[ntp:questions] Anyone ever consider running NTP in an nVidia GPU core?
tglassey at earthlink.net
Wed Dec 2 02:02:45 UTC 2009
David Mills wrote:
> Yes, my friends have tripped over all kinds of general applications
> hosted in a GPU. One of the obvious things is onboard space mission
> navigation. But, the darn things run too hot.
That's what water coolers are for. I have three very high end ones here
running and they stay comfy... Still I understand totally about the
temperature issues. But hey, if we come up with a model which would
allow nVidia to ship an optional co-processor/NTP Service model they
might adopt us (snicker) - they love this stuff and a real-time
co-processor built specifically to sidestep the pain of the
Southbridge's IO bottlenecks like the cores do, rocks.
By the way - this NTPStamp thing is getting traction in one of the
evidence groups - we have come up with a process through timestampting
to allow company's to get rid of the excess baggage SOX compliance made
them add in their roles and responsibilities models and that turns out
to be totally cool...
BTW we are working with one of the risk management groups at BofA and
they are going through all the pain of absorbing Merrill which brings me
to ask you who you worked with there. I was working with Jack Cavenaugh
in compliance and the managing partner for silicon valley, pat williams
in taking secured and audited time into their compliance model.
> Todd Glassey wrote:
>> David - I was just looking at a real-time control process a friend
>> had implemented for a controller which is handling some very fast
>> nuclear events and parts of the sensor code is actually running in an
>> nVidia GPU core and it hit me, what a neat timing server it would
>> make. I bring this up because of the other applications for the GPU's
>> (cryptography) as well.
>> The compilers are all there and it will keep time in the embedded GPU
>> module with an external reference even better. Would be an
>> interesting project.
> No virus found in this incoming message.
> Checked by AVG - www.avg.com
> Version: 8.5.426 / Virus Database: 270.14.89/2539 - Release Date: 12/01/09 19:32:00
More information about the questions