. Today I was trying to get a proof of a hypothesis, where a proof would be an exponential graph.

But here is what I just measured:

What I am thinking about now - how to modify the hypotheses? No! What is the error margin in my measurements, can it still be exponential?

26th August 2015

6:03pm: Dreaming about a tuna sandwich
At work, debugging stuff I just developed. Dreaming of a tuna sandwich

and a vending machine only sells chocolate and cheese sandwich. I think on a pic is a shoal of tuna swimming next to me. Or is it mackerel?

25th August 2015

6:40pm: Mushrooms
Here is a german instruction on poisonous mushrooms you find in the forest.

They list "свинушка" in the same list with a death cap :). There was a similar list with edible mushrooms, which did not contain волнушки&рыжики :)

24th August 2015

4:22pm: bugz
For underwater pictures and very few action pictures, I am using Olympus tg-3. According to manufacturer, it is "shockproof to 2.1m/7ft, crushproof to 100kgf". A small plastic piece broke off from a corner:
Recently Jacob was playing with it, and dropped it on the wooden floor from about 30 cm.

I've fixed it with a superglue, but I am not sure if it can still survive diving to 15 meters.. I only tried in a bucket of water.

19th August 2015

2:25pm: FPGAs
After panchul 's insightful posts on FPGA development I decided to try. So I bought an Altera university program board, bought Harris and Harris e-book (yes I know it is free in Russian, but I prefer English), and set everything up.

Boy what a mess! Latest IDE crashes on install with a cryptic message, it is well known and workaround exists.. Then it crashes now and then when working. It all looks like a spaceship controls, but when I try to follow any of "quick start guides" I can't literally follow the steps as some menu items from screenshots do not exist, and when I try to guess it all diverts..

Overall - Verilog seems easier than I thought, but tooling seems awkward. Will keep trying..

17th August 2015

4:02pm: Toys
A pic from a supermarket:

Boy's toy is more expensive, but girl's toy name is a bit embarrassing, isn't it? (First I did not realize it is a TV show..)

7th August 2015

9:17pm: Benchmarking Edison
My comment to a comment to an article on benchmarking Edison, RasPi2,etc

3rd August 2015

5:38pm: NVM
Recent announcement is one-in-a-three-decades thing. I cannot guess how it will change a computer architecture and OS/middleware programming in the following ~10 years. You cannot buy the hardware yet, but software is here.

29th July 2015

10:20am: Programmer's abacus
I just realized that my son's new toy, an abacus,

is a programmer's abacus, not layman's. You see why, right?

28th July 2015

1:30pm: gcc bug or my bad?
I just wrote the following code:
// very simplified:
int *init_impl(int *array, int size)
  printf ("n = %i\n", n);
  // initialize array

  // return pointer to last element
  return &array[size -1];

void init(int *array, int size)
    if (size < 100) return init_impl(array, size);
// if size >= 100, initialize array as chunks of size 100

int main(int argc, char *argv[])
 // malloc array
  init (array, atoi(argv[1]));

This works when compiling as -O0, but with -O3 and size < 100 init_impl call is optimized out - printf is not called and array is not initalized.

The fix:
- if (size < 100) return init_impl(array, size);
+ if (size < 100) { init_impl(array, size); return; };

Is it a compiler bug or a C feature?

23rd July 2015

5:12pm: Freediving vs scuba
After I tried scuba, it took me a while to understand why I still like freediving more. The main reason is degrees of freedom.

A scuba diver has the following degrees of freedom: buoyancy (1), forward-backward(~0.5), roll (~0.5), yaw(1), pitch (~0.5) - total 3.5
A freediver has foward-backward (~0.5), roll(1), yaw(1),pitch(1), buoyancy (~0.5) - total 4.
Plus speed of manoeuvre is so different!

Did I miss something?

22nd July 2015

1:17pm: AI is imminent and it arrives soon?
Recently some smart people noticed that strong AI will appear soon (e.g. here or here). "Soon" means in 3-4 years, and it will quickly come from academia to the industry.

I tend to disagree with timings. Let's assume some smart-ass sitting in an ivory tower indeed creates a strong AI prototype and publish a paper describing it, in 3 years from now. I recon it will still take at least 10 years after that for a full industry adoption.

