AVR 8k 8pin S8S1 Small Outline availability

Go To Last Post
55 posts / 0 new

Pages

Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


I can not find ATtiny85 in SSU version, only SU and other Sx which are not S8S1 (6x4.9mm).

Is there any 8k 8pin AVR in SSU, please.

Thanks,

Milan

 

This topic has a solution.
Last Edited: Wed. Jan 26, 2022 - 09:49 AM
This reply has been marked as the solution. 
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


grohote wrote:
Is there any 8k 8pin AVR in SSU, please.
That would be no then - your only choice is t85:

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


Thanks Cliff, true, tiny is alone... sigh...

 

ref. https://www.microchip.com/maps/microcontroller.aspx

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Try these---they all have the same size as the SSU package (at least according to Digikey)  

8-SOIC (0.154", 3.90mm Width)

 

You will need to lose weight, since these are 4K devices, rather than 8K...but we know you can do it! 

Atomic Zombie would probably find a neat way to stack 2 together!

 

 

 

There's also some pic 12 (FUN!!!)  and PIC16 with 8K, but I can't recall if they use the same "unit of measure."

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Wed. Jan 26, 2022 - 11:00 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


avrcandies wrote:
Atomic Zombie would probably find a neat way to stack 2 together!

In the Olden Days:

Top Tips:

  1. How to properly post source code - see: https://www.avrfreaks.net/comment... - also how to properly include images/pictures
  2. "Garbage" characters on a serial terminal are (almost?) invariably due to wrong baud rate - see: https://learn.sparkfun.com/tutorials/serial-communication
  3. Wrong baud rate is usually due to not running at the speed you thought; check by blinking a LED to see if you get the speed you expected
  4. Difference between a crystal, and a crystal oscillatorhttps://www.avrfreaks.net/comment...
  5. When your question is resolved, mark the solution: https://www.avrfreaks.net/comment...
  6. Beginner's "Getting Started" tips: https://www.avrfreaks.net/comment...
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Calm, calm, please- no pigback or pigpic. The T85 code is at 77% with a huge table (512x16b) which counts for 12% and can not be moved to EEPROM.

Thanks for suggestions- sorry, no solution for me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

avrcandies wrote:
but we know you can do it

 

Hope you are right, if it were Tiny13- but I never did any shrinking in any other AVR than T13,

this is not natural to me, and, yes, the code is in ASM.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

with a huge table (512x16b) 

Ah, just store the value differences, or use some other clever way, even zip ....well maybe not

Why do you need such a big table? 

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

LED light does like it, Gaussian change.

 

; Produced by bellshape_tbl8k.awk for bdm v7

jqrt00: .dw  0, 0, 0, 0, 0, 1, 1, 1, 1, 2
jqrt01: .dw  2, 2, 2, 3, 3, 3, 4, 4, 4, 5
...
jqrt49: .dw  4655, 4660, 4664, 4668, 4672, 4675, 4678, 4681, 4684, 4687
jqrt50: .dw  4689, 4691, 4693, 4695, 4696, 4698, 4699, 4699, 4700, 4700

/*
 y =  k2 * k0 to (-k3 * ((x-k4)/k5)**2) - k6
k2 = 4710
ze=exp(1)
k0 = ze
k3 = 0.00241
k4 = 9
k5 = 10
k6 = 9
*/

Last Edited: Wed. Jan 26, 2022 - 12:10 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


based on the numbers you gave, is this the equation you have in mind?  Assume you mean "to" as a power (rather than some loop)

 y =  k2 * k0 to (-k3 * ((x-k4)/k5)**2) - k6

 

What are the value for x & how does each row change something in this equation?

I plugged in some values for x, but got nothing sensible

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I wondered about "to" too!

 

It's odd that it uses both to and also **.

 

I did wonder whether it might be possible to simply calculate "on the fly" but that would presumably require an FP library which might suggest some sections being in C or at least taking the math stuff from libm.a and using avr-as as the assembler.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Linear at lesser current.

 

re FP, C++ fixed-point arithmetic (C++20, C++11)

Build | GitHub - johnmcfarlane/cnl: A Compositional Numeric Library for C++

...

