Is there something similar already in existence? A computer that

>link together
What do you mean?

>>51512645
Yes.

I don't know of anything that does what you're asking but it's certainly possible in theory. As for whether people have created such logic, my guess is no, it doesn't exist.

>>51512964
The output of a register may be used an input for another register.

>>51513081
Literally every architecture in existence?

>>51513092
But instead of having a central processor do it, the registers could be reconfigured to create a sort of assembly line for highly used algorithms, in order to increase the speed of computation.

>>51512981
Care to link us to it then?

>>51513118
Anon that sounds like a logic circuit with registers. I fail to see what you're trying to get at. Not all logic circuits need a cpu. You just need a clock.

>>51513159
It would also be clockless.
>Push reconfiguration instructions to machine
>Push Operations/Values to machine
>Any of the instructions in the above categories may be executed in any order.
>Recalculating a Result from a new input then only requires one instruction: updating a value in a register.

>>51513225
>Updating a value in a register

Do you know how registers are designed? How exactly do I update a register without a clock.

File: FINEIMARETARD.png (22KB, 762x524px) Image search: [Google]

22KB, 762x524px

>>51513254
I am misusing the word, due to a lack for a better word. Here is the "register" I am envisioning.

>>51513364
>I am misusing the word, due to a lack for a better word. Here is the "register" I am envisioning.
Looks like you need to retake EE 101.

>>51513404
>Econ Major
>Having a detailed understanding of modern architectures and terminology
Pick one...
I didn't even know what a pointer was 2 weeks ago

>>51513538
>I didn't even know what a pointer was 2 weeks ago
You're going way too fast then.
There is another thread about why clocks are necessary.
Linking two registers together without clocks would just create a hell of complexity.
To put it into terms for an Econ major, imagine one person planning the entire economy. Not a large entity, one person. It's like that.

>>51513602
Yes, it is a hell lot more complicated, which is why I've come here to /g/ for help and in search of similar machines, as I've failed to come up with a set of instruction that would have the "machine" do anything useful. However, would there not be a benefit from the reconfiguration instructions being executable in parallel?

>>51513695
>However, would there not be a benefit from the reconfiguration instructions being executable in parallel?
Yes, there would.
Manufacturing such a processor would take years of some of the best designers. Even then who's to say it will work in every logical combination. Let's not forget the floating Point error pentiums had years ago.
It's all about risk. How likely is there to be a bug once the product is released.

>>51513772
Yes, and aside from bugs, the designs themselves are also likely not to be as optimized as clocked machines due to the head start they posses. So a hardware implementation is naive, however, a crude software simulation of this machine shouldn't be too bad.

>>51513932
>however, a crude software simulation of this machine shouldn't be too bad.
Then what would the point be?

>>51514070
>Help determine the basic instructions for the machine
>Help with finding a way to deal with a disk and input/output
>Prompt more interest and research into alternative architectures?
Like I said earlier, working with just pencil and paper I have not been able to come up with instructions to do anything interesting. Even Project Euler Problem One is a giant unfinished pain in the ass.

>>51514070
>>Help determine the basic instructions for the machine
you mean the literal way we currently design run of the mill chips today? Same problem, not enough time to test every possible IO combination.

>>51512645
Holy shit are u retarded? Go back to smoking weed u dumb piece of shit. Dont post about my assembly without at least reading Intel x86 books, faggot go back to sucking dicks on reddit

>>51514794
/g/, helpful as always

>>51512645
You could easily design a processor to do that, sure

but you'd be taking up the most expensive real-estate on the processor

Also you're saving a clock cycle
on a computer running at 4ghz

>>51515835
See
>>51513225

>>51515404
go read a book nigger

>>51513118
Maybe look into FGPAs?

Your mistake is asking /g/ for a serious answer on a highly technical EE subject. /g/ doesn't have a strong community for that. They barely know what a gpu actually does on the best of days.

>>51513225
>clockless
HAHAHAAHHAAHAHA

Holy fuck /g/, business majors everyone. Trying to get into tech without knowing how the fuck things even work. S. m. h.

>>51516618
Seems to be close to what I've been thinking of, thanks.

>>51513225
You're basically describing exactly what we have right now, just without the technical knowledge to understand the specifics.

>>51512645
The only thing you would gain from being clockless is nondeterministic outcomes and weird race conditions. You can adjust the clock rate until you get to the point where increasing the clock rate more would cause logic failures. Your design wouldn't be able to work faster than that maximum clock rate anyways.

>>51513118
This is kind of like a FPGA. It is a big grid of registers, lookup tables (for math), some hard multipliers and a bunch of interconnects.

I think you mean "pipelining" by assembly line. This is a thing for both CPUs and discrete logic designs.

>>51513225
Not an EE, but how would you even be able to transfer data without a clock while also providing assurance that data isn't overwritten?

Maybe I'm not understanding this correctly.

Dataflow computers maybe?