IRC channel logs

<paulz>Good Morning

<paulz>Here in Italy is: 18 Nov 2024,

<paulz>What do you think about: "Exokernel" ?

<paulz> https://en.m.wikipedia.org/wiki/Exokernel

<paulz>Could there be a faster and more efficient architecture than the ARM64 in the future that could be used on servers, so that it consumes little electricity and at the same time carries out a good amount of data?

<paulz>Paolo Del Bene: id3rfix@gmail.com

<psydroid>RISC-V?

<azert>Hello paulz

<azert>I don’t think programs should talk directly to hardware

<paulz>Hello at all

<azert>It kind of defies the main purpose the Unix ended up having, that is to virtualize the hardware to user programs

<paulz>Anyway thank you, for your help

<azert>On the other hand, servers implementing the Hurd should definitely talk directly to hardware in order to implement device drivers

<azert>Just my opinion paulz

<azert>Im sure others will disagree

<paulz>Infact i said many thanks!!! :-)

<azert>Felt like you were leaving

<paulz>No, I only said many thanks for your support

<azert>What support?

<azert>What do you think about exokernels?

<paulz>Your reply

<azert>I can think of an high performance application talking directly to the network card, that could be helpful for certain applications

<azert>But then it should be a dedicated network card, not virtualized

<azert>Think about an mpi application using the full infiniband interface of inter server message passing

<azert>That doesn’t need a kernel, in fact for that use case a kernel is baggage

<paulz>I am interested in an advanced evolution for a new generation more fast of Arm64, which could consume little electricity, that could be used on servers for DataCenters,

<azert>Why do you think that the ISA matters for performances?

<azert>Experience shows that the implementation matters much more than the isa

<azert>Think of how the dec alpha outclasses everyone for a few years on the ‘90s. The USA was nothing special, the implementation was egregious and it ended up copied by Intel

<azert>The ISA nothing special*

<azert>An egregious implementation of RISC-V might be able to smash Arm64 to the ground

<azert>Nothing special about arm

<azert>Thing is that arm is still playing it nice with the industry

<azert>So that’s not gonna happen anytime soon

<paulz>I always was interested to them.

<azert>If one wants to work on serious optimization required, then look at the various neural engines

<azert>Particularly check what nvidia has to offer

<azert>It is beyond obvious that the industry is pozzled

<azert>And twenty years of simplifications, optimizations and standardization will follow up

<paulz>Actually are saying that Arm64 are better than the X86_64

<azert>No its bullshit

<azert>Arm64 === x86_64 don’t fall for Apple corporation shills

<azert>They are the same ‘90s tech

<azert>RISC-V also

<paulz>I am not interested to Apple Computer Copertino.inc its own computers and mobile phones are built by Slaves in Foxconn in Shenzen

<azert>Those slaves probably have a lifestyle higher then yours, but for the rest yes leave apple alone

<azert>I advise you a visit to china

<paulz>Asap

<paulz>You can see the documentary "Inside Apple"

<paulz> https://t.me/Movies_and_Documentaries/96

<gfleury>youpi: I have sent a patchset that move pthread_attr_*. I will move next __pthread_setcancelstate

<youpi>gfleury: thanks!

<azert>solid_black: hi, I have a question for you. I remember you once mentioned you implemented a 9p network filesystem client from Hurd, and you optimized beyond what can be done through libnetfs. Would you be able to resume what lesson did you learn?

<azert>There is one thing that bugs me: tmpfs is implemented with libdiskfs

<azert>why network file systems wouldn’t also be implemented that way? (vs. libnetfs)

IRC channel logs

2024-11-18.log