Just one question: _what_ load? CPUs are already underutilized as they
stand. And this is getting worse as we add more cores.
Cheers,
Cliff
--
The Internet is interesting in that although the nicknames may change,
the same old personalities show through.
My take on this is that the GPU's are getting so powerful, that they are
often "idling" so lets get the CPU and the GPU working together and really
blow things out of the water as we know it
> On 2009-11-28, Lawrence D'Oliveiro <l...@geek-central.gen.new_zealand>
> wrote:
>
>> <http://blogs.zdnet.com/hardware/?p=6289>: “...because increasingly
>> software developers are looking to the GPU to take the load off the CPU.”
>>
>> Just one question: _what_ load? CPUs are already underutilized as they
>> stand. And this is getting worse as we add more cores.
>
> My take on this is that the GPU's are getting so powerful, that they are
> often "idling" so lets get the CPU and the GPU working together and really
> blow things out of the water as we know it
But if they’re so powerful that they’re spending so much of their time
idling, then making them more powerful will mean they spend even more time
idling.
Also programme like SETI can utilise some GPUs and often the GPU can do 10x
the work of the CPU due largely to the way they're designed.
--
Shaun.
"Give a man a fire and he's warm for the day. But set fire to him and he's
warm for the rest of his life." Terry Pratchet, 'Jingo'.
> Somewhere on teh intarwebs Enkidu wrote:
> > Lawrence D'Oliveiro wrote:
> >> <http://blogs.zdnet.com/hardware/?p=6289>: "...because increasingly
> >> software developers are looking to the GPU to take the load off the
> >> CPU." Just one question: _what_ load? CPUs are already
> >> underutilized as they stand. And this is getting worse as we add
> >> more cores.
> >>
> > High power gaming.
>
> Also programme like SETI can utilise some GPUs and often the GPU can
> do 10x the work of the CPU due largely to the way they're designed.
Hi
Yup, that's why I run the cuda driver....
CUDA Device Query (Driver API) statically linked version
There is 1 device supporting CUDA
Device 0: "GeForce 8600 GTS"
CUDA Driver Version: 2.30
CUDA Capability Major revision number: 1
CUDA Capability Minor revision number: 1
Total amount of global memory: 268107776 bytes
Number of multiprocessors: 4
Number of cores: 32
Total amount of constant memory: 65536 bytes
Total amount of shared memory per block: 16384 bytes
Total number of registers available per block: 8192
Warp size: 32
Maximum number of threads per block: 512
Maximum sizes of each dimension of a block: 512 x 512 x 64
Maximum sizes of each dimension of a grid: 65535 x 65535 x 1
Maximum memory pitch: 262144 bytes
Texture alignment: 256 bytes
Clock rate: 1.46 GHz
Concurrent copy and execution: Yes
Run time limit on kernels: Yes
Integrated: No
Support host page-locked memory mapping: No
Compute mode: Default (multiple host
threads can use this device simultaneously)
BOINC then uses it as a coprocessor....
--
Cheers Malcolm ��� (Linux Counter #276890)
SUSE Linux Enterprise Desktop 11 (x86_64) Kernel 2.6.27.37-0.1-default
up 1 day 2:45, 2 users, load average: 0.29, 0.30, 0.35
GPU GeForce 8600 GTS Silent - CUDA Driver Version: 190.18
ease? Its hard to make applications truly multi-threaded and take
advantage of lots of cores....and 8 cores will be here in 2010. Its
interesting, at one stage Intel was talking of 10Ghz and beyond, now
after that (marketing hype) flopped its 80 cores and beyond. The idea
(or part) of the 80 cores is that some will be GPUs on the one
die....and you vary the quantities of each type of core depending on
the CPU's base unit's use...
I suspect that this all points to limits of humans as programmers (and
possibly return on investment, ie man hours) to get multi-core general
purpose CPUs to do what's wanted/needed....to compensate there will be
co-processors and lots of them that get called....Another indication
of this was AMD(?) commenting that their goal was to get (massively)
multi-cored CPUs to appear as one or two to the OS/application layer.
So, eealistically one CPU at 3~3.6Ghz and 4 cores would probably be
the upper limit of whats achievable for a desktop (maybe 4Ghz and 8
cores, tops)....then we have the human limitations...so you have an
envelope you have to stay inside....
regards
SSD is a big leap forward on that....at the moment its high tech
mechanical devices with their physical limits v chips and a
circuitboard...the latter to me with the huge speed advantage as well
seems the future.
regards
Great so we know in some remote spot many years ago there was some
dumb ass playing with an expensive radio....I'd rather save the CO2
emisions here thanks.
;]
regards
> SSD is a big leap forward on that..
Not if you’re accessing it through the same old SATA interface.
> On Nov 28, 2:33 pm, Lawrence D'Oliveiro <l...@geek-
> central.gen.new_zealand> wrote:
>
>> Just one question: _what_ load? CPUs are already underutilized as they
>> stand. And this is getting worse as we add more cores.
>
> ease? Its hard to make applications truly multi-threaded and take
> advantage of lots of cores...
It’s not exactly easy to do SIMD programming of GPUs either.
There is lots of work being done to make it easier to write software with
concurrency though.
On two fronts, the rising popularity of languages that support functional
programming, immutability etc. that already lend themselves well to
concurrency, and also at a library level, Grand Central Dispatch for OS X,
Parallel Extensions for .NET et al.
On a related note some recent games coming out for PS3 make excellent use of
the multi core architecture, so at least a subset of the development
community seems to have gotten its collective head around it.
> Just one question: _what_ load?
A question underscored by the death of Larrabee
<http://blogs.zdnet.com/hardware/?p=6383>.