Memory compression for data line

[MaximT]

 

THE CHATPLACE TECHNICS

 

In honor of Laplace, I'm building this imaginary space that will be a complete screen of symbol to be read and encode by the two end computer that will transfer the images. In the imaginary space of 4K, there is the place for all characters of this planet in 20X20 format pixels, of about 20,736 completed characters. All the screen possibility will be encoded materially into a memory read only device, that will trig the data stored into by the call of the transfer screen number. In the case of Mars, that should be a 7 X 7 low dispersion laser array.

 

This system is now improved, by the use of single pixel (bit) in different screen space (memory available), for the different use.

PARENTHESIS ON THE POWER GENERATED BY THE CHATPLACE TECHNICS

 

The list of device that follow doesn't include the CHATPLACE device, because it does not exist actually

• SDRAM:

  • 4 G bytes / 10 ns : 1,5 W

• EEPROM:

  • 512 k bytes / 70 ns : 0,1 W

  • 1 M bytes / 70 ns : 0,1 W

  • 1 M bytes / 200 ns : 0,05 W

• LPDDR4:

  • 32 G bytes / 0,25 ns : 1 W N.B.: This 0,25 ns could theoretically be increase by burst data to 0,032 ns by adding a memory complex to compare with the packet within a 4096 Bus width, all that modifiable

 

My approximate evaluation for the power consumption, will include the others parts of the device, that will be estimated by doubling the power consumption of the memory. For a cell phone, with a factor CHATS-LAPINS of 2048X, it should be about 2 W, for a 4 G bytes memory, remembering the reader that the ratio shall be calculated with bits values...

I must remember the reader, that this ratio is applied on the real bandwidth, so, without parallelism of some devices, you won't achieved any increase in effective data rate in the phone, but only reduced the cost of the transmission by 2048

With a 5 M bytes transmission rate device, the cost of decreasing the rate and maintaining the total phone data rate, should be:

 

• With a 32 G bytes memory unit

  • A ratio of 2048, that mean 64 CHATPLACE unit parallel

  • 64/44 of the data: 1,6X

• With 4 X 32 G bytes memory unit

  • A ratio of 2048, that mean 256 CHATPLACE unit parallel

  • 256/40 of the data: 6,4X = 32 M bytes / second At the cost of 8 W and 4 mm X (10 X 10) mm, that will be restore to the phone by lowering the WiFi bandwidth by 320X, even more

• With a 32 G bytes memory unit

  • A ratio of 1092, 256 parallel

  • 256/21: 12,2X, but this time 160X on the WiFi, and 2 W, and 1 mm X (10 X 10) mm

SOME ELECTRONICS

 

To achieved better efficiency, there are some designs specifications to be described:

 

• The two different Bus of the single chip IC, that will included memory, are as usual address and data bus.

• Since that scanning comparison procedure will be incremented one by one in the same way every time.

• There is no need to input the addresses, but only to trigger the process, for synchronization purpose.

• The total memory size of the chip IC, will be (2^m/2 + 2) * (2^m/2) bits. For storing the compared data packet.

• That storage capacity is justify, by the use of very huge internal Bus width, for speed (overall time).

• The comparison could be achieved by multi “Equal” transistorized parallel system, for # of cycles too.

• Multi signal (cables or frequency) merge inside a separate chip IC (or the same one: heat consideration)

 

The final cycle time (to produced a “m” width “integer”) : Max ( 0,25 ns X 2^m/2 X 2^m/2 / Internal Bus Width, 2^m/2 (from more than one signal) / achievable input data Bus width (physical chip width) / 100 sub memory packets (0,002 ns transistor switching time, on a pin synchronization, within the physical input data Bus: theoretical) )

 

The maximum theoretical input: 5X 100 G bit/s input, but we won't achieved that The device looks to be limited only by the comparison process...

 

At the end: 0,25 ns X 2^m/2 seconds per cycle, with an internal Bus width of 2^m/2 = 61 kHz per parallel device

 

With the 100 G bits/s cable: 51,230 devices in parallel, to saturate it... with an astonishing 1,64E15 bytes of memory

[Strange]

12 hours ago, MaximT said:

THE CHATPLACE TECHNICS

What is "chatplace"?

12 hours ago, MaximT said:

I'm building this imaginary space that will be a complete screen of symbol to be read and encode by the two end computer that will transfer the images.

If it is imaginary, how are you building it?

What "symbols" are you talking about? The Latin alphabet? ASCII? Unicode?

What "images" are you talking about?

12 hours ago, MaximT said:

In the imaginary space of 4K

4K what?

12 hours ago, MaximT said:

there is the place for all characters of this planet in 20X20 format pixels,  of about 20,736 completed characters.

Unicode can represent over 1 million characters using 20 bits. 

You are representing 20K characters using 400 bits each (assuming 1 bit pixels). Over 100 times less efficient.

12 hours ago, MaximT said:

All the screen possibility will be encoded materially into a memory read only device

So, 8 MB to store 20K characters. Not very efficient.

12 hours ago, MaximT said:

In the case of Mars, that should be a 7 X 7 low dispersion laser array.

