[Beowulf] Re: gamers: evil or just useless? ;)
steve_heaton at iinet.net.au
Wed Aug 9 17:13:39 PDT 2006
Evil *and* useless... ;)
If I see one more nice looking bit of kit (CPU, mobo etc) tested under
some gaming benchmark I'll spit! OK, there's a couple of Linux hardware
sites around and some of the bigger sites occasionally run a Linux test
but it's slim pickings.
Slashdot used to be worth reading until the gamers took over...
Somewhat more seriously, yes, I understand 'market forces'. But as Mark
and gl & hf point out catering to a specific market can be a double
edged sword. A physics coprocessor is a great idea but their cross
market ability is hampered by limiting their focus to just gaming.
Overall I suspect 'Wulf Keepers come out in front thanks to gaming
pressure. We're never likely to command anywhere near as much industry
FYI, AMD is expecting to drop the ATI brand they now own.
beowulf-request at beowulf.org wrote:
> Send Beowulf mailing list submissions to
> beowulf at beowulf.org
> To subscribe or unsubscribe via the World Wide Web, visit
> or, via email, send a message with subject or body 'help' to
> beowulf-request at beowulf.org
> You can reach the person managing the list at
> beowulf-owner at beowulf.org
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Beowulf digest..."
> Today's Topics:
> 1. gamers: evil or just useless? ;) (Mark Hahn)
> 2. Re: gamers: evil or just useless? ;) (sNAAPS eLYK)
> 3. Woodcrest Memory bandwidth (Bill Broadley)
> 4. Re: Woodcrest Memory bandwidth (Joe Landman)
> 5. Re: gamers: evil or just useless? ;) (Mark Hahn)
> Message: 1
> Date: Tue, 8 Aug 2006 18:41:14 -0400 (EDT)
> From: Mark Hahn <hahn at physics.mcmaster.ca>
> Subject: [Beowulf] gamers: evil or just useless? ;)
> To: beowulf at beowulf.org
> <Pine.LNX.4.64.0608081743090.8763 at coffee.psychology.mcmaster.ca>
> Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed
> as you know, the video card industry was taken over by gamers
> several years ago, and now the technical demands of FPS games
> call all the shots (so to speak). ATI and NV still sell "engineering"
> cards, but it's always hard to tell what their value proposition is,
> especially since they come in so much less sexy shipping boxes.
> first GPUs, now NICs? http://www.bigfootnetworks.com/ appears to be
> an actual linux-based coprocessor marketed as an offload NIC for gamers.
> weird, but true! from their whitepaper, I can't tell whether their
> performance numbers are sane (2500 UDP "calls" per second sounds very
> underwhelming...) will game vendors actually write code specific
> to this nic? obviously, a lot of effort is spent writing GPU code,
> but presumably that's because GPUs provide multiple orders of magnitude
> but having a linux coprocessor with 64M local memory and a GB port, hmm.
> imagine doing RDMA this way, or performing reductions in the coprocessor.
> or using Gamma on top of it. sort of a shame it doesn't come with 2+
> separate GB interfaces, actually...
> regards, mark hahn.
> Message: 2
> Date: Tue, 8 Aug 2006 20:40:23 -0400
> From: "sNAAPS eLYK" <3lucid at gmail.com>
> Subject: Re: [Beowulf] gamers: evil or just useless? ;)
> To: "Mark Hahn" <hahn at physics.mcmaster.ca>
> Cc: beowulf at beowulf.org
> <5a1205b30608081740q4b1b32f4n47892a52e5288ea8 at mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> As a gamer myself, and hearing about that stuff a few months ago - I
> also thought it was quite silly, useless even. I believe they are
> trying to profit from the general ignorance of most Windows users
> [read "gamers"] - however mildly beneficial the technology might be.
> Benefits, IMO, would be better seen from the technology behind
> nVidia's new onboard NIC - giving priority to certains packets, called
> FirstPacket <http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2764&p=3>.
> Before you bash gaming too much, however, I would suggest you think of
> some of the advantages that gamers bring to the industry! :P
> ->Masses of people buying hardware, encouraging competition between
> chipmakers, and keeping prices down through mass production
> ->R&D boosted by gamer's demands, leading to improvements that can be
> used elsewhere
> ->Newest hardware is tested by gamers
> gl & hf
> Message: 3
> Date: Tue, 8 Aug 2006 19:26:41 -0700
> From: Bill Broadley <bill at cse.ucdavis.edu>
> Subject: [Beowulf] Woodcrest Memory bandwidth
> To: beowulf at beowulf.org
> Message-ID: <20060809022641.GA12481 at cse.ucdavis.edu>
> Content-Type: text/plain; charset=us-ascii
> I have a theory that the 533 MHz FB dimms have a severe performance
> penalty on 1333 MHz FSB.
> Does anyone out there have a 1333 MHz FSB dual woodcrest (pick your
> vendor) with 4 667 MHz FB-dimms? A 1066 MHz FSB with 4 533 MHz dimms
> would also be interesting.
> Could you run: http://www.cs.virginia.edu/stream/ ?
> Ideally compiled with a F90 or C compiler (Intel, Pathscale, and I imagine
> many others) that support OpenMP so you can run 4 threads (one per core)
> in parallel.
More information about the Beowulf