Slashdot Mirror
AMD Catalyst Linux Driver Catching Up To and Beating Windows
An anonymous reader writes: Along with the open-source AMD Linux driver having a great 2014, the AMD Catalyst proprietary driver for Linux has also improved a lot. Beyond the open-source Radeon Gallium3D driver closing in on Catalyst, the latest Phoronix end-of-year tests show the AMD Catalyst Linux driver is beating Catalyst on Windows for some OpenGL benchmarks. The proprietary driver tests were done with the new Catalyst "OMEGA" driver. Is AMD beginning to lead real Linux driver innovations or is OpenGL on Windows just struggling?
1998 called and wanted its joke back
AMD's Linux drivers are catching up to, and beating, the Windows drivers? That shouldn't be hard, given that the Windows drivers are a steaming load of fetid moose crap. The drivers are the reason I switched back to Nvidia. Their Linux drivers may be proprietary and a little fidgity, and the FOSS Linux drivers may be worse than junk, but at least I don't have to nuke a whole system install just to upgrade Catalyst, and once they're installed the friggin' work!
Rawr
you whippersnappers these days, we had to ctrl alt delete 5 times a day and liked it! no one had ever even gotten one to run for 40 days uptime anywhere. did i mention our clock speed was 120mhz and we had to reinstall windows from 54 floppy diskettes and had a 1 gb hdd!
Well, the xbmc team did completely give up on trying to use hardware acceleration with the AMD prioprietary fglrx driver and switched to the OSS radeon driver in late 2013.
you whippersnappers these days, we had to ctrl alt delete 5 times a day and liked it! no one had ever even gotten one to run for 40 days uptime anywhere. did i mention our clock speed was 120mhz and we had to reinstall windows from 54 floppy diskettes and had a 1 gb hdd!
You hipsters - we had to do a power cycle to restart, and no windows, floppies, of hard disks - just a 1mhz cpu, a couple of tape drives, and a serial port. And we had FUN!
"Transparent" is a shit show that trades on every stereotype going. A man in drag is NOT a transsexual.
Valve does. Hell, they've created their own Debian spin-off, SteamOS to try and woo developers away from Windows. And so far, I'd say they're doing a decent job as the number of games available on Linux has jumped since the announcement (let alone since the beta). Well reviewed titles like "Battleblock Theater," "XCOM: Enemy Unknown," "Super Meat Boy," "Borderlands 2," and "Borderlands: The Pre-Sequel" are all available on Linux through Steam.
Rawr
Crimney, back in my day we had a paper and a pen! And when the paper crashed we had to abandon all our work and start over! We didn't have any of that "non-volatile" storage space beyond the pressed pulp!
Rawr
Ad hominem attacks will get you nowhere. But then again I guess that's why you're an AC.
That article is from 2008 (seven years isn't exactly "nearly" a decade in my book) and reflects an issue specifically involving causes of crashes for Windows Vista about a year after the OS came out. And you'll notice that while Nvidia is the largest single contributor to that pie, less than 30% of crashes were their fault. And, actually, if you read the original article from which Engadget derived their story this is a study from specifically around the launch of Windows Vista, not its entire lifecycle. And the data is very vague, as they say in the article, "in theory, NVIDIA's proportion of total driver crashes could be inflated by a relatively small handful of systems with severe driver issues." So this statistic is actually pretty useless without additional data.
That also doesn't indicate anything having to do with non-system-crash related issues, such as non-fatal crashes or poor system performance. It's also reasonable to assume that Nvidia has since fixed that issue with Windows Vista, as I don't remember there being any kinds of crashing plagues involving Nvidia hardware in Windows 7, 8, 8.1, or even 10 Technical Preview.
From when, and in what categories? I'm not denying that AMD makes a good graphics card and they have delivered, and do deliver the most bang for your buck at certain price points, but your claim is flimsy at best since Tom's updates these almost quarterly, as GPU manufacturers release new hardware throughout the year, and across several performance/price points. So naturally when AMD releases a new GPU they're likely to take the top spot in the high performance category, just as it's likely that when Nvidia releases a new GPU they might take the top spot.
Rawr
you whippersnappers these days, we had to ctrl alt delete 5 times a day and liked it! no one had ever even gotten one to run for 40 days uptime anywhere. did i mention our clock speed was 120mhz and we had to reinstall windows from 54 floppy diskettes and had a 1 gb hdd!
