5 ms·

I think everyone in this thread underestimates this part of ARM's business (from the F-1): * Expand our System IP and SoC Offerings. To enable further improve

by RC_ITR 3y ago

I think everyone in this thread underestimates this part of ARM's business (from the F-1): * Expand our System IP and SoC Offerings. To enable further improvements in performance and efficiency, we continue to develop a broader set of configurable systems IP offerings, including proven on-chip interconnect, security IP, memory controllers, and other design IP to be used with our processors, including the integration of multiple IP technologies into a subsystem and additional information to assist in fabrication. More recently, we have invested in a holistic, solution-focused approach to design, expanding beyond individual design IP elements to providing a more complete system. By delivering SoC solutions optimized for specific use cases, we can ensure that the entire system works together seamlessly to provide maximum performance and efficiency. At the same time, by designing an increasingly greater portion of the overall chip design, we are further reducing incremental development investment and risk borne by our customers while also enabling us to capture more value per device.* ARM doesn't just give you the architecture, they give you a reference design that already works and then lets you customize it however you want. Companies like Apple don't care about that, but MediaTek, NXP, etc. use it to speed the design process meaningfully.

monocasa 3y ago
ARM already does that though. Their dev systems like the Juno boards are ARM designed chips with ARM IP through and through covering just about every accelerator and integration niche. CPUs, NoC fabrics, GPIO, sound, just about every hardware interface master and slave, AI inference, display scanout, etc. Integrators don't typically go all in on ARM IP though for a couple major reasons. 1) They want to provide some value add beyond the standard offerings or else why pick them over a competitor? 3) They have massive collections of IP blocks already that they don't have pay ARM for. Maybe it takes some engineering to convert to the next chip, but that's probably cheaper than what ARM is asking for.
- RC_ITR 3y ago
  Sure, I'm not saying that ARM is in a good position per se, but that RISC-V has way more catching up to do than people realize.
  - bhouston 3y ago
    I am giving RISC-V 5 to 10 years to catch up in my estimates. I think that is fair. The momentum behind RISC-V is massive and even then I am not saying it is fast.
    - RC_ITR 3y ago
      5 years is a flash in chip design. Initial design to tape-out is 3 years at the leading edge if you're really good. Maybe in 5-10 years, RISC-V gains share in initial designs, but the semiconductor design process is loooong. And if you're going to talk about Microcontrollers and simpler chips, keep in mind we had a bunch of other RISC architectures a decade ago that all lost to Cortex-M for simplicity/cost reasons (RIP MIPS).
      - bhouston 3y ago
        I think RISC-V eats ARM's lunch at the low end across the board over the next 5 years. Financially that is okay in the near term for ARM as the major profits are all in the high end designs. But all that will be left in a few years is the high end areas and where there is a lot of lock in to the ISA. It isn't yet clear when RISC-V will be ready to compete at the highest end of things, but that is where Jim Keller comes in. RISC-V is a classic market disruptor for ARM, just as ARM was a disruptor to Intel.
  - monocasa 3y ago
    It's less than you would think. Despite having offerings in about every IP block segment about the only ARM IP blocks that get heavy use are the pieces that have some other ecosystem thing going on. PL011 uarts are de facto required for some initial consoles. Mali GPUs were being sold in what was frankly a move that needed some anti-trust scrutiny (ARM saw Apple taking the piss out of IMG and decided to go in for the kill by offering CPUs+GPU for cheaper than CPUs on their own, overnight destroying IMG's market). Beyond that it's easier to not pay ARM for their IP. NoC fabrics and IP blocks that touch pads are better coming relatively straight from the foundry. Simple serial controllers are essentially commodities at this point. Etc. So RISC-V really only needs to focus on CPU cores to be an existential threat to ARM. At this point their biggest most is probably the patents on AMBA specs, but if they tried hard to enforce those the industry would switch to something like TileLink in a relative heartbeat.

by RC_ITR 3y ago

monocasa 3y ago

ARM already does that though. Their dev systems like the Juno boards are ARM designed chips with ARM IP through and through covering just about every accelerator and integration niche. CPUs, NoC fabrics, GPIO, sound, just about every hardware interface master and slave, AI inference, display scanout, etc. Integrators don't typically go all in on ARM IP though for a couple major reasons. 1) They want to provide some value add beyond the standard offerings or else why pick them over a competitor? 3) They have massive collections of IP blocks already that they don't have pay ARM for. Maybe it takes some engineering to convert to the next chip, but that's probably cheaper than what ARM is asking for.

RC_ITR 3y ago
Sure, I'm not saying that ARM is in a good position per se, but that RISC-V has way more catching up to do than people realize.
- bhouston 3y ago
  I am giving RISC-V 5 to 10 years to catch up in my estimates. I think that is fair. The momentum behind RISC-V is massive and even then I am not saying it is fast.
  - RC_ITR 3y ago
    5 years is a flash in chip design. Initial design to tape-out is 3 years at the leading edge if you're really good. Maybe in 5-10 years, RISC-V gains share in initial designs, but the semiconductor design process is loooong. And if you're going to talk about Microcontrollers and simpler chips, keep in mind we had a bunch of other RISC architectures a decade ago that all lost to Cortex-M for simplicity/cost reasons (RIP MIPS).
    - bhouston 3y ago
      I think RISC-V eats ARM's lunch at the low end across the board over the next 5 years. Financially that is okay in the near term for ARM as the major profits are all in the high end designs. But all that will be left in a few years is the high end areas and where there is a lot of lock in to the ISA. It isn't yet clear when RISC-V will be ready to compete at the highest end of things, but that is where Jim Keller comes in. RISC-V is a classic market disruptor for ARM, just as ARM was a disruptor to Intel.
- monocasa 3y ago
  It's less than you would think. Despite having offerings in about every IP block segment about the only ARM IP blocks that get heavy use are the pieces that have some other ecosystem thing going on. PL011 uarts are de facto required for some initial consoles. Mali GPUs were being sold in what was frankly a move that needed some anti-trust scrutiny (ARM saw Apple taking the piss out of IMG and decided to go in for the kill by offering CPUs+GPU for cheaper than CPUs on their own, overnight destroying IMG's market). Beyond that it's easier to not pay ARM for their IP. NoC fabrics and IP blocks that touch pads are better coming relatively straight from the foundry. Simple serial controllers are essentially commodities at this point. Etc. So RISC-V really only needs to focus on CPU cores to be an existential threat to ARM. At this point their biggest most is probably the patents on AMBA specs, but if they tried hard to enforce those the industry would switch to something like TileLink in a relative heartbeat.