Jump to content

fabiosun

Recommended Posts

  • Moderators
On 9/8/2020 at 5:12 PM, Driftwood said:

Big Sur Bare Metal Candle Test

(Not as Quick As Proxmox, however we are using a patched dylib library replacing BMDs own to get Davinci Resolve working under Big Sur / AMD bare metal - and AMD BM is not supported under Mac from Davinci..

 

But even with the patch, it's still very acceptable.

 

* On Proxmox I was getting apx 34fps on 66 Blur Nodes, Here around 20-24fps at moment of screen grab. Had a load of background apps going on if that didn't help.

 

 

I ran the DaVinci 2 candle tests a week or so ago and forgot to post the results. At the time, they were basically the same as your results: around 34-35 fps in VM with dual Radeon VIIs in Catalina (1 Radeon VII giving 16 fps). Meanwhile, in bare metal (Big Sur ß4) dual Radeon VIIs gave ~24 fps. (BTW, as a reference, 16 fps is a typical result when running when using one 2080 Ti in Win10.)

 

I re-ran the DaVinci 2-Candle test since I've now reduced the AMD kernel patches and added the custom DSDT. The 2-Candle test now returns values the same as VM, fluctuating between 32 and 36, mostly hovering about 34. So for my uses, I'm not seeing any advantage of VM over bare metal.

 

The screen grab was at 35 fps with 66 nodes under Big Sur ß6 bare metal, dual Radeon VIIs.

 

DaVinci-2-Candle-Test.jpg.b1f5f40528fc31a775f94f0b64fc3549.jpg

 

DaVinci-2-Candle-35-64nodes.jpg.95adf276bb67ac8c15d009389458e707.jpg

 

 

And I've repeated the C15 tests and consistently get the same results (the C15 results are same for 1 or 2 GPUs):

 

177147972_C15-BS6-Results-2.png.03017f5a9f965184dc743acfb5cce2d5.png

 

 

 

Edited by iGPU
  • Like 2
Link to comment
Share on other sites

  • Supervisor

I believe that the results shown by iGPU are correct and the only one that for me has a value out of the ordinary is that of Cinebench 15 but as long as none of you can reproduce a DSDT together with its prop devices this result remains unique and in my opinion very interesting

thanks for the tests

Link to comment
Share on other sites

16 hours ago, Driftwood said:

@dtek Aquantia shouldn't need anything. Its supported by Apple. Shouldn't need to add anything for it.

 

971728577_ScreenShot2020-09-09at15_22_58.png.ebeaa086f757600ca927b79dbf76d2d2.png

 

The Intel 10G should need a kext I believe, I replaced my wifi pic card like others here to Broadcom Mac compatible.

 

13 hours ago, iGPU said:

 

Aquantia is supported natively through Mojave (and I thought ok in Catalina), but Big Sur requires a kernel patch. The inactivation has nothing to do with SSDT or MmioWhitelist.

 

Pavo has provided this on another forum. Here it is below (as written this will work on HS thru BS; if you only want for Catalina-BS, change 17.0.0 to 19.0.0):

 

AquantiaPatch.jpg.59e4e6a32930524d9031b97941c5763f.jpg

The Aquantia card used to work for me but I don't know what changes I made and now it doesn't work anymore.  My other 2 integrated lan cards work fine.  When I plugged the ethernet cable in, there's activity light but no internet.  I started from scratch and it still doesn't work, then I tested in windows and it works fine.

 

Screen Shot 2020-09-10 at 2.28.05 AM.png

Edited by dtek
Link to comment
Share on other sites

  • Supervisor

@dtek in big Sur you mean?

It is a clean installation or you have ported it from Proxmox?

 

you can delete your network on SystemPreferences/Network, apply this changement and then add again your aquantia card (in may case it is named en5 o similar

I can't say if you need patches I see floating around because in HS I do not need of it (People using Big Sur said here also is useful)

Screen Shot 2020-09-10 at 09.00.26.png

Link to comment
Share on other sites

@fabiosun

 

Thankfully Tonymacx86 now allows discussion of AMD, it’s been a long time coming.