You hipsters - we had to do a power cycle to restart, and no windows, floppies, of hard disks - just a 1mhz cpu, a couple of tape drives, and a serial port. And we had FUN!
You really don't want to hear what booting a PDP-8/e involved. Oh, you do!
First, utter magic incantations (perhaps under one's breath or inaudibly) while turning the key switch which powered it on. Check for no error lights (hence the magic incantations) recalling that this was in the days before LEDs. Next, toggle in an address on the binary switches on the front panel and latch it. It's best if this is a fairly low address, as this will save some time. Then toggle in an instruction and latch it. Luckily, this was a 12-bit machine, so addresses and instructions were short, being limited to 0-4095. Increment the toggled address, and toggle in another instruction. Repeat for a short while, then set the toggle switches to the first address, and execute the program. Now it's in read-in mode! Toggle in an address which is above the last one used, and latch it. Then toggle in and latch a succession of instructions (read-in mode automatically increments the addresses for you). After sufficient instructions have been entered, the toggles are set to the start of this program, and execute the program is commanded!
The machine will now read in the bootstrap card (a card full of resistors and capacitors which had traces cut to provide ones and zeroes), and execute the bootstrap loader program which it finds there. On our PDP-8, this had a simple driver for a tape and would read in the OS from a tape drive. You did remember to load the correct tape, didn't you? If not, it's back to square one.
It's implied with the name "Catalyst." The FOSS drivers have never been referred to as "Catalyst" except in momentary errors.
Rawr
All that was shown here is that AMD's *OpenGL* drivers on Linux aren't too far off from AMD's *OpenGL* performance on Windows.
Considering that AMD's OpenGL Implementation on Windows is kind of a joke compared to D3D, and considering that AMD is now even dumping D3D in favor of its proprietary* Mantle platform, this article basically proved that AMD's Windows OpenGL support is also lacking badly.
* Before anyone says Mantle is "open": AMD's executives promised an SDK published by the end of 2014... didn't happen. AMD has made zero efforts to make Mantle work on any OS other than Windows... hell, while DX11 ain't an open standard at least I can go online and get docs on how to write a program using DX11 and make it work on Windows... you can't even do that with Mantle!
AntiFA: An abbreviation for Anti First Amendment.
... but no problems with AMD/ATI as of several years now (2D only)... ...and that is important to me because there are just two main vendors in the graphics category.
Slowly it's becoming three, with Intel's HD Graphics and Iris hardware. ...
[bold part by me]
If we're talking about 2d only, then it's definitely at least 3 vendors, and there are some others that are perfectly fine in that realm.
To the AC GP, It's also not at all fair to say you haven't seen any problems for a few years while qualifying that you've only used 2d. The whole damn article is about 3d performance, and that's the part of the driver that is the most complex and has the most proprietary bits. I haven't had any problems with my AMD/ATI cards on my headless servers either, but that's hardly relevant.
Both of you probably have a piece of shit power supply. Buy a new, high quality power supply and most of your crashing will go away.
If you ever want to see just how bad nvidia is in Linux, get a laptop that has their Optimus abortion. My laptop at work regrettably has that.
With stock Intel drivers, display works but there's no acceleration, so performance is shit.
With stock nvidia or nouveau drivers, performance is great but can't use external monitors (because they are tied to the Intel chip)
Getting both working at once required a kernel built from source, a backported package from the testing build, a package from a PPA from a child distro, three dependencies built from source because of conflicts between the distro packages and the bleeding edge kernel I had to use, and the nightmare that is bumblebee. I don't dare run an update on this system because fuck knows what will break.
Meanwhile, my last three laptops at home have been AMD-based. Install Catalyst, reboot, everything is beautiful. It is remarkable how far things have swung. I remember AMD being verboten back when I first got into linux because of how godawful the support was.
Emulation applies to API function call emulation as much as it does CPU instruction emulation since the function calls are translated to native system calls in the same way, and the term can even apply to network protocol emulation, HDD command emulation, BIOS emulation etc.
Using your definition a Z80 emulator is merely an environment of Z80 CPU instructions, which just shows the distinction is inconsistent with itself.
You don't get to redefine the English language just because you have a popular FOSS project, especially when the term is well established for the same process in every other domain of computing.
Why does my post history abruptly stop? I want to laugh at the stupid things I posted as a kid.