I suspect the non-VRM caps are from the same brand and series just different values. That brand/series has proven itself to fail already. Right there in front of you.
So:
1. Do a proper full recap for peace of mind and a more stable machine.

or be stubborn and
2. Quickfix and a revisit full disassemble full rebuild in a few months time: replace only the bulging ones.

I've had boards where there were more than one brand of capacitor and several of one brand are visibly failing, while none of the other brand were. In that case I was satisfied with replacing all of the brand that were failing.

I'm not an authority on this, but what you replace is entirely up to you. You don't absolutely have to change them all... or it might turn out that every single last one of them will cause stability problems and you DO have to change them all to make the board work properly.

Generally, all caps that are of the same exact type are likely to have the same underlying problem (bad electrolytic being used in caps was a huge issue from 2000-2007 roughly). Caps of the same brand but different sizes are likely not far behind the others... but there's no way to know for sure.

It's also worth noting that heat tends to make caps fail prematurely, which may be why you're seeing physical signs on the ones by the VRMs and CPU (hottest area of the board). I have an Abit IS7 board with all good name brand caps (Rubycon I think... can't remember), but one single cap near a voltage regulator is swollen. Likely, the way the board is designed, most of the heat from that regulator is absorbed by that capacitor, which caused it to fail in a way that the other similar caps won't for a long time (or maybe they're all bad... who knows). When I get around to fixing this board, I fully intend to just replace that one first. If the board is stable after that, I'm not replacing the rest of them unless I have an issue later.

If you're extremely good at replacing caps on modern boards, and you can remove all the caps from a board with a radioshack soldering iron and some braid in less than an hour... by all means, do the whole board. But if you're like me, and even with decent tools you find working on modern (lead-free solder, multi-layer PCBs, tiny solder pads) boards tedious and risky, I'd certainly minimize the amount of times you have to jab it with a hot iron. For most mortals, every component replaced comes with the potential risk of a mistake that can either ruin the board or require more repairs (lifted pads, broken traces, overheated components, etc.).

Of course, when it comes to ordering replacement parts, it's up to you to decide whether you should just order all the caps for the whole board and hope you end up not needing them all... or save the time and money and just order the ones you need, and then order more later if you need to recap others. To be on the safe side, if they are a crappy brand or are of a known bad series from a good brand, I'd replace all of the ones that are identical to the ones that are visibly failing. If any others are the same brand\series but aren't visibly failing, I'd order some of those too. I tend to not worry as much about the tiniest electrolytic caps, because I have yet to come across a board that was dead until I replaced one of those. That's just me though. I'll likely replace them on some boards eventually... I just mainly hate working on modern boards.

Worst I've ever had to work on was an Abit A8N-SLI 32X. I'm not a pro, but I could never seem to get enough heat into the solder joints to make a good connection. There is so much extra metal in that board, it just soaked it all up. I have better techniques and tools now (a heat gun to preheat the board before removing each cap) but I wouldn't wish those repairs on anyone who isn't experienced with this kind of thing.

athlon-power wrote on 2020-12-13, 17:00:

This is a pretty simple one. I know that when changing capacitors on a motherboard, you're supposed to replace them all, however, I have a WinFast nForce 4 motherboard where every single capacitor looks great (I know they can start failing without displaying physical signs), but the ones on the VRM. Would it be possible just to replace those on the VRM, instead of every single capacitor on the motherboard? Thanks for any suggestions you guys might have.

You need to look at more detail in order to assess your risk. What MFR, series, and (optionally) value are the caps that failed? MFR, series, and value for the caps that are good?

You don't have to replace capacitors just because they are old/from a certain brand. You need to replace them all because it's impossible to tell if a capacitor is good or bad without removing it first. Once it's removed it's safer to replace it with a new, more efficient model.

It depends. The capacitor hipster brigade will tell you to shotgun it of all capacitors and replace them with polymer models or at least the lowest ESR ones you can find. There's a lot of people who do feel that way in our retro PC community.

My attitude is to use some logic and selectivity. One obvious culprit is heat. It's not surprising to find say that an 85C rated capacitor pressed up against a heatsink has failed. Likewise, low ESR capacitors used in switch mode power supplies and the VRM on motherboards tend to have fairly short lives. On the other hand, general purpose capacitors, even pretty off brand ones, routinely last without incident for forty or more years. However some capacitors are leak prone and can even destroy a motherboard in short order. Early SMD and subminiature radial capacitors are notorious for this failure mode... and they may leak to the point of PCB damage before a device stops functioning normally. Some devices probably should be recapped regardless of any operational issues because this failure mode is frequent and likely to total the machine.