Alternatively, CNL is a header-only library so you can simply point to the include directory

...

due to The Embedded Muse 361 - Tools and Tips

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

There is gawk program, it can be educational- just provide gawk.exe and try:

 

# Create intensity tables for Cor12
# The shape is bell-like by a gaussian formula
# use:
# c:\Users\MILAN\Documents\avr\_awk\_gawk\gawk.exe -f bellshape_tbl8k.awk > tbl8k_4700.asm

BEGIN {
  k2        = 4710  # 2204 bdm7l
  ze        = exp(1)
  k0        = ze
  k3        = 0.00241
  k4        = 9
  k5        = 10
  k6        = 9
   
  tp("")
  tp("; Produced by bellshape_tbl8k.awk for bdm v7")
  tp("")
  for (i=51 ; i>0 ; i--) {
    s1 = ""
    for (j=10; j>0; j--) {
      x1    = i * 10 + j - 1
      x11   = x1 - k4
      x12   = x11 / k5
      x13   = x12 * x12
      x14   = -x13 * k3
      x15   = exp(x14)
      x16   = k2 * x15
      y     = int(x16) - k6
      if (y<0) y = 0   # case first result is -1
      s1    = (s1 ", " y)
    }
    sub(/^\,/, "", s1)
    s2    = "jqrt"
    yp    = 51 - i
    if (i>41) s2 = (s2 "0")
    sp    = (s2 yp ": .dw " s1)
    tp(sp)
  }
  tp("")
  tp("/*")
  tp(" y =  k2 * k0 ** (-k3 * ((x-k4)/k5)**2) - k6")
  tp("k2 = "k2)
  tp("ze = exp(1)")
  tp("k0 = ze")
  tp("k3 = "k3)
  tp("k4 = "k4)
  tp("k5 = "k5)
  tp("k6 = "k6)
  tp("*/")
  tp("")
  tp("; eof")
  exit
}
 
END {
  exit(0)
}

#func sconv(nconv) {
#  return "" nconv
#}

func tp(tps) {
  print tps
}

# eof

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

How many t85's do you need?

Jim

 

FF = PI > S.E.T

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Not many.

On the second thought, SU/8S2 is not that bad, but, yes, SSU is very cute.

T85 is "low power" option for me, because I will use TQFP32 chips for full power.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

 

What I was wondering, was if you know the shape of one row (say i=1), then maybe the other rows are the same values  perhaps with  a scaling or shifting factor.

I have not looked at it enough to say.  The store one row as the shape and use it for the others?

 

Edit:

Well, I looked at it--a little tough

There is a trick:

Please note that you have 10*I +J in the exp 

I is a range 1-51, J is a range 1-10

 

Put J aside, don't use right now

Also leave K6 (the 9 subtraction) for the very final result step  

 

Now you have 51 values that you can put in a table, call them group V 

V=exp^(power based upon 1 ....to ....51)  

 

power is -0.00241* ((10*I-10)^2)/100

 

remember  that A^(B+C)  is (A^B)*(A^C)

EXAMPLE  3^(6+2)=3^8=6561   (3^6)*(3^2)= (729)*(9)=6561

 

Now bring J back in (it is our "C")

 

So you have a form like   V*(A^J)   , except the J power is actually -0.00241*(J*J/100)

A is the same exp base you used when forming V

So these ten A^J values can also be calculated ahead of time, call this set P

 

Thus, you only need 51 V values multiplied by 10 different P values

That is 61 values needed to be stored

result= K2*Vselected*Pselected -K6

 

At least that is my theory!! 

There is probably more work to scaling the independent wide-ranging numbers for storage---that could be an issue

 

your formula

edit:

 

arrrrg...I think my idea is a flop, due to the square term

you cant  pick I and then come back later and add J back in independently

(I+J)^2 = I^2 +2IJ + J^2    the 2IJ term means the result has a simultaneous dependency 

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Wed. Jan 26, 2022 - 08:44 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

                                                                 sum1...10/10
jqrt00: .dw  0, 0, 0, 0, 0, 1, 1, 1, 1, 2                                 0.6
jqrt10: .dw  72, 74, 76, 77, 79, 81, 83, 85, 86, 88                       1.6
jqrt20: .dw  455, 462, 469, 476, 484, 491, 498, 506, 513, 521             5.6
jqrt30: .dw  1618, 1634, 1651, 1668, 1684, 1701, 1718, 1735, 1752, 1769  15.1
jqrt40: .dw  3509, 3528, 3546, 3565, 3583, 3602, 3620, 3638, 3656, 3674   6.5
jqrt50: .dw  4689, 4691, 4693, 4695, 4696, 4698, 4699, 4699, 4700, 4700   1.1

 

I have tested several exponential types, but the Gaussian is far better for LED light start-middle-end.

