We had the same at HP - if you got the bottom ranking twice in a row you were asked to leave. We had a stable team of 10 engineers, all of which were good at their job but one had to be ranked as incompetent.
We working through the list alphabetically, so everybody got it once in a while but never twice in a row.
Come back to me when we have no tech bubbles, or property bubbles or dotcom bubbles, or facebook being propped up by the inderwriters, or Ponzi schemes, or businesses that are declaired "too big to fail".
If markets were truely connected to 'fundimentals', there would be no predictable noise for HFT to exploit.
I chatted with some guys on an FPGA forum about this. They were convinced that HFT was good, as it increased the liquidity of the market.
I ran the line that it's bad, as it exploits that over the short term all players in the market do not have complete information on the state of the market, and is therefore a highly sophisticated form of insider trading.
In truth it it is just a mechanism to suck wealth away from those who actually create it (or invest in stocks of companies that create wealth), and does more harm than good.
I've been playing with my Raspberry Pi today (just twiddling with 'ncurses' under C). I see it being excellent for learning it is perfect as the standard reference platform for a lot of CS courses from "Introduction to Programming" up - but maybe a bit out of it's depth at OS the design level.
For around the same cost as a text book everybody it ensures that everybody will have the same hardware, the same OS with all the same toolsets. This will avoid the "Jimmy owns a Mac, and I have 32 bit XP, and Bob has an Android tablet" problem. As a bonus it also has zero product licensing issues...
Sure, you wouldn't want to compile a big project on it, but for anything you would do in school it would be fine.
It has long been known that one kg of fat is 7000 (kilo)calories (sorry about mixed units).
So if you eat just 100 more calories per day than you expend, you put on a about a pound a month.
And another rule of thumb - for each kg of body weight you need one (kilo)calorie to travel a mile - regardless if you walk, jog or run. So my 6 mile run burns 540 calories - and if I do that for 7 days without eating more I'll loose 500g.
I don't think that you have any more to learn from picking up a new language - either challenge yourself with something like Project Euler (learn something interesting every night!). Your 'old' languages will get a real workout, becoming tools again.
Or, if you want to go into hardware then get a cheap FPGA board, and design your own stuff.
If this is for a hobby, and you want to keep challenged, buy an FPGA development board (e.g. a Digilent Basys2 or a Papilio One) and learn a HDL. It will cost a little bit of money ($60) but you will get months of play time out of it.
Once you've programmed in 10 or so languages they are pretty generic, but the jump from programming to Hardware Designing is a complete mindfsck, but one you grok it it is very satisfying. Everything happening in parallel in hard real time....
Build your own 'soft' CPUs, invent the next big thing!
I don't know the details of this product, but in most cases 'C' to hardware tools are used to optimize the inner-most portion of critical loops,
One way is to build either a custom CPU instruction - for example a programmable "bitshuffle" for use in crypto.
Another is to build a custom on-chip peripheral where "my_code(arg1, arg2)" maps to "start arg1 in port X, store arg2 in port Y, read answer from port Z" and the custom logic transforms X and Y into Z. The ports might even have FIFOs allowing many operations to be queued or in flight at the same time.
One other common configuration to have the custom logic interface with the DMA subsystem and few control registers (e.g. for an RAID XOR engine).
Usually all these systems have a generic CPU to do the generic stuff - the idea is that the whole design implemented in the same (or almost the same) toolset.
We all know that it is stupid, but one the "next big thing" ideas for FPGA technology will be using them for ultra-low latency high frequency share trading.
The idea being that if you can bypass switches, routers, NICs, buffers, IRQs, CPU contet switches and so on you will be able to issue your trade requests before the whole data packet has finished coming off the wire, allowing you to get a big jump on your competitors.
One assumes that the "buy, buy, buy" or "sell, sell, sell" packets will need to be generated in the finial formats needed by the market, which will most probably need something to be converted from bInary to ASCII characters.
High frequency traders dream that it would be possible to turn a trade around within a few nano-seconds of the market data arriving.