Also, I would agree that many capacitor failures will not be visible. But many will, leaking seems pretty common unfortunately.

Miphee wrote on 2020-12-13, 21:54:

You don't have to replace capacitors just because they are old/from a certain brand. You need to replace them all because it's impossible to tell if a capacitor is good or bad without removing it first. Once it's removed it's safer to replace it with a new, more efficient model.

That's a very paranoid (and expensive) way to do things. As an extreme example, if the failed caps are Ruby ZLG or Panny FR the board has likely had some combination of long service and heat stress and the risk of cap failure elsewhere is high. If they're Chemi Con KZG, the board will probably have seen less service and there's a strong chance the other caps are still fine if they're not similarly prone to failure.

gdjacobs wrote on 2020-12-13, 23:48:

That's a very paranoid (and expensive) way to do things.

And if you don't want to guess which capacitor is going to explode next it's just better to replace them all and not cut corners to save a few cents.
Brand is completely irrelevant because my top 5 failed cap brands are:

#1: Fuhyyu
#2: Teapo
#3: Panasonic
#4: Sanyo
#5: Rubycon

Paranoia or not, recapping old electronics is still the way to go if you expect something to work reliably for a long time.

Can we just refer to the other threads about this same topic scattered all over the internet? Badcaps.net , vcfed.org , vogons.org , and many many others.

It's all been talked about before.

At one end of the spectrum are people who say you're lazy and cheap if you don't replace every capacitor in every piece of equipment in your home (I'm exaggerating) because any cap that was manufactured more than a year ago is no longer new and should be incinerated for the good of humanity.

The other end of the spectrum has people BUYING vintage capacitors with crud coming out of them to get some "vintage" quality they expect their device to have. Listen to the warm sound of those 60 year old caps! 🤣

Most of us are somewhere in the middle, and for most of us, replacing all the capacitors on a motherboard isn't just a matter of spending $25 on a capacitor order (and hours trying to figure out what you need)... it's a matter of doing the job properly without damaging the board, and not spending 16 hours doing it.

I will say that I have posed questions in the past asking for examples of cases where replacing lots of visibly-okay electrolytic caps fixed a problem, and I got very few replies saying people had this experience. Most of the time I think people replace all the caps before having a problem, so there's no indicator as to whether it made any difference in their lifetime. Sure, it's preventative maintenance, but so is disassembling your car's engine and transmission every year and cleaning all the parts... but for most of us, the time and risk isn't worth the possibility of getting a million miles on a car. Routine maintenance, being aware of common problems and staying alert to changes in performance works with $30,000+ vehicles. For a $30 motherboard you got on ebay, it's usually hard to justify a lot of effort and expense.

You don't have to replace all the capacitors and you don't have to use the same brands. Just buy capacitors with the same capacitance and similar voltage rating as the originals. If you get capacitors with really high voltage tolerance then they can be physically larger which means they might not all fit in the through holes.

The bigger capacitors are more important than the little guys. Replace the large electrolytic capacitors first before you do the smaller ones. Ceramic capacitors don't dry out like electrolytic capacitors, so if they look ok, they are probably ok. You'll know if they fail because they pop and let out smoke.

I've read more than one time that a simple rule is to replace capacitors starting from 1000uf and up. Fron my little experience that worked well. I have soltek boards from the socket 462/370 era that had very bad caps from GSC and I replaced only the bigger ones and never had any problem, those where the worst loking caps I've seen. And here the Gigabyte GA-7DX series where also popular and also had problens with the bigger caps.

As a counterexample I have a Abit NF7-S that had bulged 13 out of 21 smaller caps (1000uF/10V Nichicon HM) when I got it, and upon power-up another one or two instantly bulged and spewed their electrolyte. In contrast, bigger caps (Rubycon ZL and MBZ) near the CPU VRM were in perfect shape and I ended up using them to repair other boards, as the NF7-S underwent a full poly-mod.

