|
Smythe posted:another savage sick update: We're sorry, but your operating system is not currently supported by SlingPlayer for web. Please use either Mac OS X or Windows OS . Click here for more information. better do what it says
|
# ? Oct 23, 2015 01:08 |
|
|
# ? May 10, 2024 02:53 |
|
loving libGL.so.1 also i installed the prop nvidia drivers, lost all my xfce theme stuff and text rendering. booted up yesterday and magically it's all back. thanks debian
|
# ? Oct 23, 2015 14:57 |
|
my name is Reginald P Linux
|
# ? Oct 23, 2015 19:26 |
|
hte p is for pighammer
|
# ? Oct 23, 2015 19:57 |
|
b0red posted:loving libGL.so.1 debian has packages for nvidia proprietary drivers, and those packages handle the libGL.so.1 and all other diversions just loving fine if you choose not to use the packaged drivers, and just run a universal shell script installer written by an intern at nvidia ten years ago, you should not be at all surprised when random poo poo breaks all over the system. you're trying to fix a watch with a hammer
|
# ? Oct 24, 2015 00:05 |
|
The Debian packaged version won't handle Optimus cases where you have more than one GPU on a laptop, which is becoming increasingly common. NVIDIA wrote libglvnd for this case, but Debian won't ship it.
|
# ? Oct 24, 2015 00:21 |
|
Suspicious Dish posted:The Debian packaged version won't handle Optimus cases where you have more than one GPU on a laptop, which is becoming increasingly common. NVIDIA wrote libglvnd for this case, but Debian won't ship it. do you know if they've got any plans of supporting wayland?
|
# ? Oct 24, 2015 01:54 |
|
Ever since I left Red Hat, I lost contact with them, so I'm just as in the blue as you guys are. In May, we saw an update to their driver that added a "nvidia-modeset" device. Whether it works or not, I have no idea. Otherwise, there are plans, but I don't know any update besides last year's XDC talk. http://www.x.org/wiki/Events/XDC2014/XDC2014RitgerEGLNonMesa/
|
# ? Oct 24, 2015 04:05 |
|
Suspicious Dish posted:The Debian packaged version won't handle Optimus cases where you have more than one GPU on a laptop, which is becoming increasingly common. NVIDIA wrote libglvnd for this case, but Debian won't ship it. debian jessie is still shipping nvidia version 340 drivers in the meanwhile we have bumblebee, which renders off-screen and then uses the intel driver to copy it. works fine
|
# ? Oct 24, 2015 04:54 |
|
was wondering a couple nights ago how many times faster than a SPARCstation 5 the original Raspberry Pi is, how many orders of magnitude would it be in raw CPU and graphics perf then I wondered what kind of difference there'd be with that SS5 running SunOS 4.1.3 or Solaris 2.6, and that same SS5 running Linux. how much worse would Linux be?
|
# ? Oct 24, 2015 07:55 |
|
SPARCstation 5 had either a 110 MHZ microsparc-II or 170 MHZ TurboSPARC. WIkipedia says that the turbosparc was comparable to a 120MHZ Intel Pentium. In-order execution. Up to a whopping 16 kb of L1 and 1MB of L2 cache. Raspberry PI Model A had a 700 MHZ ARM11 processor. Out-of-order (debatable) capabilities, SIMD instructions, and hardware accelerated decoding make it tough to do anything except direct comparisons on specific operations. This is especially true if the SIMD instructions would be useful. Orders of magnitude, likely. As for Linux vs SunOS, I'm confident that Linux would hold its own and be as good, if not better. As long as you strip out all of the trash processes, of course. For example, the Titan Supercomputer was the top supercomputer when introduced three years ago and runs Cray Linux. The distribution is different for login nodes and compute nodes. If there was a chance that sparc would have been faster, I'm sure they would have brought it on board. Even a tiny advantage would end up being impressive due to the sheer scale of it all.
|
# ? Oct 24, 2015 09:30 |
|
so yeah, we got the big data and small data servers and clouds locked up, what about the desktop (yes, I know mainframes still exist but are you really gonna be that pedantic guy who says that xeon racks will always be scrub tier?)
