AE2 Controller Optimization.

[namiasdf]

A couple facts:

• Each side of the controller can handle up to 32 channels
• Additional channels can be added to network via placing additional controllers adjacent to current controller (with a few structure rules)
• Dense lines from each side must be completely independent of another, or else the network interferes
• i.e. 64 channels, when dense lines interfere will be reduced to effectively 32 channel

Because of this mechanic, the fact that additional controllers must touch, and their additional dense lines may NOT touch, else interference... How do you suppose it is best to optimize your controllers?

Eventually the ones in the middle of your controller multi-block will no longer have available sides to attach a dense line to. Is the 7x7x7 restriction just design oversight, or is there a purposeful reason for this? i.e. Did they just say the max is 7x7x7, or did they have something in mind with 7x7x7 specifically. Is it totally arbitrary?

 

[Maelstraz]

Two words: colored cables. :3

 

[ljfa]

You can color your cables. Different colors connect to uncolored (fluix) cables, but not other colors.

7x7x7 is quite big if you ask me. There has to be a limit to the total number of channels somewhere

 

[Chaka]

You can color your cables. Different colors connect to uncolored (fluix) cables, but not other colors.

7x7x7 is quite big if you ask me. There has to be a limit to the total number of channels somewhere

New goal: make a dw20 9x9 filled with me controllers

Sent from my flux armor using the tappatalk app from the flux store

 

[Someone Else 37]

Also, you can run up to a dense cable's worth of channels through a single P2P tunnel, compressing 32 channels down to only two, one for the tunnel attached to the controller, and one attached to a Dense Cable somewhere else.

The most efficient controller probably only has two controller blocks adjacent to any other two controller blocks (i.e. a long line, or a folded-up line) with a P2P tunnel or dense cable attached to every single exposed face. The P2P tunnels themselves are hooked into a separate network with its own controller, and each is connected to exactly one other P2P tunnel with a dense cable on it elsewhere in the secondary network.

This way, you have one network for all the things you's normally use AE for, with 128 available channels for each controller block, plus 64 more from the ends of the line; and a secondary network dedicated to connecting the parts of the primary network to the controller. Every 32 channels in the primary network that aren't connected directly to the main network via dense cable require a pair of P2P tunnels, and therefore two channels in the secondary network.

However, I prefer to forego the secondary network and have all the P2P tunnels directly tie into the main network. This means that each channel-using device connected via P2P effectively requires a minimum of 1+1/16 channels, which I think is a small price to pay for not having to deal with two different networks. In addition, I try to stay away from dense cable, which means that each device actually uses closer to 1+5/32 channels. Somewhat less efficient? Yes. Does it really matter when you've got over a hundred channels available for each block in your controller? I don't think so.

 

[namiasdf]

Ah, did not know about P2P functionality with dense cables.

And for everybody else, how do you attach dense cables to the center-most controller in a 3x3 controller...

 

[jaquadro]

What kind of controller configuration are you referring to? A solid 3x3x3 is going to short itself out. I use a 3x3x3 "cage" controller in my designs, so the center is hollow and the center of each face of the cage is hollow. You can connect to the interior controller faces by putting a P2P tunnel on each one, and connecting with a standard cable.

 

[namiasdf]

With a 3x3x3, is it only possible to get one of the 6 inward facing faces? Can you show me this setup.

 

[jaquadro]

I'm not sure where 6 inward faces comes from, so I'm not sure we're thinking of the same thing. Here's an example though.

The center of the cube is air, and the inside of each of the 6 'donut' faces is covered in 4 P2P panels, all connected to the same standard cable. With this setup you get 11 panels per side for 352 channels, and you have 21 channels left over to connect the other P2P endpoints.

The theoretical most efficient use only requires 5 faces of this structure to be directly tied to cables for supplying P2P channels, but the wiring is much simpler if you give up 6. With 2112 channels, I don't think you'll miss the extra 48. It's also worth noting that you could remove 3 of the corner controllers and still have the same number of connected panels. I like the symmetry, though.

