Huawei's Ternary Logic Breakthrough: A Game-Changer or Just Hype?:
[Editor's Comment: The source reads as though it could have been created by AI, nevertheless it is an interesting topic and worth a discussion.--JR]
Huawei's recent patent for 'ternary logic' represents a potential breakthrough in chip technology by utilizing a three-state logic system consisting of -1, 0, and 1, rather than the traditional binary system of 0 and 1. This innovative approach could substantially reduce the number of transistors required on a chip, leading to lower energy consumption, particularly in power-hungry AI applications. The ternary logic patent, filed in September 2023 and recently disclosed, may offer significant advantages in processing efficiency and hardware design, addressing some of the physical limits faced by current chip technologies.
Ternary computing is not a novel concept; the first ternary computer was developed in 1958 at Moscow State University, indicating the feasibility of utilizing more than two states in computational logic (source). However, binary logic became the industry standard due to its simplicity and the development of compatible technologies. Huawei's pursuit of ternary logic comes amidst US sanctions, which have pressured the company to explore alternative technological paths. By reducing reliance on traditional chip designs, Huawei aims to innovate in a constrained environment and potentially gain a competitive edge in the AI and semiconductor sectors.
The commercial viability of Huawei's ternary logic chip presents an intriguing yet complex scenario for the tech industry. Ternary logic, which utilizes three states instead of the usual binary system's two, promises significant advancements in chip technology. By potentially reducing the number of transistors required, it could lead to decreases in both manufacturing costs and energy consumption, particularly in power-hungry AI applications. However, the road to commercial viability is laden with challenges. The tech industry must grapple with the transition from a binary-dominated ecosystem to one that could incorporate ternary systems. This includes revamping software and programming approaches to leverage the benefits of the new computing structure. Furthermore, the ability to mass-produce these chips economically and reliably remains unproven, leaving questions about whether the technology can achieve a cost-effective scale .
If successful, Huawei's ternary logic patent could disrupt current computing paradigms and lead to reduced energy consumption in AI technology, aligning with broader trends towards sustainability. The potential of such technology to alter chip design and improve AI efficiency could have far-reaching implications not only for Huawei but for the tech industry at large. Moreover, by possibly circumventing some effects of international sanctions, Huawei's efforts symbolize a form of technological resilience and ingenuity amid geopolitical challenges
Ternary logic represents a novel approach in computing, differentiating itself from the conventional binary logic by utilizing three distinct values: -1, 0, and 1. This trinary system aims to encode information more efficiently, offering potential reductions in the complexity and power usage of AI chips. Unlike the binary system that utilizes two states (0 and 1) to process tasks, ternary logic could lead to less energy-intensive data processing, ultimately decreasing the overall power consumption of AI-driven technologies. This innovation heralds a shift from traditional methodologies, potentially streamlining both hardware requirements and computational resource demands.
The implementation of ternary logic in modern computing could signify a breakthrough in addressing the issues associated with current chip designs. With chip manufacturing approaching its physical limits, the trinary system provides an alternative pathway to enhance processing capabilities without exponentially increasing transistor counts. Huawei's recent patent reflects this innovative direction, aiming to solve power consumption dilemmas while navigating international sanctions and fostering technological advancements in AI. Embedded in this development is Huawei's strategic response to geopolitical challenges, exemplified by their proactive patent applications that emphasize reducing dependence on traditional binary constraints.
(Score: 2) by PiMuNu on Thursday October 02 2025, @02:58PM (15 children)
> ternary logic could lead to less energy-intensive data processing
How?
(Score: 3, Informative) by epitaxial on Thursday October 02 2025, @03:13PM (5 children)
Should have kept reading
"the trinary system provides an alternative pathway to enhance processing capabilities without exponentially increasing transistor counts."
(Score: 3, Funny) by PiMuNu on Thursday October 02 2025, @03:25PM (4 children)
Thanks for the help. Much clearer now.
/sarcasm
(Score: 2) by epitaxial on Thursday October 02 2025, @04:39PM (3 children)
Fewer transistors means less power used and less heat dissipated.
(Score: 1, Interesting) by Anonymous Coward on Thursday October 02 2025, @07:40PM
Not when the remaining transistors have to do more work with extra bits
(Score: 2) by PiMuNu on Friday October 03 2025, @08:59AM
Thanks again. I assumed you were joking, I now realise you seriously think that the statement you excerpted means something.
> alternative pathway to enhance processing capabilities
Enhanced processing capabilities how? binary add and multiply are elegant pieces of math, very simple. Does ternary really simplify this?
> without exponentially increasing transistor counts
The claim is that ternary somehow means that we can solve problems whose solution goes as O(exp(n)) in linear time? I would like to see the explanation for this. Or indeed how ternary simplifies any calculation whatsoever.
==
Or is it just media bullshit written by ChatGPT with no meaning or sense?
(Score: 3, Funny) by driverless on Saturday October 04 2025, @07:25PM
DEC had ternary logic decades ago, their equipment always had three states: one, off, or busted. That third state used almost no power so was very efficient.
