• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

can 6 phase VRM boards handle 3900X? would like feedback from people who tried it

Stock EDC (short-term) limit is 140A; PBO won't take his chip far past that, you have to run fixed freq for that. And like I've said a billion times, fixed freq is not the way to go.
Sounds like you have found a way to cool 7nm density before having overclocked it to the limit.
 
Sounds like you have found a way to cool 7nm density before having overclocked it to the limit.

Sounds like you haven't read the thread. "I don't plan on hardcore OCing, but I will probably flip on PBO."

PBO isn't going to hit 200A.
 
Sounds like you haven't read the thread. "I don't plan on hardcore OCing, but I will probably flip on PBO."

PBO isn't going to hit 200A.
Sounds like 200A will not trigger FIT a while earlier, /s.
EDC is the best 'power virus' management utility. It doesn't allow heat pockets to develop in the first place.
PS: if you are interrupting pbo, you must take preventive measures not to fiddle with fixed frequency performance. One is to keep FIT at bay.
 
I found 6 that meet the requirements stated in the OP.
Every one of these will work well for the OP's needs.

not true. the gigabyte board is not good enough. i have it (see attachment). it does not feed the CPU enough power. the 3900X should score around 7100 in cinebench R20. on the gigabyte B450-i it gets between 6300 and 6800. always a random number. i thought my board was bad or the CPU was a dud. the CPU checks out fine in a low end lenovo desktop board with the same SSD* and RAM. the replacement gigabyte B450-i i got is equally worthless. does the same thing.

i also tried other benches and tried running an adobe premiere job. the gigabyte board comes up short every time. i started this thread specifically because the gigabyte B450-i that i have does not do what i need it to do.

those asrock boards are probably even worse. i've seen reviews of those boards, from both professionals and users, saying they gimp R7 CPUs. R9 need not apply. they both use the same worthless 3 phase VRM, with doublers, and terrible cooling.



back to the topic, thank you everyone for the input. there is some amazing in-depth information here. i learned more than i thought i would. i will go with the MSI board. i will only need 1 M.2 slot and don't care about the audio chipset. i will update this post to let you guys know how it handles the 3900X.
 
Last edited by a moderator:
Lots of good info here, very educational.
 
not true. the gigabyte board is not good enough. i have it (see attachment). it does not feed the CPU enough power. the 3900X should score around 7100 in cinebench R20. on the gigabyte B450-i it gets between 6300 and 6800. always a random number. i thought my board was bad or the CPU was a dud. the CPU checks out fine in low end lenovo desktop board with the same memory and RAM. the replacement gigabyte B450-i i got is equally worthless. does the same thing.

i also tried other benches and tried running an adobe premiere job. the gigabyte board comes up short every time. i started this thread specifically because the gigabyte B450-i that i have does not do what i need it to do.

those asrock boards are probably even worse. i've seen reviews of those boards, from both professionals and users, saying they gimp R7 CPUs. R9 need not apply. they both use the same worthless 3 phase VRM, with doublers, and terrible cooling.



back to the topic, thank you everyone for the input. there is some amazing in-depth information here. i learned more than i thought i would. i will go with the MSI board. i will only need 1 M.2 slot and don't care about the audio chipset. i will update this post to let you guys know how it handles the 3900X.
3900X in some circumstances could reach 7300~7400 in R20.

Out of curiosity... what Lenovo board was that? Did you test it in the same case as the B450I Aorus?

To take it further, can you screenshot a HWiNFO sensors page during R20 run?

Like this where all info available/visible...

I’m interested in this kind of info, but at the same time I like to see if I can find anything else that could be wrong, other than board’s VRM.
Although 4 phases of 50A each isn’t good enough for a 140A CPU... That’s not hard to think.
 
Last edited by a moderator:
3900X in some circumstances could reach 7300~7400 in R20.

Out of curiosity... what Lenovo board was that? Did you test it in the same case as the B450I Aorus?

To take it further, can you screenshot a HWiNFO sensors page during R20 run?

Like this where all info available/visible...

I’m interested in this kind of info, but at the same time I like to see if I can find anything else that could be wrong, other than board’s VRM.
Although 4 phases of 50A each isn’t good enough for a 140A CPU... That’s not hard to think.

the B450-i was on a test bench. not in a case. but i used the same ssd and memory. the lenovo PC is a thinkcentre. can't remember the exact model. i'll get back to you on that. it originally came with a 2700X pro CPU. it's a regular, off the shelf office PC. nothing special.

i'll try to get that HWinfo and post it here.
 
the B450-i was on a test bench. not in a case. but i used the same ssd and memory. the lenovo PC is a thinkcentre. can't remember the exact model. i'll get back to you on that. it originally came with a 2700X pro CPU. it's a regular, off the shelf office PC. nothing special.

i'll try to get that HWinfo and post it here.
For the Lenovo board it’s not all that significant. The most I wanted to know was the test conditions and if it was the same, but as it turns out the B450-I test bench is better for thermals.
 
those asrock boards are probably even worse. i've seen reviews of those boards, from both professionals and users, saying they gimp R7 CPUs. R9 need not apply. they both use the same worthless 3 phase VRM, with doublers, and terrible cooling.

The Asrock VRM is crap...it has Onsemi discrete MOSFETS whereas MSI or Gigabyte use integrated power stages here. The efficiency and heat dissipation of those discrete MOSFETS is not as good as those of power stages. And as it has only 3 real phases, they do a twin arrangement of 3x2 phases to reduce load and heat on each phase. But also MSI use discrete MOSFETS on their X570 Gaming Plus or Gaming Edge Boards...not the best solution but cheap...
The next problem is output ripple which degenerates cpu over time and which is worse with 3 real phases compared to 6 real phases.
 
