Strange sluggish graphics issue with Mojave 10.14.4 update (confirmed)

[tsialex]

While I 100% believe this bug report should be filed, I’m not so sure it will be addressed by Apple. ALL reports so far are for Xeon 5500-series CPUs which were never officially offered in MP5,1. If this impacted 5600-series it absolutely would need to be addressed.

I still think that this is more a generic NUMA/NMI related bug/race condition and not exactly CPU related.

Someone with a dual CPU board with low clock X56xx like the L5639 (Westmere Hexa 2,13GHz) will probably have it too since the fast clock will not hide the problem. @ibarnett is sending me two E5620 (Westmere Quad 2.4GHz), but will take a lonnnggggg time from Australia to Brazil.

 

[cdf]

While I 100% believe this bug report should be filed, I’m not so sure it will be addressed by Apple. ALL reports so far are for Xeon 5500-series CPUs which were never officially offered in MP5,1. If this impacted 5600-series it absolutely would need to be addressed.

Dual 5680's here with graphics issues. I originally noticed video card fan noise and sleep issues. Ran a benchmark today to confirm reduced performance with 10.14.4. This problem may actually be more widespread than we think.

 

[tsialex]

Dual 5680's here with graphics issues. I originally noticed video card fan noise and sleep issues. Ran a benchmark today to confirm reduced performance with 10.14.4. This problem may actually be more widespread than we think.

Yep, I’ve been sceptic of the initial diagnose of being Gainestown processors the culprit, to me it’s just the colateral symptom. This is NUMA/NMI/QPI related and not a problem with the type of processor being used. Faster clocks/faster QPI/bigger caches just hide the problem.

 

[bsbeamer]

Dual 5680's here with graphics issues. I originally noticed video card fan noise and sleep issues. Ran a benchmark today to confirm reduced performance with 10.14.4. This problem may actually be more widespread than we think.

Absolutely “need” more of these MP5,1 reports with 5600-series sent to Apple. They cannot simply ignore authentic configuration machine exhibiting issues caused by an update, even if the root was part of a security vulnerability patch.

Is there anything to reliably reproduce on 5600-series? I have not at all noticed issues on MP5,1 with X5690.
[doublepost=1555714354][/doublepost]Re-reading this thread... is this still only being reported on MP4,1>MP5,1 hacks or has an authentic MP5,1 also reported issues?

 

[cdf]

Is there anything to reliably reproduce on 5600-series? I have not at all noticed issues on MP5,1 with X5690.

Perhaps run a benchmark to compare performance between 10.14.3 and 10.14.4.

 

[bsbeamer]

Perhaps run a benchmark to compare performance between 10.14.3 and 10.14.4.

Is that a reportable issue to Apple vs. possible impact of a security patch? We really need repeatable and able to reliably reproduce issues triggered by something specific on authentic 5,1 configs with 5600-series to better make the “case” it is an actual issue they need to fix. Not at all saying it shouldn’t be addressed regardless, just trying to make the best “case” possible that it deserves priority attention.

 

[tsialex]

Re-reading this thread... is this still only being reported on MP4,1>MP5,1 hacks or has an authentic MP5,1 also reported issues?

This is not MP4,1>5,1 related, happens with @Riff_Al even after changing his BootROM to mid-2012 standard and happens with my real MP5,1 when using my dual CPU tray and E5520, just happens to be easy to notice the problem with the basic early 2009 dual CPU Mac Pro.

What’s needed is a easy and full proof way to trigger it, with my Macs it happens a lot less than with @Riff_Al MP4,1>5,1, he sent me a video demonstrating with his Mac, and I don’t fully understand why the discrepancy yet.

 

[krakman]

tested flurry screensaver 2010, 5.1, 10.14.4, rx480, two 32 inch monitor connected via DP...... works fine, also the dock appears and hides as normal

edit: its a dual cpu 2x 2.4 E5620 xeon

 

[tsialex]

tested flurry screensaver 2010, 5.1, 10.14.4, rx480, two 32 inch monitor connected via DP...... works fine, also the dock appears and hides as normal

It's not easy to detect with some Macs.

I already have some iTunes downloaded music files that are easy to trigger the stuttering, but I only see the problem 2 or 3 times a minute, some people have it seconds apart.

I'll receive two E5620 soon, I hope.

 

[bsbeamer]

tested flurry screensaver 2010, 5.1, 10.14.4, rx480, two 32 inch monitor connected via DP...... works fine, also the dock appears and hides as normal

edit: its a dual cpu 2x 2.4 E5620 xeon

Try the Garageband test, that seemed to show the issue for everyone with E5520:
https://forums.macrumors.com/thread...1-mojave-10-14-4.2176792/page-4#post-27278302

 

[handheldgames]

That seems to jive with the audio sluggishness reported in another thread. I wonder if it could be related.

^^^^ Absolutely... perhaps apple devs and qa have been running on single core systems with the assumption it's the basically the same platform.

 

[h9826790]

^^^^ Absolutely... perhaps apple devs and qa have been running on single core systems with the assumption it's the basically the same platform.

But only test the 142.0.0.0.0 firmware on dual processor model. And assume it will be compatible to the W series Xeon

 

[amp829]

Hello i noticed that my mac pro running Mojave is jerking when playing high quality videos , there are sound interrupts and when panning also like losing frames ... i have aN rx 580 8gb GPU ....i noticed this behaviour also when playing the screen saver photos, the motion is sluggish . Can anyone tell me what is probably wrong ?