An interesting thought exercise would be to expand this to a 7x7x7 structure, but just counting the available faces is a chore, let alone all the tricky wiring on the interior structure.

 

[PhoenixSmith]

This is part of the reason I miss AE1... Does anyone else feel like they just made AE2 more annoying but not really harder in a good way. Im all for time sink = more enjoyment and time spent getting good stuff. But the meteorite gathering, the compressing of different resources and channels all seem like tedious nerfs, rather than objective, smart ones. The whole P2P thing though gives me hope.

 

[namiasdf]

Ah, thanks for that. Keep forgetting that everything is multiparts. Is that how you have transferred your channels semi-long distances, throughout your base? i.e. instead of running dense cable everywhere, you branch off and can run 8x32 channels, off of a regular cable, by using 8xP2P tunnels which are attached to sides of a controller?

Dense cable into regular cable, insert P2P, back into dense cable... If that makes sense.

 

[jaquadro]

This is part of the reason I miss AE1... Does anyone else feel like they just made AE2 more annoying but not really harder in a good way. Im all for time sink = more enjoyment and time spent getting good stuff. But the meteorite gathering, the compressing of different resources and channels all seem like tedious nerfs, rather than objective, smart ones. The whole P2P thing though gives me hope.

I'm the opposite. I really enjoy the additional thought that needs to go into AE2. The meteors are a one-off thing, unless you really like the skystone as a building block like I do, then you'll want to hunt for more. The unusual crafting mechanics are all automatable which feels good when you get it working. And the channels are just more interesting to work with, with P2P tunnels making them much more palatable.

Ah, thanks for that. Keep forgetting that everything is multiparts. Is that how you have transferred your channels semi-long distances, throughout your base? i.e. instead of running dense cable everywhere, you branch off and can run 8x32 channels, off of a regular cable, by using 8xP2P tunnels which are attached to sides of a controller?

Dense cable into regular cable, insert P2P, back into dense cable... If that makes sense.

Yes. In fact, I only put 8 P2P tunnels on each of those cube sides (using a dense cable in the interior space to supply their channels). That way I can just run a normal 8-channel cable off of them and tap it in various places throughout my base. Here's my actual controller room.

The top side is reserved for the ME Drive data center in the room directly above, and I run a dense cable because I'm mapping multiple P2P tunnels on the drive stacks to single tunnels on the controller. The bottom side is reserved for quantum network bridges, of which I can embed up to 8 in the floor and dedicate a tunnel to each one. The remaining 4 sides feed the rest of the base, though the back two are currently unused.

 

[namiasdf]

Hmm very insightful. So in this case, you are only using the 8 sides that you can see, per side of the cube? I see that each of the two visible sides have an 8-channel cable pulling out of there.

Is what you're saying, with the drive area, that you are running a dense cable because you are utilizing the interior faces of the controllers (the top face, if I read correctly). Which would give you 8 + (4 x 6) = 32, which is the max for a dense cable. Very efficient.

So one output uses dense cable, with 32^32 channels, the rest use 8-channel cable, for 8x32 channels. You just have to add a cover to prevent short-circuiting.

 

[jaquadro]

I don't use the interior faces on the top side. Up above, I have capacity for 6 banks of 32 drives. Since machines interconnect and mess up all your channels, you need to use P2P tunnels in strategic places to make sure you don't exceed 8 channels through a single block. With that arrangement then, I have 4 P2P tunnels on each bank of 32 drives, which the 4 tunnels delivering 8 channels each and mapping to a single tunnel on the controller providing 32 channels. 24 channels across all the banks (4 x 6) + 8 channels for the tunnels on the controller itself gives you 32 on the dense cable, which is connecting to the interior faces just like the other sides are.

The arrangement you're thinking of would work, except that you'd only have access to 5 x 4 faces, since the dense cable itself must pass through one of the openings and you can't put tunnels on a dense cable segment. You also need to give up a few faces somewhere else on the cube to provide enough channels to the P2P tunnels themselves.

 

[namiasdf]

Yeah, I had just realized that. Hmmm, I'll actually have to make this and play around with optimizing it. I guess 32x8 channels is more than sufficient. You can draw out the channels required for the P2P channels from the middle. You don't have to have dense cables attached to your controllers, to still have 8 channels right? Though, you already have 8 exposed faces per side, which is the side of a non-dense cable bandwidth.

Ugh. Creative is your best friend.

 

[jaquadro]

You can attach a regular cable and you'll get 8 channels. But 8 channels are only enough to handle 4 tunnel pairs. Dense cables are attached to the controllers inside the donut holes. They carry 16 channels (8 for the tunnels on the controller + 8 for corresponding tunnels in your base).

 

[Someone Else 37]

I think that the most efficient controller possible doesn't have any loops- that is, the controller blocks form a long, perhaps branching, chain.

My attempt at a maximally-efficient 3x3 controller has an 'H' shape of controller blocks on the bottom layer, with two 'X' shapes on top of it. I would post a screenshot of it and the mess created by wrapping it in P2P tunnels, but for some reason, the forum here won't let me upload it. It just says "There was a problem uploading your file" a split second after I hit the "Open" button in the file browser dialog, without even telling me what happened, other than that it didn't work. I am running a Macbook Pro with Retina, so maybe the resolution is too high?

In any case, there are two maxed-out dense cables' worth of P2P tunnels, plus three more, leaving 29 channels open on one of the dense cable faces. This means that there are 67*32 + 29 = 2173 channels available for remote P2P tunnels and the things they connect to.

Edit: Seems that the thing I heard a while back about P2P tunnels not being able to carry channels from other P2P tunnels is, in fact, the case now. Grrr... I guess I'll have to take off a couple of tunnels then, and replace them with dense cables. No big loss, though- it should free up a couple more channels.

 

[namiasdf]

I think AX2's intention was to make people use more subsystems. I was going to make separate processes (managed by their own AE system) talk to others via storage buses. Barrels, and a general chest. My main AE system would contain all the drives, but would have access to the inventories of other subsystems which only stored their items in storage bus compatible devices, such as chests, so the main AE system could access them as well.

You would also transfer any liquids/energy from each "system" via P2P. So a tree farm + biofuel energy system managed by one AE system would store everything in barrels/chests. With the main AE system and tree farm AE system both having storage buses attached to them, and a P2P channel to transfer the energy.

Or even an energy accepter interface, to ensure that all power systems are powering your main AE system, etc.

I still have to play with this for a while. I'll try to submit photos of my idea once I get started.

 

[namiasdf]

I think AX2's intention was to make people use more subsystems. I was going to make separate processes (managed by their own AE system) talk to others via storage buses. Barrels, and a general chest. My main AE system would contain all the drives, but would have access to the inventories of other subsystems which only stored their items in storage bus compatible devices, such as chests, so the main AE system could access them as well.

You would also transfer any liquids/energy from each "system" via P2P. So a tree farm + biofuel energy system managed by one AE system would store everything in barrels/chests. With the main AE system and tree farm AE system both having storage buses attached to them, and a P2P channel to transfer the energy.

Or even an energy accepter interface, to ensure that all power systems are powering your main AE system, etc.

I still have to play with this for a while. I'll try to submit photos of my idea once I get started.

 

[xTordX]

This is part of the reason I miss AE1... Does anyone else feel like they just made AE2 more annoying but not really harder in a good way. Im all for time sink = more enjoyment and time spent getting good stuff. But the meteorite gathering, the compressing of different resources and channels all seem like tedious nerfs, rather than objective, smart ones. The whole P2P thing though gives me hope.

Just like thaumcraft... (Flippin' warp!)