About Msi B450i

[eatmc7]

Hi everyone, a newbie here. I have this board and 2200G for 2 weeks now, first time building my own pc and first time meeting with overclocking.

At first i used it only by OCing the Igpu to 1300mhz and after learning about the frequency hole between 1300 and 1500Mhz on this vega gpu im currently using it at 1500Mhz @ 1,1750volts for 3 days now. Also the cpu is at 3.8Ghz @ 1,3375volts. For Cpu i tried offset mode at first when i didnt want to OC the Cpu and wanted it to not have high voltages and it looked like that offest mode was not working at all. Then i cancelled that plan and just wanted it to have low stable voltage so i gave it a 1,3300 volts at stock speeds but this time i experienced a weird a voltage reading ( had a topic in reddit about it ) Appearently my bios ignored the value i typed this time or it was a wrong sensor reading. So at the end i ended up with 3.8 @ 1,3375 volts.

I have a g.skill 3200Mhz Cl16 16Gb ram kit but running them on 2993Mhz because windows felt like its running weird at 3200Mhz.

In this baord it only allows you to go up and down by 0,0125 volts. After i watched this video im afraid to even come close to what most people says safe.

My main concern is that the mosfets in this board that are feeding the integrated gpu doesnt have a heatsink, is there a way to know if the temperature im seeing on the hwinfo is average temperature of all vrms or it shows the highest one or something? Because im really afraid of cooking the vrms feeding the Igpu over time that doesnt have heatsink.

I will leave my bios settings and and hwinfo values of a 2-3 hours of gaming-surfing session and please guide me if any value seems dangerous or wrong.

(In the post preview pictures seem large and tried whats been said in faq, sorry if they are still stretched)

My bios settings;


Values;

 

[Johan45]

From what I see your temps look fine and you still have some room to push the CPU up another 1-200 MHz IMO
Now the memory that was a thing with the first-gen ZEN, you most likely have Hynix and it can be hard to deal with on that platform. Have you ever updated the BIOS?

 

[trents]

eatmc7, as you know from our email correspondence I have this same motherboard, but am running a 2700x at stock on it, though I have experimented a little with overclocking.

If you want to be able to adjust the voltages in finer increments than .0125 you can use the CPU LLC which is found in a subset of DigitAllPower when the overclock mode is set to expert. That is the only way to achieve finer grained adjustments. Click on the DigitAllPower line and you will see it. The lack of granularity in the main CPU voltage adjustment is likely due to storage space limitations on the 16mb bios chip, a common issue on lower end socket AM4 motherboards with this chipset because of the addition of all the AEGESA coding necessary to make them compatible with Zen 2. The CPU LLC has 9 manual steps with 1 being the highest voltage boost. Not sure what bios version are running but I am running AB which I think is the latest.

I cannot answer your questions about which set of mosfets is being referenced in what HWInfo64 reports, or whether that is the hottest one, an average or whatever. But the reading to look at is VR MOS. From what I have read the mosfets on this board are safe to at least 100c but I wouldn't push that for long periods of time.

This motherboard has a robust VRM section for a board in it's price range (6+2 I think) but like essentially every ITX board out there there isn't a heat sink on the VRM component to the north of the socket, the ones that you say cover the IGP. There just isn't room for it.

Please create a Signature (Go to Settings and look along the left side for Edit Signature) to list your system components. That way, they travel with every post you make.

 

[eatmc7]

From what I see your temps look fine and you still have some room to push the CPU up another 1-200 MHz IMO
Now the memory that was a thing with the first-gen ZEN, you most likely have Hynix and it can be hard to deal with on that platform. Have you ever updated the BIOS?

First thing i did was updating bios. In stress tests CPU was around 85 degrees as far as i remember so i didnt go further with it. And when i had the rams at 3200 it made me feel like it was more about the CPUs memory controller...

eatmc7, as you know from our email correspondence I have this same motherboard, but am running a 2700x at stock on it, though I have experimented a little with overclocking.

If you want to be able to adjust the voltages in finer increments than .0125 you can use the CPU LLC which is found in a subset of DigitAllPower when the overclock mode is set to expert. That is the only way to achieve finer grained adjustments. Click on the DigitAllPower line and you will see it. The lack of granularity in the main CPU voltage adjustment is likely due to storage space limitations on the 16mb bios chip, a common issue on lower end socket AM4 motherboards with this chipset because of the addition of all the AEGESA coding necessary to make them compatible with Zen 2. The CPU LLC has 9 manual steps with 1 being the highest voltage boost. Not sure what bios version are running but I am running AB which I think is the latest.

I cannot answer your questions about which set of mosfets is being referenced in what HWInfo64 reports, or whether that is the hottest one, an average or whatever. But the reading to look at is VR MOS. From what I have read the mosfets on this board are safe to at least 100c but I wouldn't push that for long periods of time.

This motherboard has a robust VRM section for a board in it's price range (6+2 I think) but like essentially every ITX board out there there isn't a heat sink on the VRM component to the north of the socket, the ones that you say cover the IGP. There just isn't room for it.

Please create a Signature (Go to Settings and look along the left side for Edit Signature) to list your system components. That way, they travel with every post you make.

