Slashdot Mirror

If these real-world benchmarks are to be believed, mission accomplished indeed. C++ will be replaced by Rust soon.

https://www.phoronix.com/scan....

[Rust Servo kicks Gecko C++'s ass.]

There are a large number of middle budget laptops from HP and Dell that comes with rebranded AMD SI and CIK cards which have only GCN 1.0 or 1.1 supported. I have a HP 15-BS576TX which is a June 2017 launch which has AMD 520 discrete card. So, does many models with AMD 520/530 cards which are recognized in amdgpu as HD8xxx cards. The driver performance is most of the times laggard compared to radeon driver. I hope the developers kindly give some attention to this also. The only hope is amdgpu project with these cards. I have pkppa (Padoka stable) repo enabled in my Ubuntu 17.10 system. https://www.phoronix.com/forum...

A few from the front page of goog about the AMD Secure Processor. It does, apparently, run its own OS and have its own flash/memory.
https://hothardware.com/news/amd-confirms-it-will-not-be-opensourcing-epycs-platform-security-processor-code

https://www.phoronix.com/scan.php?page=news_item&px=Linux-4.14-Crypto-AMD-SP

https://www.anandtech.com/show/11551/amds-future-in-servers-new-7000-series-cpus-launched-and-epyc-analysis/3

The biggest question for me when deciding on any hardware purchase is how well the manufacturer supports the development of Free, open-source drivers for their hardware, either through the availability of specifications or actually contributing to driver development. (...) I used to be a big fan of AMD, but it feels like they have not kept up, particularly since they purchased ATI.

Huh? ATI was totally closed source, as bad as nVidia or worse. AMD is releasing tons of open documentation, here for example is a 239 page guide describing the Vega ISA and as for the drivers Phoronix said:

It's phenomenal seeing the open-source driver support one day-one and that for Linux OpenGL games the performance even surpasses AMDGPU-PRO. This Vega launch is easily the most successful discrete GPU launch ever where it's backed by fully-open drivers.

That said, the problem for open source graphics on Linux is that the gaming market share is falling despite /. posting about how many indie games there are. The latest Steam survey says 0.60% and a whooping 96.6% for Windows, it used to be about 1%. It's hard to get AMD to spend more of their very limited cash on a near non-existent market.

It works pretty well, slight slowdown.

https://www.phoronix.com/scan....

Without Linux drivers? Hah...

What? AMD's opensource driver framework is actually better performing than their closed Linux drivers (although the opensource version currently lacks fancier features such as OpenCL 2.0 and Vulkan). You install it the same way we used to do with most regular GPU drivers back in the day: select the appropriate part in vanilla kernel config, and then install the specific X/Mesa packages. Well, there's the firmware bit too, but in any case there's no need to reinstall anything upon kernel updates. I.e. no worrying about the latest kernel breaking the binary driver.

Example: https://www.phoronix.com/scan....

They did a comparison between the highest end Intel chips and the EPYC 7601. Not to spoil it but EPYC blew the panties straight off of Intel's chips while using less power. It's no wonder Intel has been flailing in the media.

With v4.13 just released!
It's a really new OS, then!

And you seem to be fairly out of date..
https://www.phoronix.com/scan....

Sure, i would like a chip without any ME at all, but being able to disable it is atleast a step in the right direction.

Radeon RX Vega On Linux: High-Performance GPUs & Open-Source No Longer An Oxymoron

It has been confirmed to be a processor bug, not a software bug.
BSD kernel developer Matt Dillon sent AMD a reproducible test case back in April.
You can read more about it here.

Since Slashdot doesn't actually cover important technological news anymore, here's a professionally run website that exposed a serious bug in RyZen and got an official response from AMD: https://phoronix.com/scan.php?...

"With the Ryzen segmentation faults on Linux they are found to occur with many, parallel compilation workloads in particular -- certainly not the workloads most Linux users will be firing off on a frequent basis unless intentionally running scripts like ryzen-test/kill-ryzen. As I've previously written, my Ryzen Linux boxes have been working out great except in cases of intentional torture testing with these heavy parallel compilation tasks. Even when carrying out other heavy, non-compilation (GCC or Clang) parallel workloads in recent days, from server tasks to scientific processing, my Ryzen test boxes have been stable. I'm still using Ryzen 5 on my main desktop system without any faults in day-to-day use on Fedora 26 Linux."

