Do Not Use Z490 Motherboard Auto Settings: Excessive Voltage, Power, & Heat
video description
Date: 2020-05-29
Related videos
Comments and reviews: 10
Islam
Something I noticed with 3400G APU on ASUS B450-MA is the very high peak voltage 1.568 on default settings which does not look safe at all. The plan was to assemble parts on a new chase open bios of the new rig to start memory OC + disable CPU boost for more stable memory OC+ set multiplier to x37 for best gaming performance, but at the first look at the bios 1.586V Vcore. That is insane. I had to disable PBO Which was something new since Ryzen 3000 release last year to notice that voltage has become normal around 1.2 V. Recently AMD pushed a new microcode for my ASUS mobo which has disabled PBO by default, set peak voltage to 1.45 which looks more reasonable and set default multiplier to x41 and can boost to 4.2 GHz. again I had to set multiplier to x37, redo memory OC which took a week to be done again and lower Vcore voltage for more power saving. I noticed that CPU does not work at all if Vcore peak voltage is droped to 1.2V at whatever multiplier is. also using -offset seems to make mobo never boost at -0.15 which should put peak voltage 1.3V so looks like AMD has changed adaptive voltages by frequency and other factors that idk so I had to use Ryzen master for lowering peak voltage only to 1.2375V. I don't like how AMD engineer this crap. There is a reason why I love Intel, but AMD has better price to performance which makes sense to buy AMD What if I as like any regular customer who bought this APU and placed it on this ASUS mobo. Regular customer won't even look at bios settings and tweak the insane 1.568 or 1.586 Vcore not sure. His CPU would be dead now and price to performance will suffer as a result not to mention headaches. AMD rushes their technology without carefulness and we customers would pay the price if it stays like that. I don't recommend AMD for anybody unless they know enough about tweaking. Now after your video I think there are some troubles with intel now.
reply
Something I noticed with 3400G APU on ASUS B450-MA is the very high peak voltage 1.568 on default settings which does not look safe at all. The plan was to assemble parts on a new chase open bios of the new rig to start memory OC + disable CPU boost for more stable memory OC+ set multiplier to x37 for best gaming performance, but at the first look at the bios 1.586V Vcore. That is insane. I had to disable PBO Which was something new since Ryzen 3000 release last year to notice that voltage has become normal around 1.2 V. Recently AMD pushed a new microcode for my ASUS mobo which has disabled PBO by default, set peak voltage to 1.45 which looks more reasonable and set default multiplier to x41 and can boost to 4.2 GHz. again I had to set multiplier to x37, redo memory OC which took a week to be done again and lower Vcore voltage for more power saving. I noticed that CPU does not work at all if Vcore peak voltage is droped to 1.2V at whatever multiplier is. also using -offset seems to make mobo never boost at -0.15 which should put peak voltage 1.3V so looks like AMD has changed adaptive voltages by frequency and other factors that idk so I had to use Ryzen master for lowering peak voltage only to 1.2375V. I don't like how AMD engineer this crap. There is a reason why I love Intel, but AMD has better price to performance which makes sense to buy AMD What if I as like any regular customer who bought this APU and placed it on this ASUS mobo. Regular customer won't even look at bios settings and tweak the insane 1.568 or 1.586 Vcore not sure. His CPU would be dead now and price to performance will suffer as a result not to mention headaches. AMD rushes their technology without carefulness and we customers would pay the price if it stays like that. I don't recommend AMD for anybody unless they know enough about tweaking. Now after your video I think there are some troubles with intel now.
reply
Jeffrey
This CHANGES MY RECOMMENDATION for many people I bought the R9-3900x on launch last year but if I'd been buying now I'd go with the i9-10900K. The initial reviews I saw showed the 240W+ and to me that was unacceptable as heat isn't just about cooling but also ROOM temperature as it gets hot in my small office and would for most rooms. My other issue was SECURITY concerns but it's not clear now if the Ryzen 3000 is more secure or not than the newer Intel CPU's.. not clear to me anyway... which then leaves PERFORMANCE. As Intel wins the GAMING hands down than only leaves PRODUCTIVITY. Now it's easy to conclude the R9-3900x wins here but it turns out it's far more complex than reviews would have you believe. For example in HANDBRAKE the settings I utilize average about 80% CPU utilization not 100%. That means I'm using just under 10 cores and would thus be faster on the Intel i9-10900K. Many other productivity applications are limited more by the speed of each core than having an extra two cores... I hate to paraphrase Intel but real world performance not benchmarks is what matters... I don't regret my purchase as Intel had nothing to compete, but if I was buying now it may be different... I'd also ensure to get a PCIe v4.0 motherboard. I had a tough time deciding which way to go but at the time of launch CHIPSET FAN noise issues were a concern. I maybe should have waited but I'd waited far longer than I'd wanted too already and didn't know if waiting would add MONTHS due to backorders (I waited a MONTH anyway despite ordering the same day it became available).