 

I did post the link to this site/page, credit where credit is due 😀


I have never really liked insanelymac. I was once suspended and then banned for disagreeing with a few things that were said. For example, someone had run an update to macOS and was recommending everybody update asap. However that update broke a number of things so I suggested it would be wise to mention doing a full CCC backup (and testing you could boot from it) before running the update. I received a mouthful of abuse saying I was trying to undermine the OP and that he knew better, and everybody who didn’t run his full EFI, without any personal changes, were fools as it was ‘perfect’. An update to that update was quickly released by Apple that fixed the issue. I mentioned it in the thread and was banned after replying, with the same sort of language, to the further abuse I received from the OP.

 

Thankfully we get a better class of people on here 😂

  • Like 1
  • +1 1
Link to comment
Share on other sites

1 hour ago, fabiosun said:

@dtek in big Sur you mean?

It is a clean installation or you have ported it from Proxmox?

 

you can delete your network on SystemPreferences/Network, apply this changement and then add again your aquantia card (in may case it is named en5 o similar

I can't say if you need patches I see floating around because in HS I do not need of it (People using Big Sur said here also is useful)

Screen Shot 2020-09-10 at 09.00.26.png

So I moved my GPU and 10g lan card to a different slot to make room for a wifi/bluetooth card but now I can't get it to boot anymore.  I'm stuck at lost battery power and [PCI configuration begin].  Any idea how to fix this?

Edited by dtek
Link to comment
Share on other sites

21 minutes ago, fabiosun said:

I have my GPU on first slot

acquantia card in slot 3

Thunderbolt card in slot 4

Uefi On

4g on

MMIO list said before,,but you can try with nothing in MMIO list only use DevirtualizeMMIO on

 

To get it working again I had to make a few changes. 

1.Added bootarg ncpi=2000 and the original ssdt files,

2.disabled 4g 

 

To my surpise, the  acquantia card, wifi and bluetooth all work now but it has to rely on those settings and files.   I can't remove them anymore.  Also do you know why my boot menu doesn't have "clear nvram" selection?

Link to comment
Share on other sites

  • Supervisor

yes I know..but if you never answer to my question is difficult to help you better 🙂

 

however search in your config allowfullnvramreset or similar and flag it if it is not

 

if you do not see again in boot menu press space and see

 

Link to comment
Share on other sites

21 minutes ago, fabiosun said:

yes I know..but if you never answer to my question is difficult to help you better 🙂

 

however search in your config allowfullnvramreset or similar and flag it if it is not

 

if you do not see again in boot menu press space and see

 

Sorry I might have missed your questions.  What do I need to do to get old settings back?  My build is from a fresh Catalina install.

Edited by dtek
Link to comment
Share on other sites

6 hours ago, iGPU said:

I've now reduced the AMD kernel patches and added the custom DSDT. The 2-Candle test now returns values the same as VM, fluctuating between 32 and 36, mostly hovering about 34. So for my uses, I'm not seeing any advantage of VM over bare metal

 

So Im probably not going to see any improvements on mine unless I get power table DSDT and GPU water-cooling which is making you improve back to Proxmox? Actually do you mind posting your latest config for me to analyse? There may well be something I have forgotten to do. Thanks @iGPU

41 minutes ago, dtek said:

Sorry I might have missed your questions.  What do I need to do to get old settings back?  My build is from a fresh Catalina install.

Did you check your debug boot log to seee if anything changed in the hex addresses for MMIO?

Edited by Driftwood
Link to comment
Share on other sites

Interesting. I just swapped out my BMD 4K Monitor PCIe card for the 8K Decklink card and MMIO has changed to a skip 0 even though its YES in config. Which means, that slot 2 where I regularly swap round these two cards is using a hex address 0xE3300000 (child 3 out of 0 to 18) . The 100 pages for that address if I look further down could (I say could) indicate then that the next 100 page address is my Firewire card is seen at slot 4 on xFEA00000 (another 0x100 pages, 0x8000000000000001).

 

 

With 4K Decklink

0xE3300000 (0x100 pages, 0x8000000000000001) skip 1

 

With 8K Decklin card

devirt 0xE3300000 (0x100 pages, 0x8000000000000001) skip 0

 

I need to do some more tests... as it could be address error.

