Top 10 Simple ways to speed up your Redstone Computer – Minecraft Java/Bedrock

Hello the donkeys344 here and today in this video i’ll just be going over 10 like simple ish ways to well basically speed up your redstone computer so a lot of the time when you’re building a computer uh they can get rather big and also consequently rather slow

And just yeah when you’re first starting out making computers uh it’s just kind of annoying having to wait like two hours for just like a simple operation to uh take place and yeah so you just kind of the more computers you make the faster you really want them to go because

Having a slow computer is just kind of annoying to perform simple functions or just really anything so without further ado the first point is using dedicated components for your computer so over here this is a computer i built a while ago and this basically is just a linear graph plotter

It’s a very slow one i didn’t really do any timings or optimizations as you can see here’s the clock it’s really slow but basically so this isn’t really a good example but say say you wanted to do multiplication uh which this actually requires multiplication instead of actually having multiplication in software

So you can either implement multiplication like in an actual piece of code you can do a multiplication by like repeated addition or do it by like um Shifting left or right and then adding extra depending if the number’s odd or even yeah you can there’s a lot of things you can do basically with multiplication however instead of doing that you can always build a dedicated component like this this is a hardware multiplier

And it’s only it’s only a three bit multiplier so you can num you can multiply a three bit number by a three bit number and as you can see it’s a bit large however this is going to be much faster than a um the actual program multiplier and

Save space on the multiplication because this multiplier only uses one line where a program multiplier like one implemented in software would use multiple lines and which means multiple clock cycles slowing your computer down however there is one thing to consider and that will be on the next point so

With this it will also slow your cpu down because of course the multiplier hardware multipliers hardware dividers and like gpus and stuff uh or lookup tables you could use is a dedicated thing so instead of like trying to calculate out like a sine cause or tan

Value you can just have like a simple lookup table having a lot of values in where you just look up a value basically instead of having to calculate it uh but yeah i think i think actual uh real like calculators you use in school and stuff actually use the lookup table

For signed constant values like just a massive lookup table because it’s just a lot easier that way so the next point is a reduced wiring this is a very common issue i’ve had a lot of my computers especially seeing as i make computers or relatively 2d as you can see

But basically uh reduce wiring and put components like more next to each other so this is a computer that i made a very long time ago now and it’s an avid computer with like really terrible specs only three like memory cells of ram and as you can see

Most of the actual computer is just data buses bringing data to all the different components that’s really inefficient because as you can see each of these data buses has all these repeaters on and each repeater adds a redstone tick making it slower and as you can see the final clock

Is gets quite slow and this is a very simple computer and yet there’s uh a quite a long clock that’s like one two so four eight twelve uh 16 20 24 plus one so 25 then you times that by two because you invert it so that’s basically a 50

Um 50 tick clock for a very simple computer this this clock should really be like 10 ticks but it’s 50 ticks because of all these extra wires so reduce reduce wiring on your computer because i could easily have brought this wire underneath instead of having to go

Around that would save me on repeaters which would mean i could probably clock the computer faster that’s also another point is make sure your actual components aren’t just 2d so you can see this computer is just basically flat as you can see there’s not really much under or on top

Which means all the components uh are in a 2d plane where this ram could really have been on top of this and that would save a lot of space all of this space over here would be saved and it would be much faster because the components are closer so that that also

Basically means put the components uh that are used a lot very close to each other so this aru is always accessing this ram so if this ram is close to the alu then it will be faster because there’s less distance to travel so that’s that’s the second point

Reduce wiring so the third uh thing that you can do to speed up your computer is to use pipelining and uh i’m not very i still haven’t really looked at fight learning in detail so i’ll leave uh a video by matt back wings where he covers it in like 100 seconds uh

100 seconds not that much detail but it’s good enough for beginners and i’m sure there’s a lot of guides about it but basically if you pipeline your cpu it basically means while if you have idle components in your computer so here’s a good example so say say my alu takes four ticks to

Compute and my uh my ram takes three ticks to compute and that’s accounting for like latency as well i’ll talk about that a bit later but then so say this operation first so say that aou first uh does a computation basically it does like two plus two which equals four