reply
This CHANGES MY RECOMMENDATION for many people I bought the R9-3900x on launch last year but if I'd been buying now I'd go with the i9-10900K. The initial reviews I saw showed the 240W+ and to me that was unacceptable as heat isn't just about cooling but also ROOM temperature as it gets hot in my small office and would for most rooms. My other issue was SECURITY concerns but it's not clear now if the Ryzen 3000 is more secure or not than the newer Intel CPU's.. not clear to me anyway... which then leaves PERFORMANCE. As Intel wins the GAMING hands down than only leaves PRODUCTIVITY. Now it's easy to conclude the R9-3900x wins here but it turns out it's far more complex than reviews would have you believe. For example in HANDBRAKE the settings I utilize average about 80% CPU utilization not 100%. That means I'm using just under 10 cores and would thus be faster on the Intel i9-10900K. Many other productivity applications are limited more by the speed of each core than having an extra two cores... I hate to paraphrase Intel but real world performance not benchmarks is what matters... I don't regret my purchase as Intel had nothing to compete, but if I was buying now it may be different... I'd also ensure to get a PCIe v4.0 motherboard. I had a tough time deciding which way to go but at the time of launch CHIPSET FAN noise issues were a concern. I maybe should have waited but I'd waited far longer than I'd wanted too already and didn't know if waiting would add MONTHS due to backorders (I waited a MONTH anyway despite ordering the same day it became available).
reply
Kryten42
I couldn't agree more Steve. As an Engineer, I know that without a known, reliable baseline for any comparison, the comparison is flawed & all but useless. Even with stated caveat's. There are so many variables that it's an excessive use of your time to be forced to diagnose & set a motherboard to a common baseline for an accurate, fair & honest comparison. I suspect that some reviewers, especially with stringent time constraints, won't do so for that reason. Intel (and AMD) should at the very least, force manufacturers to state clearly on all marketing material & packaging what they have done that is outside the guidelines (which I think should actually be made recommendations or recommended standard settings . They should also publish a list of all Motherboards and whether they meet, or exceed, those recommendations). The MoBo manufacturers also need to make clear what the minimum CPU & system cooling & power requirements are if they go above the CPU maker's guidelines or standard settings , especially since this will increase the cost of a new build. I won't be holding my breath for any of that. shrug It is most definitely a case of caveat emptor (Let the buyer beware).
reply
I couldn't agree more Steve. As an Engineer, I know that without a known, reliable baseline for any comparison, the comparison is flawed & all but useless. Even with stated caveat's. There are so many variables that it's an excessive use of your time to be forced to diagnose & set a motherboard to a common baseline for an accurate, fair & honest comparison. I suspect that some reviewers, especially with stringent time constraints, won't do so for that reason. Intel (and AMD) should at the very least, force manufacturers to state clearly on all marketing material & packaging what they have done that is outside the guidelines (which I think should actually be made recommendations or recommended standard settings . They should also publish a list of all Motherboards and whether they meet, or exceed, those recommendations). The MoBo manufacturers also need to make clear what the minimum CPU & system cooling & power requirements are if they go above the CPU maker's guidelines or standard settings , especially since this will increase the cost of a new build. I won't be holding my breath for any of that. shrug It is most definitely a case of caveat emptor (Let the buyer beware).
