Overclock.net banner
6901 - 6920 of 17356 Posts

·
Old crazy guy
Joined
·
3,534 Posts
Tested again the LN2 switch with F12a; I don't get a POST. Either is switching to the backup BIOS or the settings are reset.
It's the same profile which I used to test it with F12b so seems specific to this version.
The board gets in a loop of debug codes 07..14..15.. guess it's the memory training failing.
 

·
Registered
Joined
·
794 Posts
Hi

And you set pcie auto/bifuration 2x8 (bios show You 1slot pcie x8 and 2slot pcie x8) ?
What performance of these rtx 2080 ti for example in the witcher 3?
Just wanted to mention that your SLI performance metrics are so poor due to the 2nd GPU running at X4. SLI only works (properly) when both cards at x8. When 1 is at X4 you have a 50% reduction in bandwidth.

As for a fixed resistor causing pci issues, idk. Please pull both GPUs out and test individually to see if they both pass benchmarks/ stability tests at the full X16 in the top slot then repeat for 2nd slot to see if both will run at x8. Also as Bluechris mentioned you can pull out all non essential hardware and test again if still having issues when testing above. Unless when you say "fixed" you jury rigged it somehow instead of de-soldering/ soldering a replacement resistor. Do you damage any traces on the pcb? I dabble a bit in soldering myself and those tiny resistors are very tricky to get 100% perfectly on the tiny pads.

Having pcie issues using sli is not unheard of I would just hate to have you assume a replaced resistor is the issue and spend $ sending out to get fixed and have the same issue once the GPU is returned. Only way to know for sure is to test each card and slot independently.

For all we know the MB slot could be defective or the IO die on the CPU as that controls PCIE lanes. If it were me I would want to know for sure my CPU/ MB are in perfect working order.

My brother was in a rush to take the CPU mounting plate of of his i7 920 MB. So he took a screw driver to pry it off, forgetting that it was screwed down, and damaged the traces along the backside of the MB. His board had 6 Dimms and the traces he damaged were for the middle channel. So as long as the board only used the 1st and 3rd memory channels the board ran fine. Yeah, I'm not sure he'll ever live that down. Point is 5h1t happens! Live and learn :)

Anyways, keep us posted.
 

·
Registered
Joined
·
2 Posts
Wanted to update the thread with a success on using an ASUS Essence STX soundcard. Installed in the bottom PCIE16 slot, using BIOS F11 and the UNi drivers it's working flawlessly. No changes to the BIOS, no crashes.
 

·
Registered
Joined
·
367 Posts
I guess the FCLK on my 3900x is just on the borderline of stability. I posted previously how I would occasionally get WHEA errors about "Corrected Machine Error - Bus/Interconnect", and I thought adjusting VSOC had solved this but they are showing up again.

The oddest bit is that I cannot get these to occur under load, only when the system is idle. Sometimes I have to leave the system sitting idle 10+ hours to even see one, sometimes they occur in 1-2 hours.

Adjusting FCLK/MCLK to 1867/1867 has left the logs without any errors for about 3 days now... I am not sure if there are any voltages other then SOC/VDDP/VDDG I can try to play with, or if this chip is just not 100% capable of sustaining 1900 FCLK without error. I've already attempted to push those about as high as I am comfortable with but the problem persists.

Even when it does error, it doesn't impact anything as apparently since it is "Corrected", but I don't consider it stable that way. Anyone have any other ideas?
 

·
Old crazy guy
Joined
·
3,534 Posts
I guess the FCLK on my 3900x is just on the borderline of stability. I posted previously how I would occasionally get WHEA errors about "Corrected Machine Error - Bus/Interconnect", and I thought adjusting VSOC had solved this but they are showing up again.

The oddest bit is that I cannot get these to occur under load, only when the system is idle. Sometimes I have to leave the system sitting idle 10+ hours to even see one, sometimes they occur in 1-2 hours.

