[Cuis-dev] VectorGraphics and ARM64

Eric Gade eric.gade at gmail.com
Sun Nov 21 16:25:26 PST 2021

Hi Ken,

On Sun, Nov 21, 2021 at 6:12 PM <ken.dickey at whidbey.com> wrote:

> This sounds like the plugin is not being loaded.
> In a workspace, what value is returned for
>      VectorEngineWithPlugin isPluginAvailable.
> ?

This was my initial thought too, but it's responding true unfortunately.

On Sun, Nov 21, 2021 at 5:25 PM Phil B via Cuis-dev <cuis-dev at lists.cuis.st>

> VectorGraphics is going to be slow(ish) on any ARM SBC...

That's what I was afraid of! Oh well

On Sun, Nov 21, 2021 at 6:35 PM Phil B via Cuis-dev <cuis-dev at lists.cuis.st>

> Ken,
> On Sun, Nov 21, 2021 at 6:24 PM <ken.dickey at whidbey.com> wrote:
>> On 2021-11-21 14:24, Phil B via Cuis-dev wrote:
>> > VectorGraphics is going to be slow(ish) on any ARM SBC or phone
>> > unless/until it is OpenCL/Vulkan accelerated and maybe even then.
>> Juan can clarify, but Cuis SVG does _not_ use the GPU because setting up
>> the graphics pipeline has too much overhead to be useful.
> That's entirely possible, I was just making the statement that short of
> that it simply isn't possible to even get in the realm of x86 performance
> as GPU acceleration is how the majority of applications get there on ARM.
>> > It's a very computationally intensive subsystem (at least an order of
>> > magnitude more so than raster graphics) and most ARM mobile chips (i.e.
>> > pretty much everything non-Apple) are generally 1/10-1/20th the speed
>> > of a fast x86 computer as they are typically inferior in every respect:
>> > manufactured on larger process nodes, lower clock speeds, lower bus
>> > bandwidth, weaker floating point units, smaller and slower cache,
>> > slower RAM etc etc.
>> I don't want a flame war, but I have always hated Intel's "push pop"
>> architecture and prefer RISCs.  I get the feeling that on Intel I can't
>> run but must hop and Intel compensates by letting me hop very fast.
>> Yes, I have written assembler for a number of architectures (68k, 88k,
>> ARMs, RISC-V..).  Intel over the years has had quite a number of
>> floating point hardware bugs compared to other processor architectures.
>> MIPS per WATT power consumption is horrible. 'Nuf said.  Different
>> measures of goodness.  Please, forgive me and let's move on..  Thanks
> I'm not much of an x86 fan myself, but unfortunately it's the high
> performance platform... for now.  I am encouraged that Apple is showing the
> industry that it (high performance non-x86) can be done and hope at some
> point it snaps out of its x86 coma.  That said, I'd rather see an open ISA
> like RISC-V really take off but right now that's more a dream than
> reality... it's x86 in most of the (non-Apple) PC world and ARM everywhere
> else.  That said, I literally just ordered my new CPU today (Ryzen) as
> apparently my Sandy Bridge i7 is no longer state of the art (or even enough
> to do what I need) and the industry still isn't making the CPUs I really
> want... ;-)
>> Good on ya,
>> -KenD
> Thanks,
> Phil
> --
> Cuis-dev mailing list
> Cuis-dev at lists.cuis.st
> https://lists.cuis.st/mailman/listinfo/cuis-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cuis.st/mailman/archives/cuis-dev/attachments/20211121/57e0bb6e/attachment-0001.htm>

More information about the Cuis-dev mailing list