Do you think that VRM bug affects TRX40 desinare as well (prime95 is stable here with suggested settings but everything is stock and still on air (temps hit 93C… on the NH12 I temporarly use).
Possibly but more “no” than “yes”. The VRM-Issue is very noticeable under high load. If you have stable high-load scenarios, then you can be sure you dont have an issue.
Contrary to common believe, high temperatures are not bad for your CPU. Well they are, but in a way that is certainly not relevant for you.
I found thermal paste application extremely difficult for me. The chip is so big, that the contact pressure is quite low - leading to too thick of a paste layer since it is not squeezed out to the sides. If you used generous amounts, try a little less. (rip the cooler of, and clean one side but not the other, just put the cooler back on).
On the other hand, its a 300W part. Holy shit thats an incredible heat density for the chip size. At some point it does not matter what you put on it, it is not an issue of cooling capacity but it becomes an issue of heat density at the thermal interface. So 93°C in Prime is likely fine. (imo)
If you throttle too much (below 3.5GHz during Prime95) and you are stable, just try some undervolting 
1 Like
Its the first PC I assembled myself, so in regard to thermal paste I used the most conservative ‘spread a thin’ layer method. Idle temps are around 40C so its not that bad. The cooler is undersized (The NH-U14S doesn’t fit due to combination of high profile ram and case not really mean for air cooling (the 011D XL)).
The Heatkiller iV is underway though to deal with that together with MORA-420 
In HWiNFO I see that package limit is at 280W, no throttling, and for some reason the sum of CPU power and SOC power is about 240W. I haven’t researched where the extra power goes.
BTW, do you know what the safe VRM temps are? On that board, someone from gigabyte thought it would be a very good practical joke to put 1100RPM (yes that is not a typo) fan on the VRM and set a default fan curve to hit that speed when VRM reaches about 60C, which means that even at idle, when VRM is at about 40C, the fan sounds like a vacuum cleaner (about 5000-6000 rpm), so I set a custom fan curve to deal with that but I wonder how far I can push VRM without overheating it.
Haha yes, the first thing i did was to disconnect the VRM fan with the original bios. That this fan is controllable was not until a later bios version. It is ridicules. I have set mine to enable above 70°C.
I am not sure where the temps are measured so there is no educated answer i can give. But 70°C and above are fine.
Lets put it this way: the VRM has an over-temperature shutdown. If you manage to trigger that, you might consider changing a setting. It is impossible to destroy the VRM through heat, so just go for it.
Btw: I have heatkiller iV + 420. Dont be disappointed if your temperatures rach 85+degC 
VRMs are OK until 90-95 °C.
Technically most powerstages have thermal limits higher than that, 100-135 °C , but the capacitors around the powerstage will not be too happy with those kinds of temperatures for extended periods of time.
Yeah i know, mosfets don’t like high switching frequency at all.
Because it produce a tremendous amount of heat.
So yeah in terms of heat production a lower switching frequency is better for the fets.
However the down side of lower switching frequency is that it increases ripple current.
Because the slower the switching cycle, the higher the ripple amplitude.
And with the skipping cycles, then yeah…
It will increase more loss in the inductors then indeed.
It’s a pretty complex story that i don’t want to go too deeply into right now.
Because that would be something for a another time, lots of write up work.
But yeah gigabyte likely did something wrong with their bios.
And will likely being fixed with the new bios update.
When Gigabyte eventually does come out with a statement,
then it will likely be a pile of nonsene.
Still of course there is a problem at two sides.
Because AMD also comes with a Agesa update.
So yeah we will see when the initial bios update comes,
out with the fix and new Agesa.
A processor running hot has NOTHING to do with VRM control loop behavior. (which this problem was)
I was getting about 72 °C on the most intense P95 torture test with a NH-U14S TR4-SP3 + a second NF-A15 fan, in an open case.
I’m temporarily on an ASUS Zenith II Extreme Alpha (I have it, it’s a much inferior board to the GIGABYTE TRX40 Aorus Xtreme IMHO, despite what most reviews say) and I’m now getting way higher temperatures, around 80 °C I think (I will retest), with the same cooler and an open case.