So after the aiu has uh computed the number the number will have to be safe to ram so after the fortix has happened which is there you compute the number will travel all the way to the ram so accounting for latency and everything then it will save right in ram and say

That or take three ticks over here so in total that takes four ticks plus three ticks which is seven ticks in total however while it is saving and traveling to ram the alu is not doing anything because there’s just a number going through so what we can do is pipeline the cpu

Meaning that while the aou is idle because the number that it just computed is saving to ram we can put more numbers in the aou to do more computations and this means that as soon as the aou has no data in because it’s just put the wrap put the number and sent it

To ram then we can send in more data into the aru to do another computation so in an ideal world since the alu only takes three ticks over here to actually compute something every three ticks we’d have another number going in and seeing as the ramen takes one tick to save

Every a one tick we’d have something saving however we can’t really do that because it takes three ticks so every three ticks we’d take we’d have it saving which means instead of seven total ticks which should take per clock cycle it’d only take three ticks and yes there’d be a bit of

Latency from uh starting it starting the clock up and having all the data reach all the components but that latency isn’t too much of an issue if you’re doing like a bunch of calculations because you just have to wait for a bit and then it’ll do all the

Calculations really fast so that that’s kind of a brief description of pipelining and then it’s kind of a bad description so hopefully the video i link is a bit better so for the fourth point uh only use serial data when necessary over here i have a computer that i made a

While ago and i have i’m using my serial protocol to transfer 12 bits of data down one line and the design i’m using is by my my zoomer games i think i might link it in the description if i can remember and this basically

Has a one bit a tick data rate so in total uh not accounting for latency or any uh wires or anything this would be able to theoretically transfer 12 bits of data in 1.2 seconds that may seem fast but if we did this in parallel where instead of

Just having one wire i had 12 separate wires i would be able to do it instantly but not accounting for any repeaters or latency of course which would mean that i’d be able to get data from my computer to my display a lot faster if i did it in parallel

Instead of serial because the problem with serial even though it’s one line and it’s very good if your computer is right next to your display and you have a lot of space there is no reason whatsoever to to do in serial yes it looks nice it’s only one line and it

Makes things a lot less crowded however if you just have 12 lines going to this i honestly some of these are redundancy i only need uh two of those lines and the extra four bits so six lines in total not even 12 lines it would make the actual display a whole lot faster

Because you wouldn’t have any encoding latency and you wouldn’t have any decoding latency so use parallel data as much as possible and if you really do serial make sure the data you’re sending through serial is either short or you have serial lines in parallel that means instead of just

Having one serial line like i have over here have multiple serial lines like this yes it will still save space because you only have a few say you’re sending 200 bits of data and uh just divide that by two and you only so you have two lines of

Serial and that means you’re taking half as long to transmit it and it only means you have two extra lines because you can transmit 100 bits down this line and 100 bits down this line so if you really do have to use serial on like massive

Sets of data say for like a display or something so you’re using a serial display like with like a a 16 by 16 or like a 32 by 32 and you want it to be fast then paralyze your data we’ll alternatively use comparators and timing and stuff and use analog signals instead

But analog signals are kind of finicky down long lines but you can still send them using pulse lengths so yeah use use serial only when really necessary so for the fifth point that i’m going to bring up today is use other people’s designs as uh whenever you want so say yeah say

You’re trying to build a cpu or something and you’re kind of stuck on a component you’ve been trying a long time and you just uh it’s just too big or too slow well i’d recommend looking at some other people’s designs and taking inspiration or using them in your own

Computer however when i say this always leave credit like whenever you’re using someone else’s design if it’s if it’s just a private thing that it’s only your computer then it’s fine however if you ever show your computer off remember to leave credit which of the components that you used for example

In my latest computer that i’ve been building i’ve been using this ram design over here as you can see i’ve kind of changed it from the original but it’s pretty much the same thing as the original design and it’s it’s for vertical vertical computer and basically this round design was from

I found in this video which is we built the lc3 in minecraft redstone and this was by cole fusion i’d recommend checking these guys out because this is the aou from it which i’m not actually using i’m using one by benny’s cube that he modified from someone else but uh yeah remember to

Give credit if you’re ever using other people’s designs but as you can see i used this guy’s design just because it’s it’s small and compact and it looks pretty cool there are other designs i could have used and i can probably design my own which i actually did

