How To Top (not as obvious) mistakes made when selecting parts for a Custom PC.

PC Tailor

Illustrious
Ambassador
I had created this for somewhere else, but thought it might be handy as a quick info topic for some visitors here. And also, a good discussion point for anything else! Let me know if you have any thoughts.

We've compiled a list of the lesser known but common mistakes many often make when building a custom PC, so that you can try and avoid them! (This doesn't look at the completely obvious, but you can see a list of those honourable mentions at the bottom.)

1) Pick a cheap (and poor quality) power supply.


Far too often this happens. Because most put all of their money into the key components, right? CPU, GPU, and Motherboard but then try and cut corners on the other elements. When in fact, we often believe that the power supply is the LAST item you should scrimp on. Don't fall into the trap of "it has enough wattage, so it is fine" - wattage is useless without quality.

This is the sole component that supplies power to the rest of your system, and cheap power supplies usually means it's made with cheap parts, and you should never go cheap on safety.

Some may think we're over-exaggerating, but we're really not, the number of times that computer components have been damaged or the computers stability has been compromised because of a poor power supply is countless. And if you want to see what really poor power supplies can do, just check this out!



Now imagine what that's doing invisibly to the power distributing round your expensive components! If a PSU blows and takes out other components (which Cheap PSUs often do) - any warranty you have will not cover the rest of your components.

Poor quality power supplies cause endless problems, and even top-quality power supplies deteriorate over time. Trust us when we say, put some money into the Power Supply and you won't regret it!

As a note: A good brand does not equal a good PSU.


2) Think that it's just the socket of the motherboard that dictates what CPUs are compatible.

This happens countless times. Many of us know that if we want a new processor, we also need a compatible motherboard (some people still get this wrong, but with the amount of choice, it can be difficult to know what you're looking at if you're a novice!).

However, one thing despite that, that is missed so often, is the actual motherboard generation and BIOS that goes with it, and this is even more important in newer generations. Many will go to the motherboards website and see "Compatible with 3rd Generation Ryzen Processors" and then click the buy button, forgetting that it's only compatible when you have the latest BIOS installed too, then we watch their disappointment when they find out their PC won't boot.

Not all is lost though, as many motherboards can be shipped with the latest BIOS installed, you just simply need to ask the seller what BIOS it ships with, equally some motherboards are "out of the box" compatible, so they don't need a newer BIOS. And some are now allowing you to update the BIOS without a CPU at all. (@WildCard999)

Equally, some motherboards can have the same socket, and simply not be compatible. It's just making the point that the socket of the motherboard, is not the only thing that dictates compatibility.

3) Pick a monitor that can't utilise the capabilities of your components (or vice versa).


Yes, this is more gaming based - but still happens a lot. Many may go straight to buying an RTX 2080, and then still sit with a 1080p monitor and wonder why their GPU utilisation sits at 30% all the time. Yes, you may get higher frame rates at lower resolutions, but it's still a large waste on the potential of the component.

The RTX 2070 and RTX 2080 for example are 2 GPUs that are fantastically geared to achieve high results in 1440p or up to 4K gaming (especially the 2080/2080Ti), and this is where their capabilities shine.

It's not a BAD or DAMAGING problem, but one that means you could have probably saved yourself a good amount of money, or you could further improve your experience. If you just want excellent 1080p gaming, go slightly below top tier, and you still won't be disappointed, and you'll maximise the usage of your components (For example a 1070).

Equally, it's common for someone to buy a GPU, then become frustrated when they realise their GPU has a DisplayPort and DVI, but their Monitor only has HDMI.


4) Buy 2 packs of RAM, as it's cheaper than buying the same amount in 1 pack.


Memory is sold in packs for a reason, and only the memory modules sold in the same pack are guaranteed compatibility. This is because the manufacturing process for the modules change frequently and sometimes substantially. So theoretically you could have 2 of the exact same memory modules, but from 2 different packs, that don't behave well with one another.

This is why manufacturers only guarantee compatibility in the form sold (from the same pack). Making sure that they are the same model, or timings, or latency, is just a way of trying to minimise the risk, not eliminate it.

