gnufx

8 ms·

91.
by gnufx 2y ago
For very small sizes on amd64, you can, and likely should, use libxsmm. MKL's improved performance in that region was due to libxsmm originally showing it up, but this isn't an Intel CPU anyway.
92.
by gnufx 2y ago
That's indeed what's typically done in HPC. However, substituting a parallel BLAS can help the right sort of R code simply, for instance, but HPC codes typically aren't bottlenacked on GEMM.
93.
by gnufx 2y ago
The papers referenced in the BLIS repo are the authoritative reference to understand this stuff. I've no idea why people think optimized BLASes aren't performant; you should expect >90% of CPU peak for sufficiently large matri
94.
by gnufx 2y ago
Do the people who are going on about fail2ban know whether that's even ported to, and included in, OpenBSD? I suspect not.
95.
by gnufx 2y ago
> SSH Certificates are vastly different then the certificates you are referencing. And the SSH maintainers will refuse offers of X.509 support, with a justification.
96.
by gnufx 2y ago
Some would argue that in an organization where you'd consider SSH certificates, it's best to use Kerberos and have general SSO. (Some of the GSSAPI functionality is patched in by most distributions, and isn't in vanilla OpenS
97.
by gnufx 2y ago
You don't just want to measure CPU consumption, but whole-system power is only useful for application performance if only one application of significance runs on it. I'd expect to measure it anyway for system management purposes.
98.
by gnufx 2y ago
Two systems I know from HPC that more usefully expose various architectures' RAPL etc. to userland via a daemon for application profiling are https://variorum.readthedocs.io/ and https://hpc.fau.de/rese
99.
by gnufx 2y ago
I haven't followed Plan 9 for ages, but I'm puzzled why Cox & co wrote "Plan 9", then. However, the point was more about the capability-oriented security in a Unix successor, and how you can use file handles as a so
100.
by gnufx 2y ago
No, but that's the case I've overwhelmingly seen over the decades. Anyway, are you going to redesign ssh not to require a user, for instance? I assume you wouldn't want sshd running as the putative single user. [I'm al
101.
by gnufx 2y ago
Visit the research computing environment sometime, for instance. The libzma SSH compromise was considered very worrying, after all.
102.
by gnufx 2y ago
In fact, if factotum were implemented on Unix along with an analogue to the Plan 9 capability device, venerable programs like su and login would no longer need to be installed ‘‘setuid root.’’ — https://plan9.io/sys/doc
103.
by gnufx 2y ago
That seems like a feature; mounting SMB is done on a local system on the basis of password, and it's horrible. (I assume you could, in principle, use some other GSSAPI mechanism.)
104.
by gnufx 2y ago
No, per https://news.ycombinator.com/item?id=40112697
105.
by gnufx 2y ago
How are we not there? The only real issue I know is allegedly requiring host keys for gssd (e.g. "joining the domain"), but rpc.gssd(8) documents "anyname" principals.
106.
by gnufx 2y ago
I saved a requirement from an old MIT AI lab job on Usenet, perhaps from the early '90s: "Applicants must also have extensive knowledge of C and UNIX, although they should also have sufficiently good programming taste to not consi
107.
by gnufx 2y ago
There was something like this using Scheme, maybe 20 years ago, but I don't remember any details and don't think I ever tried it. Edit: see http://www.abstractnonsense.com/schemix/
108.
by gnufx 2y ago
Not arguing about the merits of capabilities and the open interface, but file handles are often cited as capability-like, and the ...at calls are in POSIX; but there's the global namespace. (Security in Plan 9 does say their capabil
109.
by gnufx 2y ago
> Access control lists are provably insecure. If it's not too late to ask, where can we read about that? I've only ever seen ACL v. capability discussed in terms of trades-off as far as I remember.
110.
by gnufx 2y ago
I don't know how complete it is -- it doesn't list DeVault's Helios -- but various projects are listed at http://www.microkernel.info/
111.
by gnufx 3y ago
"Quantum what?" you might ask, apart from whether it will even work (practically).
112.
Unlocking the potential of quantum: £45M investment(gov.uk)
1 points by gnufx 3y ago|1 comments
113.
by gnufx 3y ago
I don't understand the lack of interest in a better garbage collector, rather than pratting about with the limitations of the current one. The port to the Boehm generational, incremental collector many years ago was just ignored.
114.
by gnufx 3y ago
Rather different (less daft?), but also working via email: https://delta.chat/
115.
by gnufx 3y ago
I don't follow. Assuming you're talking BLAS/LAPACK (libflame is the LAPACK equivalent using BLIS) I don't see why the implementation matters, and if you're dependent on MKL for some reason you can't run on so
116.
by gnufx 3y ago
It's probably worth saying that ~10% differences in BLAS performance are probably lost in the noise of run time variance on an HPC system, especially as things don't typically spend most of their time in BLAS. Also, you don'
117.
by gnufx 3y ago
No, and ATLAS is dead. BLIS is a systematic evolution of the "Goto" approach.
118.
by gnufx 3y ago
That's for very small matrices, but I forget the dimensions. It was only because of libxsmm that MKL got it. (You can use libxsmm to dispatch to OpenBLAS or BLIS for non-tiny cases.) BLIS has "small"-dimension support in m
119.
by gnufx 3y ago
That sort of support has come and gone in different versions, though I couldn't tell you which. But, really, why bother? There is too much mythology around Intel stuff.
120.
by gnufx 3y ago
I don't understand why people use MKL on AMD hardware rather than AMD's (current) library, which is basically upstream BLIS for those architectures; their old one from the Operton era wasn't good. OpenBLAS has been largely c

More ›