reply
gracej79
Motherboards should always default to settings that retain the CPU warranty. Settings that retain warranty but increase power limits I'm less worried about, however would prefer user action to enable (Press X on boot, or boot should show a menu until the user clicks a use these settings in the future button/checkbox) and/or clear labeling so the buyer is informed. I think most users will fall into a few basic categories: (1.) No risk appetite (ie: corporate) or thermal/power limited. Everything should be Intel spec. (2.) Maximum performance within warranty, (3.) I don't care about CPU warranty. A motherboard should never default to (3.), it should require user action unless clearly disclosed that the motherboard will void the CPU warranty. Motherboards should always offer (1.) as an easy to load option. Whether the motherboard boots into (1.) or (2.) I'm less fussed about. I don't expect a motherboard with Extreme in the name to boot to (1.) for instance.
reply
Motherboards should always default to settings that retain the CPU warranty. Settings that retain warranty but increase power limits I'm less worried about, however would prefer user action to enable (Press X on boot, or boot should show a menu until the user clicks a use these settings in the future button/checkbox) and/or clear labeling so the buyer is informed. I think most users will fall into a few basic categories: (1.) No risk appetite (ie: corporate) or thermal/power limited. Everything should be Intel spec. (2.) Maximum performance within warranty, (3.) I don't care about CPU warranty. A motherboard should never default to (3.), it should require user action unless clearly disclosed that the motherboard will void the CPU warranty. Motherboards should always offer (1.) as an easy to load option. Whether the motherboard boots into (1.) or (2.) I'm less fussed about. I don't expect a motherboard with Extreme in the name to boot to (1.) for instance.
reply
Pillokun
GN thinks it is hard for them to set parameters in bios for cpu reviews? Well it seems they have missed out when putting the same cpu in different mobos resulted in different perf because of what chipset it used, and even what math/fpu co processor was used. GN has it so easy in modern times yet they complain! :P I dont see an issue at all here if you simply need to set up parameters so that the cpus perform as intended . Motherboard manufacturers are selling their product using a cpu from either AMD or Intel to power it products. Look at it this way with this analogy. A certain sport car manufacturer is selling their sweet handling car using the same engine as a Toyota, why should they not be able to dress the engine up how they see fit. If you want base spec perf then buy a cheap board, those boards will probably adhear more to the intel/amd guide lines :P Like I said, if it is a cpu comparison then it is on GN to config the system accordingly.
reply
GN thinks it is hard for them to set parameters in bios for cpu reviews? Well it seems they have missed out when putting the same cpu in different mobos resulted in different perf because of what chipset it used, and even what math/fpu co processor was used. GN has it so easy in modern times yet they complain! :P I dont see an issue at all here if you simply need to set up parameters so that the cpus perform as intended . Motherboard manufacturers are selling their product using a cpu from either AMD or Intel to power it products. Look at it this way with this analogy. A certain sport car manufacturer is selling their sweet handling car using the same engine as a Toyota, why should they not be able to dress the engine up how they see fit. If you want base spec perf then buy a cheap board, those boards will probably adhear more to the intel/amd guide lines :P Like I said, if it is a cpu comparison then it is on GN to config the system accordingly.
reply
Bara
I wonder, do people really look at CPU + different Motherboard performance Reviews/Comparisons and buy their MB that way? I have built computers for the past 15 years and i have never met anyone who asks or looks for that kind of info. Reliability/Confidence in the Brand, price, formfactor, RAM support, IO, VRM Quality for better OC, Number of PCIe Slots etc. sure but nobody ever asked or cared for stock CPU on MB performance charts. Quite frankly, most people assume stock is indeed stock. Maybe its my ignorance who knows but i do not see the necessity for this kind of behaviour of the MB manufacturers. Auto OC and all that BS is ofc unacceptable and creates alot of issues for reviewers who advertise fake numbers in the process.
reply
I wonder, do people really look at CPU + different Motherboard performance Reviews/Comparisons and buy their MB that way? I have built computers for the past 15 years and i have never met anyone who asks or looks for that kind of info. Reliability/Confidence in the Brand, price, formfactor, RAM support, IO, VRM Quality for better OC, Number of PCIe Slots etc. sure but nobody ever asked or cared for stock CPU on MB performance charts. Quite frankly, most people assume stock is indeed stock. Maybe its my ignorance who knows but i do not see the necessity for this kind of behaviour of the MB manufacturers. Auto OC and all that BS is ofc unacceptable and creates alot of issues for reviewers who advertise fake numbers in the process.