"AMD was also able to confirm this issue is not present with AMD Epyc or AMD ThreadRipper processors, but isolated to these early Ryzen processors under Linux."

"Ryzen customers believe to be affected by the problem can contact AMD Customer Care. Some of those who have contacted customer care about the segmentation faults have in turn been affected by thermal, power, or other problems, but AMD says they are committed to working with those encountering this performance marginality issue under Linux. AMD will also be stepping up their Linux testing/QA for future consumer products."

t's not as serious as you make it out to be, it only effects EARLY ryzen chips, not zen or threadripper, and only in certain, specific scenarios. everyone here is talking about threadripper, so you might as well be talking about intel bugs, which there are also many.

Since Slashdot doesn't actually cover important technological news anymore, here's a professionally run website that exposed a serious bug in RyZen and got an official response from AMD: https://phoronix.com/scan.php?...

running that in order for VP9 hardware decode.

No. VA-API on Linux has had accelerated VP9 encoding and decoding for a long time.

The need for hardware accelerated decode is overstated anyway. 1080p VP9 video from YouTube works fine in Firefox on an 11 year-old dual core desktop. My iPhone 7 plays VP9 video just fine in VLC for iOS.

For anyone who's curious about this, here is a brief overview.

It started on Linux as Lunatik.

Nope Intel offers millions in bribes....err i man "advertising partnerships" to get benchmark companies and triple A games they think are gonna be big to use ICC.

If you want to see how big a difference using ICC makes over a CPU agnostic compiler like GCC makes? Look at these Vishera benchmarks and then go look at the other sits using windows benches...funny how many of those using windows benches have such horrible scores (one even had the i3 beating the 8350, yeah and my mama is a car) while Linux magically has the FX-8s performing exactly as you would expect them to right between the i5 and i7 depending on how heavily threaded the load is.

Or if you want something really in depth? Just type in "Intel cripple compiler" into your favorite search engine and you will find dozens of pages of researchers testing the Intel compiler and seeing how badly it rigs the code, hell one even took a Via CPU (the only CPU you can softmod the CPUID) and just by changing it to "Genuine Intel" suddenly that Via chip magically gains 35% in benches!

The hilarious part? Intel tried to claim they were merely "optimizing for their own designs because they know what they can do" but the researchers blew that all to hell because the first chip to be crippled? NOT an AMD but...an Intel! The first P4s were getting its ass royally kicked by the cheaper P3s running at the same clock speeds, Intel releases ICC, pays a ton to get benchmark companies to use it and...voila! The next year the exact same P4s against the exact same P3s and the P4s won by nearly 30%! Isn't that amazing?

There were two things that I didn't know about, so I figured I'd share those links:

F2FS is a flash filesystem: https://en.wikipedia.org/wiki/...
BFQ is a scheduler for I/O: http://www.phoronix.com/scan.p...

AMD Ryzen also seems to have a similar bug, related to hyperthreading that happens only in very special circumstances.

Quite a few Ryzen users have experienced instability problems during heavy compilation loads under Linux, especially those using compile-based distros such as Gentoo, but also under the Ubuntu subsystem on Windows.
There has been some debate whether the problems would have been caused by an actual bug, or if the people who experienced them simply had an unstable overclock - the latter being something that has also cropped up in forums recently.

Matthew Dillon, of Dragonfly BSD fame (and Amiga fame before that...) does believe that he has found a reproducible bug. He sent a test case about it to AMD in April.
This is not the first time Dillon has found a hardware bug in a AMD CPU. He found one for an earlier AMD CPU back in 2012 which was fixed in a microcode update.

I expect this to be fixed in a BIOS/microcode update soon, if not already in AGESA 1.0.0.6 - but I have yet to see any confirmation that it would have been fixed.

Unity is dead.

X11's main flaw is that it's supposed to be inefficient.

It's so much more than just inefficient.

