• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Crappy VRMs on B350s, few X370s as well - inadvisable Ryzen 7 overclocking over 1.32-1.35V

Dave_K

New Member
Joined
Aug 28, 2017
Messages
22 (0.01/day)
Hello, i am seeing all of the recommendations for B350 if you dont need SLI etc. but these are just wrong. Most manafacturers (Mainly MSI and Gigabyte) cheaped out on the VRMs on these boards and they are mostly from cancer components. I know that our glorious overclocker buildzoid did the breakdowns etc. MSI uses 4+2 design on all AM4 mobos except Xpower which has the same cheapo mosfets but 6 phases (but still runs hot AF). The used mosfets are Nikosem PK616BA for the highside and PK632BA for the lowside. Tomahawk-class trash motherboards use 4 highside vs X370s and B350 Carbon/Krait using 8 highside, both have 8 lowside. The main problem there is the PK616BA which is very slow (Tr,Tf) and has huge switching losses, even though they are doubled on some mobos, they still output similiar amount of heat which is combined with even worse "heatsinks" (on Carbon they are solid pieces of aluminium only like 2-3cm high). Gigabyte uses highside one 4C10N and two lowside 4C06N, the 4C10N is bad part as well, so avoid them, they are on all AM4 GBT mobos except K7/ Gaming 5. Best VRM on B350 is on Asus B350 Strix, Prime Plus, TUF B350M Plus, ASRock's "6" 3 phase on Pro4s and on the Carbon but it has shitty BIOS and memory support so i dont recommend it.
For comparison, MSI small 4+2 outputs approximately 19-20W of heat while running R7 at 1.42V (100A), MSI big 4+2 outputs approximately 17-18W, in reality runs cooler because of the doubling and surface area (as you will see in the graphs) and Asus´s solution (4phase, one 4C09B high and 2 lowside 4C06B) approximately 17-18W of heat, the heatsinks on the Strix are bit more "heatsinks" than on the carbon but still nothing groundbreaking (I have the Strix, had the Carbon). I am including the graphs and thermal imaging.
MSI X370 Pro Gaming Carbon - no airflow - applies to B350 Carbon as well
7589b791_16699-1494428954-bf410e94f4f11577572e72afb1b3ff01.jpeg

MSI B350 Tomahawk - should be avoided thanks to its crappy BIOS and memory support
900x900px-LL-f48c18f4_16440-1493410045-4fbfb96a842844a8d4ba89d3cb5a4913.jpeg

Thermal Image - Asus Strix B350-F - overvolted at 1.44V (at 1.35-38V i would expect 80-90°C on the VRM, 70-80°C on the heatsink)
350x700px-LL-546aa05b_629fa047-s.jpeg

Thermal image - Xpower Titanium at 1.402V. Really hot for motherboard expensive more than Crosshair 6 Hero
350x700px-LL-2bc33821_d5c076ca-s.jpeg

For comparison - Asus Crosshair 6 Hero
900x900px-LL-1e572632_16918-1495471218-6eb687d88f931519f46cc03af56fcbe8.jpeg


Hello, i see you use the Pictures from our Teammate and Overclockermember Saibot from www.Hardwareinside.de

https://www.hardwareinside.de/msi-b350-tomahawk-im-test-25145/
https://www.hardwareinside.de/community/threads/review-des-msi-x370-gaming-pro-carbon.4635/

Next Time pls, make a Sources for Copyrights. Thank you
 
Last edited by a moderator:
Most manafacturers (Mainly MSI and Gigabyte) cheaped out on the VRMs on these boards and they are mostly from cancer components.
This is not just for AM4 boards, these manufacturers also do some questionable decisions on FM2, AM3, LGA1150 and 1151 boards. AsRock does the same on the low-mid end.

It's not something new or previously unknown, but definitely worth pointing out.
 
This is not just for AM4 boards, these manufacturers also do some questionable decisions on FM2, AM3, LGA1150 and 1151 boards. AsRock does the same on the low-mid end.

It's not something new or previously unknown, but definitely worth pointing out.
Yeah. It is interesting that Gigabyte and MSI use complete trash on most while Asus and ASRock use pretty fine on few and very good on most.
 
Those temps still look within what I would consider safe , no idea on the quality of the mosfets and such.

But honestly running a 1700 at almost 1.5V is seriously exaggerated regardless of the board and quality of VRMs not sure what's the point of that.
 
Those temps still look within what I would consider safe , no idea on the quality of the mosfets and such.

But honestly running a 1700 at almost 1.5V is seriously exaggerated regardless of the board and quality of VRMs not sure what's the point of that.
The graphs show different voltages so you know what is the limit of your board. Of course 1.48V is too much.
 
Well , there are always going to be boards with bad VRMs.

I'd say the situation at least wont be as dire as it was back in the day when people were trying to run FX 9xxx chips on 4+1 phases. Since Ryzen doesn't clock as high anyway and has very good efficiency I am fairly confided we'll see far less "shit I fried my board cases" now.
 
Well , there are always going to be boards with bad VRMs.

I'd say the situation at least wont be as dire as it was back in the day when people were trying to run FX 9xxx chips on 4+1 phases. Since Ryzen doesn't clock as high anyway and has very good efficiency I am fairly confided we'll see far less "shit I fried my board cases" now.
Actually somebody already fried Tomahawk and it is the biggest problem, the VRMs wont go up in flames but at 100°C operating temperature their lifespan will decrease and they will eventually die
 
Let's face it, the only reason we don't see this on Intel too much is because its locked down and with AMD its not. It does not however change the reality that you need a solid board and good VRM to push any kind of OC if you want your gear to last. OC on B350 is just waiting for the inevitable fail.
 