In fact i designed this over here however it’s just uh it’s always good to leave credit and stuff so yeah remember to uh use leave credit and use other people’s designs if you’re really stuck because they normally are a lot better and faster so you can make a

Very fast cpu with other people’s designs of course the architecture will be yours but yeah other people’s designs can be good too sometimes so for the sixth point this kind of goes back to the first uh the first point i made where i said use the decoder

Components like this i or you over here so this this fifth point i mean six point is use io as much as possible so in this computer over here i showed off the the hardware multiplier which i said in the first point use a dedicated component as much as possible because

It’s faster than software however the problem with this component is since it’s integrated into the cpu you also have to account the amount of time it will take to actually compute and as you can see that makes the clock extremely long the problem with this then is it makes

Your well it just makes your whole cpu slow so say your aru you took two ticks to calculate a number but your hardware multiplier took 10 ticks to calculate a number your your cpu would be bottlenecked by your hardware multiplier or hardware divider or whatever like hardware like big dedicated component you’re using

So if you use i o yes i know this kind of goes back on use serial parallel however you can have parallel i o devices so that’s good but if you if you set this as an i o thing what you can do is whenever i do a multiplication i can

Send my data as well as commands to my hardware multiplier it can then multiply while i’m doing other stuff on my cpu and when it’s finished multiplying whatever numbers i sent in then i can send that data can be sent back to my computer it can issue a hardware interrupt so

Stopping whatever the computer is doing at the moment save the data to ram and then the computer can start again do whatever it’s doing then take the data from ram and use that for the multiplication that means uh uh the the clock speed can be greatly increased because you don’t have any

Super slow components in it so it’s it’s very smart to do stuff like that that also kind of means say you had uh it kind of like links into the amount of ram you have so you had like a a bunch of ram but it’s very slow well

Don’t include that in the computer have that as an io device and only have like a small cache or a small amount of ram for your actual computer if you’re not if you’re not using that extra ram or you’re only using it very infrequently it will slow down your computer so if

You want to speed your computer up you’re just going to have to use it as an io device so yeah that’s that’s the sixth point for my seventh point uh i would say use vertical computers uh more than like horizontal or traditional design computers because vertical computers are just slightly more compact

And can be faster a lot of uh new computer designs like say a big trungus 2 or the qpu like the basically the best of the best computers use vertical designs just because it’s slightly more compact because the way everything is set up you can do timings are a bit easier to do

And then components are slightly more smaller because they the space they take up is uh vertical instead of horizontal like well it’s it’s more of a vertical space than a horizontal space which is is is better because it’s easier to do stuff normally on a computer you’re space constrained in a

2d axis instead of a 3d axis where vertical computers you’re not really space constrained as you can see this vertical computer is more or less square as you can see where my older computers which are more traditional uh just horizontal designs are very flat and 2d and don’t really make space or i

Make use of the available space and so a vertical computer is just a lot more efficient however i i say use them a lot more but there are a few good examples of a very fast horizontal horizontal computer so don’t actually just completely make vertical computers horizontal

Computers can also be good and fast if you make them properly because i have seen i’ve seen a i think it was a 0.7 tick clock horizontal computer that could do like a lot of stuff like just as much as the qpu i think so yeah i’d i’d recommend just having an

Experiment with both types and finding out what is best for you however i would heavily recommend vertical vertical computers because that’s basically what most of the redstone engineers are adopting on the open redstone engineering server or or as it’s called uh most people nowadays create vertical computers so if you want

To actually use existing designs a lot of the designs will be vertical so yeah that’s that’s the seventh point so for my eighth point one of the most things i see common with people making computers is they over spec do not over spec your computer only spec

It up what you’re actually going to be doing it’s almost like real world computers you do not need an rtx 3090 to play minecraft it would be nice and you can have epic shaders but but you don’t need it basically so you do not need 128 gigabytes of ram on your resident

Computer not not that’s actually even physically possible with the game engine but basically only only build what you actually need but yes you can have a few extra cells but uh if you’re only running like the fibonacci sequence on a computer say you only need