We're not saying it never works, it's just a gamble, and the last thing you'd want to do, is spend a lot of money, to find out they don't work well together. This is also why modules in the same pack are slightly more expensive. The same logic applies for upgrading your current RAM.


5) Pick the wrong M.2 SSD Interface or size.

4686b97c24214d51915e543eb0687fc5.jpg

M.2 SSDs are all the rage now (which we slightly disagree with, but that's another story). However, many forget it's not just as simple as M.2 port = M.2 drive. Each M.2 drive has a different interface (as well as key and physical size). SATA and PCIe.

NVMe M.2 = PCIe Interface = Uses PCIe bus (faster data transfer)
SATA M.2 = SATA Interface = Uses SATA bus (same data transfer rate as a normal 2.5" SSD)

A NVMe M.2 does not work in a SATA M.2 slot and vice versa, and each M.2 slot can be one, the other, or both. If it takes both, then fantastic, if it doesn't, then you need the right M.2 to suit it.

Along with this, the 2280 number you see after the M.2 model, is the physical size. You also want to make sure your motherboard supports the size of the drive, so that it has the appropriate standoff location in place for you to secure the M.2.

Equally just for the record, in real world application, M.2 doesn't offer that many benefits over standard 2.5 SATA, they transfer data quicker, and use less cables (which is a bonus!), if you're looking for faster load times, you may as well just stick to whatever is the cheaper interface.


6) I need good temperatures, so I MUST get water cooling!

Hold fire, slow down. Water coolers nowadays are great yes, but remember that not all water cooling is alike. An AIO water cooler (All-In-One) which is the most common on the market, is not necessarily any more effective than air cooling. In fact, the top end air coolers can often perform the same or better as AIO (Custom loops are a different matter altogether - but most don't need these).

Yes, water coolers tend to look nice, and yes, they are much more useful if you're a little tight on space in the centre of the case, but if you're JUST looking for performance, then an air cooler is perfectly suitable.

You can see a pretty good comparison here: https://www.tomshardware.com/reviews/best-cpu-coolers,4181-2.html

We find LinusTechTips also did a good and informative brief analysis that whilst a couple years old, still holds true:


We're not saying they're all like for like - just you don't have to immediately jump to Water Cooling. As is well put here by another member:
Below 250w, air = aio only in its respective ranges.

Corsair H45/55/60 = CM hyper212 evo. Same capacity, negligible difference in temps. Same for any 120mm aio, you are looking at @ 140w, same as all those budget air coolers. The differences in temps will be due to the efficiency of the cooler and effectiveness of the fan.

It's only when getting to 250w output that even the largest air coolers such as the Cryorig R1, Noctua NH-D15 etc peak out, and you'd need a 280mm/360mm AIO, simply due to capacity. You want to run a i9 9900k flat out, don't bother with air, that cpu hits 250w with all core 5GHz no issue.

Noise: created by fans. Doesn't matter if it's air or aio, crappy fans = loud. No point comparing an old h100i to a NH-D15, the difference in fans speaks volumes.

Leaks: yep. Sure do. 99% caused by installer putting pressure on the fittings. Factory leaks happen as often as heatsinks with warped bases or leaking heatpipes. All about quality control.


7) Buying an unlocked processor to overclock, then buying a motherboard that can't really overclock.

You've just bought your 9600K unlocked processor and want to do some serious overclocking to get every ounce of performance, so you're just about to go into the BIOS of your "B" motherboard.

Different motherboards are accustomed to different capabilities, and many make the mistake of getting more budget friendly low-end boards that will either not let you overclock at all, or are severely limited in stability in overclocking.

A simple way of knowing the difference in modern chipsets, is knowing that the "Z" or "X" boards are much overclocking friendly. This doesn't necessarily cause harm, but it just means you spend more on an unlocked processor, to find you can't do much with the "unlocked" feature on it.

Typically cheaper motherboards are cheaper for a reason, so don't always expect a cheap motherboard with poor power phases / VRMs etc to overclock well.

8) Not looking at all of the headers/ports on your motherboard.

