<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;" class="">
<font face="Calibri" style="font-size: 14px;" class="">Vulkan allows the application to decide whether it wants the allocation to be host visible and device local. </font>
<div class=""><font face="Calibri" style="font-size: 14px;" class="">If we drop the flag, what will happen if we did not set the NO_CPU_ACCESS flag?</font></div>
<div class=""><font face="Calibri" style="font-size: 14px;" class="">Will it fail the map in case the allocation could be placed in invisible heap then?</font></div>
<div class=""><font face="Calibri" style="font-size: 14px;" class=""><br class="">
</font></div>
<div class=""><font face="Calibri" class=""><span style="font-size: 14px;" class="">Thanks.</span></font></div>
<div class=""><font face="Calibri" style="font-size: 14px;" class="">Best Regards,</font></div>
<div class=""><font face="Calibri" style="font-size: 14px;" class="">David</font><br class="">
<div>
<blockquote type="cite" class="">
<div class="">On 29 Jun 2017, at 11:03 PM, Marek Olšák <<a href="mailto:maraeo@gmail.com" class="">maraeo@gmail.com</a>> wrote:</div>
<br class="Apple-interchange-newline">
<div class="">
<div class="">Do you have any concern if we also stop using the CPU_ACCESS flag on radeon?<br class="">
<br class="">
Thanks,<br class="">
Marek<br class="">
<br class="">
On Thu, Jun 29, 2017 at 4:51 PM, Christian König<br class="">
<<a href="mailto:deathsimple@vodafone.de" class="">deathsimple@vodafone.de</a>> wrote:<br class="">
<blockquote type="cite" class="">Yeah, I was thinking something similar.<br class="">
<br class="">
See the intention behind CPU_ACCESS_REQUIRED is to always guarantee that CPU<br class="">
access is immediately possible.<br class="">
<br class="">
If you ask me that is not really useful for the UMD and was never meant to<br class="">
be used by Mesa (only the closed source UMD and some kernel internal use<br class="">
cases).<br class="">
<br class="">
I would like to keep the behavior in the kernel driver as it is, but we<br class="">
should really stop using this as a hint in Mesa.<br class="">
<br class="">
Regards,<br class="">
Christian.<br class="">
<br class="">
<br class="">
Am 29.06.2017 um 16:41 schrieb Marek Olšák:<br class="">
<blockquote type="cite" class=""><br class="">
Hi,<br class="">
<br class="">
Given how our memory manager works and the guesswork that UMDs have to<br class="">
do to determine whether to set the flag, I think the flag isn't<br class="">
useful.<br class="">
<br class="">
I'm proposing that CPU_ACCESS_REQUIRED:<br class="">
- will be deprecated.<br class="">
- It will remain to be accepted by the kernel driver, but it will<br class="">
either not have any effect, or it will serve as a hint that might or<br class="">
might not be followed.<br class="">
- The only flag that UMDs are expected to set with regard to CPU<br class="">
access is NO_CPU_ACCESS.<br class="">
<br class="">
The main motivation is the reduction of "virtual" heaps for UMD buffer<br class="">
suballocators and reusable buffer pools. A higher number of heaps<br class="">
means that more memory can be wasted by UMDs.<br class="">
<br class="">
Opinions?<br class="">
<br class="">
Thanks,<br class="">
Marek<br class="">
_______________________________________________<br class="">
amd-gfx mailing list<br class="">
<a href="mailto:amd-gfx@lists.freedesktop.org" class="">amd-gfx@lists.freedesktop.org</a><br class="">
https://lists.freedesktop.org/mailman/listinfo/amd-gfx<br class="">
</blockquote>
<br class="">
<br class="">
<br class="">
</blockquote>
_______________________________________________<br class="">
amd-gfx mailing list<br class="">
<a href="mailto:amd-gfx@lists.freedesktop.org" class="">amd-gfx@lists.freedesktop.org</a><br class="">
https://lists.freedesktop.org/mailman/listinfo/amd-gfx<br class="">
</div>
</div>
</blockquote>
</div>
<br class="">
</div>
</body>
</html>