[SOLVED] Is it worthwhile to upgrade my motherboard?

Dec 28, 2020
4
0
10
I recently upgraded my i3 8100 to an i7 9700k because they went on sale for $200 at Microcenter. I'm still using the same Aorus B360 Gaming 3 Wifi motherboard I used with the 8100. I already have a Cooler Master ML240 AIO. As of now, the 9700k is turbo-ing to around 4200MHz consistently. The VRM on the motherboard is definitely bottlenecking the CPU, considering the CPU temps stay around 50C under load. If I were to upgrade to a Z series mobo I would aim for a 5GHz OC. Are the performance gains from buying a ~$160 Z390 board worth it, or is it more worthwhile to just keep my old mobo? I would hate to feel like I'm leaving a significant amount of performance on the table. My other components are a RTX 3060ti, 16GB DDR4 2666 and a 650 watt PSU. Thanks for the help.
 
Solution
Ok, so here's the deal.

Your motherboard VRMs use both the 4C10N and 4C06N mosfets. Looks like the maximum temperature for those is recommended to be 150°C so technically you are not exceeding, or even coming close to, the maximum VRM temperature based on the specifications for the mosfets on your board. There is a problem though.

Your CPU SHOULD be boosting to 4.6Ghz with all cores at full load, and it is obviously not doing that. You are at about 3.6Ghz all core clock speed and that indicates a problem of some kind to me. Not sure WHAT, exactly, but the obvious inference here would be that the motherboard or cooling system are not capable enough to deal with the demands of that boost profile.

Couple of things I'd check first...
I recently upgraded my i3 8100 to an i7 9700k because they went on sale for $200 at Microcenter. I'm still using the same Aorus B360 Gaming 3 Wifi motherboard I used with the 8100. I already have a Cooler Master ML240 AIO. As of now, the 9700k is turbo-ing to around 4200MHz consistently. The VRM on the motherboard is definitely bottlenecking the CPU, considering the CPU temps stay around 50C under load. If I were to upgrade to a Z series mobo I would aim for a 5GHz OC. Are the performance gains from buying a ~$160 Z390 board worth it, or is it more worthwhile to just keep my old mobo? I would hate to feel like I'm leaving a significant amount of performance on the table. My other components are a RTX 3060ti, 16GB DDR4 2666 and a 650 watt PSU. Thanks for the help.
Well, 500+MHz gain is nothing to sneeze at. It's also limiting you RAM speed
 
50°C is NOT causing any kind of thermal throttling. That wouldn't happen, or even BEGIN to happen, until you were hitting somewhere over 85°C, and most probably not until you got much closer to something over 90°C, as far as the CPU temps are concerned.

The VRM throttling on the other hand, who knows. Not the greatest board obviously, even for B360, which itself is obviously not the performance chipset. Not sure exactly what the VRM configuration on that board looks like but generally speaking the Gaming series boards from Gigabyte (Gaming 3, 5 and 7) typically have pretty decent VRM configurations on the Z boards. Yours has a 4+3 VRM configuration and for the stock boost profile that OUGHT to be ok.

Have you actually taken at the VRM temperatures (And whether it specifically SAYS "yes" in the thermal throttling section) by downloading HWinfo (NOT HWmonitor, or some other utility), installing it (Or running the portable version) and choosing the "sensors only" option while disabling the "summary" option and then looking at the VRM temperatures?

While obviously there are some benefits to having a Z board, I wouldn't even THINK about overclocking, at all, much less to "5Ghz" using any of the 240mm coolers out there unless you want to be religiously annoyed by the constant sound of your radiator fans. For stock operation it would be fine, and with that cooler that's exactly how I'd run it, but obviously your mileage may vary and what you do in terms of overclocking is up to you but honestly I wouldn't even bother spending the money on a Z board unless you can:

Verify for sure that there is throttling going on due to the VRMs.

And.