Both nasty effects (which I called) "jerk-start" and "drop-bag-end" are eliminated.

 

Answer to your question: 

It may be possible to have a table of 128 x-elements, each valid for 8 x-elements.

Instruction mul may be necessary (which is missing at T85).

 

Last Edited: Wed. Jan 26, 2022 - 07:54 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Please note I made a response above (#16)---you might be able to do it with 61 values, though I think the wide range of scaling could be an issue.

 

It is based on A^(b+c) = (A^b) * (A^c)

 

See what you think!

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Wed. Jan 26, 2022 - 08:26 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think you can be right. (I have had many math semesters in college, I can read math of yours).

By looking at those awkward coefficients in the formula, any splitting can be a problem.

 

So I did it, by a table. Any other solution is possible- not as easy as two lpm, but much harder.

The answer is: not with T85. It is crippled enough because of lack of mul, and change to 4k is far away.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Can you really notice a difference 

4689, 4691, 4693, 4695, 4696, 4698, 4699, 4699, 4700, 4700

 

these are a few counts moving out of 4700 (like 2/4700 = 0.04 percent)...very small

Maybe use only a 0-255 stored range, gives 1 part in 255 (0.39%)???  Is that enough?

Maybe in some places the change is too rapid...or maybe ok?

Many years ago I calculated log & exp using AVR asm with no mul instruction...was tough but doable.

 

you might like this book

https://www.ebay.com/itm/1339554...

Do big math on small machines Write fast and accurate library functions Master analytical and numerical calculus Perform numerical integration to any order Implement z-transform formulas Need to learn the ins and outs of the fundamental math functions in

 

Math Toolkit for Real-Time Programming

I hope this book doesn't start a chain reaction conversation crying

 

 

 

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Wed. Jan 26, 2022 - 09:47 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It is even 'worse' at the begin. Both begin and end does have a weak gradient, which at the middle of the table does become impressive.

 

All this, because our eye works logarithmic, detect a change only in such steps, for example, 1, 1.4, 2, 2.8, 4, 5.6, 8, 11 ... and similar is the table.

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You might want to look at the recent thread by Kerim about delta encoding. If the step from one value to the next in the data is never great you might be able to encode the deltas into just 1/2/4 bits per entry and pack it into a much smaller space than using 16 bits per entry. 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

(6x4.9mm)

Are you looking for a particular package, or a particular overall size?

Some of the new "XTiny" chips are available in TSSOP14 (6.4x5mm)  (up to 32k flash!)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


Thanks, westfw, can you tell me a reference, please.

None XTiny is in searching tool #3.

 

 

Edit: also in Mouser of my country:

 

Last Edited: Thu. Jan 27, 2022 - 06:36 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


These the TSSOP-14...Digikey makes it easier to search  ATTINY20, ATTINY424, ATTINY824, ATTINY1624, ATTINY3224,

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

About the LUT

 

How many clk's is it aloud to take?

 

if <10 no solution other than the LUT

but if max 50 I guess 1/4 size

if max is 1000 clk 1/32 in size (remember the tiny85 don't have mul).

 

This is if it should be 100% correct, if small errors is ok then even smaller.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Clock number should not be a problem (for me), there are only 12 devices to calculate the next LUT position, 30 times per second.

 

Simplification of the table to 128 x-division is first to come to mind (and avrcandies insists), will reduce a table size to 1/4.

 

Let x1<x<x2 where x1 and x2 points to two of 128-wide LUT positions, access is : Z <- Xn/4, then lpm

(Does not needs mul math for 1, 2, 3):

 

 x1 = x and 1FC  (clear 2 LSB)

 (x2 = x1 + 4) not necess.