There are several aspects or examples to this:
  • Buying a motherboard with USB 3.0 headers, and a case with no USB 3.0 ports.
  • Buying several case fans when your motherboard only has 2 extra fan headers.
  • Using 4 pin fans on 3 pin headers (so you lose control of the fan speed).
  • Your motherboard having power connectors that your PSU doesn't have.
  • When using graphics (applies to dedicated GPUs too), you forget your monitor only has HDMI, where your graphics port has DVI.
  • Thinking you can fill all the PCIe slots. Cards take physical width and PCIe lanes. Depending on mode, some slots may be deactivated.
  • Not having enough expansion slots (like cheaper motherboards often do) You may not be able to add as many SATA drives, m.2 SSD, or PCIe cards as you want to.
And so on, again, this rarely causes harm, but it's wasted potential and wasted money on features that you may not even be able to use.


9) Not getting an SSD! Or getting too small an SSD for your needs.

This one usually goes without saying, but it still happens frequently. You won't catch a mid to high end PC without an SSD now, and frankly, there is no reason to NOT have an SSD, yes, they cost more, but the price for performance and capacity is getting better by the day.

Honestly, just invest in an SSD and you won't regret it.

That being said, don't just jump on the cheapest one at 240GB without thinking about it - because if you're a gamer or editor (if not, less of a worry), who wants to maximise programs on an SSD, the 240GB will disappear in just a week or two. 480GB is minimum recommended, otherwise you'll have a tireless battle with saving space. We're not saying 240GB or below is pointless, jusy think about it properly before you jump into it! As a couple of our readers suggested excellently:
Great post, but could you add more to #9? I believe if are putting games and other big programs on an SSD, than yes, a 480GB is the minimum. However, if you are just using it for the OS and other small applications, a 240GB would be just fine.
I always use smallest possible and practical disk for OS. Only OS and most important programs on it but data always on another disk,. Why ? Because it's relatively easy to replace/OS and programs in case of corruption or even malware, but data may be irreplaceable. With relatively high SSD (specially fast and NVM2) it also makes sense to use another SSD or even HDD for data and some programs/applications that do not benefit from fast disk.

And whatever you do, don't get an RGB SSD - just. Why?!


Also on that note, you don't HAVE to get NVMe, currently unless your performing constant large file transfers, you're probably not going to see any benefit outside of artificial benchmarks. Just be wary what gains you hope to achieve, as you can pick up some SATA SSDs for much cheaper (but if the NVMe is the same price, then no loss).

10) Cutting corners because you're on a budget. Or equally buying more than you ever really need.

This goes along with the power supply problem, if something is dirt cheap, there is usually a reason. That's not to say all cheap components are bad, just that in many cases, they quite often can be.

If you can't afford good quality, reputable products now, wait until you can, this will save you money in the long run, as you won't have to spend more just to upgrade later.

But along with that, we see all too often people who pay thousands for the top end specification of computer, when the application they use barely touch the surface of utilisation. There is no point getting a 9900K and an RTX 2080 Ti when you mostly play Counter Strike.


Other mistakes often made:
  • Buying a GPU that's too big for the case.
  • Overthinking thermal paste.
  • Buying more cooling components than you need.
  • Buying more RAM than you need.
  • Buying a computer that is capable of far more than you need.
  • Forget that warranty is important. (Look for the best warranties too!)
  • RGB Overload (RGB = heat).
  • Forgetting that your unlocked Intel processor doesn't come with a cooler.
  • "I'll re-use my DDR3 RAM on my DDR4 motherboard"
  • Forgetting to buy windows.
 
Last edited:

WildCard999

Titan
Moderator
I want to add to number 2 that some motherboards allow you to update the BIOS without a CPU so that may be worth looking into before buying a more expensive (& possibly unneeded) board.

As for the SSD size choice, it really comes down to the uses. Right now I'm only playing two games actively so realistically a 240gb SSD would be enough for my games & the OS with room to spare. But with that being said the prices for SSD's has come down pretty far in last couple of years so unless the budget is tight I would agree more towards a 480/500gb as a min.

Overall I think it's a good pre-buying read for new builders.
 
  • Like
Reactions: PC Tailor

PC Tailor

Illustrious
Ambassador
I want to add to number 2 that some motherboards allow you to update the BIOS without a CPU so that may be worth looking into before buying a more expensive (& possibly unneeded) board.

