Question can I overclock or change mosfet?

[siam siddique]

I have gigabyte b450 gaming x containing the worst mosfet of human history for low side 4c06n & for higher side 4c10n.......
can I overclock 5800x extremely with direct airflow over the vrm?? & is it possible to replace the mosfet with 4co24n & 4c029n??

 

[drea.drechsler]

I have gigabyte b450 gaming x containing the worst mosfet of human history for low side 4c06n & for higher side 4c10n.......
can I overclock 5800x extremely with direct airflow over the vrm?? & is it possible to replace the mosfet with 4co24n & 4c029n??

What do you mean by overclocking "extremely"? Using LN2? Is it just for demonstration purposes to see if it's possible? or for 24/7 use?

First thing to realize is it's not a good idea to overclock a 5800X on any board in a conventional sense as it's effectively pushed to the silicon's limits by the bin-sort AMD uses and the boosting algorithm. People with "extreme" setups might get very slight all-core improvements but it usually results in light threaded performance loss which is most important for gaming. Better all-round performance is usually found by using PBO curve optimizer and a slight VCore negative offset.

2nd...yes, it's technically possible to graft on better VRM circuit but it will not be easy. You can't just change FET's, you have to also change the controller's operating parameters and it's supporting circuitry; that means also a modded BIOS that initializes the VRM controller with the correct parameters. It's far, far, far easier to just buy a better motherboard. B550's have universally better VRM's, but of course some are better than others.

If the FET's are overheating with what you have then locate a fan blowing on them. I've mounted a fan (50-80mm) to a bracket I made from a removed I/O plate before and attached it to the rear case fan then pointed it to blow across the VRM FET's. It worked very well.

But if you're really serious about this you might be interested in some of BuildZoid's youtube vid's. He does this sort of thing with GPU's mainly but also mods some motherboard VRM's.

View: https://www.youtube.com/watch?v=wlm_lMQCRqY

 

[siam siddique]

What do you mean by overclocking "extremely"? Using LN2? Is it just for demonstration purposes to see if it's possible? or for 24/7 use?

First thing to realize is it's not a good idea to overclock a 5800X on any board in a conventional sense as it's effectively pushed to the silicon's limits by the bin-sort AMD uses and the boosting algorithm. People with "extreme" setups might get very slight all-core improvements but it usually results in light threaded performance loss which is most important for gaming. Better all-round performance is usually found by using PBO curve optimizer and a slight VCore negative offset.

2nd...yes, it's technically possible to graft on better VRM circuit but it will not be easy. You can't just change FET's, you have to also change the controller's operating parameters and it's supporting circuitry; that means also a modded BIOS that initializes the VRM controller with the correct parameters. It's far, far, far easier to just buy a better motherboard. B550's have universally better VRM's, but of course some are better than others.

If the FET's are overheating with what you have then locate a fan blowing on them. I've mounted a fan (50-80mm) to a bracket I made from a removed I/O plate before and attached it to the rear case fan then pointed it to blow across the VRM FET's. It worked very well.

But if you're really serious about this you might be interested in some of BuildZoid's youtube vid's. He does this sort of thing with GPU's mainly but also mods some motherboard VRM's.

View: https://www.youtube.com/watch?v=wlm_lMQCRqY

ok so u r saying overclocking 58010x isn't worth it will hardly improve performance? then may I can boost 5800x on this board & run it 24/7 ???& if I cannot run 5800x 24/7 on this board then is it worth changing the vrm cz I'm very poor to buy a new motherboard??

 

[drea.drechsler]

ok so u r saying overclocking 58010x isn't worth it will hardly improve performance? then may I can boost 5800x on this board & run it 24/7 ???& if I cannot run 5800x 24/7 on this board then is it worth changing the vrm cz I'm very poor to buy a new motherboard??

Have you got a 5800X already? if not get a 5600X as it will be cheaper and perform just as well in gaming. It will also not stress the VRM nearly as heavily.

With either CPU enable PBO in the latest BIOS for your motherboard and work with the Curve Optimizer and a slight negative offset core voltage. That keeps the CPU cooler so it boost higher. A cooler CPU means it's drawing less current on the VRM so it too should run better.

Watch VRM temps. FET's aren't CPU's or GPU's, they are way more temperature tolerant. So only if they're getting well above 105C...like 110C, maybe even 115C or some FET's as high as 125C... will it be a problem. If they do get hot then put a fan blowing on the VRM section, like I said. Even with a 5800X running in PBO I would be surprised if your VRM goes over 100C.

 

[siam siddique]

with a 5800X running in PBO I would be surprised if your VRM goes over 100C.

so place a fan on the side panel of the case I mean directly over the vrm will work just fine??? & if b450 tomahawk max can overclock 5800x extremely why my mobo can not? I mean they both have 4+2 vrm...my vrm has 4mW resistance so it will produce a lot of heat where tomahawk's low side mosfet's resistance is 2.8mW so it will produce less heat...other than temperature they should have no difference right??

