Re: R10000-150mHz_vs_R12000-300mHz...(?) - SGI

This is a discussion on Re: R10000-150mHz_vs_R12000-300mHz...(?) - SGI ; arie van schutterhoef writes: > I wonder if it's worthwhile to replace a R10000-150mHz with a > R12000-300mHz in an o2. > Is the doubling increase reality or just bogus? Depends on the price, I suppose. > Just wondering whether ...

+ Reply to Thread
Page 1 of 2 1 2 LastLast
Results 1 to 20 of 28

Thread: Re: R10000-150mHz_vs_R12000-300mHz...(?)

  1. Re: R10000-150mHz_vs_R12000-300mHz...(?)

    arie van schutterhoef writes:

    > I wonder if it's worthwhile to replace a R10000-150mHz with a
    > R12000-300mHz in an o2.
    > Is the doubling increase reality or just bogus?


    Depends on the price, I suppose.

    > Just wondering whether audio applications will benefit
    > from this.


    What exactly are you doing?

    --
    Atro Tossavainen (Mr.) / The Institute of Biotechnology at
    Systems Analyst, Techno-Amish & / the University of Helsinki, Finland,
    +358-9-19158939 UNIX Dinosaur / employs me, but my opinions are my own.
    < URL : http : / / www . helsinki . fi / %7E atossava / > NO FILE ATTACHMENTS

  2. Re: R10000-150mHz_vs_R12000-300mHz...(?)

    In article , Atro Tossavainen
    wrote:

    > Depends on the price, I suppose.

    -90 euro's (which is affordable, I guess...)

    > What exactly are you doing?

    -Running PD and doing FFT's for realtime audio performances.
    I'm not so sure how benchmarks relate to this, in the
    sense of pure number crunching.
    Because I found this link regarding PD running on
    different platforms:
    http://iem.at/mailinglists/pd-list/2003-11/015206.html
    I wa quite surprised about the outcome.
    I also made me wonder about SGI policy itself, regarding
    OS and hardware. But that another topic...

    AvS

  3. Re: R10000-150mHz_vs_R12000-300mHz...(?)

    arie van schutterhoef writes:

    >> Depends on the price, I suppose.

    >
    > -90 euro's (which is affordable, I guess...)


    If you can get the CPU upgrade from 150 MHz R10k to 300 MHz R12k for
    90 euros, I would already have done it.

    >> What exactly are you doing?

    >
    > -Running PD and doing FFT's for realtime audio performances.


    My gut instinct is that realtime FFT will certainly benefit from the
    increased CPU power. I'm not sure about Pd, I'm totally unfamiliar
    with that application.

    The overall system design differences between the O2 and the Octane
    may mean that you won't be able to get quite as much out of your O2
    as Olaf does on his Octane with an identical CPU. What kind of
    results are you getting now (if you do the same test that Olaf did)?

    > I wa quite surprised about the outcome.


    Hey, all SGI aficionados know it's not the MHz that counts. :-)

    I should probably try compiling Pd and its friends with the MIPSpro
    compilers rather than gcc which is what Olaf used to see if we can't
    get even more oomph out of it :-)

    --
    Atro Tossavainen (Mr.) / The Institute of Biotechnology at
    Systems Analyst, Techno-Amish & / the University of Helsinki, Finland,
    +358-9-19158939 UNIX Dinosaur / employs me, but my opinions are my own.
    < URL : http : / / www . helsinki . fi / %7E atossava / > NO FILE ATTACHMENTS

  4. Re: R10000-150mHz_vs_R12000-300mHz...(?)

    > The overall system design differences between the O2 and the Octane
    > may mean that you won't be able to get quite as much out of your O2
    > as Olaf does on his Octane with an identical CPU.

    -The question is probably more or whether the o2's UMA makes
    a difference in comparison to the Octane's switch. In particular
    when the system contains two cpu's.
    Also secondary cache will influence the outcome.

    AvS

  5. Re: R10000-150mHz_vs_R12000-300mHz...(?)

    In article <190320051742498187%arsche@xs4all.nl>,
    arie van schutterhoef wrote:

    : > The overall system design differences between the O2 and the Octane
    : > may mean that you won't be able to get quite as much out of your O2
    : > as Olaf does on his Octane with an identical CPU.
    : -The question is probably more or whether the o2's UMA makes
    : a difference in comparison to the Octane's switch. In particular
    : when the system contains two cpu's.
    : Also secondary cache will influence the outcome.

    The point that was trying to be made is that the R10k O2s are essentially a
    graft job, and the CPU performance suffers because of it even when the rest of
    the system isn't anywhere near its limits.

    You can read Ian's CPU performance tables for examples showing how the FP
    performance especially is completely neutered in the R10k O2 compared to the
    exact same CPU in the Octane or even the Indigo2:




    If you don't use the UMA features in the O2 then you're *much* better off with
    an Octane.


    Cheers - Tony 'Nicoya' Mantler

    --
    Tony 'Nicoya' Mantler -- Master of Code-fu -- nicoya@ubb.ca
    -- http://nicoya.feline.pp.se/ -- http://www.ubb.ca/ --

  6. Re: R10000-150mHz_vs_R12000-300mHz...(?)


    > You can read Ian's CPU performance tables for examples showing how the FP
    > performance especially is completely neutered in the R10k O2 compared to the
    > exact same CPU in the Octane or even the Indigo2:
    >
    >

    -I've read them some time ago and they give a general
    indication, but I'm not so sure to what extent they relate
    to audio processing. They all seem to graphics -oriented.
    Though maybe it's better to speak in general signal processing
    terms.

    > If you don't use the UMA features in the O2 then you're *much* better off
    > with an Octane.

    -I presume with UMA's strengths, you talk of the I/O benefits?
    As far as being better of: the o2 is going to be used for
    concerts and an Octane is far too noisy with all these fans.
    The o2 is much quieter, so otherwise the Octane would be the favorite.

    AvS

  7. Re: R10000-150mHz_vs_R12000-300mHz...(?)

    In article <200320051805445510%arsche@xs4all.nl>,
    arie van schutterhoef wrote:

    : > You can read Ian's CPU performance tables for examples showing how the FP
    : > performance especially is completely neutered in the R10k O2 compared to
    : > the
    : > exact same CPU in the Octane or even the Indigo2:
    : >
    : >
    : -I've read them some time ago and they give a general
    : indication, but I'm not so sure to what extent they relate
    : to audio processing. They all seem to graphics -oriented.
    : Though maybe it's better to speak in general signal processing
    : terms.

    The benchmarks on those pages use a number of algorithms to test various aspects
    of floating point and integer CPU/System performance respectively.


    : > If you don't use the UMA features in the O2 then you're *much* better off
    : > with an Octane.
    : -I presume with UMA's strengths, you talk of the I/O benefits?
    : As far as being better of: the o2 is going to be used for
    : concerts and an Octane is far too noisy with all these fans.
    : The o2 is much quieter, so otherwise the Octane would be the favorite.

    UMA's strengths are primarily in the areas of 3d texturing using video as
    texture or using jumbo textures. The O2 also has the benefit of having very
    cheap and available video IO.

    The Octane can be a bit noisy if the ambient temperature is above 25c, otherwise
    it's not too bad.


