@All

Lately we have found that RadeonHD 3.x driver on x5000 behave badly, and when you fill second 256mb of GPU memomry (first ones taken by workbench and co) it simple lockup and crashes. Some of you may remember that Huno's port of RCTW Reboorn crashes on high-details : that was it, it fill first 256mb of GPU memory and then crash.

So, now need to know is it x5000 only, or it happens in one or another way on another platofmrs. What we need there , is to check on

1). x1000
2). x5000
3). sams (any versions which have RadeonHD)
4). tabor

So, anyone, who have any of that hardware and RadeonHD in, plz, give test binary a go.

Download it plz from there:

http://kas1e.mikendezign.com/aos4/radeonhd/radeonhd_check.lha

There you will find binary "test" (together with sources, if one need to check what it do).

Binary doing only that : by usage of ogles2.library and warp3dnova it fill the GPU memory by some little textures. I.e. if you run it like this:

work:radeonhd_check:> test 300

then it will put about 300 textures whole size of which will be ~100MB , so 100MB of GPU memory will be filled.

Then if you do "test 600" , it will take then 200mb of GPU memory.

So, on x5000, once you fill first 256 and then go to the next chunk , it simple lockups. That around 765 textures.

What i need from anyone who willing to help, is to run it just like this:

test 300 : works or not (should everwhere)
test 600 : works or not (should everywhere)
test 900: works or not (now we fill already 256mb of GPU, and already at ~300)

then:

test 1600 : works or not (there we test if 512mb of GPU memory is used fine)
test 2400 : works or not (there we more than 768mb of GPU memory)
test 3100 : works or not (there we more than 1024mb of GPU memory)

So that probably will be enough to check if first 1GB of GPU video memory working/not working.

By working it mean it will after few seconds create a simple empty window, which you can then close after. No crashes or lockups should be (on x5000 it is, but let's see).

The easy way to monitor how much of GPU data is taken while example works, is to use zzd10h's Sysmon v6.3 from Z-Tools , and with hitting "auto-refresh" button in the System/system tab. That how it looks like:

http://kas1e.mikendezign.com/aos4/radeonhd/sysmon.jpg

There i filled about 2GB of GPU memory by doing "test 6000". (see field "Used GPU"). That in RadeonRX 570 (polaris 10) , which didn't lockup in compare with RadeonHD (that another point that RadeonHD probably have a bug).

Remember, those tests only for 3.x version ! Only ! And ogles2.library should be installed together with warp3dnova.library, of course.

So when doing tests plz write back there with name of your hardware, graphics card and all results from those tests.

Thanks !

Last edited by kas1e on 28-Sep-2019 at 07:43 PM.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

X5000 lockup after test 900

HD7700 (R7 250)

Last edited by Skateman on 28-Sep-2019 at 08:17 PM.

_________________
AmigaOne X5000 -> 2GHz / 16GB RAM / Radeon R7 250 / M-Audio 5.1 -> AmigaOS 4.1 FE / Ubuntu Linux
Amiga 1200 -> Recapped / 68ec020 ACA 1221ec / CF HDD / RetroNET connected to the world

@Kas1e

test 2400 on X1000 with Radeon 7700 (1GB memory) : no crash, all memory eaten during test and given back after closing the window.

The problem seems to only occur on X5000 (as was seen by Huno with RTCW).

_________________
PowerMac G5 2,7Ghz - 2GB - Radeon 9650 - MorphOS 3.14
AmigaONE X1000, 2GB, Sapphire Radeon HD 7700
FPGA Replay + DB 68060 at 85Mhz

@Skateman
Radeon R7 250 right ?

@K-L
Try also plz other tests like 300, 600, 900, etc till 2400 by +300. Just so we will know, that such a bug didn't happens in "any chunk" of GPU on x1000.

@All
Need also Tabor and Sams tests

Last edited by kas1e on 28-Sep-2019 at 08:16 PM.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

tested all from 300 to 3100 on X1000, no crash.
can't test on x5000.

Strange, now, after running the test, moving icons on Workbench is totally sluggish.
Even after a hard reboot.
Too tired to test any further now. I'll turn the x1000 off until tomorrow.

@Amigo1

3100 its more than 1gb will be , so if you have 1gb of GPU, then it all going after to system memory => usuall slow down. In other words after hardcore power off / waiting a little / power on all will be back on track.