First the there is the security issue. X security is inexistent. The problem in the last link is by design and it is simply unfixable.

X is big, as in really big, old, poorly written and it typically runs as root. Any buffer overflow or unchecked parameter is a potential privilege escalation vulnerability. Do yourself a favor and don't that dinosaur in your servers.

X is bloated. It has stuff in there like font rendering, drawing primitives, printers, mode setting, dozens of extensions and things that nobody uses. Nowadays, toolkits render their own fonts, and manage their own windows and essentially hand over a large image to the X server. The X server does not even render the image, it handles it to the compositor. X does not set the video mode, the kernel does that. X does not handle keyboard and mouse, it sends the data to the compositor, the compositor tells which window should receive the input and X sends the input event to the corresponding application. In short all X does nowadays is IPC, and there are millions of lines of code just sitting there doing nothing right next to this IPC.

It turns out that it is even terrible at IPC. An application does hundreds of requests to the X server just to display a simple window. You know what the worst case is for that? yeap, remote display. All the back and forth is really disastrous over a slow connection like the internet.

People say network transparent. Well, more like network capable. DRI does not work over the network at all.

Something supported by Mac OS X and Windows for decades like VSync so your don't see tearing is not doable in X.org, there is no way to wait for a screen refresh or prevent X.org from rendering your partially drawn window even with double buffering.

I am guessing you embrace the unix philosophy? if so, what is the one thing X does and how is it doing it well?

Heck, the wayland developers are the same X.org developers. They did miracles to keep X.org alive for so long, but they are moving on to greener pastures.

I would highly recommend this talk by Daniel Stone. He does an excellent job explaining why X.org is obsolete and unfixable.

See here for OSS AMD vs proprietary nVidia.
Yes they are perhaps slower, but not slow as hell, and depending on your usage still fast enough. But YMMV.
For me personally, OSS AMD worked fine.
Now I have nVidia, but their OSS drivers are unusable, so I need proprietary.
nVidia proprietary gives me screen tearing during videos, a LibreOffice bug, unable to change refresh rate thru xrandr, no highres text mode, no KMS.
See also the famous Torvalds statement.

Nothing wrong with the FX series, as long as you remember programs compiled with the Intel Cripple Compiler will be slower (but that is true of Ryzen as well) but if you look at benches made with GCC instead of ICC you see exactly what you would expect, the FX-8 sits between an i5 and an i7 depending on how multi-threaded the program is. I personally see absolutely no reason to upgrade from my FX-8320e as I'm getting better than 60fps on all my games (most better than 90fps) and even on heavy workloads like multitrack audio DSP processing it frankly works faster than I can.

As for Hillary? Anybody who could read the numbers knew she would lose a good year before the election. When she announced her candidacy her approval rating? 15%. After nearly a year and a half of the DNC rigging primaries for her, the MSM practically falling over themselves to kiss her ass and even running stories by her for approval? Her approval rating was...14%. Nobody in history has EVER won the POTUS with numbers that low so no shit she lost, she was about the worst possible candidate the DNC could possibly run, a rich influence peddling Wall Street insider that Wikileaks showed was getting her ass kissed by the MSM and cashing big fat checks from nasty dictators like the Saudi princes and if that wasn't enough? Her team pushing a narrative that she deserved it because she had a vag instead of anything she had actually done and that it was owed to her ("Its her turn" which they hung up on giant signs at all her rallies) just made her look even more arrogant and condescending.

Im stuck with windows for my VR development gear, at least for now...

Unreal Engine 4 handles VR quite well (disclaimer: my day job is developing VR using UE4) and SteamVR will be coming to Linux very soon.

The 1800X seems to be extremely strong in compile benchmarks: https://www.phoronix.com/scan.php?page=article&item=ryzen-1800x-linux&num=1

Too bad Ryzen only supports Windows 10

Source? My sources say GNU/Linux runs on it.

Along the same lines, I was wondering how would the Linux support would be, and duckduckgoing a bit, it seems that AMD has been pushing out kernel updates for some time. Looks promising. Some links for the lazy ones.

The thing is it is all too easy for a company like Apple to deceptively jump in bed with a huge number of external devs and then turn over and leave with an insane amount of knowledge now headed for the proprietary development of their own system.