// Make value 17 bits long
for(i = 0; i != 5 ; i++)
{
digit[i] = '0';
if(value >= 80000) { value -= 80000; digit[i] |= 8; }// Extract the bits for the current digit - In 'C' I' need to use the |= operator as I never did get the hang of addressing bits in 'C'
if(value >= 40000) { value -= 40000; digit[i] |= 4; }
if(value >= 20000) { value -= 20000; digit[i] |= 2; }
if(value >= 10000) { value -= 10000; digit[i] |= 1; }
value = value*8 + value*2;// Prepare for extracting next digit. - the *4 and *2 would be implement by the wiring into the adder
}
Advantages: * No divide/mod operator * Extracts digits from most significant to least significant (if you want to stream out the digits) * Can be unrolled or pipelined to meet timing / throughput requirements
Sorry about any syntax/typos/errors in the code... it is a comment!
Most of these technologies 'C' to hardware technologies are overhyped and under-deliver.
* It is definitely not ANSI C. It might share some syntax elements but that is about it * C programmers do not make good hardware designers (C programmers will disagree, HDL programmers won't) * The algorithms used in software by software developers do not translate well into hardware * If you want "good" hardware developed, use hardware design tools.
If you don't agree with me on these points, post how you would convert "short unsigned value" into ASCII in char digits[5] and I'll show you how to do the same if you were designing a chip...
Slashdot Mirror
Lots of interesting stuff to read!
If only their USB stack was robust enough that I could play MP3s without it locking up when I change volume with "amixer"...
(But I have yet to try the latest Raspbian, so it might be fixed now...).
The internet is great for all businesses, but it better not improve the productivity of :
- drug traffickers,
- child predators
- religious fundamentalist (except Christians of course!)
- unauthorised file sharing
- white power groups (except those in the Southern USA, where it is a tradition).
- anti governmental uprisings (except in Egypt and Syria - those uprisings are OK)
- or scammers and spammers (except those Himalayian Gojo berries and commercial Vitamin pills - those are real businesses)
- those promoting the views on "Global Warming/ Climate change", on either side of the debate
- school kids who "dis" their school
- People who believe that endless economic growth is impossible and ultimately unsustainable - the end is near!
A bit harsh... I've got a RPi, cost NZ$53 (US$40ish).including shipping. and 15% tax, so US$36.90 before tax.
Works a bit better than the other ARM boards I've got (the software support is really crap on some no-brander ARM boards).
I've managed to use the GPIO pins for simple stuff too (e.g. with a LVTTL serial to USB) without frying them.
We had the same at HP - if you got the bottom ranking twice in a row you were asked to leave. We had a stable team of 10 engineers, all of which were good at their job but one had to be ranked as incompetent.
We working through the list alphabetically, so everybody got it once in a while but never twice in a row.
NIce dogma...must take some grooming :-)
Come back to me when we have no tech bubbles, or property bubbles or dotcom bubbles, or facebook being propped up by the inderwriters, or Ponzi schemes, or businesses that are declaired "too big to fail".
If markets were truely connected to 'fundimentals', there would be no predictable noise for HFT to exploit.
Even the info in Wikipedia should give you cause to question your stance. http://en.wikipedia.org/wiki/Stock_market#The_behavior_of_the_stock_market
As Kenny Rogers said: "You got to know when to hold them, know when to fold them, know when to walk away, know when to run".
I bet other HF trader recounts the day as he just cleaned somebody out of $150M.
I assume that most politicians have money in Stocks / bonds / commodities and other investments (maybe in blind trusts).
They would never impose a tax that they had to pay!
I chatted with some guys on an FPGA forum about this. They were convinced that HFT was good, as it increased the liquidity of the market.
I ran the line that it's bad, as it exploits that over the short term all players in the market do not have complete information on the state of the market, and is therefore a highly sophisticated form of insider trading.
In truth it it is just a mechanism to suck wealth away from those who actually create it (or invest in stocks of companies that create wealth), and does more harm than good.
Obligatory XKCD reference.
I've been playing with my Raspberry Pi today (just twiddling with 'ncurses' under C). I see it being excellent for learning it is perfect as the standard reference platform for a lot of CS courses from "Introduction to Programming" up - but maybe a bit out of it's depth at OS the design level.
For around the same cost as a text book everybody it ensures that everybody will have the same hardware, the same OS with all the same toolsets. This will avoid the "Jimmy owns a Mac, and I have 32 bit XP, and Bob has an Android tablet" problem. As a bonus it also has zero product licensing issues...
Sure, you wouldn't want to compile a big project on it, but for anything you would do in school it would be fine.
It has long been known that one kg of fat is 7000 (kilo)calories (sorry about mixed units).
So if you eat just 100 more calories per day than you expend, you put on a about a pound a month.