Okay i cheated a bit with yours about signature, dont mind me about it please

I think i will have a look about those LLC settings in the future but i dont have the courage with the lack of information of myself at the moment.

And about those VRMs.. I hope something like the actual board voltage being higher than what we type in the bios is not happening in our boards and im not coming close to 100c on those mosfets with 1,1750volts then..

 

[trents]

Are your VR MOS temps showing variance between idle and load? If so, the temp sensor in that area would seem to be working. I would worry more about temps than voltages. I think it's the watts that count, especially if you are not sure the mosfet voltage is being reported correctly.

Though I have a different CPU than you, here's a pic of HWMonitor reports for my system after running a bit of OCCT to compare yours by:

 

[eatmc7]

When i do cpu intensive loads VR MOS value increases like how a Vrm with a heatsink on it with a little airflow would heat up but for example when i fire up furmark VR MOS changes like 1c after 2 minutes so i am assuming VR MOS doesnt cover the Vrms that are feeding
the IGpu at all or it does show the average temparture of all the Vrms.

Also i think this is not important anymore because i think i found that these 2 are the seperate charts for those 2 Vrms for the IGpu; (screenshot is only to show which readings i am talking about, not after stress tests)


Because the Vr Loop2 reading on the first one and Vr Loop1 reading on the other rises up like how you would expect a Vrm without a heatsink on it with a little airflow would rise up. I did a fire up cinebench r20 and furmark at the same time and these 2 reading ( which i am almost certain that they show the 2 Vrm Of Igpu) hit maximum of 85 ish degrees in 5 minutes. When i stop the furmark they are back to 50ish degrees in 10-15 seconds.

Only thing makes me a little confused about this now is that in your last picture your Vr Loop1 reading is also high even tho you dont have an Igpu, maybe they show the highest temperature of Vrms?
Edit; i mis-saw it as VR MOS was 89 and Vr Loop was 90 not the vice versa, so that means Vr Loop is not showing the highest for sure. So can it be that those +2 phases feeding the cpu as well when there is no Igpu? I thought they are dedicated for Igpu if theres one till now but i dont know.

 

[trents]

From what I have read, the +2 VRM section on most motherboards provides power for everything other than the CPU. That would include but not be limited to the IGP. And this arrangement may not be true for all motherboards.

https://www.maketecheasier.com/what-is-vrm/
"VRMs are typically sold as something like “8+3” or “6+2.” The number before the plus indicates the number of phases dedicated to cleaning power for the CPU. The number after the plus indicates the VRM phases left to power other motherboard components like RAM."

http://blog.logicalincrements.com/2019/02/what-are-how-many-motherboard-vrms-power-phases/
"There are two separate groups of power phases in the VRM. One is used for the CPU cores, and the other is used by other parts of the CPU, like the integrated GPU. On your typical motherboard, the power phases used for the CPU cores (the ones we care about most) are found left of the CPU while the others are above it, but this isn’t always the case, especially for smaller motherboards."

Cinebench R20 does not utilize the IGP for rendering images. I would suggest you run a fairly long stress test with Realbench which does use the GPU and the CPU for processing the images. Then post back with an attached pic.

Last night I added a fan inside my case to blow on the VRM heatsink and it knocked down my VR MOS temp about 6c from what it had been after 12 minutes of OCCT. On a longer test the net gain was more like 3-4c. Heat saturation happens over time and is very dependent on case ventilation and volume as after a while, even when you have fans blowing on things, they are just blowing hot air onto them.

What case make and model are you using?

 

[eatmc7]

I run cinebench and furmark at the same time when i do stress it. Case is as in the signature. It has a fan on top of the cpu fan which provides some direct airflow to the Vrms that comes out of the back of the case.

 

[trents]

Okay, your Sig is now showing. It wasn't until now.

 

[eatmc7]

Yea it started to show this morning for me and interstingly its not shown under my older posts

 

[trents]

I did some monitoring HWInfo64 on another AM4 system I have with a mATX ASRock motherboard running a 3000G APU. The line item labels in HWInfo64 were very different than on my main system with MSI b450i and 2700x. There was no VR Mos label at all. I suppose different motherboards have different sensors in different locations. A couple of the line items were showing like 110-111c temps but I touched all relevant components with my finger tip and a laser temp gun as well and nothing was any where near that hot.

 

[eatmc7]

Which motherboard was it exactly?

Below are the results after 13 minutes of cinebenchr20 + furmark(900p with x2 antialiasing). And i also uploaded a 46 seconds video( 9 Mb) of what i think are the temperatures of those +2 Vrms. You can judge better to decide to agree with me or not. (The video starts just when i end the stress tests; https://drive.google.com/open?id=1rP-MLfstNHRU43_xGRPkMzVP-bv3825R )

https://prnt.sc/r2pfmy This is the image of my 120mm case fan sitting on top of the motherboard just on top of the cpu fan but a bit larger towards the Vrms, i think its providing around 15-16 cfm airflow.

 

[trents]

ASRock B450M HDV v.4

- - - Auto-Merged Double Post - - -

Your max CPU tdie/tctl temps are essentially equivalent to mine but your VR MOS temps are much lower. My 2700x is putting much more strain in the VRM section.