Adjusting FCLK/MCLK to 1867/1867 has left the logs without any errors for about 3 days now... I am not sure if there are any voltages other then SOC/VDDP/VDDG I can try to play with, or if this chip is just not 100% capable of sustaining 1900 FCLK without error. I've already attempted to push those about as high as I am comfortable with but the problem persists.

Even when it does error, it doesn't impact anything as apparently since it is "Corrected", but I don't consider it stable that way. Anyone have any other ideas?
Did you try with Global C-States disabled?
Would also be worth to check if the RAM is the culprit; set the IF at 1900 and the RAM at a much lower frequency and relaxed timings, not the XMP profile.
 

·
Registered
Joined
·
794 Posts
I guess the FCLK on my 3900x is just on the borderline of stability. I posted previously how I would occasionally get WHEA errors about "Corrected Machine Error - Bus/Interconnect", and I thought adjusting VSOC had solved this but they are showing up again.

The oddest bit is that I cannot get these to occur under load, only when the system is idle. Sometimes I have to leave the system sitting idle 10+ hours to even see one, sometimes they occur in 1-2 hours.

Adjusting FCLK/MCLK to 1867/1867 has left the logs without any errors for about 3 days now... I am not sure if there are any voltages other then SOC/VDDP/VDDG I can try to play with, or if this chip is just not 100% capable of sustaining 1900 FCLK without error. I've already attempted to push those about as high as I am comfortable with but the problem persists.

Even when it does error, it doesn't impact anything as apparently since it is "Corrected", but I don't consider it stable that way. Anyone have any other ideas?
Some one posted in one of the other forums that bumping up the chipset vcore can help with 3800/ 1900 IF stability. I have not been able to confirm myself as I only run 3600/ 1800. When I tried playing around with 3733/ 1866 I had an issue or bug where the bclk adjusted itself to 107. I'm just lucky that I didn't corrupt any data on my NVME/ ssds. Not sure how that would happen as I never mess with bclk. Needless to say, I was thoroughly disgusted.

Anyway, it's worth a shot let me know if it helps you.
 

·
Old crazy guy
Joined
·
3,534 Posts
Did you try with Global C-States disabled?
Would also be worth to check if the RAM is the culprit; set the IF at 1900 and the RAM at a much lower frequency and relaxed timings, not the XMP profile.
Also which BIOS versions did you tried with?
So far the most stable and "bug free" for me it's F12a.

What about the SOC LLC setting? Did you try with Low or Medium while lowering the voltage?
With lower SOC voltage you can also try lowering the VDDP/VDDG, it can improve reliability for IF and RAM.
I found the SOC voltage set to Normal plus offset seems more stable than a fixed value.
 

·
Registered
Joined
·
367 Posts
Some one posted in one of the other forums that bumping up the chipset vcore can help with 3800/ 1900 IF stability.
It isn't clear to me, these are the PM_* values in the BIOS?

Also which BIOS versions did you tried with?
So far the most stable and "bug free" for me it's F12a.

What about the SOC LLC setting? Did you try with Low or Medium while lowering the voltage?
With lower SOC voltage you can also try lowering the VDDP/VDDG, it can improve reliability for IF and RAM.
I've only every used F11, still waiting for an "official" F12 release. I tried a number of various settings for SOC (Fixed 1.1, Offset, etc.) + SOC LLC without much luck, though this is incredibly difficult to test because it can take so long for one of these errors to show up in the Windows event log. The only thing I can say for sure is lowering VDDG below about 1000mv @ 1800 FCLK will result in actual crashing.

As far as the memory it is using custom timings (baseline from DRAM calculator)... which has passed multiple runs of Karhu running 12+ hours or more... since memory testing keeps the system at least lightly loaded the WHEA errors have never appeared while it is running.
 

·
Registered
Joined
·
794 Posts
It isn't clear to me, these are the PM_* values in the BIOS?



I've only every used F11, still waiting for an "official" F12 release. I tried a number of various settings for SOC (Fixed 1.1, Offset, etc.) + SOC LLC without much luck, though this is incredibly difficult to test because it can take so long for one of these errors to show up in the Windows event log. The only thing I can say for sure is lowering VDDG below about 1000mv @ 1800 FCLK will result in actual crashing.