Unless you plan to swap that memory out for a higher speed kit, because that is where the majority of gain will come from switching to a Z board is the ability to run much faster memory kits. If you are going to stick with the 2666mhz memory kit, then it's almost pointless unless there is some other specific feature you need that you don't currently have. Even overclocking isn't going to give you much because there's not that much overhead there to begin with on the 9700k beyond what it already will boost to on most samples and any gains will definitely be dampened by that slower memory kit.
 
Dec 28, 2020
4
0
10
50°C is NOT causing any kind of thermal throttling. That wouldn't happen, or even BEGIN to happen, until you were hitting somewhere over 85°C, and most probably not until you got much closer to something over 90°C, as far as the CPU temps are concerned.

The VRM throttling on the other hand, who knows. Not the greatest board obviously, even for B360, which itself is obviously not the performance chipset. Not sure exactly what the VRM configuration on that board looks like but generally speaking the Gaming series boards from Gigabyte (Gaming 3, 5 and 7) typically have pretty decent VRM configurations on the Z boards. Yours has a 4+3 VRM configuration and for the stock boost profile that OUGHT to be ok.

Have you actually taken at the VRM temperatures (And whether it specifically SAYS "yes" in the thermal throttling section) by downloading HWinfo (NOT HWmonitor, or some other utility), installing it (Or running the portable version) and choosing the "sensors only" option while disabling the "summary" option and then looking at the VRM temperatures?

While obviously there are some benefits to having a Z board, I wouldn't even THINK about overclocking, at all, much less to "5Ghz" using any of the 240mm coolers out there unless you want to be religiously annoyed by the constant sound of your radiator fans. For stock operation it would be fine, and with that cooler that's exactly how I'd run it, but obviously your mileage may vary and what you do in terms of overclocking is up to you but honestly I wouldn't even bother spending the money on a Z board unless you can:

Verify for sure that there is throttling going on due to the VRMs.

And.

Unless you plan to swap that memory out for a higher speed kit, because that is where the majority of gain will come from switching to a Z board is the ability to run much faster memory kits. If you are going to stick with the 2666mhz memory kit, then it's almost pointless unless there is some other specific feature you need that you don't currently have. Even overclocking isn't going to give you much because there's not that much overhead there to begin with on the 9700k beyond what it already will boost to on most samples and any gains will definitely be dampened by that slower memory kit.
Thanks for the suggestion. I ran a Prime95 stress test for about 30 min with HWinfo on. Although the VRMs did get pretty hot, they didn't throttle the CPU at all.
View: https://imgur.com/a/6Tdsqyv

I think I'm just going to stick with the B360 board unless I can find a really good deal. Thanks again for the help.
 
Of course, that is your prerogative, but I might have to reverse my earlier opinion based on what I see in your HWinfo window. I see the VRM MOS maximum temperature of 99°C and that is WAY too high, and tells me that at least at some points the VRMs are getting pretty warm. I'd say, without looking up the specifications of the specific mosfets used on your motherboard, that you really don't generally want to see your VRM mosfet temperatures to go beyond 105°C, MAX, in most daily driver configurations. Because at that temperature you can expect that some kind of throttling probably IS happening even if you're not seeing it in HWinfo.

And I'm not totally sure you're not because I don't see it there. Do me a favor. Run HWinfo, put your mouse in between the first and second columns and drag it to the right a bit to widen the first column which has the description of what sensor is being measured and monitored there so that we can see the full description name. Next, run Prime95 Small FFT. Do not run Blend, or Smallest FFT, or Large FFT. Only run "Small FFT", AND, be sure in the first window that pops up when you run Prime95 to disable ALL AVX options that present themselves. Usually you have to disable them one at a time in order to disable the next one which will UN-gray once you've disabled any other AVX option that is visible. So the end result is that you run Small FFT without AVX at all.