As for the SSD size choice, it really comes down to the uses. Right now I'm only playing two games actively so realistically a 240gb SSD would be enough for my games & the OS with room to spare. But with that being said the prices for SSD's has come down pretty far in last couple of years so unless the budget is tight I would agree more towards a 480/500gb as a min.

Overall I think it's a good pre-buying read for new builders.
Good point, I'd neglected to mention the CPU-less BIOS update - which I've now added.
I've also re-emphasised the SSD size element, I agree, thus why I put more direction to gamers/editors specifically, but I may not have made the link clear enough, so I have enhanced it slightly.
 
  • Like
Reactions: WildCard999

Karadjgne

Titan
Ambassador
I'd like to add to #6. Below 250w, air = aio only in its respective ranges.

Corsair H45/55/60 = CM hyper212 evo. Same capacity, negligible difference in temps. Same for any 120mm aio, you are looking at @ 140w, same as all those budget air coolers. The differences in temps will be due to the efficiency of the cooler and effectiveness of the fan.

It's only when getting to 250w output that even the largest air coolers such as the Cryorig R1, Noctua NH-D15 etc peak out, and you'd need a 280mm/360mm AIO, simply due to capacity. You want to run a i9 9900k flat out, don't bother with air, that cpu hits 250w with all core 5GHz no issue.

Noise: created by fans. Doesn't matter if it's air or aio, crappy fans = loud. No point comparing an old h100i to a NH-D15, the difference in fans speaks volumes.

Leaks: yep. Sure do. 99% caused by installer putting pressure on the fittings. Factory leaks happen as often as heatsinks with warped bases or leaking heatpipes. All about quality control.
 
  • Like
Reactions: PC Tailor

PC Tailor

Illustrious
Ambassador
I'd like to add to #6. Below 250w, air = aio only in its respective ranges.

Corsair H45/55/60 = CM hyper212 evo. Same capacity, negligible difference in temps. Same for any 120mm aio, you are looking at @ 140w, same as all those budget air coolers. The differences in temps will be due to the efficiency of the cooler and effectiveness of the fan.

It's only when getting to 250w output that even the largest air coolers such as the Cryorig R1, Noctua NH-D15 etc peak out, and you'd need a 280mm/360mm AIO, simply due to capacity. You want to run a i9 9900k flat out, don't bother with air, that cpu hits 250w with all core 5GHz no issue.

Noise: created by fans. Doesn't matter if it's air or aio, crappy fans = loud. No point comparing an old h100i to a NH-D15, the difference in fans speaks volumes.

Leaks: yep. Sure do. 99% caused by installer putting pressure on the fittings. Factory leaks happen as often as heatsinks with warped bases or leaking heatpipes. All about quality control.
Very valid point. I had neglected to add my other comment that we're not saying Water is worse, just that it's not the first thing you need to jump to. So I have quoted your brilliant detailed post accordingly and also clarified my point:

"We're not saying they're all like for like - just you don't have to immediately jump to Water Cooling. As is well put here by another member:"

Thank you my friend!
 

PC Tailor

Illustrious
Ambassador
Also added a little more info to point 1 following some conversations I just had around quality vs quantity.
If anyone would like to actually add some points too as opposed to enhance the current ones, I'm also open ears :)
 

mangaman

Honorable
Great post, but could you add more to #9? I believe if are putting games and other big programs on an SSD, than yes, a 480GB is the minimum. However, if you are just using it for the OS and other small applications, a 240GB would be just fine.

In fact, I'm still using my 120GB SSD and have no issues with space, as it is used for my OS and other small programs. I just don't want new PC builders to get the wrong idea and spend more money on a big SSD. More than likely, they'll be putting games on a large hard drive if the game take up a lot of storage (if they choose to get a secondary storage device).

Oh and it's best to stay away from counterfeit Chinese SSD's on AliExpress. I believe Linus did a video on that, but I forgot the episode name.

