5 ms·

Didn't also raise of Linux sped Sun's demise. In early 2000's the private compute farm of the org I was working had a mix of Solaris and Linux machines, with So

Didn't also raise of Linux sped Sun's demise. In early 2000's the private compute farm of the org I was working had a mix of Solaris and Linux machines, with Solaris hosts outnumbering Linux hosts. Within couple of years with aim of uniformization all hosts were standardized to Linux.

PeterStuer 4y ago
Not as i recall. Linux at that time was seen as an interesting project, but not yet suitable for the typical workloads ascociated with a Sun workstation.
- lizknope 4y ago
  Strongly disagree. I posted elsewhere in this thread but the semiconductor ran on Sun / Solaris / SPARC. We switched to Linux x86 practically overnight in 2002. Sun was slightly relevant for a few jobs requiring 64-bit >4GB memory. When AMD released the 64-bit Opteron that was the exact moment that Sun died and was completely irrelevant to the semiconductor industry. 99% of our software was text based and could be recompiled for Linux by typing "make"
  - PeterStuer 4y ago
    This might be where our experiences differ. Our workloads often had strong graphical / visialization components. We ran a mix of Sun and Apple machines. We tried some SGI as well but found them less suited. We moved to x86 with (for that time) highend GPUs because of pricing as wel as better/cheaper connectivity and toolchain support for what would now be called IoT. We ran some Linux, purely as basic unix stations.
    - lizknope 4y ago
      I agree, our use cases were quite different. In my other comment I said that around 1997 we put Linux x86 and 21" monitors running at 1800x1440 on all the engineers desks and used them basically as X terminals to the Suns in the closet. Our Sun graphics cards and monitors were 19" (I think) limited to 1152x900 so everyone thought Linux and the monitors was a big upgrade. We did use graphical programs like waveform viewers and chip layout programs but they were all simple 2D graphics and worked okay via remote X11 with 10 Mbit ethernet. We upgraded to 100 Mbit ethernet and it was virtually the same as running locally. By 2002 we were running all the software on Linux x86 directly We did have some Creator 3D cards in the Ultra 2's but it was mostly overkill for what we were doing. Those cards did have an interesting feature for programs hard coded to X11 8-bit 256 PseudoColor. We had an internally written program like that and we would basically run out of colors if you also tried to run a web browser or evern a second copy of the program. The Creator 3D card had one 24-bit display and then 4 separate 8-bit displays combined together. We could run 4 copies of that poorly written internal program before having colormap issues. I remember running netscape with the -install option to get a private colormap. The screen would basically look like a rainbow except in the netscape window then move the mouse outside and netscape would turn into a rainbow. On Linux we could use the extra virtual consoles to start multiple X servers so sometimes I had one 16-bit X server running for normal stuff and two 8-bit X servers for that ol d program. Luckily we didn't need to copy paste between all that much if ever.

PeterStuer 4y ago

Not as i recall. Linux at that time was seen as an interesting project, but not yet suitable for the typical workloads ascociated with a Sun workstation.

lizknope 4y ago
Strongly disagree. I posted elsewhere in this thread but the semiconductor ran on Sun / Solaris / SPARC. We switched to Linux x86 practically overnight in 2002. Sun was slightly relevant for a few jobs requiring 64-bit >4GB memory. When AMD released the 64-bit Opteron that was the exact moment that Sun died and was completely irrelevant to the semiconductor industry. 99% of our software was text based and could be recompiled for Linux by typing "make"
- PeterStuer 4y ago
  This might be where our experiences differ. Our workloads often had strong graphical / visialization components. We ran a mix of Sun and Apple machines. We tried some SGI as well but found them less suited. We moved to x86 with (for that time) highend GPUs because of pricing as wel as better/cheaper connectivity and toolchain support for what would now be called IoT. We ran some Linux, purely as basic unix stations.
  - lizknope 4y ago
    I agree, our use cases were quite different. In my other comment I said that around 1997 we put Linux x86 and 21" monitors running at 1800x1440 on all the engineers desks and used them basically as X terminals to the Suns in the closet. Our Sun graphics cards and monitors were 19" (I think) limited to 1152x900 so everyone thought Linux and the monitors was a big upgrade. We did use graphical programs like waveform viewers and chip layout programs but they were all simple 2D graphics and worked okay via remote X11 with 10 Mbit ethernet. We upgraded to 100 Mbit ethernet and it was virtually the same as running locally. By 2002 we were running all the software on Linux x86 directly We did have some Creator 3D cards in the Ultra 2's but it was mostly overkill for what we were doing. Those cards did have an interesting feature for programs hard coded to X11 8-bit 256 PseudoColor. We had an internally written program like that and we would basically run out of colors if you also tried to run a web browser or evern a second copy of the program. The Creator 3D card had one 24-bit display and then 4 separate 8-bit displays combined together. We could run 4 copies of that poorly written internal program before having colormap issues. I remember running netscape with the -install option to get a private colormap. The screen would basically look like a rainbow except in the netscape window then move the mouse outside and netscape would turn into a rainbow. On Linux we could use the extra virtual consoles to start multiple X servers so sometimes I had one 16-bit X server running for normal stuff and two 8-bit X servers for that ol d program. Luckily we didn't need to copy paste between all that much if ever.