Keep in mind that this isn’t a fair, scientific comparison of the two boards, first of all because I’m also temporarily on another 3970X, and because there are necessarily some variations between the setups, e.g. silicon lottery, thermal paste application, etc.
UPDATE: After 20 minutes of P95 in Small FFTs test on the ASUS board, the CPU is at 76 °C (Tdie) and the chipset is at 56.3 °C. Ambient is at 20 °C.
The Franz who who wished had Thunderbolt,
just wouldnt try to undervolt .
His Processor running hot,
does him bother not
Maybe you have to turn your processor around,
until another solution is found.
I know you guys think my post just was pathetic,
i just really feel poetic
4 Likes
I´m not sure who was talking about that?
But the cpu running hot has nothing to do with this nope.
Anyways i´m going to wait and see if the initial bios update,
with the new Agesa is going to fix the said problem for everyone.
BTW, note that I thought that prime95 was stable from few runs in windows.
Later on I run it in Linux, and volla the same error (with the highest power benchmark)
Thankfully BIOS F4C does fix that issue!! Phew.
I have the Designare TRX40
Unfortunately Gigabyte did not fix the issue for now.
It is borderline crazy what an chaos this company must be.
Remember: they have the fix, they proved it in an earlier BIOS. That bios was taken offline since it did not include the AGESA 1.0.0.3B fix by request from AMD (understandable action at the time).
They simply fail to include both fixes in one Bios. I really cant understand that. This is not due to Corona. This is incompetence and a lack of pride for the product.
Obviously they can push out bios versions in rapid succession when they want.
In my post above/earlier, i was positive about the future, but they simply stopped working on it, it seems. The crazy thing is: i went through my 3rd AORUS MASTER now. ALL had the same VRM problem. Clearly nothing to care about. What a joke. The whole product has a major (and consistent) problem in power delivery with devastating results to stability.
The frustrating thing is, that AMD is shielding Gigabyte. I have never had contact with them. And AMD is currently not working effectively (Corona), so this problem might likely to persist for now…
Maybe its time that Wendell makes a clickbait scandal video about which state of the product is deemed acceptable at Gigabyte.
1 Like
I think you’re forgetting something…the Intel 10th gen mainstream desktop launch. It’s quite possible that they have a skeletal crew and it makes sense to divert all of their resources towards making their Z490 launch as successful as they can, after all it’ll make them exponentially more money than the TRX40 mobos.
1 Like
Ha, i fully understand the issue. It is clearly not a technical problem, it is about not giving a shit…
Hello all,
I am in the process of purchasing a Gigabyte Aorus Extreme trx40 mainboard along with an AMD TR 3970X. I would appreciate if somebody could confirm that the stability issue under heavy AVX2 load solved. Thanks in advance !
I don’t believe so; the most current BIOS version on Gigabyte’s website is still listed as F5c, which DerAlbi has reported above as being unstable (including the AGESA fix but missing the VRM voltage fix)
Hello shawarma_bees,
Thanks for letting me know, I really appreciate it.
@shawarma_bees … I am pretty new to a lot of this, so please pardon my ignorance. But where do you see a BIOS Version F5c on Gigabyte’s website for the Aorus Xtreme TRX40? I can only find F2, F3, and F4d. I see that you wrote that reply 20 days ago (as I write this). Did they used to have a newer version that has since been removed? Or is there another page someplace that I do not yet know about?
Thank in advance!
-pvs
EDIT: Sorry, I just realized I posted in @DerAlbi’s thread (he’s running the Aorus Master) and not in mine. Apologies for the confusion.
I think @shawarma_bees may have mixed the Aorus Master and the Aorus Xtreme, as far as I recall BIOS versions differ between the two models.
I just noticed an instant ago that there’s a new BIOS version for the Aorus Xtreme, F4l:
I didn’t try it yet. I’m still running an unreleased beta BIOS, which is the only one that fixes the stability/correctness issues mentioned in this thread.
Hi @FranzB. No worries. And yes, F4l has been out for a few days now. It replaced a short-lived F4k, which has since been removed. I am already running it, though I do not see any differences. I’d love to know if it fixes the issues you discovered.yt
2 Likes