Let's face it, the only reason we don't see this on Intel too much is because its locked down and with AMD its not. It does not however change the reality that you need a solid board and good VRM to push any kind of OC if you want your gear to last. OC on B350 is just waiting for the inevitable fail.
The better B350s are absolutely fine for Hexacore, not for 8 core. The idea of B350 being good for overclocking is given by people like Kyle (BitWit) or ScienceStudio who overclock on their review samples and have absolutely no idea about the VRMs or the temps of the VRMs. Thats just bad.
 
Hmm but it is Military Class 4, Guard-Pro: Latest evolution in high quality components for best protection and efficiency. lol.
tf6-pcb-military-class-4.png
 
I think this is all fairly common sense tbh. If you're buying a low end board you obviously aren't getting the same quality of components, so you know it's going to be hotter. Also if you're overclocking a 6-8 core CPU then you should know it's going to be consuming a huge amount of juice. You should be monitoring temperatures when oc'ing. And you definitely should have airflow going over the motherboard.

So hypothetically if you bought the bottom of the barrel AM4 motherboard paired it with an R7, put it in a case with no airflow, dialed the voltage up to 1.45v and it barbecued itself.. well, no sympathy really!
 
I think this is all fairly common sense tbh. If you're buying a low end board you obviously aren't getting the same quality of components, so you know it's going to be hotter. Also if you're overclocking a 6-8 core CPU then you should know it's going to be consuming a huge amount of juice. You should be monitoring temperatures when oc'ing. And you definitely should have airflow going over the motherboard.

So hypothetically if you bought the bottom of the barrel AM4 motherboard paired it with an R7, put it in a case with no airflow, dialed the voltage up to 1.45v and it barbecued itself.. well, no sympathy really!
Well, 8 core eats as much power as 7700K, i just want to point it out because i've seen the bad recommendations here on the forum
 
3.8 GHz, auto overclocking, I don't see temps that high.
This is after 30 min at full load with only the CPU fan active.
upload_2017-9-2_21-35-1.png
 
3.8 GHz, auto overclocking, I don't see temps that high.
This is after 30 min at full load with only the CPU fan active.
Yeah, these sensors tend to be inaccurate. Dont use auto overclocks. If you can, try to measure with IR thermometer
 
You'll note that ASRock was the only OEM who sent me AMD boards for review. I cannot recommend any other brands because of that.. clearly they didn't want me covering these particular products... because I got the Intel boards Z270 and X299, just not AMD. If it hasn't been covered on the front page of TPU, you should not buy it. ;) That includes Threadripper CPUs.
 
I think this is all fairly common sense tbh.
Not when you buy a motherboard which says "Overclocking", "Highest performance" and "Everlasting quality" all over the box.
Or names of famous overclockers, like Fatal1ty (damn you AsRock).

This is the reason I gave up explaining to customers why they should not overclock G3258 on a $45 H81 board.
They simply don't listen, because:
a) The manual says you can do it
b) Some dude on youtube says you can do it
c) Conventional wisdom (or as I call it sheep syndrome, or mass stupidity). "My friend did it, so who are you to tell me not to!"
 
I think this is all fairly common sense tbh. If you're buying a low end board you obviously aren't getting the same quality of components, so you know it's going to be hotter. Also if you're overclocking a 6-8 core CPU then you should know it's going to be consuming a huge amount of juice. You should be monitoring temperatures when oc'ing. And you definitely should have airflow going over the motherboard.

So hypothetically if you bought the bottom of the barrel AM4 motherboard paired it with an R7, put it in a case with no airflow, dialed the voltage up to 1.45v and it barbecued itself.. well, no sympathy really!

Those who buy the mid or lower boards should add direct airflow over them or water. This has been the same news for 990FX vs 970. I didnt go with a Crosshair but with a Sabertooth and the chip/temps are within normal range for load.

By the way Fatal1ty is not an Overclocker but a well known Gamer...
 
This is the reason I gave up explaining to customers why they should not overclock G3258 on a $45 H81 board.
They simply don't listen, because:
a) The manual says you can do it
b) Some dude on youtube says you can do it
c) Conventional wisdom (or as I call it sheep syndrome, or mass stupidity). "My friend did it, so who are you to tell me not to!"

Not to mention you used to be able to overclock anything on anything.
 
Not when you buy a motherboard which says "Overclocking", "Highest performance" and "Everlasting quality" all over the box.
Or names of famous overclockers, like Fatal1ty (damn you AsRock).

Honestly after dozens of years of advertisements people should know better.

And if they don't , nothing is going to change that now.
 
You'll note that ASRock was the only OEM who sent me AMD boards for review. I cannot recommend any other brands because of that.. clearly they didn't want me covering these particular products... because I got the Intel boards Z270 and X299, just not AMD. If it hasn't been covered on the front page of TPU, you should not buy it. ;) That includes Threadripper CPUs.
I dont see a single reason why you shouldnt recommend them. Their quality is good all across the spectrum, if i had the cash i would probably buy Taichi any day of the week.
 
Good old MSI uses Nikos mosfets on all of their AM4 lineup, even the top end X370 Titanium. I have those on my MSI 970 gaming too, what a waste of money that board is.

 
Good old MSI uses Nikos mosfets on all of their AM4 lineup, even the top end X370 Titanium. I have those on my MSI 970 gaming too, what a waste of money that board is.
See the thermal image how frickin hot it runs. I watched that video and even messaged builzoid if my calculations are correct so i can perform them like he does
 
Back
Top