Otherwise, great post. :)
 
  • Like
Reactions: PC Tailor
I always use smallest possible and practical disk for OS. Only OS and most important programs on it but data always on another disk,. Why ? Because it's relatively easy to replace/OS and programs in case of corruption or even malware, but data may be irreplaceable. With relatively high SSD (specially fast and NVM2) it also makes sense to use another SSD or even HDD for data and some programs/applications that do not benefit from fast disk.
My example: I use Samsung 960 evo, 250GB for Windows and Office, CAD etc. but games are installed on a cheap 500GB SATA SSD. Other SATA SSDs , 120GB a piece are for W10 insider version and 2 Linux distros.
I also use couple of HDDs in removable trays, 1 and 2 GB for data and backups. Another reason for dividing data and OS disks is convenient backup of each one that can be done on different schedules.
Speaking of BACKUPS, It would be very wise to have a disk of at least same size of OS disk/partitions and make it removable so it can be kept offline until needed, either using removable trays or USB.
 

PC Tailor

Illustrious
Ambassador
@mangaman and @CountMike
Thanks for the input - and actually yes I agree with you, which is why I tried to put more emphasis towards particular types of people. The reason why it was stated is because let's face it, people like you and I probably consider all of the risks above before buying, so we dont necessarily fall into those caveats, however in all of the systems I see, I am advising far more often to people who bought a 240GB and are wondering why they have no space.

I guess the point is it is geared slightly at a different audience upon writing. So I have adapted it and quoted your great info to clarify this, and I've re-emphasised it's a "think about it". I agree it's not a blanket sweep that 240GB are bad, there is a reason why they exist and people use them (as I have for the same reason you have @CountMike) - It's simply that I am often advising on the former!

So I have quoted you both to re-emphasise the point behind it - thank you and very valid points that should now make the post clearer! Let me know if you think it is sufficient my friends :)
 
Last edited:
Jul 13, 2019
12
1
15
should i buy a good computer even my monitor sucks so if i would buy another one i could play with good frame rates (future-proofing the pc?)
 

PC Tailor

Illustrious
Ambassador
should i buy a good computer even my monitor sucks so if i would buy another one i could play with good frame rates (future-proofing the pc?)
Yes I would say it makes sense. I don't think you should ever limit yourself just because your monitor is bad, because otherwise you'll always be limited.

But it's worth remembering that if you spend money on high end components, your monitor might not be able to utilise them, so you will be disappointed until you upgrade. But some excellent monitors can be picked up for not much, and they go on sale regularly, so if you want a high end PC now, get it, and put some money aside for a good money that will fit the bill too. Remember to ensure that your monitor can take the display cable type that the GPU has, as adapters are never that great.

If you want my honest view though, save up before buying the PC to get the monitor AS WELL, at the same time, then you will see just how brilliant the difference could be. That and then you know you're buying components that will all work with each other, for the best deal at the time :)
 
Last edited:
  • Like
Reactions: WildCard999

brokeBuilder2019

Prominent
Sep 14, 2019
75
15
545
you could have 2 of the exact same memory modules, but from 2 different packs, that don't behave well with one another.

Hi, first of all, great article with a lot of good info ! I learned a lot.

Now, referring to RAM, I'm assuming that, in addition to compatibility, performance is also a concern ?

Can you quantify the difference in performance between 1 kit of 4 matched sticks VS 2 kits of 2 sticks each (assume dual channel mobo) ? Are we talking 1%, 10%, 50% ? Are there any numbers available for this ?

^ Assuming both pairs are of the same brand, same model, same speed, and same timings. But just bought say 6 months apart.
 
  • Like
Reactions: PC Tailor

USAFRet

Titan
Moderator
Hi, first of all, great article with a lot of good info ! I learned a lot.

Now, referring to RAM, can you quantify the difference in performance between matched/unmatched pairs ? Are we talking 1%, 10%, 50% ? Are there any numbers available for this ?

^ Assuming both pairs are of the same brand, same model, same speed, and same timings. But just bought say 6 months apart.
Unmatched RAM may not work at all.
Always always always buy RAM in the capacity you need, all in one set. Not 8GB now, and another 8GB in 6 months.
 

brokeBuilder2019

Prominent
Sep 14, 2019
75
15
545
Unmatched RAM may not work at all.
Always always always buy RAM in the capacity you need, all in one set. Not 8GB now, and another 8GB in 6 months.

I understand. When they are not compatible, they may not work at all.