 Read y2 and y1 by lpm

 dy = y2 - y1

 dx = x and 3  (= x-x1:   1, 2, 3)

 y = y1 + dy * dx / 4

 

 

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


 

 

here are your curves

 

 

Below, I did look at modeling as a pot wiper with  resistors in parallel with the top & bottom halves of the pot....I could get a shape that is "somewhat" similar (its really a sigmoid curve)

In reality they only crudely compare to each other.

The calc for the curve was just product over sum, ratios, and simple clalcs...no exp , no powers (that's the main benefit)  

Two resistor values set the "shape" curve inflection points

So you'd supply two resistor values, X then calc the Y value result. 

Of course with the tiny...that might be fun!  

 

The shape depends on which parameter is varied as the x value.   Here I varied the pot "position" as X, but I could have instead varied one of the parallel resistors, or maybe both.  

 

 

for you, here is a generic visual...I will ignore the 4-way connection!

 

 

 

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Thu. Jan 27, 2022 - 10:25 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

grohote wrote:
None XTiny is in searching tool #3.
tinyAVR® Devices | AVR® Instruction Set Manual

Core is AVRxt.

Instruction Set Summary | AVR® Instruction Set Manual

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

 

Thanks, avrcandies, it seems to be very close to exponential reality.

 

Here is my Gaussian curve: 0-4700 on Y-axis (200 per division), 0-510 on X-axis (50 per division).

 

 

 

Last Edited: Thu. Jan 27, 2022 - 11:38 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Perhaps it is my eyesight but doesn't the majority of that curve look pretty linear for the majority of the middle part? If so then that part could easily be boiled down to a simple y=mx+c couldn't it? You just need sample tables at either end (and again it could be me seeing things - but they almost look like the curves are mirror image - so perhaps only points for one end actually need to be stored?)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

 

No, not linear, not even the smallest part is, but- if you divide in 128 sections, then, yes, it can be done.

 

 

Last Edited: Thu. Jan 27, 2022 - 04:51 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Why do you need ten 51 point curves?  

I was looking to find how to take one & turn it into one of the others...but haven't figured that yet (it is not just a K scaling)

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Thu. Jan 27, 2022 - 06:48 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I need 510 points, to allow a continuous increase or decrease of light, up to 15 seconds.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Why do you have 4700 as top ? (I would expect 4095 (2**n-1) )

 

If you can show the all the 512 numbers you need I will look into a solution (I take it as a crossword puzzle).

 

The problem with a smaller LUT is that the code gets bigger

My guess is that LUT+code will be about 70-100 bytes

Last Edited: Thu. Jan 27, 2022 - 09:49 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


 

 

 

I need 510 points, to allow a continuous increase or decrease of light, up to 15 seconds.

Ok, though your table generates ten 51 point curves, not a 510 point curve...which one  is your intention? 

 

Oh, maybe you pick one as a final value curve, and ride it for 15 sec?

 

Maybe you could have only one curve and just ride it until you get to the desired level?

 

Note ---I noticed the plotting did not match the data I generated---I have reposted the plots

 

I believe this is the data you are using:

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Thu. Jan 27, 2022 - 10:30 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

does this mean that it's always one step up or down and you don't need GetValue[200], just GetNextValue ?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

avrcandies wrote:
Ok, though your table generates ten 51 point curves, not a 510 point curve...which one  is your intention? 

 

This is an assembler table, and access is simple. It is just one table, not a family of them (I do not know where you get this idea).

 

jqrRefresh1d:  ; in: X=phase Y=dev adr
;                out: w1=zLvlL, w2=zLvlH
    ldi     ZH, high(2*jqrt00) ; jqrTableLog
    ldi     ZL, low(2*jqrt00)  ; jqrTableLog
    add     ZL, XL
    adc     ZH, XH
    add     ZL, XL
    adc     ZH, XH
    lpm     w1, Z+
    lpm     w2, Z
    ...

Forming of Phase in X is another story, it can go up or down and be changed as needed (mean in  1, 2, 4, 8... steps) each 32ms.

Number 4700 means 4700*0.25= 1175us of light.