In recent 10 years I was very often using fresh academia research to create PoCs/prototypes. The gap between source used in a research paper and a prototype I can show in a trade fair was always very noticeable. Even bigger gap exists between PoC and an alpha version of customer ready solution. Then there is a gap between alpha and proper 2.0 release. Then there is a gap between 2.0 release of a successful but novel software, and wide market adoption. If the technology is trivial, each gap can be filled in months, giving ~2-3 years in total.

If the technology is not so trivial but still not too complex, total goes to 5-7 years. Here is an example: this research paper ignited a SDN revolution in networking in 2007. SDN just became mainstream (around 2014). It took 7 years...

Neural networks/deep learning renaissance happened in 2000s, making a leap forward in weak AI and providing a potentially useful building block for a strong AI. However I think that expecting a breakthrough research in 3 years from now is also quite optimistic. So my guesstimate is ~10 years until a Strong AI described in a paper, then 10 more years for adoption. My baby will be 21 by then, so it is not that far :)

Of course, if AI is immediately used to improve itself and drive market adoption it could help but only a bit as it has to convince humans/IT directors/CEOs :), and they use Gartner hype cycle to estimate technology adoption speed.

20th July 2015

10:41am: CPU
Which CPU is it?

Unfortunately I was not careful enough when opening it, and damaged ~4k transistors out of 35k.

15th July 2015

11:04am: Onion after watering time lapse
Yesterday evening I found that because of very hot weather, an onion growing on our balcony was thirsty. I watered it and tried to make a time lapse video by putting a heavy math book on my camera's shutter button.

shutter and aperture was constant so it becomes quite dark in the end, but onion is back to normal.

29th June 2015

11:26am: Dumb dust
When we returned from Israel, we found that floor of our balcony is covered with something yellow. Our first idea was that it is pollen from the flowers that grow in the garden, but after I collected a bit of it into a box,

it appeared to be sand. Indeed a sand storm somehow brought sand from Sahara via mediteranian, Italy, Alps!!! all the way to Munich.

25th June 2015

1:32pm: Journalists..
Every couple of years I read news from my hometown.

The quality of writing goes down. Here are today's headlines":
"Три пассажира бесправного водителя получили травмы в ДТП.."
"Вчера салют начался на 15 минут раньше, из-за чего многие зрители на него опоздали."
"В январе-мае рождаемость снизилась на 5%."
"Житель Чебоксарского района напал на двоих спящих с топором, которых ранее приютил в своем доме."

And [almost] a whole article:
"Прогноз погоды: Жара усиливается - будьте внимательны!

В жару не рекомендуется есть продукты, согревающие изнутри, например, мед и сало. Перед сном лучше отказаться от тяжелых продуктов вроде мяса. Охлаждают организм и помогают адаптироваться к жаре продукты с легким вяжущим вкусом и все овощи-фрукты зеленого и белого цветов. Необходимо увеличить потребление жидкости. Носить лучше одежду из светлых натуральных тканей.

В частности не рекомендуется есть продукты, согревающие изнутри, например, мед и сало. Перед сном гражданам советуют отказаться от тяжелых продуктов вроде мяса, поскольку переваривание тяжелой пищи вызывает повышение температуры тела и усиливает потоотделение. Охлаждают организм и помогают адаптироваться к жаре продукты с легким вяжущим вкусом (хурма или бананы) и все овощи-фрукты зеленого и белого цветов. В жару из-за усиленного потоотделения в организме нарушается водный баланс, поэтому нужно выпивать не менее 3,5 литров жидкости в день."

Even better sample under cut:
Read more...Collapse )

23rd June 2015

11:20am: Deja vu
In my new old job, everything seems the same. I will optimize for Haswell and Skylake a compute kernel that once I optimized for Nehalem. When I installed a RHEL on my test servers I encountered a bug in the installer I remember from old days: if HDD ever used to be in RAID, it is not detected unless you do dd if=/dev/zero of=/dev/sdx

10th June 2015

9:38am: Poing
Munich zoo entrance fee is ~14 euros and you can't feed animals, and in Poing wildpark 2 adults can enter for 15 euros, and you can feed animals (except for carnivores.)

But we were lucky to see how they feed bears.

The ration is simple: a piece of meat, a loaf of rye bread, an apple and a carrot.

A wolf.
Read more...Collapse )

Storks feeding babies.
Read more...Collapse )

Me giving a carrot to a deer.
Read more...Collapse )

8th June 2015

3:41pm: Flame!
This is a nice one from Theo de Raadt. A bit Linus Torvalds style, but even more insults.
Powered by LiveJournal.com