MP 1,1-5,1 No graphic card at cold boot cMP

[superparati]

Hello!,

Since I've updated my OS to 10.14.6 last week my computer at cold boot does not activate my screen.
I check via screen sharing and I can load my session, but I can see in the preference system no graphic card detected.

I have to reboot in order to activate my screen and therefore the graphic card detected.

Any one is reporting the same issue?
I'll do this evening the second 10.14.6 released yesterday and see if there is any progress.

Thanks

 

[superparati]

Hello,

Yesterday I did some test switching powers from 8-pin id1, id2 and mixing the power coming from the ODD bay and the 2x6-pin powers BoostA and BoostB.

From what I saw if ht 8-pin id1 is powered by the DVD bay at cold boot the graphic card won't show up.
moving the power to the 8-pin id2 seems to solve the issue.

I've tried different way to supply the Vega 56 by mixing the powers and the results are strange (hardware monitor and Istat Menu).


Test 1

Hardware Monitor

8-pin id1=> power with 2x6-pins from the motherboard under load BoostA-41w and BoostB-44w, at idle 1~2w each.
8-pin id2=> power with 2xSATAs from the ODD bay under load 31.7w, at idle 17w
PCI slot 1 => under load 22~24w, idle 6~8w
CG total avg max: less than 140w (PSU 373w)​

Istat Menu

8-pin id1=> power with 2x6-pins from the motherboard under load Boost A 3.4A, Boost B 3.71A, at idle BoostA 0.09A and BoostB at 0.17A
8-pin id2=> power with 2xSATAs from the ODD bay under load 2.66A, at idle 1.34A
PCI slot 1 => under load 1.97A, idle 0.61A
CG total avg max: 10.7A -128w (PSU 30A - 360w)​

Test 2

Hardware monitor

8-pin id1=>power with 2xSATAs to 6-pin->8-pins from the ODD bay under load 31.8w, idle 3.6w
8-pin id2=> power with 2x6-pins from the motherboard under load BoostA 53.6w & BoostB 43.6w, idle BoostA 8.5w & BoostB 7.7w
PCI slot 1 => under load 23.1w, idle 7.4w
CG total avg max: 152.4w (PSU 365w)
​

Istat Menu

8-pin id1=>power with 2xSATAs to 6-pin->8-pins from the ODD bay under load 2.66A, idle 0.29A
8-pin id2=> power with 2x6-pins from the motherboard under load BoostA 4.46A & BoostB 3.64A, idle BoostA 0.70A & BoostB 0.62A
PCI slot 1 => under load 2.03A, idle 0.61A
CG total avg max: 12.79A (PSU 31.79A)
​

Test 3

Hardware monitor

8-pin id1=> took SATA power 6-pin cable + one 6-pins from the motherboard, ODD bay under load 20.2w and BoostA at 66w, at idle ODD 0.2w & BoostA 2.1w
8-pin id2=> BoostB 6-pin to 8-pin, under load max 91.8w and idle 15.5w
PCI slot 1 => under load 23.7w, idle 7.1w
CG total avg max: 201.7w (PSU 374w max)​

Istat Menu

8-pin id1=> took SATA power 6-pin cable + one 6-pins from the motherboard, ODD bay under load 1.72A and BoostA at 5.55A, at idle ODD 0.02A & BoostA 0.19A
8-pin id2=> BoostB 6-pin to 8-pin, under load max 7.64A and idle 1.29A
PCI slot 1 => under load 2A, idle 0.59A
CG total avg max: 16.91A (PSU 28.95A max)​

Test 4

Hardware monitor

8-pin id1=> 6-pin BoostB to 8-pin, under load max 85w, idle 12w
8-pin id2=> took SATA power 6-pin cable + one 6-pin BoostA from the motherboard, ODD bay under load 22.6w and BoostA at 73w, at idle ODD 3.2w & BoostA 3w
PCI slot 1 => under load 24w, idle 7.2w
CG total avg max: 205.2w (PSU max 404w)​

Istat Menu

8-pin id1=> 6-pin PCIe BoostB to 8-pin, under load max 7.04A
8-pin id2=> took SATA power 6-pin cable + one 6-pin BoostA from the motherboard, ODD bay under load 1.88A & BoostA at 5.99A, at idle ODD 0.27A & BoostA 1A
PCI slot 1 => under load 1.98A, idle 0.59A
CG total avg max: 16.89A (PSU max 33A)​

We can see test 3 and 4 are close to the Vega 56 TDP but the test 1 and 2 we are very phare !
Each time I ran a LuxMark bench to stress the card and each time the score were identical ~26900 at 1% ~ 2% differences.

 

[Macschrauber]

get stable 12 volts from the Power Supply by some kind of Pixlas mod. You may burn your backplane board by overloading.

 

[superparati]

I'm considering it and in the meantime I would like to understand this significante difference between all my settings.

If the PSU is at the end is rated with the same power consumption ~30A max, looks like there is some kind of bug with those softwares.
Therefore I don't know which setup is the best except Pixlas mod.

 

[tsialex]

I'm considering it and in the meantime I would like to understand this significante difference between all my settings.

If the PSU is at the end is rated with the same power consumption ~30A max, looks like there is some kind of bug with those softwares.
Therefore I don't know which setup is the best except Pixlas mod.

Unless you are using controlled lab power supplies to power and monitor your GPU you won't detect the power spikes, or the initial startup/burst consumption, with monitoring apps.

Since MP5,1 backplane don't offer a lot of headroom for modern GPUs, just eliminate it with a secondary power supply or with a Pixla's mod.

 

[superparati]

And the Vega / any new GPU as you are saying are playing a lot more with the power than before (overclocking boost method).

I've completed my tests this evening and looks like the best configuration with a good power mix is with my setup test 4.
I'll see if the screen / GPU are loaded at cold boot tomorrow.

I reiterate my finding, test 1 and 2 are weird, low power consumption vs what the GPU should eat while doing a stress test (regardless the power spikes).

I've everything to do the Pixla's mod missing just some time to do it and the energy .