    Cheers - Tony 'Nicoya' Mantler

    --
    Tony 'Nicoya' Mantler -- Master of Code-fu -- nicoya@ubb.ca
    -- http://nicoya.feline.pp.se/ -- http://www.ubb.ca/ --

  8. Re: R10000-150mHz_vs_R12000-300mHz...(?)

    In comp.sys.sgi.apps arie van schutterhoef wrote:
    >
    >> You can read Ian's CPU performance tables for examples showing how the FP
    >> performance especially is completely neutered in the R10k O2 compared to the
    >> exact same CPU in the Octane or even the Indigo2:
    >>
    >>

    > -I've read them some time ago and they give a general
    > indication, but I'm not so sure to what extent they relate
    > to audio processing. They all seem to graphics -oriented.
    > Though maybe it's better to speak in general signal processing
    > terms.
    >


    Signal processing will likely involve FFTs, and these benefit greatly from
    a decent sized cache. The Octanes will cope much more easily than the
    O2. They also have the digital audio as a standard rather than an extra.

    If you are reading or writing to disk I have always found the Octanes to
    be better are shoving large amounts of data across the system, so that
    would be another advantage.

    Either one has good quality audio hardware.


    >> If you don't use the UMA features in the O2 then you're *much* better off
    >> with an Octane.

    > -I presume with UMA's strengths, you talk of the I/O benefits?
    > As far as being better of: the o2 is going to be used for
    > concerts and an Octane is far too noisy with all these fans.
    > The o2 is much quieter, so otherwise the Octane would be the favorite.
    >
    > AvS


    If you are worried about the noise, you could always put it in a
    (ventilated) box. The Octane is not too bad, really. Not silent, but not
    unbearable. They do weigh quite a bit more than the O2. Never fun to shift
    them about.


    --
    Dr Tristram J. Scott
    Energy Consultant

  9. Re: R10000-150mHz_vs_R12000-300mHz...(?)

    Tony 'Nicoya' Mantler writes:

    > The Octane can be a bit noisy if the ambient temperature is above 25c,
    > otherwise it's not too bad.


    I guess that means the sensors are off on every single Octane I've ever
    laid hands on (that's four R12K 300 MHz ones and seven R10K ones between
    175 and 250 MHz, one of them dual CPU). They are noisy as hell at our
    normal room temp of 20'C or so. I'd hate to learn that's the quiet mode.

    How many fans are there in the Octane? Are they reasonably standard
    components and is it possible to replace them with quieter ones of an
    otherwise identical spec? I'm not afraid of getting my hands dirty,
    particularly with the one I've got at home.

    --
    Atro Tossavainen (Mr.) / The Institute of Biotechnology at
    Systems Analyst, Techno-Amish & / the University of Helsinki, Finland,
    +358-9-19158939 UNIX Dinosaur / employs me, but my opinions are my own.
    < URL : http : / / www . helsinki . fi / %7E atossava / > NO FILE ATTACHMENTS

  10. Re: R10000-150mHz_vs_R12000-300mHz...(?)

    In comp.sys.sgi.apps Atro Tossavainen
    wrote:
    >
    > How many fans are there in the Octane? Are they reasonably standard
    > components and is it possible to replace them with quieter ones of an
    > otherwise identical spec? I'm not afraid of getting my hands dirty,
    > particularly with the one I've got at home.
    >


    They have three fans. The one on the PSU is easy to replace if you have
    the Cherokee one with the silver handle. On the older ones (with the black
    handle) it requires disassembly of the PSU. As for finding a quieter one
    with the same spec, I have not managed to do so with any level of success.
    I put in one made by Y.S. Tech, FD12387S-1I. Supposedly a bit quieter, and
    with slightly greater throughput. When installed it turned out to be
    louder than the Panaflo FBA12G12U1A I was replacing.

    I think the problem was that the noise is not just from the fan itself, but
    from all the air movement throughout the machine. The new fan had a
    slightly greater airflow, and hence (I guess) greater total noise.

    In the end I stuck it in a corner with a pillow over it to muffle it, but
    still allow sufficient airflow that it doesn't overheat.

    Given that mine doesn't run with fast fan ever (dual 250, SSI, 1.5GB RAM) I
    might perhaps drop in a lower spec fan. If it overheats then it can always
    go to fast fan mode...