 

[drea.drechsler]

so place a fan on the side panel of the case I mean directly over the vrm will work just fine??? & if b450 tomahawk max can overclock 5800x extremely why my mobo can not? I mean they both have 4+2 vrm...my vrm has 4mW resistance so it will produce a lot of heat where tomahawk's low side mosfet's resistance is 2.8mW so it will produce less heat...other than temperature they should have no difference right??

Tomahawk has what's termed a 'weak 8 phase' VRM that's actually 4 phases with 8 current paths. There are 8 FET's total, two lo-side and two hi-side for each phase. It also has a massive and well-finned heatsink covering the FET's, tightly secured with screws not plastic clips. The doubled FET's on each phase and huge heatsink give it the power handling equivalent to an 8 phase while keeping cool; that gives good voltage stability, critical when heavily overclocked at low voltages in heavy all-core workloads.

Being true 8 phase would make it nicer for transient response. That' important when processing loads change suddenly, as from low load to all-core high load causing overshoot or high to low, causing undershoot .

That gaming X is also 4 phases....it might have 2 lo-side but almost for sure it has 1 hi-side FET and skinnny little block of aluminum they call a heatsink. Don't let the 8 chokes fool you, they're just paralleled on each phase. Even if you COULD "upgrade" the FET's to low RDS-on devices it would still have that architecture.

It doesn't need a large fan, all it really needs is air movement to get hot air out of the pocket that can be created under the rear case fan. I also found that putting a fan on the output side of my tower heatsink did the same thing (dropped VRM temps 5-10 degrees under Handbrake transcoding loads). I think because it creates a lot of spinning, turbulent airflow on the output side of the fan.

 

[siam siddique]

Even if you COULD "upgrade" the FET's to low RDS-on devices it would still have that architecture.

chokes don't play any important role it only stabilizes the current so it's not that important.....& ISL95712 is it's controller a pretty decent controller for 4+3 vrm configuration ... so if I change all the high side & low side mosfet I should have huge performance improvement ...
so what is the thing that will lag me behind from performance improvement??

 

[DSzymborski]

There's barely any reason to aggressively overclock a Ryzen chip and even less reason to do so if you're short on money. The performance improvements will be negligible. This is a bad idea.

 

[drea.drechsler]

...
so what is the thing that will lag me behind from performance improvement??

Mainly your expectations.

Performance improvement gained from all-core, manual overclocking of a 3rd or 4th gen Ryzen is little if any. And it usually degrades the single thread performance that's most important for gaming. That's on premium motherboards with highly capable VRM's, like B550 Tomahawk with 10 true phases of highly efficient smart power stages.

You'd probably do well enough with PBO overclocking using curve optimizer though. Just be ready for toasty warm VRM' temperatures in heavy all-core workloads like rendering or transcoding; gaming won't get it too hot as it's not that heavy of a CPU processing load.

And don't minimize the effect of paralleling inductors: doing it might increase the current handling but it reduces the effective inductance of the circuit. The circuit design requires a certain inductance so that means they have to use higher value inductors that have other undesireable effects (like earlier saturation in the small package size needed). I can't believe any circuit designer would do that if they have single inductors available that can do the job....which they do since most every other motherboard with 4phase VRM's use them. Make no mistake about it, putting 8 inductors on a true 4 phase VRM is pure marketing as the only purpose I can see it serving is to mislead buyers into thinking they're getting an 8 phase VRM.

 

[siam siddique]

There's barely any reason to aggressively overclock a Ryzen chip and even less reason to do so if you're short on money. The performance improvements will be negligible. This is a bad idea.

so if i do not have a good motherboard then I will not get higher performance ??

 

[DSzymborski]

so if i do not have a good motherboard then I will not get higher performance ??

There's very limited headroom on modern processors, which are aggressively binned. You could likely overclock a little more with a better motherboard, but if you don't have the cash for a new motherboard, then you shouldn't be pushing a CPU that sees only minor benefits. Nor should you be tinkering with the MOSFETs on your current motherboard.

Leave PBO on and leave it alone. That's it. No good reason here to do anything more.

 

[siam siddique]

And don't minimize the effect of paralleling inductors: doing it might increase the current handling but it reduces the effective inductance of the circuit. The circuit design requires a certain inductance so that means they have to use higher value inductors that have other undesireable effects (like earlier saturation in the small package size needed). I can't believe any circuit designer would do that if they have single inductors available that can do the job....which they do since most every other motherboard with 4phase VRM's use them. Make no mistake about it, putting 8 inductors on a true 4 phase VRM is pure marketing as the only purpose I can see it serving is to mislead buyers into thinking they're getting an 8 phase VRM.

I know chock is not important at all I am asking that if I change fets vrm quality should improve ...why it will not improve??

 

[siam siddique]

There's very limited headroom on modern processors, which are aggressively binned. You could likely overclock a little more with a better motherboard, but if you don't have the cash for a new motherboard, then you shouldn't be pushing a CPU that sees only minor benefits. Nor should you be tinkering with the MOSFETs on your current motherboard.

Leave PBO on and leave it alone. That's it. No good reason here to do anything more.

am I ok with this mobo or a better mobo can provide significant improvement??

 

[DSzymborski]

am I ok with this mobo or a better mobo can provide significant improvement??

