Fact Finder
Top 20
I have thought about your statement:Hi FF,
it really is comparing chalk to a laser printer.
The short answer is yes, it would double the frame rate. This is because, statistically half the data bits will be zeros and half will be ones, so by ignoring all the multiply by zeros, you reduce the number of operations to be performed by half. You only ignore the the zeros in one of the numbers, the multiplier.
Caution is needed in comparing fps depending on whether the function is as a camera or as a projector. What Akida does is classify camera images. Nvidia can act as a projector in displaying images or it can be used in an AI manner comparable with Akida.
But, alas Nvidia are still stuck in the 20th century with MACs, CNN, ALUs, 8+bits (although I did see somewhere that a couple of competitors (can't recall who?) had caught on to 4-bit quantization) ... eg:
US2020364508A1 USING DECAY PARAMETERS FOR INFERENCING WITH NEURAL NETWORKS (Priority: 20190514)
View attachment 20623
1 . A processor, comprising:
one or more arithmetic logic units (ALUs) to be configured to identify one or more digital representations of one or more objects based, at least in part, on one or more neural networks trained using one or more decay parameters.
2 . The processor of claim 1, wherein the one or more ALUs are further to be configured to:
apply the one or more decay parameters to prior state information maintained for the one or more neural networks and used for identifying the one or more objects, a weighting of the prior state information being reduced according to the one or more decay parameters.
3 . The processor of claim 2, wherein the one or more ALUs are further to be configured to:
store the state information external to the one or more neural networks and providing the state information to the one or more neural networks for each set of input to the one or more neural networks.
5 . The processor of claim 1, wherein the one or more ALUs are further to be configured to:
determine the one or more decay parameters using a hyper-optimization process and a selected decay function.
[0070] … The training manager 712 can be responsible for training the data, such as by using a LARC-based approach as discussed herein. The network can be any appropriate network, such as a recurrent neural network (RNN) or convolutional neural network (CNN), among other such options.
[0072] … If permitted and available, user data may also be collected and used to further train the models, in order to provide more accurate inferences for future requests. Requests may be received through a user interface to a machine learning application 726 executing on the client device 702 [#### ie, a software app ####] in some embodiments, and the results displayed through the same interface. The client device can include resources such as a processor 728 and memory 730 for generating the request and processing the results or response, as well as at least one data storage element 732 for storing data for the machine learning application 726 .
It isn't really fair to compare legacy equipment with the SOTA.
“It isn't really fair to compare legacy equipment with the SOTA.”
So assuming the State of the Art technology you are referring to is AKIDA technology if I decide not to be unfair then I will probably never need to post again on this subject.
But I reluctantly have to agree with you and I will resist the temptation to point out that compared to these up graded Nvidia Jetson entries in the Edge technology race:
1. AKIDA 1.0 is at least 50 to 150 times more power efficient,
2. AKIDA 1.0 is at least one eighth the cost of the base model Jetson,
3. AKIDA 1.0 is despite the fact that the Jetson has doubled its frames per second to 200 AKIDA 1.0 is already reaching 1670 fps,
4. AKIDA 1.0 is still unchallenged with its ability to process completely on chip without connection while one shot, few shot and incrementally learning, and
5. AKIDA 1.0 is scalable and available as IP.
Really none of these things need to be stated and of course it would be very unfair to do so, so I will say nothing further on the subject of Jetson.

My opinion only DYOR
FF
AKIDA BALLISTA