What is the relevance of Mars?

What is the relevance of a 7x7 laser array?

12 hours ago, MaximT said:

This system is now improved, by the use of single pixel (bit) in different screen space (memory available), for the different use.

You mean it was once worse than this?

12 hours ago, MaximT said:

PARENTHESIS ON THE POWER GENERATED BY THE CHATPLACE TECHNICS

How is a storage system for characters supposed to generate power?

12 hours ago, MaximT said:

The list of device that follow doesn't include the CHATPLACE device, because it does not exist actually 

• SDRAM:

  • 4 G bytes / 10 ns : 1,5 W

• EEPROM:

  • 512 k bytes / 70 ns : 0,1 W

  • 1 M bytes / 70 ns : 0,1 W

  • 1 M bytes / 200 ns : 0,05 W

• LPDDR4:

  • 32 G bytes / 0,25 ns : 1 W N.B.: This 0,25 ns could theoretically be increase by burst data to 0,032 ns by adding a memory complex to compare with the packet within a 4096 Bus width, all that modifiable

What are all these memory devices being used for?

12 hours ago, MaximT said:

To achieved better efficiency, there are some designs specifications to be described:

You could start with a sane idea, instead.

 

In summary: I have absolutely no idea what you are jabbering on about.

 

Maybe you should start from the beginning and say what you are truing to achieve.

[MaximT]

1 hour ago, Strange said:

In summary: I have absolutely no idea what you are jabbering on about.

 

Maybe you should start from the beginning and say what you are truing to achieve.

It's a data compression technics, that use a lot of memory. That memory data, will be throw in the data line on the reception of the integer address where it is stored, I'm using imaging to try explain all that. Sorry about being unable to achieved it easily, but I will complete today...

[Strange]

3 hours ago, MaximT said:

It's a data compression technics, that use a lot of memory.

Is it? Perhaps you can explain how taking 100 times more space to store data is "compression"?

What is your compression algorithm?

Is it it lossless or lossy

How does it compare with other compression algorithms?

What data sources have you tested it with?

[MaximT]

2 hours ago, Strange said:

Is it? Perhaps you can explain how taking 100 times more space to store data is "compression"?

In a cell phone with a ratio of 2048X, it will takes (1 mm X (10 mm X 10 mm)) / (7.5 mm X 200 mm X 100 mm) X 100 = less than 0.1%

 

3 hours ago, Strange said:

What is your compression algorithm?

Your right, it's not really a compression algorithm, it's material, with the ratio 2^m/2 / m

 

3 hours ago, Strange said:

Is it it lossless or lossy

It's lossless 100%

 

3 hours ago, Strange said:

How does it compare with other compression algorithms?

It can't be compare, it's incomparable!

 

3 hours ago, Strange said:

What data sources have you tested it with?

It's completely theoretical, the biggest issue is the time in nano second of the memory chip, to complete a cycle of comparaison, but with that ratio, it does not matter! 

[Strange]

7 hours ago, MaximT said:

In a cell phone with a ratio of 2048X, it will takes (1 mm X (10 mm X 10 mm)) / (7.5 mm X 200 mm X 100 mm) X 100 = less than 0.1%

You need to explain what you are talking about. I have no idea what these numbers mean or wherevthey come from. So they do nothing to support your case.  

I can equally well say that in an elephant with a ratio of 42Q it takes 5 miles / 2 light years = more than 200 kg. 

It is nonsense.

7 hours ago, MaximT said:

Your right, it's not really a compression algorithm, it's material, with the ratio 2^m/2 / m

Where does this ratio come from? You appear to have made it up. 

7 hours ago, MaximT said:

It can't be compare, it's incomparable!

Incomparable, incomprehensible and unbelievable. 

7 hours ago, MaximT said:

It's completely theoretical, the biggest issue is the time in nano second of the memory chip, to complete a cycle of comparaison, but with that ratio, it does not matter! 

So it is pure fiction.

[Sensei]

16 hours ago, Strange said:

You are representing 20K characters using 400 bits each (assuming 1 bit pixels). Over 100 times less efficient.

He made that (nonsense) value from 4K... 3840 * 2160 / ( 20 * 20 ) = 20,736...

@MaximT

20 x 20 bits = 400 bits. There is 2^400 combinations of possible 0 and 1 in it... (2.58*10^120)

More than quantity of atoms in the entire Universe..

[Ghideon]

On 7/28/2019 at 5:14 AM, MaximT said:

I'm building this imaginary space that will be a complete screen of symbol to be read and encode by the two end computer that will transfer the images.

 

15 hours ago, MaximT said:

It's a data compression technics, that use a lot of memory.

 

8 hours ago, MaximT said:

it's not really a compression algorithm, it's material

I’m just trying to guess what the core of your idea is. My experience from encryption and communication is insufficient to dechipher your descriptions.

Are you trying to store all possible combinations of data that can be transfered, one copy at each computer taking part in the communication? And then transfer a reference to pre-stored data rather than transfering the data?

I’ll comment more once the idea is explained.