Edited by Driftwood
Link to comment
Share on other sites

  • Supervisor
7 minutes ago, Driftwood said:

Interesting. I just swapped out my BMD 4K Monitor PCIe card for the 8K Decklink card and MMIO has changed to a skip 0 even though its YES in config. Which means, that slot 2 where I regularly swap round these two cards is using a hex address 0xE3300000 (child 3 out of 0 to 18) . The 100 pages for that address if I look further down could (I say could) indicate then that the next 100 page address is my Firewire card is seen at slot 4 on xFEA00000 (another 0x100 pages, 0x8000000000000001).

 

 

With 4K Decklink

0xE3300000 (0x100 pages, 0x8000000000000001) skip 1

 

With 8K Decklin card

devirt 0xE3300000 (0x100 pages, 0x8000000000000001) skip 0

 

I need to do some more tests... as it could be address error.

 

check in debug log if that MMIO is the same

because skip 0 means or you have a wrong conversion or it misses now (if yes is in your config I mean)

Link to comment
Share on other sites

12 minutes ago, fabiosun said:

 

check in debug log if that MMIO is the same

because skip 0 means or you have a wrong conversion or it misses now (if yes is in your config I mean)

Yeah, its a memory address error. Damn! thought I was onto something for a minute.

 

In Catalina I hadn't fixed an error from the other week on that address. Thats from using BSS so long!

 

Hopefully soon Algrey can nail down what these addresses relate to.

Edited by Driftwood
Link to comment
Share on other sites

On an open quest to discover common traits of our MMIO addresses in our boards HPET always keeps cropping up for address 0xFED00000

 

"The BIOS often sets up the HPET incorrectly in the ACPI, initializing it correctly only in Intel 8253 mode. If the ACPI is not set up correctly the OS cannot list the HPET. And the BIOS and OS developers don't see the need to get Real Time support. So the HPET is only there to satisfy the system's high speed needs. If the HPET is set up correctly in the ACPI from BIOS then the ACPI MMIO page of the first HPET chip should be at 0xFED00000 and the second HPET at page 0xFED80000"  WIKIPEDIA

 

So I started investigating the other MMIO addresses to find any common values out there in webworld as nobody seems to know what they relate to!

 

Therefore we have in our debug of my Asrock TRX40 Creator an estimate of what these values mean:-

 

child 0 MMIO devirt 0xCB100000 -  AppleACPIPCI PCI Bridge Peg 0

child 1 MMIO devirt 0xD7180000 - Registered nosave memory ?

child 2 MMIO devirt 0xE3180000 - Registered nosave memory ?

child 3 MMIO devirt 0xE3300000 - device interrupt (10?)  memory

child 4 MMIO devirt 0xEF100000 - device interrupt(20?)memory

child 5 MMIO devirt 0xFEA00000 -  device interrupt (e.g. interrupt 16 at device 0.0 on pci4)

child 6 MMIO devirt 0xFEC00000 - RTMPTable IOAPIC (The address in the physical address space of the I/O APIC. If set to zero, the default address 0xFEC00000 is assumed)

child 7 MMIO devirt 0xFEC10000 - IOAPIC

child 8 MMIO devirt 0xFED00000 - HPET1

child 9 MMIO devirt 0xFED40000 - Trusted Platform Module (TPM)  ?

child 10 MMIO devirt 0xFED80000 - AMD ACPI MMIO address / General Purpose Input Output (GPIO) address (Southbridge?) / HPET2

child 11 MMIO devirt 0xFEDC2000 - Reserved no save memory ?

child 12 - MMIO devirt 0xFEDD4000 - Reserved ?

child 13 - MMIO devirt 0xFEE00000 - LAPIC - LAPIC registers are usually placed on address 0xFEE00000, and I/O APIC registers on address 0xFEС00000

child 14 - MMIO devirt 0xFF000000 - Int Mask

Child 15 - MMIO devirt 0x10000000000 acpipci0 pcimem
Child 16 - MMIO devirt 0x2BF40000000 acpipci3 pcimem
Child 17 - MMIO devirt 0x47E80000000 acpipci2 pcimem
Child 18 - MMIO devirt 0x63DC0000000 acpipci1 pcimem

As you can see in the device interrupt areas these memory addresses are probably going to change between motherboards and devices on the PCI/PCIe lane.

Anyways, research goes on... please dont take the above as gospel!