 

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I am using a kind of GetValue[XH:XL]. To scan all 4700 values in 2s, X is changed by 8.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

 

Oh I see, since you use I (1-51) & J (1-10) to generate data, I thought you were making 10 curves in an iteration

 

If I wanted to calculate 510 points on one line, I'd just use one parameter & vary it from 1 to 510

Then I would have 510 points ready to work with

 

I see now, you are working to break the data into shorter rows & did that during the calculation itself!

Now it is clear.

 

Here is the plot of the 510 points (all the rows together)

I bet if you broke this into 20 segments & linear interpolated, you would be plenty accurate.  The segments don't have to have the same width.  

I've done it in asm many times, years ago, just using integer math.

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Thu. Jan 27, 2022 - 11:20 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Do recheck, please- a curve of mine does cross y=1000 at x=200 (approx).

Is your formula correct?

Edit: hold on- I was wrong, my x-division is 50.

Last Edited: Thu. Jan 27, 2022 - 11:40 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

 

grohote wrote:

Do recheck, please- a curve of mine does cross y=1000 at x=200 (approx).

Is your formula correct?

  

 

This is my table I posted before, but arranged in one column of 510 points.

go look at my previous table---you will see it matches exactly some .dw numbers in rows you posted early on (in #17)

 

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Thu. Jan 27, 2022 - 11:35 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes, I will correct 40 into 50, sorry.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

avrcandies wrote:
The segments don't have to have the same width.

 

That would be a problem for T85. See, my example in #27 is easy to use - multiplication with 1,2 and 3.

For mul-capable AVR, agree- any width is possible.

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes there are some advantages to same width segments---you only need to  store y, since all the X are uniformly known:

 

(0, 25)

(10, 78)

(20, 123)

(30, 160)   ...no need to store 0/10/20/30 in the avr

------------------

(0,25)

(5,60)

(16,106)

(21,111)

(23,133)

(35,206)   ...variable widths

...now you need to store x & y, more storage, to get both dx & dy

This is what I usually do, but not always.

Sometimes variable width can actually result in fewer points needed on the line & actually save memory...it depends!

 

I like your idea spacing x by a power of 2 so the divide can be done by a shift  & no multiply needed...maybe even use 8...that is still about 63 segments...a lot! 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Fri. Jan 28, 2022 - 12:06 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

These the TSSOP-14...Digikey makes it easier to search  ATTINY20, ATTINY424, ATTINY824, ATTINY1624, ATTINY3224,

Sorry; I thought "xTiny" was in common use - that refers to the "new" Tiny-0, Tiny-1, and Tiny-2 architecture chips.
They're still just called "ATTINYxxny", where "xx" is memory size, n is 0,1, or 2, and y is a package designation.

 

I also thought that there were more of these 14pin chips than essentially just the ATTINYxx24 :-(

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

is the board already ready ?

if yes have you sonsider/have room to create small PCB   qfn20->soic with plated half holes in the side and solder a newer tiny806 for example qfn20 on top of this ?

 

if pcb is in development phase design a newer case with new "xtinys" from the ground up :)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks, plouf. It is not (only) board question... this Tiny806 is too advanced for me.

I like to repose in the security of 8pin smd, qfn is too small... not to me.

(a hint to Moderators: could you consider a special Forum for this devices?)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

avrcandies wrote:

you might like this book

https://www.ebay.com/itm/1339554...

Do big math on small machines Write fast and accurate library functions Master analytical and numerical calculus Perform numerical integration to any order Implement z-transform formulas Need to learn the ins and outs of the fundamental math functions in

 

Math Toolkit for Real-Time Programming

I hope this book doesn't start a chain reaction conversation 

 

Thanks for the link, just ordered.

Happy Trails,

Mike

JaxCoder.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

MATH Toolkit for Real-Time Programming, Jack Crenshaw | Even More Book Reviews by Jack Ganssle

[third paragraph]

Most of us who went through engineering school once knew this stuff, ...

"Numerical Methods" was the course title.

... but likely forgot the details immediately after the final exam. ...

laugh Indeed though still have the textbook.

 


Jack Crenshaw, Author at Embedded.com

 

Conversely

Math Toolkit for Real-Time Programming [With CDROM] | IndieBound.org

Only 74 miles (119 km) away in a relatively nice city (IMHO)

 

Numerical RecipesTM : The Art of Scientific Computing

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Pages