Board was stone-dead with the bulged smaller caps (and came back after the poly-mod), so the smaller ones certainly matter.

quicknick wrote on 2020-12-14, 23:00:

As a counterexample I have a Abit NF7-S that had bulged 13 out of 21 smaller caps (1000uF/10V Nichicon HM) when I got it, and upon power-up another one or two instantly bulged and spewed their electrolyte. In contrast, bigger caps (Rubycon ZL and MBZ) near the CPU VRM were in perfect shape and I ended up using them to repair other boards, as the NF7-S underwent a full poly-mod.

Nichicon had some known bad batches of series HM along with other ultra low ESR lines in the early oughts.

Miphee wrote on 2020-12-14, 08:04:

Brand is completely irrelevant because my top 5 failed cap brands are: […]
Show full quote

Brand is completely irrelevant because my top 5 failed cap brands are:

#1: Fuhyyu
#2: Teapo
#3: Panasonic
#4: Sanyo
#5: Rubycon

Interesting factoid, but meaningless in terms of reliability without further context. Also objectively wrong. Some series, or even whole brands, are known to fail at elevated rates in certain applications compared to other caps known to perform according to spec. In certain cases we know this because the manufacturer issued service bulletins outlining the problem.

Brand may not always be the most reliable way to judge the quality of a capacitor, but the idea that all capacitors are pretty much the same is definitely false. I am the furthest thing from an expert, but I have the best impressions of nichon and rubycon. When shopping for capacitors, I use heat tolerance and rated hours as an indicator of quality - after capacitance and voltage.

gdjacobs wrote on 2020-12-15, 01:35:

Also objectively wrong.

That's what I call a religious belief based on nothing, an argument for the sake of the argument. Not my problem, it's OP's decision.

Ozzuneoj wrote on 2020-12-14, 19:17:

For a $30 motherboard you got on ebay, it's usually hard to justify a lot of effort and expense.

I agree because the majority of guys here won't use a mainboard for more than a few weeks at most.
So why would they replace all caps? Too much money, too much work.
But!
I'm talking about doing a good job and that requires diagnostics, not just visual inspection. Bulging-leaking caps are easy to find but that's just the beginning.
If it was an old TV-set and not a mainboard people would immediately tell OP to change all caps because it solves the majority of the problems and prevents bigger problems like blown flyback drivers or failed deflection circuits.
It's essentially the same with computer parts, failed caps will result in failed power phase or even worse. People just refuse to remove and diagnose these caps because they are mighty hard to remove without proper tools (like you mentioned before). So they choose to half-ass it instead and call it a day.
That doesn't mean that the rest of the caps (not just the leaky-bulgy ones) shouldn't be changed.
It's like repairing acid damaged board without removing the affected chips, connectors and sockets first. A half-assed job that will work... for a while.
Anybody with a little pride in their work would consider a full recap when bad brands and faulty caps are involved.

Miphee wrote on 2020-12-15, 07:27:

Anybody with a little pride in their work would consider a full recap when bad brands and faulty caps are involved.

... and the time, money, experience, and tools... right? 😀

Ozzuneoj wrote on 2020-12-15, 07:52:

Miphee wrote on 2020-12-15, 07:27:

Anybody with a little pride in their work would consider a full recap when bad brands and faulty caps are involved.

... and the time, money, experience, and tools... right? 😀

Right. We don't start repairing a PSU or a gas stove without the proper tools and knowledge either. 😉

Miphee wrote on 2020-12-15, 08:11:

Ozzuneoj wrote on 2020-12-15, 07:52:

Miphee wrote on 2020-12-15, 07:27:

Anybody with a little pride in their work would consider a full recap when bad brands and faulty caps are involved.

... and the time, money, experience, and tools... right? 😀

Right. We don't start repairing a PSU or a gas stove without the proper tools and knowledge either. 😉

Sure, but given a general lack of commercial repair services in the US, and a desire to keep equipment running a bit longer, you can reasonably expect people with minimal electronics backgrounds to attempt to do their own board level repairs. For most items, the practical reality is that someone repairs it themselves, or the item can become a shelf queen. As for the power supplies, many older systems have proprietary form factor units for which there is no generic fit, and even for normal cases, electrically and aesthetically compatible grey steel ATX 1.3 or earlier units have not been made for decades.So people are going to rebuild those too, understanding of theory be damned. Have we actually had any casualties in the hobby from this, say a small fire, electric shock incident, .etc?