| Thread overview |
|---|
January 10, 2012 Re: Vector performance | ||||
|---|---|---|---|---|
| ||||
Attachments:
| awesome sauce... Incidently I got given a 48MHz arm dev board today....
On Wed, Jan 11, 2012 at 12:14 AM, Manu <turkeyman@gmail.com> wrote:
> Just thought I might share a real-life case study today. Been a lot of talk of SIMD stuff, some people might be interested.
>
> Working on an android product today, I noticed the matrix library was
> burning a ridiculous amount of our frame time.
> The disassembly looked like pretty normal ARM float code, so rewriting a
> couple of the key routines to use the VFPU (carefully), our key device
> moved from 19fps -> 34fps (limited at 30, we can now ship).
> GalaxyS 2 is now running at 170fps, and devices we previously considered
> un-viable can now actually get a release! .. Most devices saw around 25-45%
> speed improvement.
>
> Imagine if all vector code throughout was using the vector hardware
> nicely, and not just one or 2 key functions...
> Getting the API right (intuitively encouraging proper usage and
> disallowing inefficient operations), it'll make a big difference!
>
| |||
January 10, 2012 Re: Vector performance | ||||
|---|---|---|---|---|
| ||||
Attachments:
| On 10 January 2012 16:35, Danni Coy <danni.coy@gmail.com> wrote:
> awesome sauce... Incidently I got given a 48MHz arm dev board today....
Sweet, what's your plan for it? :)
These 'embedded' (should probably read 'non-x86 system', since 'embedded' is quickly becoming 'standard') systems really do benefit a lot more from good SIMD implementation than x86.
| |||
January 10, 2012 Re: Vector performance | ||||
|---|---|---|---|---|
| ||||
Attachments:
| On Wed, Jan 11, 2012 at 12:42 AM, Manu <turkeyman@gmail.com> wrote: > On 10 January 2012 16:35, Danni Coy <danni.coy@gmail.com> wrote: > >> awesome sauce... Incidently I got given a 48MHz arm dev board today.... > > > Sweet, what's your plan for it? :) > It came with a 3 axis accelerometer and a 2 axis gyro.... Haven't not exactly sure what I will do with it -- maybe some form of musical instrument. > > These 'embedded' (should probably read 'non-x86 system', since 'embedded' is quickly becoming 'standard') systems really do benefit a lot more from good SIMD implementation than x86. > | |||
January 10, 2012 Re: Vector performance | ||||
|---|---|---|---|---|
| ||||
Attachments:
| On 10 January 2012 16:48, Danni Coy <danni.coy@gmail.com> wrote:
> On Wed, Jan 11, 2012 at 12:42 AM, Manu <turkeyman@gmail.com> wrote:
>
>> On 10 January 2012 16:35, Danni Coy <danni.coy@gmail.com> wrote:
>>
>>> awesome sauce... Incidently I got given a 48MHz arm dev board today....
>>
>>
>> Sweet, what's your plan for it? :)
>>
>
> It came with a 3 axis accelerometer and a 2 axis gyro....
> Haven't not exactly sure what I will do with it -- maybe some form of
> musical instrument.
>
Well that sounds like a fair start :) .. That, or a wild kinky bedroom toy...
| |||
Copyright © 1999-2021 by the D Language Foundation
Permalink
Reply