    The topic of replacing fans in the Octane has been discussed many times in
    these groups. I'm not aware of anyone reporting astounding success, with
    the exception of people who stuck the thing in a padded box.

    --
    Dr Tristram J. Scott
    Energy Consultant

  11. Re: R10000-150mHz_vs_R12000-300mHz...(?)

    In article ,
    Atro Tossavainen wrote:

    : Tony 'Nicoya' Mantler writes:
    :
    : > The Octane can be a bit noisy if the ambient temperature is above 25c,
    : > otherwise it's not too bad.
    :
    : I guess that means the sensors are off on every single Octane I've ever
    : laid hands on (that's four R12K 300 MHz ones and seven R10K ones between
    : 175 and 250 MHz, one of them dual CPU). They are noisy as hell at our
    : normal room temp of 20'C or so. I'd hate to learn that's the quiet mode.

    I assume you've checked to be sure that the honeycomb vents aren't blocked.

    My Octane is a dual 250 with SE graphics and the PCI shoebox. It should be noted
    that installing some options will force the system into fastfan mode, which
    overrides the temperature sensors. (I think the MXI/MXE is the most common
    culprit)


    : How many fans are there in the Octane? Are they reasonably standard
    : components and is it possible to replace them with quieter ones of an
    : otherwise identical spec? I'm not afraid of getting my hands dirty,
    : particularly with the one I've got at home.

    The fans are all standard parts. As I recall, the power supply fan is a 120x25,
    the drive cage fan is an 80x25, and the frontplane fan is a 60x10.

    There's two things that cause fan noise: bearing noise and turbulence noise.
    Sadly the Octane mostly suffers from the latter, so upgrading to higher quality
    fans won't buy you any silence.

    The hard drives in most Octanes will tend to be fairly noisy compared to more
    modern units. You might want to try powering on the system with the drives
    removed to see if it would be worth replacing them.

    Even in ideal conditions the Octane is pretty far from the modern standard of
    quiet computers. It is, however, quite livable if it's kept cool and out of
    fastfan mode.


    Cheers - Tony 'Nicoya' Mantler

    --
    Tony 'Nicoya' Mantler -- Master of Code-fu -- nicoya@ubb.ca
    -- http://nicoya.feline.pp.se/ -- http://www.ubb.ca/ --

  12. Re: R10000-150mHz_vs_R12000-300mHz...(?)

    removed to see if it would be worth replacing them.
    >
    > Even in ideal conditions the Octane is pretty far from the modern standard of
    > quiet computers.

    -Which is a point well made. Still the Octane is a wonderfull machine
    considering it's age and for that time quite revolutionary for
    a desktop workstation. I've one in my working place where it
    can produce all the noise it wants, but you can't use it
    on stage, with a (albeit closely miked, thus amplified)
    violin playing delicate noises (which are analysed by a SGI
    machine and subsequently send out control signals with OSC to
    a Apple G4 for signal procesing).
    So probably it end up being an upgraded o2, conected to
    other machines over a local network. One option still
    under consideration is putting one or more Octanes backstage
    and connections the desired I/O throught ADAT and ethernet
    to the appropriate places. But that is the next item on the menu...

    avS

  13. Re: R10000-150mHz_vs_R12000-300mHz...(?)

    Tony 'Nicoya' Mantler writes:

    > I assume you've checked to be sure that the honeycomb vents aren't blocked.


    Yes.

    > My Octane is a dual 250 with SE graphics and the PCI shoebox. It should be noted
    > that installing some options will force the system into fastfan mode, which
    > overrides the temperature sensors. (I think the MXI/MXE is the most common
    > culprit)


    We don't have MXI/MXE on any of them. My home machine is a single 175
    with SE graphics and no shoebox.

    > There's two things that cause fan noise: bearing noise and turbulence noise.
    > Sadly the Octane mostly suffers from the latter, so upgrading to higher quality
    > fans won't buy you any silence.


    Right. Won't bother then.