Two or three uh ram cells you don’t need 64 ram cells so only really build what you need because of course with ram and with program memory the more you have the typically slower the computer gets so a good example is this computer i have right here

This only has eight cells of ram it’s a four bit computer and it uses every single one of these eight cells however some of these cells are reused throughout the program and it does not ever need more than eight that’s the same as the actual program memory it only has 64

Lines of program memory however most of the program memory is actually used so far i i still need to finish the program in this but so far the program has got up to line 40. as you can see there were this would be double the size so say

All of this would be basically uh this extra long length so it basically be double the size if i had 128 lines but why would i need 128 lines my program is going to be less than 64 lines and any program that i’d write on this

Would always be less than 64 lines so there is no point whatsoever having more than 64 lines of program memory so that’s a that’s a very important point that i’d make sure because the more the more lines you have or the the bigger uh the the more ram you have or the

Bigger number of bits you have the slower the computer will just get that’s that’s kind of like this is only a 4-bit computer because it only needs to do 4-bit operations i would recommend when building your computer if you’re very new to resident computing build a four bit computer don’t bit

Don’t build one under four bits because four bits is about right for a beginner however if you just uh kind of advance and stuff build an eight bit one or a 16 bit one don’t really go over 16 bits or because it gets way too

Big even 16 bits itself is quite a lot of computing power and it will just make things quite slow an 8-bit computer is still just as powerful as a slow 16-bit computer because you can still do 16-bit operations on an 8-bit computer you just have to do

One after the other you just have to implement it in software and make it a bit it’ll make it a bit slower but it might just be as fast as a 16-bit one and you’ll rarely need to use 16 bits anyway most of the time because an 8-bit

Computer can do a lot of things so don’t really over spec is basically the point so for my ninth point reduce your instruction set basically what what i mean by that is the actual instructions that your program memory stores uh basically saying like increment or any conditionals loading into ram or rom

And just or not loading into rom reading from rom basically keep those instructions as little as possible over here on this computer i only have 16 instructions for it and it is a four bit computer that is capable to do conditional branching as well as loading from a rom and has eight

Uh eight cells of ram so for only 16 lines that is quite a lot of stuff it can do and so basically what i’m saying make sure your instruction set is as small as possible because if it’s too big then of course you’ll increase the the amount of

Program memory you need to actually store the instructions a good way to reduce your instruction set is to actually use micro code or multiplex your instructions so like for an alu say my aou can do add subtract and say and you would not need to run three separate

Lines you could just run two separate lines and encode it in binary then when it gets to the ao you decode it so the program memory would only contain two separate lines and you could have like a a very simple truth table so say you have say zero zero equals nothing

And then you could have zero one which equals add one zero which equals subtract and one one which could equal and or whatever operation you wanted to do so that’s only two lines you’d actually need instead of three lines if you’re doing each of the things so that’s what

You’d do to reduce your instruction set then you can also use microcode i don’t really have a good uh example of microcode in any of my cpus but uh it’s it’s very useful and if you if you look up on youtube you can find some very good videos explaining

How to do it effectively and properly and it’s basically just encoding your instructions and then having like a pre-defined uh like code that you can send in and it’ll do a list of instructions and it just reduces your instruction set by a lot so i’m going to actually show you a good

Example of why you want to reduce your instructions and i’ll show you right about now so here’s a very good example of why you’d want to reduce instructions so this is a 8-bit cpu that i was working on a while ago now i kind of just gave up on it because it’s quite

Time-consuming uh because it has so many different things it can do but basically look every single one of these lines over here each of these is an instruction as you can see there are a lot of instructions and that means oh well just look how big the

Program memory is this is only 64 lines of program memory and i there’s almost uncountable instructions there’s like one two three i’m not going to count them all out but you get the picture compared to only 16 instructions on that four bit one which is one two three four five six

Seven eight nine 10 11 12 13 14 15 16. so this is how many instructions the 4-bit one would be and this all of this is my 8-bit one as you can see it’s very unoptimized and if i used micro code or just a better more efficient ways of actually doing

Instructions and stuff this would be a lot smaller and a lot less complicated because an 8-bit computer using a much more efficient instruction set should only only really take up say 24 instructions because if you do three 8-bit micro code things then it’d only take up 24 instructions or around roughly that