Last edited:
The next problem is output ripple which degenerates cpu over time and which is worse with 3 real phases compared to 6 real phases.
This is absolute nonsense. Voltage ripple does not degrade CPU's unless excessive voltage is being applied, but then you have a much bigger problem.
 
Sure, if you have 25mV ripple and you need for a stable oc 1.375V then you have to set 1.4V to get stable. So the Chip degrades at 1.4V but you only would need 1.375V.
 
That was only an example, with higher voltage the effect is still there.
 
That was only an example, with higher voltage the effect is still there.
Whether or not voltage ripple is present is not what I'm taking issue with. Voltage ripple ALWAYS exists because everyone in the world runs on AC power grids. What defines how great the effect is depends on the PSU and mobo quality. A solid PSU will make even a low quality board run well, where as a crap PSU will make even the best board run like crap.

The OP is not going to match a solid board to a crap PSU, so voltage ripple will not be an issue. And for the record, 3x2 phase power is electronically as a stable a 6x1. The VRM part quality is more important than that of the end configuration.
 
Last edited:
The more phases you have the less the ripple effect, also the higher your pwm controller is switching the less the ripple effect or you have crap or good inductors may also be a point here. Why do oc mainboards have the Feature to adjust PWM switching frequency? One point is to reduce ripple.
A real 6 phase vrm is way better than a 3x2 or 3 phases doubled vrm.

Edit:
It depends not on the PSU and a solid PSU doesn‘t compensate a crap vrm in any way.
 
Last edited:
I can confirm that PBO+ works fine with a 3900X using both an MSI B450M Mortar Max (2+2 phase) and Asrock B450 Pro (3x2 phase). Ryzen Master shows power draw of between 135 and 140W with my two samples.

These are both relatively low end boards for a 3900X but with half-decent VRM and proper heatsinks they don't seem to have any issues with PBO+

I'm not sure I'd be going for overclocking records with them but IMO there's really no point pumping huge amounts of juice into Zen2. It's 98% of the way there using stock voltages on PBO+ and whilst you can definitely push it higher than that, throwing silly amounts of voltage at them will get you a hundred MHz maybe, but an earful of fan noise and potentially short lifespan of the chip.
 
I can confirm that PBO+ works fine with a 3900X using both an MSI B450M Mortar Max (2+2 phase) and Asrock B450 Pro (3x2 phase). Ryzen Master shows power draw of between 135 and 140W with my two samples.

These are both relatively low end boards for a 3900X but with half-decent VRM and proper heatsinks they don't seem to have any issues with PBO+

I'm not sure I'd be going for overclocking records with them but IMO there's really no point pumping huge amounts of juice into Zen2. It's 98% of the way there using stock voltages on PBO+ and whilst you can definitely push it higher than that, throwing silly amounts of voltage at them will get you a hundred MHz maybe, but an earful of fan noise and potentially short lifespan of the chip.
Try running cinebench r20 Back to back a few times, let me know what score you get and if it’s consistent or not.
 
If one of these boards then take the Mortar, stay away from the Asrock. The Mortar has 4 phases and 2 Low/High Side MOSFETs per phase, each capable of 80A max. They have also less resistance.
The Asrock only 3 phases, twin arrangement, with 2 Low/High Side MOSFETs per phase, each capable of 65A max and higher resistance, so MSI should stay cooler here.
 
Last edited:
Crank up voltage and LLC, move some air over VRMs and let er rip. . It only gets warm if you let it :D Guys are saying it runs, kick it in the nuts it'll do something for awhile. If you want to want to run your overclocks 24/7 you should be buying a good board anyways to run those overclocks.
 
You're sticking these parts into an ITX case, are you really going to be stressing it that much?

if the motherboard buckles under stress from a 10 second benchmark, then it will not last long. this is a complicated build and if i have to take it apart one more time i might just flip out.
 
If one of these boards then take the Mortar, stay away from the Asrock. The Mortar has 4 phases and 2 Low/High Side MOSFETs per phase, each capable of 80A max. They have also less resistance.
The Asrock only 3 phases, twin arrangement, with 2 Low/High Side MOSFETs per phase, each capable of 65A max and higher resistance, so MSI should stay cooler here.
Your opinions are high subjective. The MSI has a much older audio chipset.

if the motherboard buckles under stress from a 10 second benchmark, then it will not last long. this is a complicated build and if i have to take it apart one more time i might just flip out.
Fair enough.
 
Last edited:
I have an AsusROG Strix B450-F (4 phase) and an Asus ROG Strix X470-F (6 phase) and they both worked very well with my 3700x.
And from the info I found here:

The B450-I and X450-I and both a step above the B450-F and a slight step down from my X470-F as far as VMs go so either should run the 3900x just fine with PBO set on auto and decent case airflow.
 
Your opinions are high subjective. The MSI has a much old audio chipset.

Ok, that could be true, didn't look for other components of those mainbaords as OP asks a vrm question.

The B450-I and X450-I and both a step above the B450-F and a slight step down from my X470-F as far as VMs go so either should run the 3900x just fine with PBO set on auto and decent case airflow.

The Asus X470 and the MSI B450I should be vrm wise nearly the same, the Asus has a slightly better SOC vrm as far as I can see. As he looks for an ITX board I woulod recommend the MSI.
 
Back
Top