141–150 of 162 posts
Earlier quoted context omitted.
Nvidia are supposed to be using RISC-V inside future GPUs.
Yes, as a replacement for their internal proprietary logic controllers which operate in the periphery. That has nothing to do with the RISC-V ISA being the right choice for core functionality like shader cores.
Earlier quoted context omitted.
The biggest issue is lack of huge caches, and the extreme latency to main memory due to the speed of light. GPU’s are fine or even better for a wide range of compute tasks see CUDA etc, they just end up being very slow at a subset of common tasks.
Circuits don't operate at speed of light, though. People like to say this over and over again, but electron mobility through a circuit is only around 2/3s the speed of light, IIRC. They may have very tiny mass, but they still have mass.
Earlier quoted context omitted.
Why would we want RISC-V on laptops?
Just reading up on it, the big thing pushing this is that the instruction set is completely open source. That's why someone might want it. https://hackernoon.com/why-we-need-risc-v-f94e3929891b
High performance microarchitecture is never going to be open source. They cost hundreds of millions to design and the design stays relevant only 4-5 years.
So there are basically 2 possibilities. It is possible for Huawei, and Chinese companies in general, to build a successful alternative without using American technology (for the purposes of this post, I'm defining successful as equivalent to their American tech using competitors). It's not possible. What the actions on Huawei have done, however, is ensure that if that possibility exists, very significant chunks of Ch…
> without using American technology Eh, RISC-V was developed at UC Berkeley. In the United States.
Earlier quoted context omitted.
Actually, what's to stop Huawei from just copying and incorporating whatever they need into their IC designs regardless of patents? I can imagine foreign foundaries refusing orders from Huawei to avoid manufacturing IP infringing designs, but are there foundaries in China that can produce the chips necessary while ignoring foreign patents? As an aside to that, how do foundaries know whether their client designs infri…
Infringement is detected by display of patented features or potentially reverse engineering relevant portions of an IC. There is no practical way to police infringement using the disparate array of design inputs mapped onto proprietary gate libraries.
Earlier quoted context omitted.
>Generational benchmarks show big leaps in every new arch on userbenchmark, That is because GPU scales very well with Transistor count, throw in more transistor and memory bandwidth and you are good. On the CPU you could have doubled the transistor budget and get marginal improvement in IPC.
Why can't a GPU power a computer?
Earlier quoted context omitted.
> GPUs have sort of plateaued except for the raytracing component What? As far as I understood it, CPUs had plateaued outside of efficiencies, GPU power still follows Moore's. Generational benchmarks show big leaps in every new arch on userbenchmark, whereas CPUs from 7 or 8 years ago are only marginally slower (SC) than modern ones.
>Generational benchmarks show big leaps in every new arch on userbenchmark, That is because GPU scales very well with Transistor count, throw in more transistor and memory bandwidth and you are good. On the CPU you could have doubled the transistor budget and get marginal improvement in IPC.
IPC improvements are still happening, which is why Intel has been on 14nm for years and still makes CPU's every year that are a little bit faster, but they are hard to do and you only get incremental improvements.
If someone could figure out a way to write regular software that easily parallelized to hundreds of cores, CPU's could also get the same leaps in performance that GPU's have been enjoying. It's a software problem more than it is a hardware problem.
I think that RISC-V has a long way to go to be actually competitive with ARM on a IPC basis. Although it is great to have more competition in the market I believe that this is more of a 5-10 year goal rather than distinctly near term.
https://www.amazon.com/RISC-V-Reader-Open-Architecture-Atlas...
In the book the authors compare the expressiveness (and thus code density) of code implemented with the RISC-V ISA with instructions for x86, ARM and MIPS. They show that the RISC-V ISA is able to give ARM a good fight in terms of IPC, and that it beats MIPS. (Mainly due to the delay slots which can't be filled in many real world cases.)
They also show why the R5 should make implementations scale better in clock speed compared to ARM. This is the kind of stuff Patterson has been doing research on the last 3-4 decades, and it is quite interesting to see how this experience has guided the design of the RISC-V ISA.
Earlier quoted context omitted.
>Generational benchmarks show big leaps in every new arch on userbenchmark, That is because GPU scales very well with Transistor count, throw in more transistor and memory bandwidth and you are good. On the CPU you could have doubled the transistor budget and get marginal improvement in IPC.
To expand on that, this is my understanding: process shrinks used to lead to faster clocks AND more transistors, but a wall was hit and since then they only lead to more transistors. GPU's can use those extra transistors because their workload is embarrassingly parallel (thousands of parallel things), but CPU's can't because their workload is mostly sequential (developers have to explicitly write code for every threa…
Amdahl's law [1] , and while we have demonstrate parallelised code can speed things up, generally speaking Software performance still have lots of low hanging fruit without even going deep into parallel and concurrency world, it is just a matter of cost whether it is worthwhile to improve or fix it.
Earlier quoted context omitted.
It's a lot of work, and I don't know that there's any support for the effort to port V8 and ART to RISC-V. When I was starting on V8 port, I didn't see much in terms of interest in providing material support for it. I got one interview out of it, but I feel like maybe the Foundation is the right place for that. They'd have to delegate the hiring and management to a member, but I think it could work. Huawei has their…
Absolutely, giants like Google or Huawei could pull this off, I really don't know if we'll see Java without someone like that. Personally I'm keen to see Mozilla's JavaScript VM (is it still called SpikerMonkey?) ported, but I doubt it will happen until we see more widely available/affordable RISC-V hardware.