(Score: 4, Informative) by VLM on Thursday October 02 2025, @03:19PM (2 children)
VERY theoretically 2 ** 64 = some 19 digit decimal number and 3 ** 40 is some 19 digit decimal number so in theory with much spherical cows you could build 40 trit adders and 40 trit buses and similar instead of 64 bit adders and 64 bit buses. So chips would be about 62% the size, maybe very handy wavy
(Score: 2) by Undefined on Thursday October 02 2025, @03:49PM
That's what happens when you calculate using only two fingers but are afraid to make a peace sign.
I use a dedicated preprocessor to elaborate abbreviations.
Hover to reveal elaborations.
(Score: 3, Interesting) by sfm on Thursday October 02 2025, @05:12PM
Trit adders are going to be significantly more complicated than
current bit adders. More transistors, less noise immunity and
slower (which MAY be offset by fewer required operations).
This idea is very old, think 1950's. If if it improves computation
efficiency in edge cases, I could see it being implemented, but not
at all likely to replace binary adders.... EVER
(Score: 5, Insightful) by hopdevil on Thursday October 02 2025, @03:20PM (4 children)
I read as much of the original article as I could muster, it is completely devoid of any actual information. Just the same AI goop over and over. Why do AI slop articles like this exist?
(Score: 3, Interesting) by VLM on Thursday October 02 2025, @03:30PM (2 children)
The irony is if the internet really is 95% bots as per Dead Internet Theory, an AI article that repels humans with ad blockers will make the company more money because the bots watch and click ALL the ads.
(Score: 2) by janrinok on Thursday October 02 2025, @06:49PM (1 child)
Whatever they are advertising will not sell anything though - bots don't tend to buy much, do they?
So the reduced revenue means that advertisers will begin to look elsewhere to advertise. I wonder if there is a sweet spot where we get our internet back, but it isn't worth the advertisers spending much to sell their product?
OK, I can dream, can't I?
(Score: 4, Insightful) by hopdevil on Friday October 03 2025, @08:33AM
This is very much AI generated (quantity over quality). Everything on the opentools site reads like this. Reminds me of the scientific paper fraud issue; flood the world with garbage but why? What are they (presumably opentools) selling?
It could be:
• selling ads. It's possible, I don't know anything about the ad economy. I looked at the site, adverts crammed into every nook
* a promotion piece of Huawei. propaganda of some kind. There are other disinformation-y articles there. Other similar articles recently struck me with the same vibe https://soylentnews.org/article.pl?sid=25/09/25/1231232 [soylentnews.org]
* general misinformation, maybe related https://soylentnews.org/article.pl?sid=25/09/21/2243223 [soylentnews.org]
I don't blame janrinok for posting it. SN editors and staff have enough on their plate and I applaud their efforts.
But I'm very unsettled by the fact that this is the most commented on article in the past few days. A discussion about the merits of the concept of this technology even, though a lot of the comments are confused, almost as if the red herring is working. I feel obligated to comment on this mostly out of concern that AI crap is flooding our attention space. This is the dark future we are heading towards, but who is pushing us there and why?
(Score: 0) by Anonymous Coward on Thursday October 02 2025, @07:35PM
Profit! What else is there?
(Score: 4, Interesting) by bzipitidoo on Thursday October 02 2025, @08:07PM
3 has the lowest radix economy of all the integers. The most efficient radix is actually e.
I like to think of it this way. What denominations should coins have, so that any amount can be reached with the fewest possible coins? Answer: ought to have a 3 (cent) coin.
(Score: 4, Interesting) by BsAtHome on Thursday October 02 2025, @03:50PM (2 children)
In ternary designs you have two dead zones (-1 ↔ 0 and 0 ↔ 1) which may prove hard to cope with and generally require additional transistors.
You need complex and completly new glue logic to interface to the mostly binary reality.
Additionally, you create an intermediate 0 value when going -1 ↔ 1 and these transitions will cost twice the amount of energy (two-value hops) and require extra time to settle.
So, I don't really see the savings here. The effective scale saving from moving from 2^N to 3^M, where M < N, will probably not be enough to compensate for the, on average, larger than one energy unit per change (because of the two-value hop described above) and added glue. The calculation speed is also limited by the two-value hop.
(Score: 2) by HiThere on Thursday October 02 2025, @06:40PM (1 child)
More to the point, the original Fortran IV has if statements with three way branches. Usually only two of them were used. So the three way if was eventually dropped from computer language designs. They were occasionally useful, but not often enough.
Javascript is what you use to allow unknown third parties to run software you have no idea about on your computer.
(Score: 2) by bzipitidoo on Thursday October 02 2025, @07:58PM
But 3-way made a comeback, with the "spaceship" operator, <=>. It was added to C++ in C++20.
(Score: 3, Interesting) by VLM on Thursday October 02 2025, @04:02PM
I'm a little confused about the transistor count.
