temac

38 ms·

421.
by temac 6y ago
CISC and RISC are mostly an ISA concept. Most chips are "microcoded", at least here and there. The term does not define a whole micro-architecture, esp in modern complex ones, but I'm not really sure how you would implement o
422.
by temac 6y ago
Yeah CISC vs RISC thing barely makes sens anymore. It could only matter in the context of hand-programmed and low frequency processors. What remains of "it" (at least in some people's mind) that is still relevant today on maj
423.
by temac 6y ago
It was the PPro and OOO processors were being developed by lots of companies and teams at that time. Wikipedia indicates it was featured in the PowerPC 601 (93), SPARC64 (95), PPro (95), MIPS R10000 (96), etc. Hard to see where was the adva
424.
by temac 6y ago
I'm not saying that the exact situation you describe can never happen, but I'm not 100% convinced it is all that frequent. But if all your descriptions are precisely correct, then my opinion is that your conclusions broadly are to
425.
by temac 6y ago
> The substantive points of practical difference are far less important IMO the substantive points of practical difference are vastly important: the obligations of GPLv3 vs Apache 2.0 are not the same... The whole document is a side-disc
426.
by temac 6y ago
3 years latter good commit messages are extremely useful, and most of the time very short ones and/or giant commits are mostly useless. I occasionally read back multiple pages commit message I wrote and wished I dumped even more info
427.
by temac 6y ago
I kind of agree with you. I was thinking of people perfectly able to do/finish the job. If that's not their job to begin with, they have another one, and can be evaluated on it. If it's just that their time is too precious to
428.
by temac 6y ago
Yet you can do a few simple thought experiments. What would those fast-dev do in a vacuum? Or with only clone of themselves coworkers. What happens when one of those fast-dev quit? -- If everything is/would be fine, then yes , they ar
429.
by temac 6y ago
OTOH I'm tired of people who don't "have the time or knowledge to complete [their ideas]". Ideas are cheap. Show me the code. Asking, explicitly or implicitly, for others to "implement/finish" their ideas
430.
by temac 6y ago
Part of this reflection is weird. Office is already running on Mac OS. M1 does not change anything in that regard. So the only meaning that remains that would make sense is: how would HP, Asus, Dell, and others react if MS sold non-OEM lice
431.
by temac 6y ago
Plus, in some cases it is perfectly fine to integrate technology by soldering a discrete package on a PCB. It is not superior to a SoC in all cases, but neither is the inverse true. In that sense Intel, AMD, or Nvidia are absolutely conveyi
432.
by temac 6y ago
Switch to B-Trees if you ever need to reject all candidates :P (And astonishingly, a comprehensive description of them including the tools for complexity analysis in a form practical for real-world implementers - not just CS student writing
433.
by temac 6y ago
> She's even tweeted that she suspects there was a team of people behind the decision to accept her resignation. Would that be somehow bad? I'm not even entirely sure of what does that even mean. If I submit a resignation lette
434.
by temac 6y ago
MS x86 emulation will likely be far slower if they don't have access to the memory model trick.
435.
by temac 6y ago
Functions are about code reuse, abstracting variables and easy handling of their lifetime, and most importantly invariants and specifying reasonably sized pieces of code (even if you "specify" only in your own head). And then mayb
436.
by temac 6y ago
You are right although in my mind it is frozen, and given Linux continues to move, I see WSL1 has almost deprecated. For a few years it may continue to be useful, however soon enough you won't be able to run a modern distro on WSL1.
437.
by temac 6y ago
Well you don't have any of the AVX512 nonsense, but probably the OOO of the M1 uses more transistors than on an Intel chip. And Google tells me that a quad core i7-7700K has 2.16 B.
438.
by temac 6y ago
Not really the best idea for the kind of speed baselines and differences discussed here. You can use better ideas like putting GPU memory first in the fast part then the rest in the slow area. You know, like XBox Series does.
439.
by temac 6y ago
I don't really get what you are talking about. Everybody has been doing MMIO for a while now (and by for a while I mean multiple decades), and IO is usually not an issue in personal computers anyway; OOO, compute and the memory hierarc
440.
by temac 6y ago
> An insane amount of good engineering went into this CPU. I agree but lets not overblow the difficulties either. > For example, the delay in the decode stage itself scales quadratically with the width of the decoder. That could be ir
441.
by temac 6y ago
There are tons of programs that do not go through "the" libc under Linux, everybody and their dogs seem to reimplement their own locks recently for example. Plus the libc actually makes a way greater surface to cover given it give
442.
by temac 6y ago
The previous impl used the tech that became WSL1, and I'm not sure you can in practice bypass the Linux syscall. WSL1 is now deprecated and I don't think it is actually reasonable to try to follow the fast paced dev of Linux, ev
443.
by temac 6y ago
The concept of "explicitely" not supporting graphics did not make much sense with the X architecture and X servers under Windows being available. Of course MS dis not ship one, and of course X is considered deprecated by its own m
444.
by temac 6y ago
So I see the historical NT subsystem story restated all the time and that makes little sense. WSL1 used little (if any?) of the historical NT subsystem concepts and WSL2 even less. SFU and its successor were, but that was a completely diffe
445.
by temac 6y ago
WSL2 is a VM. If you want excellent Windows compat, run Windows in a VM. If you merely need application not using very Windows specific feature, or if you need game compat, run something like Wine. Is there a need for yet another reimpleme
446.
by temac 6y ago
I would think that maybe they can save a very low proportion of chips with ECC not-working, but the large majority would be completely arbitrary market segmentation: handling of ECC is going to be limited to a small relative surface area of
447.
by temac 6y ago
I was not really thinking about Apple when writing that part, more about some weak details of Zen N vs. Intel, that do not matter in the end (at least for most workloads). Be it inter-cores or intra-core. I think the logical design space is
448.
by temac 6y ago
If I remember correctly the M1 has around 600 reorder buffer entries, and I just checked and Anantech estimate the int register file at around 354 entries. That's still big, but not 600.
449.
by temac 6y ago
> Apple's decision to make a 8-wide decoder engine is a decision, a compromise, which will make scaling up to more-cores more difficult. Apple's core is simply the biggest core on the market. > Whereas AMD decided that 4-wid
450.
by temac 6y ago
I think he is a little overrated, and the only reason he is that much known is because it is actually rare to see former Intel engineers opening a Youtube channel with some Intel content. His take about how to fix Spectre made completely no

More ›