Last edited by kas1e on 28-Sep-2019 at 08:54 PM.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

X1000, HD7950, 3GB ram

test 300-ok- used 100MB (fromSysMon)
test 600-ok- 201MB
test 600-ok- 301MB
test 900-ok- 301MB
test 1600-ok- 535MB
test 2400-ok- 803MB
test 3100-ok- 1037MB

btw, Huno's latest port of Return to Castle Wolfenstein flies on my X1000

Last edited by klx300r on 28-Sep-2019 at 09:29 PM.
Last edited by klx300r on 28-Sep-2019 at 09:26 PM.

_________________
____________________________
c64-2sids, A1000, A1200T-060@50(finally working!),A4000-CSMKIII
! My Master Miggies- Amiga 1000 & AmigaOne X1000 !
mancave-ramblings
X1000 I BELIEVE

@kas1e

Well, only X5000 with Radeon HD is affected. I don't know if Hans is still working on the Radeon HD driver but this really needs to be fixed.

_________________
PowerMac G5 2,7Ghz - 2GB - Radeon 9650 - MorphOS 3.14
AmigaONE X1000, 2GB, Sapphire Radeon HD 7700
FPGA Replay + DB 68060 at 85Mhz

@kas1e

Tested on X1000, R7 240, no problems with the 6 tests you wanted.

_________________
OS4 Rocks
X1000 beta tester, Sam440 Flex (733)

Visit the Official OS4 Support Site for more help.

It may be that your sole purpose is to serve as a warning to others.

@K-L

Quote:

Well, only X5000 with Radeon HD is affected. I don't know if Hans is still working on the Radeon HD driver but this really needs to be fixed.

At moment have almost all tests only from x1000 and x5000. Need more SAMs and Tabor, just to be 100% sure.

Polaris card do not have such problems on x5000, there i can use up to 4gb. So i somehow hope, that RadeonHD will behave bad on some Sam or on Tabor.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

Davebraco tested on his Sam460 : http://www.amigans.net/modules/xforum/viewtopic.php?topic_id=8082&forum=25#threadtop

_________________
PowerMac G5 2,7Ghz - 2GB - Radeon 9650 - MorphOS 3.14
AmigaONE X1000, 2GB, Sapphire Radeon HD 7700
FPGA Replay + DB 68060 at 85Mhz

@K-L

Yeah, that start to be interestig. Also asking in beta-test mailing list about testing that on Tabor and on x5000/040 and on some other SAMs, let's see..

What is common at moment between working/non working tests, is that x1000 and sam460 both have that "dma cpu hack" in graphics.library which speed up some graphics.library operations, so _MAYBE_ that realted (as x5000/020 didn't have that). And so Tabor and x5000/040 tests should prove if that idea is right or wrong. By related i mean maybe that "dma hack" somehow workaround that issue and do some "remapping of GPUCPU" be a bit different to not have that bug. But let's see more tests.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

Tested up to 3100, no issues.
Sam460 and Radeon HD7770.

Cheers

_________________
If your nose runs and your feet smell, you're upside down.

@All
And news for which i was in hope for : TABOR have the same issue as x5000 !

It the same can't pass 256mb barier of extendent GPU memory and just die the same.

Which mean that :

1). It is not x5000 related (both x5000/20 and x5000/040 have that issue). It also happens on Tabor.

2). pattern there are: x5000 and Tabor do not have "dma-hack" inside of graphics.library (and they fail), while x1000 and sam460 do have (and they didn't fail). That can be it, or can be not, but wild guess that remapping of memory between CPU and GPU are works when "dma-hacked" routines is used, and didn't when not. Maybe inside of RadeonHD some routines used , from which remapping is taken for granted as it happens to returns from "dma-hack" graphics.library, but its done something different when graphics.library is without "dma-hack". I can be wrong of course, but pattern is here.

So will create bug-report with all the info to RadeonHD now.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@all

bug-report created:

http://www.amiga.org/developer/bugreports/view.php?id=457

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@All

Is anyone with X1000 can use "toggledma" binary, and disable DMA, and then do the same tests with 300,600,900, etc ?

But be 100% sure that you disable DMA. I do not have x1000, so i do not know how to use that "toggledma", and how to check if it was enabled/disabled , so be sure you know what and how to do.

By that way, we try to understand if DMA is related somehow or not.

Thanks!

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites