Re: [ARMini-support] ARMini vs Iyonix

  • From: Chris Hall <chris@xxxxxxxxxx>
  • To: armini-support@xxxxxxxxxxxxx
  • Date: Wed, 13 Jun 2012 16:12:49 GMT

In message <99ad419f52.chris@xxxxxxxxxxxxxxxx>
          Chris Hall <chris@xxxxxxxxxx> wrote:

> In message <cd4ece9e52.martin@xxxxxxxxxxxxxxxxxxx>
>           Martin Wuerthner <lists@xxxxxxxxxxxxxxx> wrote:

>> In message <9c7cc09e52.chris@xxxxxxxxxxxxxxxx>
>>           Chris Hall <chris@xxxxxxxxxx> wrote:

>>> In message <9ca6a39e52.martin@xxxxxxxxxxxxxxxxxxx>
>>>           Martin Wuerthner <lists@xxxxxxxxxxxxxxx> wrote:

>>>> This is an intriguing puzzle. This particular ArtWorks file is not
>>>> even unusually complex. The only obvious performance weakness of the
>>>> ARMini compared to an Iyonix is disc access speed, and that is
>>>> irrelevant here. For all I have been told, the ARMini should
>>>> outperform the Iyonix in any area relevant to this task, in some even
>>>> by a large margin, so there should be a considerable speed-up. Even
>>>> not having a speed-up would have been surprising, but a dramatic
>>>> slowdown is absolutely unexpected. I have a suspicion what could cause
>>>> this slowdown. I will run some tests and report back.

>>> The actual file (WorldMap_500BC) loads in seven seconds from RAMFS on
>>> the Iyonix (Artworks 2.X.00) and in six seconds from the RAMFS on the
>>> Raspberry Pi and ARMimi 1000MHz (Artworks 2.XI.01). Not sure what is
>>> intriguing.

>> We were not discussing loading speed. We were discussing rendering
>> speed. It is intriguing that - assuming Nick's measurements are
>> correct and can be reproduced - the ARMini is three times slower to
>> render this ArtWorks file than the Iyonix, even though it ought to be
>> much faster than the Iyonix according to its specifications.

>> Martin
> I am getting consistent results for rendering (press f12 then return
> and time it) as follows:
> Pandaboard 1500 MHz 3.5s
> Rapsberry Pi 700MHz 4.0s
> ARMini 1000MHz      5.5s
> Iyonix 600MHz      13.0s

> Seeing as the Pi has a video processor these look consistent.
Ah! But I now see that subsequent renders (not the second or third 
strangely) on the Iyonix are down to 4.5s so it must be something to 
do with memory buffering/cacheing.


-- 
Chris Hall <chris@xxxxxxxxxx>
---
To alter your preferences or leave the group, 
visit //www.freelists.org/list/armini-support
List-related queries to info@xxxxxxxxxxxx

Other related posts: