ompi VS OpenSSL

The commercial Linux world (see https://github.com/open-mpi/ompi/issues/4349) and other open source OSes (eg FreeBSD) seem to have lined up behind little-endian PowerPC. IBM still has a big-endian problem with AIX, IBM i, and Linux on Z.

Roughly, the sets of computational problems that people used (use?) MPI for. Things like numerical solvers for sparse matrices that are so big that you need to split them across your entire cluster. These still require a lot of node-to-node communication, and on top of it, the pattern is dependent on each problem (so easy solutions like map-reduce are effectively out). See eg https://www.open-mpi.org/, and https://courses.csail.mit.edu/18.337/2005/book/Lecture_08-Do... for the prototypical use case.

One of the guys at the local LUG is one of the lead maintainers of Open MPI. He told us about a comment that ran into the hundreds of lines, all for a one-line change in the code.

But it would be very inconvenient to have to keep crediting everyone who's ever worked on it. If you look at old projects, their licenses can have like 10-20 of those lines (here's one I was recently looking into).

I have a bit of experience programming for a highly-parallel supercomputer, specifically in my case an IBM BlueGene/Q. In that case, the answer is a lot of message passing (we used Open MPI [0]). Since the nodes are discrete and don't have any shared memory, you end up with something kinda reminiscent of the actor model as popularized by Erlang and co -- but in C for number-crunching performance.

That said, each of the nodes is itself composed of multiple cores with shared memory. So in cases where you really want to grind out performance, you actually end up using message passing to divvy up chunks of work, and then use classic pthreads to parallelize things further, with lower latency.

Debugging is a bit of a nightmare, though, since some bugs inevitably only come up once you have a large number of nodes running the algorithm in parallel. But you'll probably be in a mainframe-style time-sharing setup, so you may have to wait hours or more to rerun things.

This applies less to some of the newer supercomputers, which are more or less clusters of GPUs instead of clusters of CPUs. I imagine there's some commonality, but I haven't worked with any of them so I can't really say.

[0] https://www.open-mpi.org/

I would suggest using OpenMPI because it's pretty easy to get started with. You can build OpenMPI with CUDA support, then you can pass device pointers directly to MPI_Send and MPI_Recv. Then you don't have to deal with transfers and synchronization issues.

backends from native torch distributed configuration: nccl, gloo, mpi.

I use a linux ubuntu machine with MPI (https://www.open-mpi.org/). I had a question on making my computer simulations faster. Would be better to get an older AMD 9590 machine clocked at 4.7 ghz or continue using my Ryzen 7 1700 machine clocked at something like 3.5ghz?

OpenMPI - Message passing interface implementation. BSD-3-Clause

ENV OPENSSL_PREFIX=/opt/openssl ENV SSL_CERT_FILE=/etc/ssl/certs/ca-certificates.crt WORKDIR /tmp RUN git clone --branch OpenSSL_1_0_2n https://github.com/openssl/openssl.git RUN cd openssl RUN ./config shared --prefix=$OPENSSL_PREFIX --openssldir=$OPENSSL_PREFIX/ssl RUN make RUN make install RUN rvm install 2.6.0 -C --with-openssl-dir=$OPENSSL_PREFIX ENV PATH /usr/local/rvm/bin:$PATH RUN rvm --default use ruby-2.6.0 ENV PATH /usr/local/rvm/bin:/usr/local/rvm/rubies/ruby-2.6.0/bin:$PATH ENV GEM_HOME /usr/local/rvm/rubies/ruby-2.6.0/lib/ruby/gems/2.6.0

Today, April 7th, 2024, marks the 10-year anniversary since CVE-2014-0160 was published. This security vulnerability known as "Heartbleed" was a flaw in the OpenSSL cryptography software, the most popular option to implement Transport Layer Security (TLS). In more layman's terms, if you type https:// in your browser address bar, chances are high that you are interacting with OpenSSL.

At this point I pretty much understand the entire process on how the xz backdoor came to be: its execution stages, extraction from binary "test" files etc. But one thing puzzles me: how can the ifunc mechanism be used to replace something like RSA_public_decrypt? Granted this probably stems from my lack of understanding of ifunc, but I was under the impression that in order for the ifunc mechanism to work in your code, you have to explicitly mark specific function with multiple implementations with __attribute__ ((ifunc ("the_resolver_function"))). Looking at the source code of the RSA function in question, ifunc attribute isn't present:

https://github.com/openssl/openssl/blob/master/crypto/rsa/rsa_crpt.c#L51

So how does the backdoor actually replace the call? Does this means that the ifunc mechanism can be used to override pretty much anything on the system?

OpenSSL and Go crypt/tls has no support yet, so none of the webservers that depend on them support it. Apache, Nginx, and Caddy, they all need upstream ECH support first.

- https://github.com/openssl/openssl/issues/7482

- https://github.com/openssl/openssl/pull/22938

- https://github.com/golang/go/issues/63369

If I'm understanding the draft correctly, I think the webserver you're hosting your sites on would need it implemented as it requires private keys and ECH configuration. In the example of nginx since it uses openssl, openssl would need to implement it. I found an issue on their Github but it's still open: https://github.com/openssl/openssl/issues/7482

I confirmed that the systm was on 1.1.1f with openssl version command. Hmm...... I check the openssl version in the repo with apt list... LOL package names wernt helpful. finally went to the repo pages and found that its still on 1.1.1f, https://launchpad.net/ubuntu/+source/openssl. Meenwhile I looked up the version history on https://www.openssl.org/ and saw that 1.1.1v was released at the beginning of this month... ok. I can understand it it was out less then 30 days. I looked up when f came out, end of MARCH 2020. NEARLY 3-1/2 YEARS