As far as the memory it is using custom timings (baseline from DRAM calculator)... which has passed multiple runs of Karhu running 12+ hours or more... since memory testing keeps the system at least lightly loaded the WHEA errors have never appeared while it is running.
PM or promotory refers to chipset. I'll poke around for the OP and link if I come across it. I recall him posting as a tip to help with instability running 3800/ 1900IF as from what I have seen not all samples can do 1900 IF and be 100% stable
 

·
Science
Joined
·
3,487 Posts
I apologize if this has been answered, but I am using the Aorus X570 Pro-Wifi, and am having some fan problems. I have one fan that ramps up to 1400 RPM, while the others stay at 900 RPM. This change in RPM is sudden and unpredictable.

I thought it was an old beta BIOS that I had, so I just upgraded to Bios 10D, and have tweaked with SmartFan settings several times with no luck. Is this a common thing?
 

·
Registered
Joined
·
129 Posts
I apologize if this has been answered, but I am using the Aorus X570 Pro-Wifi, and am having some fan problems. I have one fan that ramps up to 1400 RPM, while the others stay at 900 RPM. This change in RPM is sudden and unpredictable.

I thought it was an old beta BIOS that I had, so I just upgraded to Bios 10D, and have tweaked with SmartFan settings several times with no luck. Is this a common thing?
Why did you update to 10 and Not the most recent BIOS version 12e or even 11? Just curious.
 

·
Science
Joined
·
3,487 Posts

·
Registered
Joined
·
16 Posts
Hi

And you set pcie auto/bifuration 2x8 (bios show You 1slot pcie x8 and 2slot pcie x8) ?
What performance of these rtx 2080 ti for example in the witcher 3?

Mine is set to auto. I have not played the witcher 3 with the setup. It pumps out high frames on titles that can use SLI. Control gives them a work out on ultra settings.
 

·
Registered
Joined
·
247 Posts
Hi

I ordered another used MSI ARMOR OC 1080 TI because one of my GPUs that was repaired is not working properly and for this the motherboard set it to GPU pcie x4
If I break the loop I will test the repaired GPU in 1 slot pcie and how the motherboard will set pcie x4 will be the problem solved
and I bought the GPU for 1690 PLN
 

·
Registered
Joined
·
247 Posts
And I have a question :D
I do not know the Ryzen platform on the part of OC
can anyone help me with this

I did some performance tests for ryzen 3900x (turbo ~4.2 ghz during in test)
OCCT 5.5.5 avg temp 63c
Prime 95 avg temp 62c

(water temp 32c home 26c)

Temperatures are good for this CPU I can try some OC
 

·
Old crazy guy
Joined
·
3,534 Posts
It isn't clear to me, these are the PM_* values in the BIOS?



I've only every used F11, still waiting for an "official" F12 release. I tried a number of various settings for SOC (Fixed 1.1, Offset, etc.) + SOC LLC without much luck, though this is incredibly difficult to test because it can take so long for one of these errors to show up in the Windows event log. The only thing I can say for sure is lowering VDDG below about 1000mv @ 1800 FCLK will result in actual crashing.

As far as the memory it is using custom timings (baseline from DRAM calculator)... which has passed multiple runs of Karhu running 12+ hours or more... since memory testing keeps the system at least lightly loaded the WHEA errors have never appeared while it is running.
I'd give a try to F12a; I used F11 for quite a lot and at least for me it's more reliable.

On my setup if lowering VDDG below 1000 causes a crash, the issue is somewhere else.
I was able to keep VDDP/VDDG at 1000/1050 with 1800 FCLK; had to up the SOC voltage to keep reproducible scores in benchmarks.
Wasn't fully satisfied; the system is usually butter smooth and I could notice very sporadically a glitch or micro stuttering.

Now that I switched to 1900 FCLK after a little while I started having stability issues; much more often and increasing.
Not WHEA errors, for me is voltage drop on the USB ports causing device resets and in some cases crashes or reboots.

