Slashdot Mirror
Intel's Clear Linux Distribution Offers Fast Out-Of-The-Box Performance (phoronix.com)
An anonymous reader writes: In a 10-way Linux distribution battle including OpenSUSE, Debian, Ubuntu, Fedora, and others, one of the fastest out-of-the-box performers was a surprising contender: Intel's Clear Linux Project that's still in its infancy. Clear Linux ships in an optimized form for delivering best performance on x86 hardware with enabling many compiler optimizations by default, highly-tuned software bundles, function multi-versioning for the most performant code functions based upon CPU, AutoFDO for automated feedback-direct optimizations and other performance-driven features. Clear Linux is a rolling-release-inspired distribution that issues new versions a few times a day and is up to version 5700.
Intel's own compiler in many cases generated code that runs 15%-20% faster than code compiled under GCC.
"Transparent" is a shit show that trades on every stereotype going. A man in drag is NOT a transsexual.
"In support of the goal to provide an agile Linux* distribution that rapidly detects and responds to quality issues in the field, Clear Linux for Intel® Architecture includes a telemetry solution, which notes events of interest and reports them back to the development team."
https://clearlinux.org/feature...
Clear Linux ships in an optimized form for delivering best performance on INTEL x86 hardware with enabling many INTEL compiler optimizations by default
For major corporate environments, this means nothing as youve likely been married to Intel hardware for quite some time. For startups, most of these features are pointless:
Autoproxy: is compensating for departmental overhead and the bloat of a monolithic organizational structure that prevents network operations or system administrators from coherently deploying a server outside the corporate proxy.
Function Multiversioning (FMV): is intels solution for the -march=native compiler flag. Want your code to scream on haswell and crawl on bulldozer? Intel sure does, and what better way to ensure that then fucking with the compiler again.
Telemetry: something something agile...something something quality...we rewrote the linux Backtrace so in 15 years after we lose interest in clear linux, your code is still hobbled to us.
clear containers: take something an open source team worked almost a decade on and slap you brand on it. viola.
stateless: half a dozen devops tools already take care of imaging, reimaging, config management, and the rest. but lets have an Intel proprietary solution too. after all, Openstack is no fun unless youre vendor dependent all over again.
debug: did we mention we fucked with the compiler again? it was awesome when we did it to kill AMD64, and now its awesome again when we're trying to kill openstacks vendor-independent features.
Good people go to bed earlier.
Systemd? Hopefully they chuck it.
There are SO many things that you can bash Microsoft for, but this is not one of them.
They spend a lot of time testing, regression testing, and backward compatibility testing. They almost-only release updates and patches once per month. When you consider the vastness and diversity of their ecosystem and the number of issues that they constantly deal with and the very few instances of breakage that occur, you cannot reasonably deny that they do a fantastic job of developing and testing and deploying.
Be serious. I've had so many regressions and breakages by numerous mainstream Linux distribution that I could puke. And I'll bet you've never even written a Hello World without a single bug, let alone an operating system and applications.
Clear Linux is a rolling-release-inspired distribution that issues new versions a few times a day and is up to version 5700.
Big deal. Firefox will catch up with that shortly.
It must have been something you assimilated. . . .
"Clear Linux Project for Intel Architecture"
In reading a few pages, I don't think they used the word 'Intel' enough....
XML is like violence. If it doesn't solve the problem, use more.
They also claim the code to be 100% thetan-free.
Coming from a major U.S corporation who is on NSA's and the U.S gov's leash, this is an important question to ask, and something that people should look into before adopting this.
What would you expect. Intel is using a custom kernel optimized for Intel processors and chipsets. The other distros ship generic kernels to work with various processors and chipsets. If you prepare custom kernels for the specific hardware at hand, any of those distros listed in the summary will perform wickedly fast.
Linux for Scientologists. Thetan free.
Upon seeing the description of function multi-versioning I thought of three distinct ways to use that for malware in as many minutes, and the ideas are still coming. (And I don't write malware, so someone in the field would probably think of more, faster,)
It's also a great way to make competitors' processors look bad: Detect their processors and fall back on the minimalist defaults or even hand them "grinched" code that does worse, or contains odd kickers. Or just don't support THEIR accelerations. Also: Don't support their implementations of YOUR accelerations.
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
Its nice to see this however we really should, in general, have a better way for Linux programs to be able to easily take advantage of the CPU extensions available without recompile. There are dozens of permutations of CPU extensions, so distributing a binary for each permutation is not feasible. Full from source compilation takes too long for many users. Having Linux binaries being able to use the CPUs most advanced features has been a problem. One solution that I favor is to take a page from AS/400, in a variation of that, in each library file, put a copy of the machine code, but also a copy of the abstract syntax tree, the last compilation phase. If the binary is moved to a new CPU, the AST is run through the code generator to regenerate the machine code in the file according to the options the CPU supports. All done in situ. This is much better than storing a copy a binary for each CPU permutation in a library file. It makes things easy to use and is faster than compiling from source as the lexer and parser phase does not need to be repeated.
The rankings in individual benchmarks were all over the place; a composite of those benchmarks is only valid for some theoretical "average" workload that's the average of all the workloads each individual benchmark is supposed to represent; almost nobody is bound to have a workload that resembles that "average".
In fact the whole "shooutout" scenario is silly because Clear Linux is a container-centric distro. It makes no sense at all to compare it to general purpose distros like Ubuntu and plain vanilla Centos then leave out Red Hat/Centos's Atomic Host flavors.