    > The hard drives in most Octanes will tend to be fairly noisy compared to more
    > modern units. You might want to try powering on the system with the drives
    > removed to see if it would be worth replacing them.


    I can recognize drive noise from air flow noise. :-)

    --
    Atro Tossavainen (Mr.) / The Institute of Biotechnology at
    Systems Analyst, Techno-Amish & / the University of Helsinki, Finland,
    +358-9-19158939 UNIX Dinosaur / employs me, but my opinions are my own.
    < URL : http : / / www . helsinki . fi / %7E atossava / > NO FILE ATTACHMENTS

  14. Re: R10000-150mHz_vs_R12000-300mHz...(?)


    arie van schutterhoef wrote:
    > -I've read them some time ago and they give a general
    > indication, but I'm not so sure to what extent they relate
    > to audio processing. They all seem to graphics -oriented.
    > Though maybe it's better to speak in general signal processing
    > terms.


    The SPECfp95 benchmarks don't focus on graphics tasks at all. They're
    mostly numerical number crunching of various kinds. One test uses 32bit
    fp I think, but if I recall correctly it's not the MADD-style math
    found in geometry/lighting, which is what R5K is good at.


    > -I presume with UMA's strengths, you talk of the I/O benefits?


    Depends on what you're doing. In some cases, yes. O2 is good at
    complex media tasks, video as texture is cheap, accelerated imaging,
    etc.

    For example, I know one place which used an O2 to help do a live launch
    of a new mobile model. A live feed of the boss speaking about the model
    was mapped onto the 'screen' of a 3D real-time model of the mobile,
    which was displayed on a big screen. O2 is ideal for this sort of
    thing.


    > As far as being better of: the o2 is going to be used for
    > concerts and an Octane is far too noisy with all these fans.
    > The o2 is much quieter, so otherwise the Octane would be the

    favorite.

    Indeed, O2 has the noise advantage, and certain models of disk are VERY
    quiet indeed. I have a 9GB SCA in my gateway Indy which is so damn
    quiet
    that sometimes I have to walk over & listen just to be sure it's still
    on!

    Cheers!

    Ian.

    SGI Depot: http://www.futuretech.blinkenlights.nl/sgidepot/
    Email: mapesdhs@yahoo.com (eBay ID: mapesdhs)
    Backup email (send copy to this too): sgidepot@blueyonder.co.uk
    Home: +44 (0)131 476 0796 (best to call this number first)
    Mobile: 07743 495403 (usually off; leave a message and I'll call back)

    SGI/Future Technology/N64: http://www.futuretech.blinkenlights.nl/
    Doom Help Service (DHS): http://www.gamers.org/dhs/


  15. o2 upgrade (was:R10000-150mHz_vs_R12000-300mHz)

    Sometime ago I asked some questions about the validity of upgrading an
    o2 with a 150 mHz R10000 uppwards to R12000 300 mHz. Which was clearly
    answered that it was. In the meantime I managed to secure a R12000 at
    400 mHz. Now the point is that SGI Document number 007-3725-002:
    http://techpubs.sgi.com/library/manu.../pdf/007-3725-
    02.pdf published: 2000-08-07
    about this cpu upgrade mentions: 'CPU support (P/N 050-0347-002) for
    use with the O2 R12000A' and 'IRIX Overlay CD that shipped with this
    R12000 CPU hardware'.
    This R12000/400mHz doesn't come with any CD or software whatsoever....
    The question is: how necessary is this software, considering the
    publication-date of that document-2000-08-07- and the fact that this
    particular o2 runs 6.5.12, which is of a later date and might cover
    this
    upgrade software already. Before going about doing this upgrade I would
    like know more about this the likely problems I could face.