As always since the beginning, to get stability I have to find the right settings for VDDP/VDDG at 900/950.
It's recommended by 1usmus to get better IF stability and RAM overclock and he's right, at least for me.
Anything above these voltages will end up sooner or later in stability issues.

Now after 3 days of painful trials and errors seems I found the right settings.
But they are linked to a proper CPU vCore voltage; a little bit too much or to less and it's not going to work.
I'm also using the EDC at 1 bug turbo boost so it's even more challenging.

My settings are below in the BIOS screenshots; CPU LLC & current protection, SOC LLC, SOC vCore and its offset.
Give it a try maybe it can be helpful also with your setup.

Set the VDDP/VDDG on the XFR menu and leave it Auto on the CBS menu.

For system stability indeed Karhu doesn't help, not even benchmarks.
I can pass a 14000% test but yet the system wouldn't be stable.

As a quick check to verify stability I use CPU-z bench.
But you have to do it this way:

1) Open Task Manager and check nothing else is eating CPU; especially Windows Defender, Updates, Telemetry etc. Close everything else, not even HWinfo running.
2) Close Task Manager and start CPU-z
3) Run the benchmark
4) The score should be consistent; meaning should not have big drops while running. eg: starts at 5950 and goes down to 5920 is fine; bad if goes down to 5800, even if then goes again up to 5900 and more. No big drops.
5) Wait 3-5 seconds, not less, not more
6) Run again; the 2nd bench could drop values but it's pretty normal, excellent if it's not dropping
7) Wait 3-5 seconds, not less, not more
8) Repeat and it should be as 4)
9) Repeat 7) and 8), runs should be consistent

Best way to test stability is gaming with some medium to high workload; something too heavy will not trigger that easy instability.
 

·
Old crazy guy
Joined
·
3,534 Posts
And I have a question :D
I do not know the Ryzen platform on the part of OC
can anyone help me with this

I did some performance tests for ryzen 3900x (turbo ~4.2 ghz during in test)
OCCT 5.5.5 avg temp 63c
Prime 95 avg temp 62c

(water temp 32c home 26c)

Temperatures are good for this CPU I can try some OC
With these temps if you want to live dangerously you could try the EDC at 1 bug:

https://www.overclock.net/forum/13-amd-general/1741052-edc-1-pbo-turbo-boost.html

At least the effort it's worth :)
 

·
Registered
Joined
·
48 Posts
I apologize if this has been answered, but I am using the Aorus X570 Pro-Wifi, and am having some fan problems. I have one fan that ramps up to 1400 RPM, while the others stay at 900 RPM. This change in RPM is sudden and unpredictable.

I thought it was an old beta BIOS that I had, so I just upgraded to Bios 10D, and have tweaked with SmartFan settings several times with no luck. Is this a common thing?
This is kind of a known issue, though doesn't happen to everyone. I've reported it to Gigabyte, but they were not able to reproduce. It used to happen for me on SYS_FAN4 that is managed by the secondary controller (CPU, SYS_1-2 are on main controller and work just fine). Disabling ErP fixed it for me, but YMMV. Running Aorus X570 Pro (non Wifi model), stuck fan was Noctua A12x25 on SYS_4. What connector is your problematic fan attached to?
 

·
Science
Joined
·
3,487 Posts
This is kind of a known issue, though doesn't happen to everyone. I've reported it to Gigabyte, but they were not able to reproduce. It used to happen for me on SYS_FAN4 that is managed by the secondary controller (CPU, SYS_1-2 are on main controller and work just fine). Disabling ErP fixed it for me, but YMMV. Running Aorus X570 Pro (non Wifi model), stuck fan was Noctua A12x25 on SYS_4. What connector is your problematic fan attached to?
According to HWInfo, its System 5 pump. Right now, I have two intake fans (Fractal), two exhaust fans (Corsair) and the fan on my Heat sink (Noctua). The mistake may be me installing the fan on a pump pin?

The fan that is randomly jumping up and down in RPM, is a fractal fan, with no adapters, just plugged straight into the motherboard.
 
6901 - 6920 of 17356 Posts
Top