For example you can built a binary two input NAND gate using 4 mosfets. This is "old stuff".
I found a FinFET (basically dual gate mosfet, really modern tiny process stuff) design for a trinary NAND gate using a mere 10 FinFETs. This was the VLSI journal article from like 2022 "Design of Unbalanced Ternary Logic Gates and Arithmetic Circuits". So 10/4*100 = ternary gates use 150% more transistors than binary.
So a trinary system would have busses and functional unit counts about 62% the qty of binary if 2 ** 64 is about 3 ** 40
So at least as of 2022 if you replace a binary design with a trinary you have roughly 40% fewer functional units but each unit uses 150% more transistors which is a net loss. Its cool, but its a net loss.
Having not found the primary source of whatever the AI article is referencing (did I overlook it?), I would guess when it is found, they found a a sneaky way to make trinary gates that are less than 40% more transistors than binary gates making it a net win.
They can't seriously be suggesting using binary to trinary and trinary to binary bus xcvrs on each bus... can they? That sounds like an absolute nightmare.
In the long run we should scrap this new fangled "digital" stuff and implement analog computers. Why send bits down a line when you can send a nice clean 16 bit-resolution analog signal instead?
(Score: 2) by VLM on Thursday October 02 2025, @04:07PM (1 child)
How does noise margin and high speed bus voodoo work with trinary?
If you "aggressively" shove a current down a bus you can shunt to gnd/vcc at the far end and all is well for high speed bus performance overcoming capacitance.
If you implement trinary how does that middle value work?
(Score: 2) by ese002 on Wednesday October 08 2025, @06:00PM
PCIE Gen6 and 100G Ethernet use PAM4. Trinary (PAM3) would be easier. It does reduce noise margin but not as much as increasing the frequency. Still, PCIE Gen6 resorted to forward error correction to get acceptable bit error rate. Of course, using a multi level signalling on a major bus with lot of support logic behind it is quite different from doing it raw for all internal signalling
(Score: 2) by VLM on Thursday October 02 2025, @04:20PM (3 children)
I'm a little confused about encoding as the FinFET designs I've found so far work on 0, Vcc/2, and Vcc for the three values, as I understand it.
Going full dual supply op amp will be exciting if they really are going +Vcc, 0, -Vcc
LOL don't store too many +1 in your memory or the current draw will pull everything downward and some of your 0 might be interpreted as -1
The whole topic of trinary strikes me as it would be a hilarious EE Youtuber video for me to watch because it seems so rube goldberg in design and it would be entertaining to see someone try to make this work.
Imagine using a giant pile of 741 opamps to simulate finfets (or, I suppose, for the really hard core, using genuine dual gate mosfets) to make a trinary full adder or the sheer excitement of trying to make a working trinary latch.
(Score: 1) by pTamok on Thursday October 02 2025, @06:01PM
Perhaps the experience with multi-level NAND helps? Data is stored in cells at different voltages, so there are plenty of dead zones, so maybe (waving hands into a blur) the engineering is transferable?
Conventional wisdom used to be that the advantages of ternary were outweighed by the difficulties and binary was the better option. The article doesn't give any insights into what, if anything, has changed.
(Score: 1) by anubi on Thursday October 02 2025, @10:14PM (1 child)
Are they doing this for terminating transmission line phenomena?
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 3, Interesting) by VLM on Friday October 03 2025, @11:27AM
Basically, 1960s/70s T-1 carrier AMI "modulation" which amounts to flip the polarity of alternating "1". Works pretty well at eliminating DC bias if you're sure you won't get too many "0" in a row. You can also self-clock if you make each pulse of a precise 1/3 duration.
But no, AFAIK that does not seem to be their plan, according to most marketing.
(Score: 2) by Covalent on Thursday October 02 2025, @08:02PM
As others have mentioned, some binary circuits are incredibly simple, fast, and useful. But I imagine there are ternary circuits that are superior to their binary equivalents. I envision purpose-built calculators (maybe for GPUs?) that utilize both for maximum efficiency.
You can't rationally argue somebody out of a position they didn't rationally get into.
(Score: 2) by turgid on Thursday October 02 2025, @09:10PM
We could just go to four states [semiengineering.com] and have 0,1,2,3.
I refuse to engage in a battle of wits with an unarmed opponent [wikipedia.org].
(Score: 2) by jb on Friday October 03 2025, @08:33AM
How on earth can anyone get a patent on ternary logic?
Leaving aside the fact that mathematical discoveries are not patentable in the first place...
...there are simply mountains of prior art evident to anyone "ordinarily skilled in the art".
Ternary logic was discussed extensively in the literature in the late 20th century (doesn't nobody else remember the debate about whether or not to allow nulls in the relational model?)...
...and real ternary (and even higher state count) computers were actually built way back in the 1950s (pretty sure the Russian one TFS cites was not the first, but it's a good example anyway).
Has the patent office really gone completely bonkers? Or is TFA itself a hoax?