Only a small improvement. You have a misconception that you can get a lot more performance from the 5800X by either tinkering with your current motherboard or getting a new one. Just leave everything alone, stick PBO on, and enjoy the 5800X.

 

[siam siddique]

And don't minimize the effect of paralleling inductors: doing it might increase the current handling but it reduces the effective inductance of the circuit. The circuit design requires a certain inductance so that means they have to use higher value inductors that have other undesireable effects (like earlier saturation in the small package size needed). I can't believe any circuit designer would do that if they have single inductors available that can do the job....which they do since most every other motherboard with 4phase VRM's use them. Make no mistake about it, putting 8 inductors on a true 4 phase VRM is pure marketing as the only purpose I can see it serving is to mislead buyers into thinking they're getting an 8 phase VRM.

I know chock is not important at all I am asking that if I change fets vrm quality should improve ...why it will not improve??

 

[drea.drechsler]

I know chock is not important at all I am asking that if I change fets vrm quality should improve ...why it will not improve??

If such a thing were feasible, all that would improve is the waste heat output of the VRM. CPU performance itself would be unnaffected.

 

[siam siddique]

If such a thing were feasible, all that would improve is the waste heat output of the VRM. CPU performance itself would be unnaffected.

can u explain a bit

 

[drea.drechsler]

can u explain a bit

Low RDS-on FET's don't get as hot in operation, that's what makes them so good. But you can't do anything to change the basic VRM arch as that's constrained by the motherboard lay-out. If you're skilled with a hack-saw and a file you might could do something about that block of aluminum Gigabyte considers a heatsink though.

A better motherboard might have it's advantages for performance. You might have better results overclocking memory and that's a good way to boost system performance. Making 3600 at least, or higher so long as you don't de-sync the IF, can be significant. Better motherboards have better memory interconnects to the CPU and make it easier to do that.

So, you may ask why they make boards with such powerful VRM's if you can't overclock the CPU's. One reason is 12 core and 16 core CPU's, obviously. Another is people think just because it's unlocked it has the same OC headroom as what they had before, but that's simply not true. And the third reason is people watch the competitive overclockers on youtube getting huge OC's over 5.5Ghz, but on LN2. That needs massively power VRM's. You can't do that without LN2, but they want that kind of board anyway. It's all a bit crazy, IMO.

 

[siam siddique]

Low RDS-on FET's don't get as hot in operation, that's what makes them so good. But you can't do anything to change the basic VRM arch as that's constrained by the motherboard lay-out. If you're skilled with a hack-saw and a file you might could do something about that block of aluminum Gigabyte considers a heatsink though.

A better motherboard might have it's advantages for performance. You might have better results overclocking memory and that's a good way to boost system performance. Making 3600 at least, or higher so long as you don't de-sync the IF, can be significant. Better motherboards have better memory interconnects to the CPU and make it easier to do that.

better fets have low rds(on) so it will generate low heat and has better frequency so better fets will respond faster than my current fets which should help to increase the PWM signal & that is what vrm do vrm provides PWM so better PWM should mean better vrm that's why I'm expecting better performance by replacing fets

 

[drea.drechsler]

better fets have low rds(on) so it will generate low heat and has better frequency so better fets will respond faster than my current fets which should help to increase the PWM signal & that is what vrm do vrm provides PWM so better PWM should mean better vrm that's why I'm expecting better performance by replacing fets

PWM doesn't help the CPU, at least not directly. FET's have very high power losses when in transition from on-off or off-on states. So when PWM frequency is high there are more transitions in every time period. Get it too high and the VRM gets very inefficient and therefore gets a lot hotter because waste heat outuput is so high and voltage stability suffers for it. But a high PWM frequency helps with transient response. It's a trade-off that designers tweak a lot...as do LN2 overclockers. In normal useage ranges it's rarely a significant factor, not with modern VRM's. Even less capable ones.

 

[siam siddique]

which one is better for 5800x extreeeeeeeeeeeeeeeeeeeeeeeeeme overclock for a very very long time like 24 hr a day I already have gigabyte b450 gaming x should I exchange it with ASRock b450pro4-f??

 

[--SID--]

Why not a good B550 instead of a budget B450 that can hardly handle a stock 5800X and lacks PCI-e4.0?

 

[drea.drechsler]

... extreeeeeeeeeeeeeeeeeeeeeeeeeme...

with that many "e"'s it will probably need sub-ambient cooling.

Asus X570 Crosshair Hero VIII.
MSI X570 Unify
Gigabyte X570 Aorus Master-S

 

[siam siddique]

Why not a good B550 instead of a budget B450 that can hardly handle a stock 5800X and lacks PCI-e4.0?

I'm very poor to buy new mobo so I dont have much choice & what will I do with pci e zen 4 when zen 4 ssd cost more then a kidney

 

[siam siddique]

with that many "e"'s it will probably need sub-ambient cooling.

Asus X570 Crosshair Hero VIII.
MSI X570 Unify
Gigabyte X570 Aorus Master-S

I'm very poor to buy new mobo so I dont have much choice & what will I do with x570 when it cost more then a kidney my left kidney has already been booked for 5800x