Run it for about a minute and then begin taking screenshots of ALL the HWinfo sensors. It usually requires about three screenshots, scrolling down to view the next set of sensors and capture them in the window, to get them all. I want to see if there is anything at all that stands out as a legitimate concern for any of those sensors. In this case I think you can safely ignore any of the drive or graphics card sensors, so it may be that only two screenshots are necessary since those devices are normally listed toward the bottom of the HWinfo listings.

If the VRMs truly are reaching 99°C while running non-AVX Small FFT, then that is probably something to be concerned about.
 
Dec 28, 2020
4
0
10
Of course, that is your prerogative, but I might have to reverse my earlier opinion based on what I see in your HWinfo window. I see the VRM MOS maximum temperature of 99°C and that is WAY too high, and tells me that at least at some points the VRMs are getting pretty warm. I'd say, without looking up the specifications of the specific mosfets used on your motherboard, that you really don't generally want to see your VRM mosfet temperatures to go beyond 105°C, MAX, in most daily driver configurations. Because at that temperature you can expect that some kind of throttling probably IS happening even if you're not seeing it in HWinfo.

And I'm not totally sure you're not because I don't see it there. Do me a favor. Run HWinfo, put your mouse in between the first and second columns and drag it to the right a bit to widen the first column which has the description of what sensor is being measured and monitored there so that we can see the full description name. Next, run Prime95 Small FFT. Do not run Blend, or Smallest FFT, or Large FFT. Only run "Small FFT", AND, be sure in the first window that pops up when you run Prime95 to disable ALL AVX options that present themselves. Usually you have to disable them one at a time in order to disable the next one which will UN-gray once you've disabled any other AVX option that is visible. So the end result is that you run Small FFT without AVX at all.

Run it for about a minute and then begin taking screenshots of ALL the HWinfo sensors. It usually requires about three screenshots, scrolling down to view the next set of sensors and capture them in the window, to get them all. I want to see if there is anything at all that stands out as a legitimate concern for any of those sensors. In this case I think you can safely ignore any of the drive or graphics card sensors, so it may be that only two screenshots are necessary since those devices are normally listed toward the bottom of the HWinfo listings.

If the VRMs truly are reaching 99°C while running non-AVX Small FFT, then that is probably something to be concerned about.
I ran two more stress tests with Prime95. Both with Small FFT, one with AVX on and another with AVX off. To clarify, my first test was Smallest FFT with AVX on.
View: https://imgur.com/a/ZpFsoLJ

The VRM temps got way too hot, especially considering you want the max to be 105°C. I only ran the test for a couple minutes. This definitely looks like something I should be worried about.
 
Ok, so here's the deal.

Your motherboard VRMs use both the 4C10N and 4C06N mosfets. Looks like the maximum temperature for those is recommended to be 150°C so technically you are not exceeding, or even coming close to, the maximum VRM temperature based on the specifications for the mosfets on your board. There is a problem though.

Your CPU SHOULD be boosting to 4.6Ghz with all cores at full load, and it is obviously not doing that. You are at about 3.6Ghz all core clock speed and that indicates a problem of some kind to me. Not sure WHAT, exactly, but the obvious inference here would be that the motherboard or cooling system are not capable enough to deal with the demands of that boost profile.

Couple of things I'd check first though. Make sure you have the most up to date STABLE (non-BETA) BIOS version installed.

In Windows, make sure you have the Performance power plan enabled AND in the advanced power plan settings make sure that the minimum CPU power state is set to 8% and the maximum is set to 100%.

In the BIOS, make sure that Intel speed step is enabled and I would recommend that you DISABLE Intel speed SHIFT, if it is even present, which it should be.

If all that is as I've outlined and there are no other clear indications of what the problem is, then I'd suggest, tentatively, that a new board with a better VRM configuration, whether that means a Z board or not (Makes little sense to NOT go with a Z board, but it needs to be a decent one because there are still pretty crappy Z boards as well) is up to you but seems implausible to change boards and NOT go with a decent Z board if you're going to spend the money on something.
 
Solution

TRENDING THREADS