And another rule of thumb - for each kg of body weight you need one (kilo)calorie to travel a mile - regardless if you walk, jog or run. So my 6 mile run burns 540 calories - and if I do that for 7 days without eating more I'll loose 500g.
I don't think that you have any more to learn from picking up a new language - either challenge yourself with something like Project Euler (learn something interesting every night!). Your 'old' languages will get a real workout, becoming tools again.
Or, if you want to go into hardware then get a cheap FPGA board, and design your own stuff.
Just pick any language and work through Project Euler
You will think harder and learn a lot more than learning another language
If this is for a hobby, and you want to keep challenged, buy an FPGA development board (e.g. a Digilent Basys2 or a Papilio One) and learn a HDL. It will cost a little bit of money ($60) but you will get months of play time out of it.
Once you've programmed in 10 or so languages they are pretty generic, but the jump from programming to Hardware Designing is a complete mindfsck, but one you grok it it is very satisfying. Everything happening in parallel in hard real time....
Build your own 'soft' CPUs, invent the next big thing!
I don't know the details of this product, but in most cases 'C' to hardware tools are used to optimize the inner-most portion of critical loops,
One way is to build either a custom CPU instruction - for example a programmable "bitshuffle" for use in crypto.
Another is to build a custom on-chip peripheral where "my_code(arg1, arg2)" maps to "start arg1 in port X, store arg2 in port Y, read answer from port Z" and the custom logic transforms X and Y into Z. The ports might even have FIFOs allowing many operations to be queued or in flight at the same time.
One other common configuration to have the custom logic interface with the DMA subsystem and few control registers (e.g. for an RAID XOR engine).
Usually all these systems have a generic CPU to do the generic stuff - the idea is that the whole design implemented in the same (or almost the same) toolset.
We all know that it is stupid, but one the "next big thing" ideas for FPGA technology will be using them for ultra-low latency high frequency share trading.
The idea being that if you can bypass switches, routers, NICs, buffers, IRQs, CPU contet switches and so on you will be able to issue your trade requests before the whole data packet has finished coming off the wire, allowing you to get a big jump on your competitors.
One assumes that the "buy, buy, buy" or "sell, sell, sell" packets will need to be generated in the finial formats needed by the market, which will most probably need something to be converted from bInary to ASCII characters.
High frequency traders dream that it would be possible to turn a trade around within a few nano-seconds of the market data arriving.
Google just found me cunning way to implement binary to BCD conversion that works by using modified shift registers.
Very slick, Wouldn't be found by a 'C' to hardware process or 'C' programmer.
Looks like you failed to spot the character constant in digit[i] = '0'; - it is already a character....
for(i = 0; i != 5 ; i++)
{
digit[i] = '0';
if(value >= 80000) { value -= 80000; digit[i] |= 8; }
if(value >= 40000) { value -= 40000; digit[i] |= 4; }
if(value >= 20000) { value -= 20000; digit[i] |= 2; }
if(value >= 10000) { value -= 10000; digit[i] |= 1; }
value = value*8 + value*2;
}
Advantages:
* No divide/mod operator
* Extracts digits from most significant to least significant (if you want to stream out the digits)
* Can be unrolled or pipelined to meet timing / throughput requirements
Sorry about any syntax/typos/errors in the code... it is a comment!
Most of these technologies 'C' to hardware technologies are overhyped and under-deliver.
* It is definitely not ANSI C. It might share some syntax elements but that is about it
* C programmers do not make good hardware designers (C programmers will disagree, HDL programmers won't)
* The algorithms used in software by software developers do not translate well into hardware
* If you want "good" hardware developed, use hardware design tools.
If you don't agree with me on these points, post how you would convert "short unsigned value" into ASCII in char digits[5] and I'll show you how to do the same if you were designing a chip...
I like GOTOs... and so does this this guy, and this guy and the book on writing Linux devices drivers
Can you take your dogma out back and shoot it? 8-)
It may not be the project referred to, but have a look at "Ciarcia's circuit cellar, Volume 7" - it has the full design, including schematics.
It is based around Intel 8751 or and 8031 CPU - 64 of them. It was about 20x as fast as an IBM PC AT.
A 640x400 fractal took 1.7 minutes to draw. A current AMD CPU is about 6000 times faster than this 'supercomputer
.
I am sure if I was ask you and him a question about the ARM architecture I know who's answer will be correct....
He was waiting for his Raspberry Pi to arrive, so had time to kill.