|
# ? Oct 24, 2015 09:39 |
|
Suspicious Dish posted:Ever since I left Red Hat, I lost contact with them, so I'm just as in the blue as you guys are. In May, we saw an update to their driver that added a "nvidia-modeset" device. Whether it works or not, I have no idea. it's good to hear that they're actually working on it though
|
# ? Oct 24, 2015 12:10 |
|
eschaton posted:was wondering a couple nights ago how many times faster than a SPARCstation 5 the original Raspberry Pi is, how many orders of magnitude would it be in raw CPU and graphics perf graphics would be orders of magnitude. the sun graphics are essentially a dumb framebuffer, the raspberry pi has an all-singing all-dancing modern gpu architecture. cpu would be a much smaller gap than you might imagine. especially if you use an SS20 instead of an SS5 (same chips, different memory architecture). it's strong in all the places a cellphone ARM is weak: instructions per clock, memory latency, memory bandwidth. eschaton posted:then I wondered what kind of difference there'd be with that SS5 running SunOS 4.1.3 or Solaris 2.6, and that same SS5 running Linux. how much worse would Linux be? in micro-benchmarks, linux shits all over solaris on the same hardware in real world performance, it doesn't matter that much
|
# ? Oct 24, 2015 14:58 |
|
celeron 300a posted:SPARCstation 5 had either a 110 MHZ microsparc-II or 170 MHZ TurboSPARC. WIkipedia says that the turbosparc was comparable to a 120MHZ Intel Pentium. the turbosparc came out in 1996. it competed with early pentium IIs, at much higher clock rates. (badly.) turbosparc was meant to be cheap, not fast. and it didn't really turn out to be either one a more interesting comparison would be 80 MHz supersparc
|
# ? Oct 24, 2015 15:03 |
|
my main memories of sparc workstations are that they were hideously slow compared to similarly-priced x86 hardware, and the only reason we kept using them for so long was that we had a bunch of sparc binaries with ancient code that was not trivial to compile for x86 instead this was only about 10 years ago though, things may have been different 20 years ago
|
# ? Oct 24, 2015 15:41 |
|
risc/unix stopped being price competitive with x86 in the early 90s, when 386/486 became affordable the only way they limped along for another 10 years was the availability of big machines. in the 1990s, x86 smp was so bad it wasn't worth doing. so you bought your developers overpriced and underpowered sparc workstations, then deployed code on 4- and 8-processor unix boxes with ~*~ multiple gigabytes ~*~ of ram when intel released nehalem in 2008, that was the last nail in the coffin. post-nehalem, you can build 4-8-16-32 processor intel boxes that will do just as well as their risc counterparts
|
# ? Oct 24, 2015 16:46 |
|
tell us your true opinions on itanium
|
# ? Oct 24, 2015 22:01 |
|
HPUX ftw
|
# ? Oct 24, 2015 22:16 |
|
Soricidus posted:my main memories of sparc workstations are that they were hideously slow compared to similarly-priced x86 hardware, and the only reason we kept using them for so long was that we had a bunch of sparc binaries with ancient code that was not trivial to compile for x86 instead My main memories of Sparc were in college when I was limited to putting around with CDE and trying to find the drat compiler. This was in 98, when Linux was still considered to be a hobbyist OS to bring UNIX to x86 computers. To my horror and amusement, compilers for Solaris were not free and I had to do something special to gain access to it. The graphics were crisp but slow. The normal command line arguments didn't work (a good example was ps), and reading /proc taught me nothing. Still, I thought it was cool. All of the novelty was in the fact that I was web browsing on a computer that wasn't a Mac or a Windows computer, on an OS that treated the command line as a first class citizen and had a pedigree that spanned decades. Now, I probably won't even bother to learn Solaris or HP-UX even if I was paid, unless there's a good chance that it'd help my career in Linux device driver development. Also, it's pretty cool that we live in an age where even Microsoft will give away software development tools for free for commercial use.
|
# ? Oct 24, 2015 22:38 |
|
I used to hang out at a lab with a Digital workstation. Even though it had an insanely high clock speed (maybe 600 mhz or something? when the average desktop was 300 mhz) it didn't feel very fast at all. And the bus speed was really slow, to boot. Pretty funny to think about now, I guess.
|
# ? Oct 24, 2015 22:48 |
|
pram posted:HPUX ftw hot piss user experience
|
# ? Oct 24, 2015 23:17 |
|
what is the best unix
|
# ? Oct 24, 2015 23:28 |
|
Apple Operating System X 10.11 El Capitan
|
# ? Oct 24, 2015 23:38 |
|
huh, looks like there's a linux that's actually a unix these days. inspur k-ux. never heard of it before.
|
# ? Oct 25, 2015 00:53 |
|
That's niche, a Chinese RHEL clone for mainframes.
|
# ? Oct 25, 2015 00:59 |
|
i'm just surprised that someone finally bothered after all these years
|
# ? Oct 25, 2015 01:01 |
|
who pays attention to SuS, like, at all? doesn't it still mandate the really awful STREAMS interface they tried to shove into POSIX a while back? IRIX and Tru64 are SuS compliant.
