We attended the RISC V Summit last week and came away with our view largely unchanged. As we noted a few months back, the project has entered the steady but boring phase of its development. The RISC V ecosystem now has thousands of developers contributing and hundreds of different working groups – churning out extensions, drivers, patches and all the rest. This makes for some very dull headlines, but positions the project very much where it needs to be.
Overall, interest in RISC V remains high. Last week, Nvidia notably announced that it has shipped over 1 billion RISC V cores (which is not really a useful stat) and that every Nvidia GPU has RISC V cores in it (which is more more noteworthy). As has been the case for a while, RISC V remains highly popular in Asia, and some of the largest exhibitors at the show were Alibaba and Andes. All of which is to say that RISC V makes for good progress, but uneventful conferences.
That all being said, we are increasingly wondering in which markets will RISC V will end up gaining traction. We wrote last year, that for the foreseeable future, we will see a lot of RISC V cores sitting next to Arm cores in SoCs. This comes in addition to all the industrial, IoT and embedded systems where it has has its largest footprint. Beyond that we came away with more questions than answers.
One area which we think is lagging notably is mobile. Google said two years ago that they would port Android to RISC V, but we did not see any progress in mobile at this year’s conference. Last year, we pointed out the Android ports still had years of work ahead of them, and Google has, if anything, backtracked since then. Maybe there is more going on behind the scenes, but progress here is frustratingly slow.
The other big question is the data center. There is an argument that RISC V offers critical flexibility which will help hyperscalers design truly custom chips for Cloud and AI workloads, giving the project a long-term advantage in this critical market. The other side contends that RISC V is: 1) not yet powerful enough to handle the truly heavy workloads of the data center; and 2) has a software ecosystem that is not yet fully developed and optimized. It is tough for us to gauge the former, but the latter seems fairly clear.
Put simply, if you are looking for big headlines and major announcements out of RISC V, you will be disappointed. But if you are looking for steady progress and continued interest in the ecosystem that seems to be what RISC V has to offer.
Leave a Reply