shash

9 ms·

91.
by shash 2y ago
> That's only true for a subset of programming languages, and it requires you to already know how to program in at least another language of the same family. Knowing Java will not help you with Haskell, but it will help you with C#.
92.
by shash 2y ago
I was literally thinking of this article!
93.
by shash 2y ago
That's because it's actually quite sane; relying on commodity backed currencies - especially those which are _finite_ leads to deflation. You see that with BTC, where the value keeps rising and you need more and more fractional de
94.
by shash 2y ago
There’s this individual’s name which involves a clock sound: Nǃxau ǂToma[1] [1] https://en.m.wikipedia.org/wiki/N%25C7%2583xau_%C7%82Toma
95.
by shash 2y ago
The problem _there_ is that there's very little diversity in the training data - it's all NVidia designs which are probably from the same phylogenetic tree. It'll probably end up regurgitating existing NV designs...
96.
by shash 2y ago
I've never tried SystemC. But after having tried to learn Chisel and friends, and successfully learning Bluespec (and using it in professional projects), I have some insights. It's fundamentally important when doing hardware desig
97.
by shash 2y ago
In this specific case, it's hard to see how LLMs can get you from here to there. the problem isn't the boilerplate code like when you build a react website with them, but the really novel architectures (and architectural decisions
98.
by shash 2y ago
In India, coal just dipped below 50% for the first time since the 1960s. https://m.economictimes.com/industry/energy/power/coals-shar... EDIT: but apparently it went back up in April. Installed coal capacity
99.
Reevaluating Google's Reinforcement Learning for IC Macro Placement(cacm.acm.org)
2 points by shash 2y ago|0 comments
100.
by shash 2y ago
For anything programmable, _maybe_. But power management? Analog frontends? Very difficult...
101.
by shash 2y ago
Actually, https://github.com/google/android-riscv64?tab=readme-ov-file And the mailing list is pretty active too: https://lists.riscv.org/g/sig-android/topics?sidebar=true
102.
by shash 2y ago
There were two competing SIMD specs, and personally I'm glad that RVV won out over PSIMD. It's an easier programmer's view and fewer instructions to implement.
103.
by shash 2y ago
Almost all in-order processors will do pipelining, so that's there. Many are even multi-issue. Andes has an out of order core [1] and so does SiFive (though I don't know of many actual chips using these. [1] https://www
104.
by shash 2y ago
Interesting to see RISC-V described as an "incumbent" :D
105.
by shash 2y ago
> Many details of an ISA extend beyond the instruction decoder. For instance, the RISC-V ISA mandates specific behavior for its integer division instruction, which has to return a specific value on division by zero, unlike most other ISA
106.
by shash 2y ago
The P670 is a core, not a chip, so you can't really get to power numbers (or indeed, raw performance as opposed to performance / watt or performance / GHz) without actually making a practical chip out of it in some given proc
107.
by shash 2y ago
Well, Tenstorrent, Andes and others have their respective designs... On the in-order side, I can see on-par performance with the ARM A5x series quite easily.
108.
by shash 2y ago
In a sense, Freescale/NXP did it from their old PowerPC to ARM.
109.
by shash 2y ago
FFMPEG has a RISC-V port. We're yet to try it, but I did successfully compile it to target RISC-V vector extensions.
110.
by shash 2y ago
> Virtually all high performance processors these days operate on their own internal “instructions”. The instruction decoder at the very front of the pipeline that actually sees ARM or RISC-V or whatever is a relatively small piece of lo
111.
by shash 2y ago
As a user of Bluespec and a follower of his CA papers, salutes and RIP.
112.
by shash 2y ago
These guys are friends... We share investors, I've visited them at their test site, and they're neighbours at IIT Madras Research Park (where their rocket factory is). Their engines are a pleasure to hold and see in real life... C
113.
by shash 2y ago
Yeah, that's just not true - it _eases_ post-fabrication qualification, but Agnikul does indeed qualify their nozzles (among other means, by static firing them in a test rig).
114.
by shash 2y ago
More like 10,000,000,000 years at least. The closest star is 4.26 light years away. At current speeds (~65000 km/h) it’ll take 40,767,123 years to reach. _IF_ it’s going in the right direction. It’s doubtful they’ll even find the sun i
115.
by shash 2y ago
IIRC, Zen at one point had an ARM decode stage. I don't know if it was intended commercially, or they just did it to try out...
116.
by shash 2y ago
An interesting statistic from the Shakti[1] project - after building a 5-stage RISC-V core with an FPU, the same team had a reason to write an x86 decoder. The _decode_ stage of the x86 alone was bigger than the entire pipeline of the RISC-
117.
by shash 3y ago
Both of you are correct - there is a duopoly because of the complexity, and the duopoly has no real incentive to change, given that the current local optimum seems to be working for the foreseeable future. There is no pressure to innovate,
118.
by shash 3y ago
So, "founder" in the way that Elon "founded" Tesla?
119.
by shash 3y ago
Somehow, I don't see him going to Nvidia in any case. Their designs are very different from his philosophy, and he can be quite opinionated. I'd characterise Tenstorrent as his attempt to build the opposite of an Nvidia style acce
120.
by shash 3y ago
"Works for" is one way to put it - he's CEO and (I think?) co-founder..

More ›