What about when they do work ? Will they work less optimally ? If so, how less ?
 

PC Tailor

Illustrious
Ambassador
I understand. When they are not compatible, they may not work at all.

What about when they do work ? Will they work less optimally ? If so, how less ?
You could find a variety of situations, you may find that your performance is less than it should be (which you may never actually know as how would you), you may find when you change other hardware, that your RAM becomes unstable, or you may find the RAM because unstable in certain applications, and finally you may find that they don't overclock as they should.

It's one of those where it's difficult to find what potential instability is caused when the computer otherwise appears to work fine.
 
  • Like
Reactions: brokeBuilder2019

PC Tailor

Illustrious
Ambassador
Hmm. This seems to contradict what PC Tailor said.
No, US is correct as it is usually fine, but it's not unheard of to have some instability.

I've had some where mixed RAM modules have simply gone wild when they've tried to overclock for example. But generally if they boot, you're fine, but how would you notice if they aren't working correctly if it's only minor.

The manufacturing of RAM (and almost any PCB or chip) is actually a very delicate process, it is performed in very highly regulated clean rooms, and the material that goes into those components are also very delicate to change.

Each chip is cut out of silicon, and that silicon can be prone to various impurities or contamination even simply from dust, which can affect the compatibility of the modules. The point being, is they pre-determine the modules that will go in a pack, manufacture accordingly, and then test them on a system together at the end, so by all accounts, there are no guarantees thereafter.

You can actually buy 2 of the EXACT SAME MODEL RAM, but from 2 different packs, and they can (and have) not be compatible.

So does it not work all of the time? No, there are plenty of times where it does work.
But is it a gamble? Absolutely yes.
Part of the problem is, is if it DOES actually affect the performance of the RAM together, how likely would you be to notice it? Not at all unless you encounter obvious stop errors or crashes for example.

They may place nice, or they may stop your PC from booting at all, and in many cases, they work fine, but may cause instability later when you start overclocking, or changing platforms for example. And as USAFRet states, this becomes even more complicated when you mix sizes, speeds, ranks etc.
 
  • Like
Reactions: brokeBuilder2019

brokeBuilder2019

Prominent
Sep 14, 2019
75
15
545
No, US is correct as it is usually fine, but it's not unheard of to have some instability.

I've had some where mixed RAM modules have simply gone wild when they've tried to overclock for example. But generally if they boot, you're fine, but how would you notice if they aren't working correctly if it's only minor.

The manufacturing of RAM (and almost any PCB or chip) is actually a very delicate process, it is performed in very highly regulated clean rooms, and the material that goes into those components are also very delicate to change.

Each chip is cut out of silicon, and that silicon can be prone to various impurities or contamination even simply from dust, which can affect the compatibility of the modules. The point being, is they pre-determine the modules that will go in a pack, manufacture accordingly, and then test them on a system together at the end, so by all accounts, there are no guarantees thereafter.

You can actually buy 2 of the EXACT SAME MODEL RAM, but from 2 different packs, and they can (and have) not be compatible.

So does it not work all of the time? No, there are plenty of times where it does work.
But is it a gamble? Absolutely yes.
Part of the problem is, is if it DOES actually affect the performance of the RAM together, how likely would you be to notice it? Not at all unless you encounter obvious stop errors or crashes for example.

They may place nice, or they may stop your PC from booting at all, and in many cases, they work fine, but may cause instability later when you start overclocking, or changing platforms for example. And as USAFRet states, this becomes even more complicated when you mix sizes, speeds, ranks etc.

Got it. Thanks again for a detailed answer. I understand that the downside of mixing kits is hard to quantify. And I am satisfied with that answer. But I also want to briefly elaborate on why I insisted on trying to come up with some sort of numerical estimate of performance degradation.

I'm just trying to weigh the difference in cost VS the potential risks of incompatibility/system instability/performance degradation/etc with the assumption that there is a significant cost difference between 2 separate kits vs 1 whole matched sticks kit. If I have some kind of number to work with (e.g. the variance in RAM speed can be up to 10% ... from 3000MHz down to 2666MHz ... OR the probability of failure is 7% ... just hypothetical), then it gets easier for me to work this out in my mind using my own personal "algorithm".