reply
aardvaark
I wonder if these CPUs are all going to drop like flies in a few years time because: a) Motherboards manufacturers are pushing them past what Intel recommends b) Intel not enforcing limints that ensure the CPU lasts decades without electron migration killing them, and c) Intel pushing them hard from the factory anyway. Normally, Intel are super conservative with their clocks and voltage because they want their CPUs to last for decades. That is one of the main reasons they were so overclockable in the first place, and AMD chips weren't. They might've just given up on that longevity to stay competitive. It'd be really interesting to hear what someone in-the-know from Intel has to say about this.
reply
I wonder if these CPUs are all going to drop like flies in a few years time because: a) Motherboards manufacturers are pushing them past what Intel recommends b) Intel not enforcing limints that ensure the CPU lasts decades without electron migration killing them, and c) Intel pushing them hard from the factory anyway. Normally, Intel are super conservative with their clocks and voltage because they want their CPUs to last for decades. That is one of the main reasons they were so overclockable in the first place, and AMD chips weren't. They might've just given up on that longevity to stay competitive. It'd be really interesting to hear what someone in-the-know from Intel has to say about this.
reply
Guilherme
What I don't get is why the mobo makers just don't do it the other way around, just like xmp. Just ship the mobos at standard and announce an overclocking profile with up to (you know there'll be a ton of asterisks there) xx% increased performance. Anyone that's even semi intent on ocing a cpu will set the profile themselves and this alternate profile might provide a quick solution, much like xmp, to people who don't want to bother and the general consumer audience. That would provide an aditional layer of consideration for some when purchasing a mobo and maybe even a selling point for whoever has the coolest or fastest oc profile. Getting that without bs marketing is a whole other issue tho.
reply
What I don't get is why the mobo makers just don't do it the other way around, just like xmp. Just ship the mobos at standard and announce an overclocking profile with up to (you know there'll be a ton of asterisks there) xx% increased performance. Anyone that's even semi intent on ocing a cpu will set the profile themselves and this alternate profile might provide a quick solution, much like xmp, to people who don't want to bother and the general consumer audience. That would provide an aditional layer of consideration for some when purchasing a mobo and maybe even a selling point for whoever has the coolest or fastest oc profile. Getting that without bs marketing is a whole other issue tho.
reply
Xion
My opinion: If testing a CPU: Test both stock (removing any motherboard bias), and optimal stable overclocks. If testing a motherboard: Let it have its biased stock settings, and also test custom settings for optimal stable overclocks. You don't really performance review/test motherboards though; you review their feature sets, and that's about it. So of course you wouldn't let motherboards impose any bias in their settings. When I shop for a motherboard, all I give a damn about is aesthetics, layout, and IO. Being able to at least run the CPU I intend to use at stock should be a given. Features aside from the afore mentioned are kinda after-thoughts to me that don't really sell them to me.
reply
My opinion: If testing a CPU: Test both stock (removing any motherboard bias), and optimal stable overclocks. If testing a motherboard: Let it have its biased stock settings, and also test custom settings for optimal stable overclocks. You don't really performance review/test motherboards though; you review their feature sets, and that's about it. So of course you wouldn't let motherboards impose any bias in their settings. When I shop for a motherboard, all I give a damn about is aesthetics, layout, and IO. Being able to at least run the CPU I intend to use at stock should be a given. Features aside from the afore mentioned are kinda after-thoughts to me that don't really sell them to me.
reply
John
The way I see it, there needs to be a published spec that simply defines what absolutely every single 10th gen CPU can meet on absolutely every single Z490 motherboard made. This is the minimum spec for the platform and should be what every stock benchmark is done at. Anything else is extra that the MB manufactures can use for their marketing wank. On the other hand, there is a benefit to the board makers to have better performance out of the box as idiots that don't know what they are doing will get it in their head that their motherboard is better than others because the CPU runs faster out of the box. I think it should be set to the default spec out of the box, but wankers gonna wank.
reply
The way I see it, there needs to be a published spec that simply defines what absolutely every single 10th gen CPU can meet on absolutely every single Z490 motherboard made. This is the minimum spec for the platform and should be what every stock benchmark is done at. Anything else is extra that the MB manufactures can use for their marketing wank. On the other hand, there is a benefit to the board makers to have better performance out of the box as idiots that don't know what they are doing will get it in their head that their motherboard is better than others because the CPU runs faster out of the box. I think it should be set to the default spec out of the box, but wankers gonna wank.
reply
Add a review, comment
Other channel videos