    AvS

  16. Re: o2 upgrade (was:R10000-150mHz_vs_R12000-300mHz)

    arie van schutterhoef writes:

    > The question is: how necessary is this software, considering the
    > publication-date of that document-2000-08-07- and the fact that this
    > particular o2 runs 6.5.12, which is of a later date and might cover
    > this
    > upgrade software already. Before going about doing this upgrade I would
    > like know more about this the likely problems I could face.


    My guess is that the overlay that it speaks of applied to IRIX 6.3 and
    that running any variety of IRIX 6.5 is quite sufficient to get it going.

    --
    Atro Tossavainen (Mr.) / The Institute of Biotechnology at
    Systems Analyst, Techno-Amish & / the University of Helsinki, Finland,
    +358-9-19158939 UNIX Dinosaur / employs me, but my opinions are my own.
    < URL : http : / / www . helsinki . fi / %7E atossava / > NO FILE ATTACHMENTS

  17. Re: o2 upgrade (was:R10000-150mHz_vs_R12000-300mHz)

    In article , Atro Tossavainen
    wrote:
    > My guess is that the overlay that it speaks of applied to IRIX 6.3 and
    > that running any variety of IRIX 6.5 is quite sufficient to get it going.

    -Thanks for replying in the first place!I had the feeling that might be
    the case indeed. But nice that also some one else confirms this
    assumption. Do you have experience with doing these kinds of upgrades?

    AvS

  18. Re: o2 upgrade (was:R10000-150mHz_vs_R12000-300mHz)

    On 2005-05-23 16:41:32 +0100, arie van schutterhoef said:

    > In article , Atro Tossavainen
    > wrote:
    >> My guess is that the overlay that it speaks of applied to IRIX 6.3 and
    >> that running any variety of IRIX 6.5 is quite sufficient to get it going.

    > -Thanks for replying in the first place!I had the feeling that might be
    > the case indeed. But nice that also some one else confirms this
    > assumption. Do you have experience with doing these kinds of upgrades?
    >
    > AvS


    My understanding is that Irix 6.5 supports R12K out of the box.

    But even so, the age of that document is August 2000, and your overlay
    (6.5.12) is May 2001... if you need such an update, it's there already.
    The 'overlay' would be your 6.5.x release. I don't ever recall them
    mentioning adding 400MHz support to the overlays, so they might be just
    saying this as a precaution.

    Chris


  19. Re: o2 upgrade (was:R10000-150mHz_vs_R12000-300mHz)



    Chris Moxon wrote:
    > My understanding is that Irix 6.5 supports R12K out of the box.


    A suitable 6.5.x version does (6.5.6 or later I think, not sure),
    but the original June 1998 6.5 does not support R12K AFAIK.

    Ian.

    SGI Depot: http://www.futuretech.blinkenlights.nl/sgidepot/
    Email: mapesdhs@yahoo.com (eBay ID: mapesdhs)
    Backup email (send copy to this too): sgidepot@blueyonder.co.uk
    Home: +44 (0)131 476 0796 (best to call this number first)
    Mobile: 07743 495403 (usually off; leave a message and I'll call back)

    SGI/Future Technology/N64: http://www.futuretech.blinkenlights.nl/
    Doom Help Service (DHS): http://www.gamers.org/dhs/


  20. o2 upgrade-continued... (was:R10000-150mHz_vs_R12000-300mHz)

    > My understanding is that Irix 6.5 supports R12K out of the box.
    > But even so, the age of that document is August 2000, and your overlay
    > (6.5.12) is May 2001... if you need such an update, it's there already.
    > The 'overlay' would be your 6.5.x release. I don't ever recall them
    > mentioning adding 400MHz support to the overlays, so they might be just
    > saying this as a precaution.

    -In this continuing saga of upgrading:
    does anyone know, whether extra software is necessary for
    adding a R12000/400mHz to the o2 hardware.
    Busy installing 6.5.27 and do I need to do this in 'maintenance',
    or is Irx clever enough to this on its own?

    AvS

+ Reply to Thread
Page 1 of 2 1 2 LastLast