|
# ? Oct 25, 2015 01:12 |
|
quote:STREAMS was required for conformance with the Single UNIX Specification versions 1 (UNIX 95) and 2 (UNIX 98), but as a result of the refusal of the BSD and Linux developers to provide STREAMS,[citation needed] was marked as optional for POSIX compliance by the Austin Group in version 3 (UNIX 03). POSIX.1-2008 with TC1 (IEEE Std 1003.1, 2013 edition) has designated STREAMS as 'marked obsolescent'[14][15] meaning that said functionality may be removed in a future version of the specification. thanks wikipedia
|
# ? Oct 25, 2015 01:41 |
|
quote:The Windows NT kernel offered a full port of STREAMS as the streams.sys binary. NT DDK even had a chapter on STREAMS, going as late as NT4 though in NT4 DDK it was declared obsolete. The original TCP/IP stack for Windows NT 3.1 was implemented atop STREAMS by Spider Systems, and used the streams.sys binary.
|
# ? Oct 25, 2015 01:44 |
|
carry on then posted:what is the best unix for generic x86 servers? linux for particular ultra expensive servers? AIX. just don't try compiling anything primarily targeting Linux and expect anything but a total ball ache though for quite literally any other purpose? OS X
|
# ? Oct 25, 2015 01:45 |
|
pram posted:thanks wikipedia so its been irrelevant ever since 2003 much like the SuS
|
# ? Oct 25, 2015 01:59 |
|
shitface posted:for particular ultra expensive servers? AIX. just don't try compiling anything primarily targeting Linux and expect anything but a total ball ache though not even ibm believes this linux is available on their most expensive POWER boxes and they actively push it as the preferred choice
|
# ? Oct 25, 2015 02:48 |
|
pram posted:HPUX ftw for the last seven years hp has been talking about doing superdome boxes that support both xeon and itanium, so they can run linux, windows, and hp-ux workloads on the same system architecture, just changing the line items as necessary for a given order this year they announced their new superdome X. it's 288 cores in 1/4 of a rack, instead of 256 cores in two racks. and it's x86 only. and linux only. because lol @ enterprise anything on windows. itanium is dead. hpux is dead. windows is one foot in the grave. linux/x86 supremacy.
|
# ? Oct 25, 2015 02:51 |
|
celeron 300a posted:I used to hang out at a lab with a Digital workstation. Even though it had an insanely high clock speed (maybe 600 mhz or something? when the average desktop was 300 mhz) it didn't feel very fast at all. And the bus speed was really slow, to boot. dec sold some insanely fast RISC boxes with one* of the worst unix systems ever, "digital unix." it was a derivative of osf/1, so it was mach-based, with all the ugly problems that implies linux on dec alpha had 1/10th the overhead of the vendor unix. dec didn't really care about this since their core market was their proprietary operating system, vms. and vms on alpha was demonstrably faster than vms on vax. *the absolute worst-ever unix was also a dec product: ultrix. but that's another story for another day
|
# ? Oct 25, 2015 02:54 |
|
Notorious b.s.d. posted:dec sold some insanely fast RISC boxes with one* of the worst unix systems ever, "digital unix." it was a derivative of osf/1, so it was mach-based, with all the ugly problems that implies Yup, we had linux installed on that dec workstation. There were rumors of an SGI/IRIX box too, but I don't think anyone would have wanted to touch it with a 50 foot pole without a clear path to getting a working Linux on it. So, I guess it probably just sat there gathering dust, which probably just made it more valuable.
|
# ? Oct 25, 2015 03:19 |
|
Notorious b.s.d. posted:for the last seven years hp who buys hp servers
|
# ? Oct 25, 2015 03:22 |
|
pram posted:HPUX ftw Someone reset the pram, I think it's corrupted. No, I think you also have to hold down the closed-apple key. Maybe another letter? Lemme ask siri. pram posted:Apple Operating System X 10.11 El Capitan Much better.
|
# ? Oct 25, 2015 03:22 |
|
it was sarcasm
|
# ? Oct 25, 2015 03:23 |
|
|
# ? May 10, 2024 02:53 |
|
Suspicious Dish posted:who buys hp servers According to Gartner, there are enough buyers to move 534,559 shipments in Q1 2015, about 22.6% of the market (if I'm reading these things right). But yeah, if any of them are actually Itanium running HP-UX, it would be a small number. HP execs are saying that it's mission critical stuff like telecoms, with contracts to 2025. Nothing like vendor lock-in to guarantee those sweet profits. EDIT: Maybe a lot of it is HP themselves when they are providing services in general. celeron 300a fucked around with this message at 03:40 on Oct 25, 2015 |
# ? Oct 25, 2015 03:33 |