Just like for instance, imagine you're shopping for monitors ... and there are 2 you really like, but for 20Hz faster refresh rate, you are required to pay $200. Having numbers like 20Hz or $200 makes it easier to come up with a decision as to "Is it worth the extra money ?" For someone who earns money in INR and has to pay an ear to have this stuff shipped 10,000 miles, stuff like this is a practical concern. Hope this makes sense as to why I'm asking this question in this form.

But again, thank you for explaining everything so clearly. What you said makes a lot of sense. Cheers !
 
Last edited:
  • Like
Reactions: PC Tailor

PC Tailor

Illustrious
Ambassador
Got it. Thanks again for a detailed answer. I understand that the downside of mixing kits is hard to quantify. And I am satisfied with that answer. But I also want to briefly elaborate on why insisted on trying to come up with some sort of numerical estimate of performance degradation.

I'm just trying to weigh the difference in cost VS the potential risks of incompatibility/system instability/performance degradation/etc with the assumption that there is a significant cost difference between 2 separate kits vs 1 whole matched sticks kit. If I have some kind of number to work with (e.g. the variance in RAM speed can be up to 10% ... from 3000MHz down to 2666MHz ... just hypothetical), then it gets easier for me to work this out in my mind using my own personal "algorithm".

Just like for instance, imagine you're shopping for monitors ... and there are 2 you really like, but for 20Hz faster refresh rate, you are required to pay $200. Having numbers like 20Hz or $200 makes it easier to come up with a decision as to "Is it worth the extra money ?" For someone who earns money in INR and has to pay an ear to have this stuff shipped 10,000 miles, stuff like this is a practical concern. Hope this makes sense as to why I'm asking this question in this form.

But again, thank you for explaining everything so clearly. What you said makes a lot of sense. Cheers !
I can completely understand that logic, and it's a great question I think, the difficulty comes from most problems are due to an interaction of factors, rather than one factor. So the instability of one set of RAM modules can be completely different to the next, so it'd be impossible to quantify, and even if it wasn't it wouldn't be representative of any one persons systems outside of a test environment.

The main cost downfall from it all, comes from simply paying more for the RAM that then doesn't work. So i can pay
  • $100 for 2x8 or = $100 total
  • $40 x2 for 1x8 = $80 total

To then find that the second one don't work together well, so you're potentially $40 down if they don't accept a return as you have both used them, and they are completely functional. That and again, you might only find out the instability say when you go to overclock them later.

it's a difficult one, I'd love to quantify the potential instability it could cause, but i guess the easiest place to do this would be to quantify the degradation of performance when you mix different speeds / ranks / sizes. As for mixing 2 of the same model, it becomes near impossible.
 
  • Like
Reactions: brokeBuilder2019

brokeBuilder2019

Prominent
Sep 14, 2019
75
15
545
I can completely understand that logic, and it's a great question I think, the difficulty comes from most problems are due to an interaction of factors, rather than one factor. So the instability of one set of RAM modules can be completely different to the next, so it'd be impossible to quantify, and even if it wasn't it wouldn't be representative of any one persons systems outside of a test environment.

The main cost downfall from it all, comes from simply paying more for the RAM that then doesn't work. So i can pay
  • $100 for 2x8 or = $100 total
  • $40 x2 for 1x8 = $80 total
To then find that the second one don't work together well, so you're potentially $40 down if they don't accept a return as you have both used them, and they are completely functional. That and again, you might only find out the instability say when you go to overclock them later.

it's a difficult one, I'd love to quantify the potential instability it could cause, but i guess the easiest place to do this would be to quantify the degradation of performance when you mix different speeds / ranks / sizes. As for mixing 2 of the same model, it becomes near impossible.

Yes, it's helpful to look at how much money you could potentially lose if the new pair just doesn't work at all with the existing installed pair ... and use that knowledge towards making the purchase decision.

One final question, then - if it is up to us (buyers/builders) to ensure that we get whole RAM kits for our systems, does that imply that RAM manufacturers are not legally obligated to have 2 of their own (separate) kits (exact same model and specs) work nicely together ? Sorry if the answer is totally obvious.
 
  • Like
Reactions: PC Tailor