So yeah make sure to try and reduce your instruction set to save size and also time so for my 10th point uh program effectively basically so say i wanted to do a very simple program on this computer just uh say i wanted to do like a hardware multiplier i could either do repeated

Addition and use a program like that or i could use shifting the number left or right and then adding the number again if it’s a a one or a if it’s odd or even basically and depending on which type of code i choose one might be shorter than the other

And one actually might take less time on the actual cpu than the other basically choose the fastest one if if your code has a lot of unnecessary things so say like this line said read uh read a ram cell and this line said display well why can’t you just put that display

Line on that display command on that original line where you said read the ram cell if you do that that’ll make your they’ll reduce the amount of instructions you had just by that one so instead of two simple use cycles to read the ram then display what’s from that ram it will

Only take one cpu cycle if you have it all in the same cycle which will make things a lot faster because you’re reducing the amount of cycles you’re actually wasting because it is wasting when you just have unnecessary things in your code so i’d say this is one of the most

Important points to speed up your computer because if you don’t program effectively your computer will be just as slow as your actual program you’ve put in because even a really fast com computer won’t be fast if you have a really terrible program running on it

And for my 11th point this is just a bonus point that i’m doing clickbait thumbnail said 10 points but 11 points whoa multi-thread your code so this is kind of a weird term and most redstone engineers would not do this and this is actually a really terrible

Example of a computer i’m just going to go over so on and i’ll talk on the way but basically if you multi-thread your code which means that running your if you run your code on more than one computer at once it can be almost double as fast or maybe

Double as fast if you’ve programmed it effectively so what i mean by that say i had eight numbers that i wanted to add to each other so i want to add all eight numbers to each other and i have two computers that i can do

If i only do it on one computer it will have to add a number to itself or whatever eight times if i do it on two computers at the same time or two calls or threads well it’ll take half that time or at least a bit over half that time

Because it can this one can add four numbers this one can add four numbers and then this one can send it to this one and this can add the two uh four numbers though i mean the two numbers from each core so this this course number and this

School’s number together to produce the final result this would be almost double as fast of course you have that extra piece of code adding these two together and sending from this one to this one however it would still be faster than just having this one do all eight numbers in the first place

So if you multi-thread your code running on and and that’s just an example for two i could even use four calls or even eight for example eight computers or eight cores and that would just make your code a lot faster because you’re running it simultaneously at the same

Time on multiple uh processing cores or computers so you can you can either use multiple alus or just multiple computers i i’ve been calling it cores i guess because yeah the definition of core is kind of a bit vague but this also isn’t uh to to be confused with like hyper threading or

Simultaneous multi-threading as you see from intel and amd because that is only still one physical core uh or physical alu so one physical cpu well alu to be precise however uh operations uh it it pretends it’s two because in idle time that the cpus are not using anything or it’s waiting for something

Else it uses the aou to do some other operations so that’s where you have two threads per course is threads is kind of a very bad sound for anyway but yeah so hyper threading and simultaneous multi-threading is not a term you’d use in wrestling computing too much because your computer would never be

In an ideal world your computer would only take one one to ten cycles to actually compute something you’d actually never need those uh cycles in between to run other things you you’d really want the computer to be a hundred percent active all the time so if you so basically conclusion is

Split your program over multiple uh computers or cores if you can because that will just make it faster yes it will mean you’ll have to copy your computer out twice and you’ll also have to program it so that it can work on two at once however it can make it almost as

Double as fast a great example would be the program that i’m working on at the moment which is a sort of i was going to say convolutional ai but the actual intelligence part where the the filters are actually predefined so there’s no back propagation which in summary means it’s not an

Actual prop ai i don’t think it’s more of like a shape recognizer if i did that on one core it would take like 10 minutes to compute however if i split it over multiple computers or cause it would take well i’m gonna do it over eight so it’ll take eight times less

As fast so it’ll be a lot faster basically so if you split your program it’ll just make it a bit faster so yeah thanks for watching please like and subscribe i hope um your computer is a bit faster now after watching these points but yeah just try do the things i

Said and if you’re really stuck ask someone else to try and help you with your computer yeah please like and subscribe and i’m out