Has Akida got its head in the clouds?
The original Business wire article from
@Csharmo talks about SiFive and Akida
at the edge. But which edge are they talking about?
SiFive Intelligence™ solutions with their highly configurable multi-core, multi-cluster capable design, integrate software and hardware to accelerate AI/ML applications. The integration of BrainChip’s Akida technology and SiFive’s multi-core capable RISC-V processors will provide a highly efficient solution for integrated edge AI compute.
SiFive Intelligence™-based processors offer industry leading performance and efficiency for AI and ML workloads. The highly configurable multi-core, multi-cluster capable design has been optimized for the broadest range of applications requiring high-throughput, single-thread performance while under the tightest power and area constraints.
“Employing Akida, BrainChip’s specialized, differentiated AI engine, with high-performance RISC-V processors such as the SiFive Intelligence Series is a natural choice for companies looking to seamlessly integrate an optimized processor to dedicated ML accelerators that are a must for the demanding requirements of edge AI computing,” said Chris Jones, vice president, products at SiFive. “BrainChip is a valuable addition to our ecosystem portfolio”.
The original Business wire article from
@Csharmo talks about SiFive and Akida
at the edge. But which edge are they talking about?
https://www.sifive.com/cores/intelligence-x280
The SiFive® Intelligence™ X280 is a multi-core capable RISC-V processor with vector extensions and SiFive Intelligence Extensions and is optimized for AI/ML compute at the edge.
In addition to ML inferencing, it is ideal for applications requiring high-throughput, single-thread performance while under power constraints (e.g., AR, VR, sensor hubs, IVI systems, IP cameras, digital cameras, gaming devices).
There are some pretty heavy processing loads there - so not your fridge nose.
Then we get to the FB post:
The fastest CPU core announced to date from SiFive is the P650 core which has up to a 50% performance improvement over its predecessor, the P550. SiFive achieved that level of performance uplift by widening the out-of-order execution over the P550 and improving clock speeds.
Using the integer SPEC performance benchmark (SPECint) and normalizing for clock speed, SiFive says the P650 will be comparable to the Arm Cortex-A77 announced about three years ago. SiFive also offers customers the ability to build multicore complexes with up to 16 CPU cores.
[### The cortex A77 is used in the Samsung Galaxy S10 Note. So this is much more at the thick end of the edge.]
The company’s performance trajectory suggest that it will be able to meet or exceed the performance of the best Arm cores, including the Arm Neoverse core for data center applications, within the next few years. To get there, SiFive will need to build multiple CPU design teams that can deliver a constant stream of new and improved CPU designs based on the latest semiconductor process nodes and hence it needs to hire mode engineers.
There has been some discussion about Akida in the cloud before. SiFive need to be able to match or better ARM's cloud servers. So combining their CPU with power saving and acceleration from Akida may be the answer.
The question is, does this put Akida on the stairway to heaven?
It definitely didn't do my blood pressure any good
.
spectrum.ieee.org
