• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Lightelligence's Optical Processor Outperforms GPUs by 100 Times in Some of The Hardest Math Problems

AleksandarK

News Editor
Staff member
Joined
Aug 19, 2017
Messages
3,118 (1.09/day)
Optical computing has been the research topic of many startups and tech companies like Intel and IBM, searching for the practical approach to bring a new way of computing. However, the most innovative solutions often come out of startups and today is no exception. According to the report from EETimes, optical computing startup Lightelligence has developed a processor that outperforms regular GPUs by 100 times in calculating some of the most challenging mathematical problems. As the report indicates, the Photonic Arithmetic Computing Engine (PACE) from Lightelligence manages to outperform regular GPUs, like NVIDIA's GeForce RTX 3080, by almost 100 times in the NP-complete class of problems.

More precisely, the PACE accelerator was tackling the Ising model, an example of a thermodynamic system used for understanding phase transitions, and it achieved some impressive results. Compared to the RTX 3080, it reached 100 times greater speed-up. All of that was performed using 12,000 optical devices integrated onto a circuit and running at 1 GHz frequency. Compared to the purpose-built Toshiba's simulated bifurcation machine based on FPGAs, the PACE still outperforms this system designed to tackle the Ising mathematical computation by 25 times. The PACE chip uses standard silicon photonics integration of Mach-Zehnder Interferometer (MZI) for computing and MEMS to change the waveguide shape in the MZI.



It is worth pointing out that this approach demonstrates that optical computation, more specifically Lightelligence's direction, is helpful for more sets of problems compared to "just" Artificial Intelligence. These computationally expensive classes of mathematical problems, like Ising, are often found in material science, thermodynamics, bioinformatics, cryptography, circuit design, power grid optimization, and much more.

View at TechPowerUp Main Site
 
I hope it does mining, so that people can leave the GPUs alone.
 
We are still comparing a tailored ASIC with general purpose silicon that is jack of all trades. Tailored ASIC's are usually miles faster at that single math discipline it is designed for, like hardware video encoders for example.

100 times faster? It ain't that groundbreaking.

As working tech demonstrator it is... but the comparison ain't fair here, it ain't apples to apples.
 
I remember seeing cures for cancer and AIDS over the last 10-15 years too
 
Please make ASICs with this tech so the crypto crowd can FO.

It won't solve the shortage. ASICs any kind of them would be bite off the share of TSMC anyways and leaving us the short stick anyways. We need a cheap foundry alternatives like SMIC to ramp up sub 14nm nodes along with UMC to ignite competition. They are way too late to the party and hurts us all.
 
I guess this startup needs the hype to get the funds from sheeps and the "100x better than 3080" is perfect for the hype.
 
I hope it does mining, so that people can leave the GPUs alone.

mining is all about money, i imagine this to be a lot more expensive (new tech, not mass produced), some math had to be done on price vs performance
 
Slap that puppy on a gpu and sell it, asap if not sooner. I wonder how many fps it gets at 8k.
 
We are still comparing a tailored ASIC with general purpose silicon that is jack of all trades. Tailored ASIC's are usually miles faster at that single math discipline it is designed for, like hardware video encoders for example.

100 times faster? It ain't that groundbreaking.

As working tech demonstrator it is... but the comparison ain't fair here, it ain't apples to apples.


It clearly states this Optical processor beats even the purpose built asics by 25%.
 
And the questionin every gamers mind…..

will it mine way faster than gpus and be cheaper than gpus and when will it be available in mass quantities

lol
 
And the questionin every gamers mind…..

will it mine way faster than gpus and be cheaper than gpus and when will it be available in mass quantities

lol
but can it play Crysis?
 
It clearly states this Optical processor beats even the purpose built asics by 25%.

No doubt a optical switch setup is faster at switching photons, but with no ability to store data, reliance on existing tech for converting I'm fairly certain their claims are for "processing speed" of a clearly defined linear non branching data stream that has been spoon fed to their device to get more investment.

I'm sure in the right server setup it will be amazing, but its still light years away form general purpose computing and the masses.
 
"Outperform GPUs by 100 times" is a blanket statement. You can build many conventional processors that are 100 times faster than a GPU at a lot of different things, the question is, is it faster at general purpose computation ?
 
To @W1zzard When can we expect review ?
To All But can it do Halo Infinite 4k@60FPS ?
 
We are still comparing a tailored ASIC with general purpose silicon that is jack of all trades. Tailored ASIC's are usually miles faster at that single math discipline it is designed for, like hardware video encoders for example.

100 times faster? It ain't that groundbreaking.

As working tech demonstrator it is... but the comparison ain't fair here, it ain't apples to apples.
People are aware it was mentioned in the article that it still was 25 times faster than a traditional ASIC

"Outperform GPUs by 100 times" is a blanket statement. You can build many conventional processors that are 100 times faster than a GPU at a lot of different things, the question is, is it faster at general purpose computation ?
Does it matter if you only need it to do one thing?
 
Last edited:
I hope it does mining, so that people can leave the GPUs alone.
It really doesn't matter because the systems are designed to increase the difficulty every time the processing power or the amount of mined crypto exceeds certain limits
 
It really doesn't matter because the systems are designed to increase the difficulty every time the processing power or the amount of mined crypto exceeds certain limits
Designed to waste more energy and consume more to create nothing of value.
 
"Outperform GPUs by 100 times" is a blanket statement. You can build many conventional processors that are 100 times faster than a GPU at a lot of different things, the question is, is it faster at general purpose computation ?
There are thousands fields of science and engineering alone that would massively benefit from hardware that could solve certain classes of problems a 100x faster. They specifically don't use general purpose hardware for everything because they are too slow. The article specifically mentions "NP-complete class of problems". There are exist many such problems that have no efficient algorithms at all to solve them. These problems run in superpolynomial time which can be exponentially or factorially slow.
 
I want one... ^^ :D :toast:

I remember seeing cures for cancer and AIDS over the last 10-15 years too
Cancer cant survive in an alkaline environment. Drink enough pure water every day with iodine to keep your body Alkaline and try to eat organic greens, herbs etc...
 
I'm surprised no one has seen the elephant in the room: some of the hardest math problems are NP complete (or worse). Throwing 100x more HP at them does little more than squat.
 
I want one... ^^ :D :toast:


Cancer cant survive in an alkaline environment. Drink enough pure water every day with iodine to keep your body Alkaline and try to eat organic greens, herbs etc...
The calcium in your bones are the biggest PH balance in your body, alkaline water is snake oil.
 
I hope it does mining, so that people can leave the GPUs alone.
I wish mining would stop creeping into threads it does not belong in...

I remember seeing cures for cancer and AIDS over the last 10-15 years too
No, you don't. There has been progress but no hard cures.

That will likely be a perfect use-case-scenario. Now if they can only mass produce them..
How so? This doesn't do any kind of meaningful data hash that I can see.

Slap that puppy on a gpu and sell it, asap if not sooner. I wonder how many fps it gets at 8k.
0, son. Wrong kind of math.

It clearly states this Optical processor beats even the purpose built asics by 25%.
that is not unheard of for FPGAs (which are hardly purpose built) vs ASICs.

People are aware it was mentioned in the article that it still was 25 times faster than a traditional ASIC
*FPGA

Designed to waste more energy and consume more to create nothing of value.
Please, the topic...
 
Back
Top