<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 12pt;
font-family:Calibri
}
--></style></head>
<body class='hmmessage'><div dir='ltr'>Mike and Gerhard, <div>dont think the situation of the pcie powerpc is bettrer.</div><div>but compared with last years with the new kernels and last xorg on a radeonhd 4650 </div><div>i have an increase of performance about 250x ... </div><div>example QuakeSpasm was gaving 640x480157fps on Radeon 4650... Now is 380 fps </div><div>yes compared the old nvida 7800gtx on Osx 450 fps this results are less but for sure better than before.</div><div><br></div><div>The worst of last period im facing r600 radeon ring test errors and i cant use</div><div>with gpu accel now only in fbdev the 5450 and 6570 that it was perfect working before.</div><div><br></div><div>Luigi<br><br><div>> From: gerhard_pircher@gmx.net<br>> To: michael.heltne@gmail.com<br>> Subject: Re: PowerPC agpmode issues<br>> Date: Tue, 9 Feb 2016 12:52:15 +0100<br>> CC: aneesh.kumar@linux.vnet.ibm.com; michel@daenzer.net; linuxppc-dev@lists.ozlabs.org; reinhard.boris@googlemail.com; bobby.prani@gmail.com<br>> <br>> > On 9 Feb 2016 03:27, "Mike" <michael.heltne@gmail.com> wrote:<br>> > Ok, so its quirks to be added then? Something not implemented in KMS<br>> > that was in UMS?<br>> > Reports are that the same issue exsist on PPC Amiga Ones with a VIA<br>> > chipset, and the Pegasos 2 with the Artica s chipset, i posted a<br>> > mail from detailiing that.<br>> Just to avoid some confusion:<br>> Old long story short: the issues for AmigaOnes and the Pegasos _1_ with<br>> ArticiaS northbridge and VIA southbridge are that:<br>> 1. the AGP controller corrupts data transfers in AGP mode (also depending<br>> on the AGP HW request queue size). So there is no official AGP driver that<br>> would require radeon.agpmode=-1. The microA1 is supposed to have a fix<br>> for this HW data corruption, but I yet have to dig out my ArticiaS AGP<br>> driver code for some test runs...<br>> 2. At least the AmigaOne with ArticiaS chip need non-coherent DMA<br>> allocations and/or proper cache flushes to avoid corrupted DMA transfers.<br>> <br>> Nonetheless I had DRI1 working _only_ on my A1SE under Debian Squeeze (i.e.<br>> glxgears could run on the desktop with hardware acceleration), but DRI2<br>> with its very dynamic GART mapping is a no-go on every first-gen AmigaOne<br>> machine, even if the GART driver test (radeon.test=1) runs through in<br>> PCIGART mode (could it be that it uses a more or less static GART mapping<br>> for the test?).<br>> <br>> > Sure that might be it, but i get different results trying agpmode=1-2-4,<br>> > 2 gave a noisy screen before the hard crash. i find it rather impossible<br>> > to debug at all as the crash happens so fast no logs seem to be written..<br>> > I think i would need serial...<br>> > I'd personally love nothing more then to see support restored and a<br>> > default as expected working condition ought be the minimum requirement.<br>> > I use a powerbook a1106, 5,6. With a 5,8 on the way. Those are the last<br>> > two revision powerbooks in the 15" series. In swrast they become useless,<br>> > impossible to use for any productivity. Most people trying to use linux<br>> > on ppc for personal use come in macs, with the exception of the Amiga PPC<br>> > crowd now running their amcc 440/460ex or e600 based x500/5000, all of<br>> > which have of course pci-e more cores and more threads. Yet struggle even<br>> > with regressions left and right to keep up with the single core performance<br>> > of the G4's. Sure it's pushing 10 years , but it's the only alternative<br>> > if one wishes to remain mobile.<br>> swrast definitely isn't fun on 10 years old PPC machines. Current Firefox<br>> is already slow enough on these machines... :-)<br>> <br>> > On 9 Feb 2016 02:41, "Michel Dänzer" <michel@daenzer.net> wrote:<br>> > > On 08.02.2016 22:28, Mike wrote:<br>> > > Certainly 750~800 fps in glxgears vs 3000+ in debian squeeze, i cant<br>> > > bring myself to say that it's an acceptable situation no matter how<br>> > > tired i am of the problem knowing how well the setup could do. It's<br>> > > clear that the implementation is broken for everything but x86, [...]<br>> > <br>> > Why is that? It was working fine on my last-gen PowerBook. AFAIK Darwin<br>> > / OS X never used anything but a static AGP GART mapping though, so it<br>> > seems very likely that the issues with older UniNorth revisions are<br>> > simply due to the hardware being unable to support the usage patterns of<br>> > modern GPU drivers.<br>> > <br>> > That said, if you guys have specific suggestions for a "proper"<br>> > solution, nobody's standing in your way.<br>> I have to admit that I lack the knowledge of the inner workings of the<br>> TTM/radeon code (and its TTM AGP backend) to do any useful work here.<br>> I was hoping that the TMA DMA allocator could be of any help at least for<br>> non-cache coherent machines given that (IIRC) ARM is using it together<br>> with the nuoveau driver on the TEGRA platform, but I guess that would need<br>> some modifications also on the powerpc architecture side (maybe a new<br>> non-coherent DMA allocator that is not limited to 2M virtual address space<br>> for mappings). Thus I guess a lot of things could be improved/fixed, but<br>> nowadays Linux code doesn't seem to be something for the "occasional hobby<br>> hacker". :-)<br>> <br>> regards,<br>> Gerhard<br>> _______________________________________________<br>> Linuxppc-dev mailing list<br>> Linuxppc-dev@lists.ozlabs.org<br>> https://lists.ozlabs.org/listinfo/linuxppc-dev<br></div></div> </div></body>
</html>