Your assertion would be false: Apple announced Metal at WWDC 2014 (June). They even have a video of the presentation. It was released in Sept 2014 for iOS 8.

Vulkan API was not formally announced until March 2015. Version 1.0 of the specification would not be released until Feb 2016.

In any way you look at it, Apple released Metal long before the Khronos Group announced Vulkan.

Yes, and get results like this: http://www.phoronix.com/scan.p...

Personal note: I have and use Odroid -U2, -U3, -C, -C1, -C2, RPI2, RPI3, and a UDOO (original backer), mostly as micro-servers. I don't require much customization and as long as that remains true, I find them to be great machines.

The Odroids are definitely better hardware, but the story gets more complicated when the question of kernels (and the binary blobs needed for media) are updated to mainline. I've heard but not verified that the original Exynos CPUs in the Odroid-Ux are supported by mainline kernels.

The Allwinner chips in the Pines, Banana Pis, Orange Pis, etc. lack complete HW docs and need critical binary blobs (At least the Allwinner H8 has long needed a DRAM controller library blob, for example). If Allwinner were to clean up their documentation and make truly complete hardware docs available, then the overall product would be better than RPi. Until then, RPi support is so much better than their competitors that it overwhelms the otherwise obvious performance advantages.

Quoting from https://www.phoronix.com/scan....

However, the support isn't complete for the Allwinner A64 and is blocked in part by lack of proper documentation. Andre commented, "Due to a lack of official documentation and hardware availability this doesn't go any further at this moment."

The Allwinner A64 is comprised of the less-powerful Cortex-A53 cores, supports H.264/H.265 video decoding, and is widely talked about as being the "$5 ARM SoC." Hopefully this mainline kernel support will get figured out in time for the Pine A64 shipping.

...if they happen to pick Linux Mint Cinnamon...

And I quote: "without having to deal with people that tell them they are ... on the wrong distro." See? There you go already.

They don't get a holy war about whatever runs when the computer starts

Yet somehow, it's the very first question posted under this topic.

There are applications.

See? That's what I mean! You don't care what people actually need to do with their computer, there are a few applications so they should just shut up. What kind of shitty attitude is that? You didn't even ask "what do you need?", you just blindly assumed that whatever it is, surely there must be something.

Hardware does connect and 'just work'.
Including printers and video cards.

Oh, does it? Here, posted today: http://phoronix.com/scan.php?p...

Be sure to read the comments.

They don't need to ever open a shell to type commands.

Are you for f'ing real?

To me, your "not have a working GPU" remark discredits your entire post.

What's that Phoronix post about then?

Linux users can even be hardcore gamers.

Yeah. Playing Nethack.

Current, about 2% of desktop users are on Linux.

I'm not sure why you see that number as being supportive of your position, rather than mine.

And, by the way, there's a real Linux Foundation project for In-Vehicle Information (IVI) linux distributions.

They provide core functionnality, which can then subsequently be customized by vendors.

It is risky to be in the front line of a revolution. Those folks usually die to the second wave.

Oddly, nVidia succeeded by hitching their wagon to Microsoft. Since only a few nerds care about open source drivers, this really hasn't hurt them. It has been speculated that the reason that nVidia cannot release more driver source is that it is encumbered by agreements with Microsoft related to the adoption of the NV2A part for Xbox, and nVidia's subsequent opportunity to basically define Direct3D in their image for a time. Supposedly their Tegra chipset is generally in-house and unencumbered, so they can release more information about that family than they can GeForce.

I'm not up on the known details of the contract between nVidia and Microsoft on that occasion, if indeed there are any, and I would appreciate any information Slashdotters have on the subject. It would be interesting to know how nVidia managed not to wind up as a Microsoft subsidiary, and yet also succeeded.

After many years of Ubuntu use as primary desktop, the thing that drove me away was ending the support for the closed source AMD video drivers.

What does this have to do with Ubuntu? AMD ended their support.

Someone decided that the open source drivers were 'good enough'. Well, they are not, at least for what i was doing.