 

Interesting PROBE of Asrock TRX40 Creator under Linux here: https://linux-hardware.org/index.php?probe=724cb22da0

 

Edited by Driftwood
Link to comment
Share on other sites

  • Moderators
8 hours ago, dtek said:

 

The Aquantia card used to work for me but I don't know what changes I made and now it doesn't work anymore.  My other 2 integrated lan cards work fine.  When I plugged the ethernet cable in, there's activity light but no internet.  I started from scratch and it still doesn't work, then I tested in windows and it works fine.

 

Screen Shot 2020-09-10 at 2.28.05 AM.png

 

 

It is reported that Aquantia ports will stop working under macOS if their drivers are updated under Windows. Did you update the Aquantia drivers while running Windows?  (This is based on info I remember from one of KGP's threads on his X299 builds.)

 

Edited by iGPU
Link to comment
Share on other sites

  • Moderators
5 hours ago, Driftwood said:

 

So Im probably not going to see any improvements on mine unless I get power table DSDT and GPU water-cooling which is making you improve back to Proxmox? Actually do you mind posting your latest config for me to analyse? There may well be something I have forgotten to do. Thanks @iGPU

Did you check your debug boot log to seee if anything changed in the hex addresses for MMIO?

 

Here is my latest config.plist file (sans SNs). Keep in mind that I use many custom SSDTs that are not uploaded, so this config file is not plug-'n-play.

 

Also, I use the unpopular RadeonBoost as well as Radeon VII properties injected via the OC/DevProp section (the latter you can view in the DevProp section). I'll attach below v1.6 of RadeonBoost that I use.

 

Further, I've begun re-using WEG, but with the -wegbeta boot argument; the computer seems to boot more smoothly and the log-in screen is the correct resolution (not a lower resolution as without WEG). However, I do routinely set UIScale to "02" in the OC/NVRAM section.

 

(BTW, the above re-test of the DaVinci 2-Candle test was done with this config file set-up.)

 

RadeonBoost-v1-6.kext.zip config-public.plist.zip

  • Ok 1
Link to comment
Share on other sites

I have been experiencing "jumpy" mouse and "jumpy" sound lately. Every few seconds or so I'd get a stutter if if moving mouse and that would also interrupt sound. Is this common? If so what usually causes it and how do you debug it? Must have started recently as I didn't notice it a week ago. Running on latest OC 0.6.1 release. I noticed the same issue in the VM with sound.

  • Sad 1
Link to comment
Share on other sites

  • Moderators
8 minutes ago, meina222 said:

I have been experiencing "jumpy" mouse and "jumpy" sound lately. Every few seconds or so I'd get a stutter if if moving mouse and that would also interrupt sound. Is this common? If so what usually causes it and how do you debug it? Must have started recently as I didn't notice it a week ago. Running on latest OC 0.6.1 release. I noticed the same issue in the VM with sound.

 

That was sometimes seen with 5700XT (and occasionally with Radeon VII) and WEG on the Intel side. If you have WEG enabled, try running with it disabled.

 

I did notice this esp with USB, if WiFi was enabled. So after checking WEG status, try turning off WiFi (BT being on is okay).

 

Edited by iGPU
  • Like 1
Link to comment
Share on other sites

  • Supervisor
panic(cpu 1 caller 0xffffff8016d7d694): "Spinlock acquisition timed out: lock=0xffffff8017499698, lock owner 

ei Guys, try to solve this in Photoshop 2020

it happens in different conditions and it reboots completely my pc

 

have you some idea?

Photoshop 2020 or similar is fully working (after patching procedure)?

 

thanks

 

Link to comment
Share on other sites

  • Supervisor

Adobe suite Users:

for those of you who use Adobe After Effects I found that in High Sierra and with my Nvidia graphics card disabling the well known patch that partially solves the slowness in games also solves a serious lag problem in the graphical interface during some operations If you want to try there is an easy example to understand whether or not you have improvements by disabling it: new composition, new solid, create a path and try to edit it, maybe even putting a stroke effect on the unpatched improvement should be evident to you too

Link to comment
Share on other sites

  • fabiosun changed the title to [Discussion] - TRX40 Bare Metal - Vanilla Patches

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • There are no registered users currently online
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.