Yeah, most people forget that instant carry adders using pistons are 2 tick slower on the falling edge (11-XX), so most only obtain a throughput of 6 ticks. Though, I would love to see a CLE that can achieve a 2 tick throughput; it would require a 1 tick half adder (XNOR/NAND), instant inverters, and another 1 tick XNOR. Even if that is possible, the implementation must be absolutely huge...
Actually i've done this a long time ago. Here's my video. Also I should mention that any throughput limitation of anything I make is due to the limitations of the materials it is made with. E.G. torches can only react to 3 tick pulses without burning out, and so my adders using torches have a 3 tick throughput, while this adder I show in the video has a two tick throughput because it uses comparators in place of torches, which have a 2 tick throughput.
How did you manage to get the CLE adder down to 4 ticks? Did you use some kind of instant NOT or a 1 tick XNOR? I would love to see the individual gate designs in the adder, I can't get on the server to take a look because my computer will overheat.
I use 1 tick NOR to prevent propagation and the rest are standard 2 tick XNOR gates. Hopefully your computer is fixed soon so I can show you how I got this to work on ORE ;3
So with SAC would it be better to use a CLE/CLA adder or my
Current instant carry.
I personally like CLE because it can be made to be synchronized on all edges. Instant carry with pistons can have issues especially due to block drops, so I would recommend a pistonless method like CLE or CLA (really the same thing).
Bump! This needs way more views And replies as it is just awsome
Also guy can a instacarry adder be faster than a CLE.0"0
Dammit just realised how old the last post is 8hrs
D: fail
CLE caries instantly, and yes it is always faster than a device that uses pistons (in terms of throughput). The lowest latency achievable in a CLE is 3 ticks, which is faster than all non fully instant adders. Also, a fully instant adder cannot have a 2 tick throughput like CLE. So I would say that an instant carry adder (utilizing pistons in the second half adder to create instant AND) will always be slower than CLE if all edges (referring to inputs) are taken into account.
Very clever optimization you have there.
I'm currently working on a 4 bit analog multiplier of comparable size (can be expanded to 8 bit with 4 multipliers in parallel and some adders) and I'm using your 24 tick multiplier as a benchmark. So far, I've gotten the delay down to 19 ticks, and I might be able to implement pipelining. Just letting you know why I asked about the delay
Ah yeah, nice :D. The time of 24 ticks is only really fast for sequential multipliers. Combinational multipliers can be fully instant (but I'd avoid it! They tend to have glitches due to update order stuffs.)
0
Make sure you're using the correct IP, we changed domains to openredstone.org a while ago
0
I recommend ORE
mc.openredstone.org
0
You're wrong10 ticks to 1 second. It's a definition; there is no argument
0
0
:3 /me agreeeeeeees
0
Lol thanks man!
Actually i've done this a long time ago. Here's my video. Also I should mention that any throughput limitation of anything I make is due to the limitations of the materials it is made with. E.G. torches can only react to 3 tick pulses without burning out, and so my adders using torches have a 3 tick throughput, while this adder I show in the video has a two tick throughput because it uses comparators in place of torches, which have a 2 tick throughput.
0
I use 1 tick NOR to prevent propagation and the rest are standard 2 tick XNOR gates. Hopefully your computer is fixed soon so I can show you how I got this to work on ORE ;3
0
I personally like CLE because it can be made to be synchronized on all edges. Instant carry with pistons can have issues especially due to block drops, so I would recommend a pistonless method like CLE or CLA (really the same thing).
0
0
CLE caries instantly, and yes it is always faster than a device that uses pistons (in terms of throughput). The lowest latency achievable in a CLE is 3 ticks, which is faster than all non fully instant adders. Also, a fully instant adder cannot have a 2 tick throughput like CLE. So I would say that an instant carry adder (utilizing pistons in the second half adder to create instant AND) will always be slower than CLE if all edges (referring to inputs) are taken into account.
0
0
2
0
Indeed! I love the creativity here
0
Ah yeah, nice :D. The time of 24 ticks is only really fast for sequential multipliers. Combinational multipliers can be fully instant (but I'd avoid it! They tend to have glitches due to update order stuffs.)