Yep, that "someone" was AMD. They apparently decided to focus more on a new Linux driver project, as noted in the posts from AMD folks quoted in the above link. Ubuntu isn't able to offer "support" for a closed-source driver that apparently breaks with the newer versions of Xorg. (I'd note that AMD had months to prepare before the new version of Ubuntu upgraded to the newer version of Xorg, and it's been a year or more and AMD hasn't updated their driver.)

And the choice to use the drivers as released by AMD was removed

Because it might break your system.

Imho, Ubuntu, and all derivatives like Mint, suddenly alienate half their user base with that decision.

How was it Ubuntu's fault (let alone Mint's, who didn't do anything here) that AMD stopped updating their drivers for Linux? Ubuntu and its derivatives aren't the only distros that this created problems with -- anyone who is using a version of Xorg released in the past year will have the same problem. And since Xorg is standard across most Linux distros, this truly has nothing to do with Ubuntu (or Mint) per se.

So, i'm back to windows 10 which serves my need

Yep -- AMD decided to update their drivers for the latest Windows version. Ubuntu can't do so, because they don't have the source code.

Why are you angry at Ubuntu when the people who stopped the support are AMD?

I don't mean to sound insulting, but you do understand what the implications of "closed-source driver" are, right? Ubuntu would likely be happy to provide support and updates if they had the source code... but they don't, and AMD won't release it.

Not yet, but AMD Zen CPUs will have such a feature. Have some articles:

http://wccftech.com/amd-zen-en...
http://www.phoronix.com/scan.p...
http://www.redgamingtech.com/a...

Not yet, but I'm sure Lennard will correct that in due course.

One of the systemd developers tried to make changes to the kernel for systemd only, breaking other software in the process, and was politely told where to stuff it.

Yes, systemd in 2014 was flakey, just as pulseaudio was back in it's time. But it is doing well today, healthy, and meeting expectations.

systemd is a pile of horse shit that was thrown into a fan so it sprayed everywhere, touched everything and contaminated what it touched.

sysvinit is a pile of cow shit, in a field somewhere, touching only the ground it rests on. Don't go to that field and step in that pile and it won't bother you.

If there are bugs in sysvinit, they affect sysvinit. If there are bugs in systemd, its everyone else's fault and everyone else should re-write their software to handle the bugs in systemd because the systemd developers are way too important to waste their incredible talent fixing their own bugs.

Not yet, but I'm sure Lennard will correct that in due course.

One of the systemd developers tried to make changes to the kernel for systemd only, breaking other software in the process, and was politely told where to stuff it.

The general infrastructure works.It was fixed years ago. The problem with Skylake was introduced when Intel hanged their graphics architecture.Intel Skylake GPUs needs a firmware blob. Intel needs to debug the firmware and the driver. More information here: https://01.org/linuxgraphics/intel-linux-graphics-firmwares and and https://www.phoronix.com/scan.php?page=news_item&px=Intel-SKL-BXT-Firmware-Blobs

What about the multi-monitor issue that plagues Skylake? Is Kaby Lake support better? See: http://www.phoronix.com/scan.p...

Can we phrase this the other way that doesn't make Microsoft look good? Just say Windows 8.1 and older will not get updates for Intel Kaby Lake and AMD Zen.

We expect most modern OSes to do these kinds of upgrades. Only calling out Windows 10 makes it seem like these are somehow special windows features, when they are nothing of the sort. Linux already has patches available for Intel's Turbo Boost 3.0, and that's just the first example I searched for.

Congratulations! You have just figured out why they called it a new name ("Wayland") instead of calling it "X11 release 8". They wanted to clean everything up so keeping compatibility wasn't an option.

And I'm glad you told me that the code is OK. Are you an X11 developer? Because I've watched YouTube videos of X developers telling me the code is brain-damaged and they are tired of dealing with it.

Here's a short outline of why Wayland is a good idea: http://www.phoronix.com/scan.php?page=article&item=x_wayland_situation&num=1

Are you going to take over X11 and keep it going when the world moves on to Wayland? If so, good luck, and I hope you are right about the code base being OK.

