Re: Quake performance SGI vs Sun
From: Benjamin Gawert (bgawert_at_gmx.de)
Date: 03/28/04
- Next message: Benjamin Gawert: "Re: Quake performance SGI vs Sun"
- Previous message: joe smith: "Re: Quake performance SGI vs Sun"
- In reply to: UNIX Musuem: "Re: Quake performance SGI vs Sun"
- Next in thread: joe smith: "Re: Quake performance SGI vs Sun"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Sun, 28 Mar 2004 20:59:49 +0200
UNIX Musuem wrote:
> I seriously doubt the FX4 had much better OGL performance than MXI.
FX4 (which was HP's midrange workstation gfx) of course wasn't faster than
MXI (SGI's High End Workstation gfx). But FX6 (HP's Highend Workstation gfx
at that time!) with texture option kicked MXI's/MXE's *** as did FX10 with
SGIs V12.
> I
> had both... FX4 was OK, but most of all, the quality of display is
> not as good.
What do You mean with "quality"? In what respects was the FX4 worse than MXI
beside the performance of course?
The signal quality of the old FX Series (FX2/4/6) was very good, however, if
You didn't had a display with EVC in You needed one of the EVC-to-VGA
adapters. HP made two versions of these adapters, the older one (in fact a
short cable) decreased the image quality at high resolutions noticeably. The
newer adapters were quite good. And the newer FX Series (FXe/5/10) already
offered VGA and (on FX5/10) DVI also.
> Granted the HP FX series aren't bad (they do a better
> job than Sun's for sure), I still maintain that sgi OGL is a
> reference for "nice" OGL display.
In what respect? What makes an SGI displaying OpenGL better than any other
fully OpenGL compliant system?
I know SGI invented OpenGL but they gave the control over OpenGL away to a
consortium long ago. I see no reason why OpenGL on SGI should be better than
OpenGL on HP/Sun/IBM/whatever.
>>> This includes
>>> the 320 cobalt chip to a certain extent.
>>
>> Cobalt was nice, but the drivers were just buggy...
>>
> Not as buggy as the current ATI & nVIDIA cards.
Sorry but that's nonsense. The Cobalt drivers had really serious problems
with a lot of applications, something which the ATI and Nvidia drivers
didn't have. Even standard applications (i.e. Office) sometimes suffered
from the crappy drivers.
> It had some severe
> bugs/glitches (like the colormap reversal which made java3d/swing
> apps totally useless until JRE 1.4), but was of overall good quality.
For me good quality is a bit more than crappy drivers for flakey hardware.
> The apps profiles were actually a pretty good job. I still have a
> 320, one of the first ones to ship, and I'm still using it for
> high-end CAD without a problem, even though the support has been
> dropped years ago.
I have an SGI320 myself, one of the last machines made (with the new sgi
logo). A nice system but certainly not at the same quality level I get from
IBM or HP. Prone-to-fail PSUs (something like an SGI trademark) and the
certainly most stupid memory sockets ever made don't make a high quality
system. The software side is almost as worse, besides the crappy gfx drivers
without any working D3D support SGI also forgot to make working drivers for
the firewire ports which are somewhat useless. Or the 1600SW which
definitely is still a very nice display but sadly lacks brightness controls,
something which is already bad enough. But SGI also even didn't implement a
brightness control utility for the SGI320 drivers. Really great to have an
SGI320 with 1600SW but no way to bring the brightness down, and the 1600SW
has a really strong brightness. The only way to bring down brightness on the
1600SW is the Multilink (or the other adapters that are available now) or
the ancient #9 Revolution gfx card, or connecting it to an O2.
>> That's why SGI choosed ATI-based gfx, eh?
>>
> Yep, that's probably the saddest decision... Looks like sgi, like a
> lot of the industry, is now settling on price instead of quality
No. SGI has realized that a complete inhouse development of gfx hardware is
too expensive and too time consuming while the competition is going forward
in huge steps. Using modified comodity hardware which already outperforms
the current SGI-developed gfx systems is the best they could do.
>>> Once again, this is not for
>>> games, but for high-end CAD types of applications,
>>
>> We do high end CAD applications and we don't have any problems with
>> our
>> FireGLs and Quadros. The drivers are very good, especially the ones
>> from
>> Nvidia, and even the ATI drivers didn't gave us much problems (older
>> versions had some display errors, but that's long gone).
>>
> There's always display artefacts, glitches and unexplained lock ups
> with most applications. I'd be curious to know which CAD app you guys
> are running...
We're running several CAD applications in various departments. In my
department for example most engineers use Pro/Engineer, Unigraphics and
SolidEdge on Windowsxp Professional and HP-UX 11i, and also Catia on
RS/6000. Other departments also run Microstation and I-DEAS. We also use
non-CAD applications like FLUENT a lot, and of course there are also a lot
of inhouse-developed programs.
>> When we kicked SGI out we also had some people that missed the nice
>> SGI
>> desktop, but that's it. Today the same engineers are very productive
>> on
>> HP9000 UNIX-Workstations and HP x(w)-Series PC-Workstations. Most of
>> the
>> systems run 24hrs/day, and give us less trouble than we had with SGI.
>>
> No doubt on the HP-9000, these are pretty strong machines.
Right, very fast hardware, and built like a tank.
> XW, I
> would question the 24/7 statement...
> Everywhere I go, Win users are rebooting at the end of the day, if
> not several times a day...
What for? Windows isn't necessarily unstable. If You use really good quality
hardware and a solid setup (no tweaks or "tunings") with working drivers
there's no need to reboot on a regular base, especially not on Windows
workstations (server might be a different story, though).
Benjamin
- Next message: Benjamin Gawert: "Re: Quake performance SGI vs Sun"
- Previous message: joe smith: "Re: Quake performance SGI vs Sun"
- In reply to: UNIX Musuem: "Re: Quake performance SGI vs Sun"
- Next in thread: joe smith: "Re: Quake performance SGI vs Sun"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]