In any case if performance is your paramount concern, then "out-of-the-box" performance is bound to be irrelevant to you because you'll be compiling from source with your own choice of compiler and flags, as well as fiddling with all those bells and whistles exposed in the /sys interface. What's interesting would be an exploration of why various distros did better or worse on individual benchmarks.
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.
That image is not accessible to blind people. Have you reported this inaccessible image to your local branch of Intel and to disability advocates? I wonder what they'll do if reminded of National Federation of the Blind v. Target Corp.
Just because something is (fairly) new, we cannot safely conclude that it is better than traditional methods. Unfortunately, fashion is far more powerful than judicious evaluation.
"The computer industry is the only industry that is more fashion-driven than women's fashion. Maybe I'm an idiot, but I have no idea what anyone is talking about. What is it? It's complete gibberish. It's insane. When is this idiocy going to stop?"
- Larry Ellison at Oracle OpenWorld, September 2008.
I am sure that there are many other solipsists out there.
As soon as I see the words "rolling release", I swicth to another piece of software.
Let me guess: you didn't understand Katamari Damacy.
To me, an extra 0.1% performance increase, even if I am only imagining it to be faster, is certainly worth one day a week recompiling all of the latest packages from source code.
If 1 part in 1000 runtime improvement is worth 1/7 of one machine's time, then you must be running a cluster of at least 142 identical servers.
One solution that I favor is to take a page from AS/400, in a variation of that, in each library file, put a copy of the machine code, but also a copy of the abstract syntax tree, the last compilation phase.
You're thinking of distributing LLVM bitcode. Google was thinking of the same thing when designing PNaCl.
No matter what distro I use for my desktop, I always use the latest pf-kernel, with bfq scheduler, low latency, cpu optimizations, etc. I can overload the desktop, and music/video is smooth as silk, and compiling is faster. Its a real world performance boost.
I'd love to see how a pf-kernel does vs stock on each distro.
Generally agreed. One might say this is the EXTREME version of the release early and release often development model.
It has it's ups and downs and a similar development model is spreading everywhere as devops take over and developers are being given the keys to the castle. Unfortunately, in the past young an enthusiastic developers have been restrained both by more experienced programmers and experienced admins. When a young but talented programmer ignores a more experienced programmer the admins could be counted on to refuse to deploy anything that hadn't survived a test of time and good practices. The amazing capabilities offered by devops simplifies many aspects of administration to the point where developers can do it which gives people the insane idea they SHOULD do it.
I can see what results come from these models, innovative ideas realized that might not have been otherwise, but unstable and unmanageable implementations. The wheel keeps turning. When stability and consistency becomes the prized and rare commodity again those stable and consiste implementations will use better development practices but use them to implement the things that did turn out to be good ideas in the times of chaos ahead. Unfortunately, at my age that means chaos until I'm close to retirement.
At least the wheel has turned with hardware performance progression slowing and optimization and performance coming back. Next up, every layer in the stack trace that dumps for any error in your app represents at least 4 extra instructions every time that code is used * calls per day * nodes it is used on. Not to mention all that black boxing meaning you don't know how to efficiently utilize the black box and most definitely aren't. Your thousand node cluster could probably be a 200 node cluster and your ridiculously abstracted design may well avoid re-inventing the wheel but also means nobody has actually known your codebase, for real, for at least 5 years since the guy who actually wrote it and kept all those details in the back of his head. Everything carries a tradeoff, sooner or later whatever piece "is insignificant" relative to benefits will start to add up.
Clear Linux also comes packaged with spyware
From the 1st paragraph at that link:
That's what Canonical (Ubuntu) and a few others think, and it's wrong. Clever wording like "We send only minimal stuff without your knowledge" and "it's for your own good and so that we can make it better" don't change the default state of the software in question. I refuse to load Ubuntu on anything because Canonical installed their software in an always on state and hid it from consumers. I will never ever trust them again, just like I have not trusted Microsoft after their shenanigans (yeah, you have to go back pretty far for me trusting MS).
Very simply put, if it's always on and users don't have big flashing lights warning them at first boot that it's going to be on, then it's intentionally hiding the software. You can claim good intent all you want, but if you tried to hide from the start do you think I'll believe your intent is altruistic? "Fool me once", and all those quotes.
-The wise argue that there are few absolutes, the fool argues that there are no probabilities.
Well, it seems that several moderators must have marked this as "troll". I can't work out whether that's from resentment of the (perhaps) implied superiority of "before a lot of you were born", or from natural resentment that Mark permitted himself to criticize M$.
I'm standing back to back with Mark, so please moderate this reply "troll" to your heart's content. Or stop and think for a few moments about what he actually said.
I am sure that there are many other solipsists out there.
The only person using false drama is you. I never stated that it was an evil plot, you did that. You also attempted to claim I said it, which is bullshit.
Telemetry requires a consumer dig through details to find it, and to turn it off. How hard is it to do like Redhat does, and give a prompt to users during the install which asks them if they want the service on or off? Don't bother stopping to think about why Redhat does this as opposed to just turning it on, because that may be more "false drama".
You claim to agree with my position, but then type what a shill would type. It is obvious why _you_ remain hidden in anonymity and the person who is fighting for anonymity does not hide.
-The wise argue that there are few absolutes, the fool argues that there are no probabilities.
...or you just go for that warm fuzzy feeling of having the fastest distribution on the planet...
Until you get that nagging feeling that maybe switching to that other package, compiling it with Clang and set it to... (dang!)