Not interesting on price you say? Well I'd really like to you show me an Intel at $135 that gets these kinds of numbers because the last time I checked all you can get from Intel for $135 is a crappy Pentium dual or an even crappier Atom. When you look at the bang for the buck it really isn't even close, I mean you can get an octocore from AMD for around $135 that has a 4Ghz turbo clock OOTB and which can easily hit 4.4Ghz-4.6Ghz on air, you aren't gonna find anything from Intel that competes until you at least double the price. Then when you look at motherboards and see how you can get a much nicer motherboard for less when you go AMD? Its really a no brainer, you can get a hell of a lot nicer system for a lot less money by going AMD.

This isn't even bringing up the elephant in the room which is that software hasn't kept up with hardware in quite a few years so even for gaming you can pair that chip with a $200 GPU and enjoy your games at 1080P with maxxed out graphics and high FPS, and for those that don't game one of the AMD APUs will give them all the power they need while having excellent picture quality and hardware accelerated video and both will spend a lot less.

According to this Phi can be had for $200 at the low end. You can't buy a Nvidia Tesla product for that.

https://www.phoronix.com/scan....

... but Slashdot instead publishes a Microsoft advertisement, as if "Surface support" was in any way as relevant as the other big new features - read e.g. here if you want to know more. Shame on you, Slashdot!

A20 now has open source graphics driver? is it actually usable for accelerated graphics (composition in X) ?

No, there is no open source graphics driver for the A20's MALI. However the FSF will look the other way and grant an exception if the GPU is not used and graphics are processed on the CPU. This is what they have done with the one variant of the system they are having certified. They are selling (at least) two others which are not certified.

that's a rather.... that's a misunderstanding of how the FSF operates. i didn't fully understand the criteria until it was explained to me a few months ago by josh gay. the primary concern is, "will a non-technical user end up arbitrarily executing privacy-violating malware WITHOUT THEIR KNOWLEDGE"? so things like the Debian synaptics package manager providing a simple-to-use point-and-click option to enable the "nonfree" repository WITHOUT ANY WARNING WHATSOEVER OF THE CONSEQUENCES is one of their worst nightmares.

this really did surprise me, but in the case of the MALI GPU, because it is entirely memory-mapped, there is *literally* no way for the average end-user to even *know that it's there*. "lspci" returns nothing (because there's no PCI/PCIe bus on the A20). "lsusb" returns... USB peripherals the GUI-equivalents of these two programs would also return... nothing.

now, we may say "but... but... someone who is technically-aware or who could follow instructions could recompile the kernel and add mali.ko!!!! wtf????!!?!?!?!" and the answer is "that's not an average end-user, is it?".

there was a huge debate about this on phoronix a couple of weeks ago - i won't go through all of them, just drop you somewhere in the middle with a link - but basically xf86-video-fbturbo works really well with the libre-compliant, GPL-compliant 2D GPU (called G2D) on the A20, which, for the purposes of doing email, internet, libreoffice, image editing (gimp supports PDF, PS, PNG and many more) - 2D acceleration is actually *better* and more power-efficient than *3D* acceleration. if you ever tried running compiz on a standard intel laptop and wondered why the fan just won't stop spinning, you'll know what i'm talking about. anyway - here's the link to phoronix... https://www.phoronix.com/forum...

Oh really? Might want to check the benchmarks before you start screaming horseshit, as in many tests its not even close, we're talking 181fps for Win 10 to...wait for it...41 fps for Ubuntu.

Windows and Linux use about the same amount of power. Linux being less efficient sometimes.

http://www.phoronix.com/scan.p...

I think you're over estimating the ability for edge to start let alone play 4 videos at the same time while running on Linux.

Windows and Linux use about the same amount of power. Linux being less efficient sometimes.

http://www.phoronix.com/scan.p...

Sorry - but I have to argue here... I have a Dell Inspiron 15 laptop - Windows 10 gives me ~5 hours of battery life. Fedora 24 gives me nearly 8. And yes - I have measured.

Windows and Linux use about the same amount of power. Linux being less efficient sometimes. http://www.phoronix.com/scan.p...

This solution worked for me on ubuntu 16.04 with kernel from http://kernel.ubuntu.com/~kern...: https://www.phoronix.com/forum...