On upgrading your system beware of 16+ GBs RAM - Hell Corner  

So basically I am writing this (Frequently Asked Yet Wrongly Answered Question) after realizing that there are minimal knowledge available about RAMs and their optimizations :

Q - I have 8 GBs of RAM would buying 16 GBs make my system run faster ?

A - A tech would answer you "But probably you do not need all that memory and you might peek at tops 80% of 8 GB capacity so might be a useless upgrade not unless you are running high end Video/2D/3D editing software !" .

But yet here is a fact few know , upgrading to 16 GBs of RAM might actually "Slow You Down" . As you might know RAMs come claimed to be running at a certain Bus Speed (Lets say DDR4 2400 MHz) , is it running really at such speed ? Nope you see this is the Dual Channel Bus technology claimed speed , Dual Bus is a technology made so you can be reading data off both RAMs simultaneously (Commonly they give out two slots for RAMs in most systems where if more , the claimed bus speed may incrementally increase along with channels where it would be 4 channels ~ 8 channels etc) . So by claiming RAMs to be running at 2400 MHz this means a single stick speed is actually 1200 MHz .

The sad story is there had been a war for years between manufacturers of RAMs and OS makers about what is called "Load Balancing" of data i.e altering the writing to both sticks so upon reading you get the actual Dual Bus speed value , the OS makers believe this should be handled by a smart chip on a RAM stick while manufacturers believe this is OS problem choosing what data goes where as OS is all about control .

The result of this sad story is : We are left with sequential writing i.e the OS keep filling the RAMs based on first free address it finds along the path . Hence one RAM stick needs to be filled first before the other light up .

So guess by now you are starting to gather the plot , if you are actually hardly pushing the use of your RAM capacity to 80% while on 8 GBs , upgrading to 16 GBs might actually get you to use 1 stick the whole time , meaning running at Single Channel Bus (1200 MHz in this example) capacity the whole time which is something about 50% speed loss . So before considering over-clocking your 16 GBs RAMs which would get you marginal MHz speed boost , do consider a down grade to 8 GBs to gain a significant 100% speed boost .

Cheers


--Edit--
I have noticed the surprise on some , but let me tell you I am not posting this to start a debate but to rather raise awareness , people on reading fact sheets and tech review posts fail to realize this is all not so different to when is said SATA 3 Bus speed/Transfer rate is "theoretically" 600 MB/s this doesn't make each single drive run at such speed .

For RAMs there is a cap to each stick and that's the single channel speed it is marked at in its own specs . The bus bandwidth however incrementally adds bandwidth equal to the speed of each stick added and this is to mark it can utilize/read/write each stick independently , but that doesn't mean you can squeeze data out of a single stick faster by throwing in more rams but rather that the pool of installed RAMs can rather "Theoretically" operate at such incremented bus speed collectively . Hence on running benchmarks no uniform result can be noted and that is related to how many sticks were utilized during the benchmark .

--Edit 2--
There is some intel I landed on regarding the situation . Seems by design of dual bus (or higher) technology there seemed to be a setting designed specifically to aid with the horrible effect of sequential writing of RAMs that almost all manufacturers of RAMs over looked until not so long ago , it is called CL , you see CL allows a latency like to read or write to RAMs , it is like the stick of RAM introduce a pause so not to accept reading or writing for a certain period of time (tiny fraction of a second) then un-pause to continue . The aim for such trick was to allow more sticks to alternate in pausing and un-pausing so to evenly distribute data over both sticks (pausing one stick would lead the bus controller to check the other stick and see if it is paused or not to write to which) . For years brilliant manufacturers had been producing RAMs with fixed CL (i.e both sticks would pause simultaneously and un-pause simultaneously rendering this feature useless and that RAMs continue to operate in sequential writing mode that renders Dual Bus useless until first RAM is congested and second RAM lighting up .

First to realize a use for such feature were Kingston , they managed to release a line of RAMs under code-name HyperX , what they introduced was that both RAMs would randomly flicker their CL time hence by luck their pauses are out of sync leading to both RAMs filling at (near) balanced order (it can not be claimed its a perfect balance since the process is left to randomizing latency within a region of course) . Surprisingly Kingston named this feature "Automatic Overclocking" and the internet is praising the speed gain of these RAMs not understanding that they did nothing but sticking to the design and yet the naming was a little biased as hardly there is any bus speed change . Weather or not other manufacturers had followed that trend is beyond the intel I had but then I thought to keep you updated .

nIGHTmAYOR said:

So basically I am writing this (Frequently Asked Yet Wrongly Answered Question) after realizing that there are minimal knowledge available about RAMs and their optimizations :

Q - I have 8 GBs of RAM would buying 16 GBs make my system run faster ?

A - A tech would answer you "But probably you do not need all that memory and you might peek at tops 80% of 8 GB capacity so might be a useless upgrade not unless you are running high end Video/2D/3D editing software !" .

But yet here is a fact few know , upgrading to 16 GBs of RAM might actually "Slow You Down" . As you might know RAMs come claimed to be running at a certain Bus Speed (Lets say DDR4 2400 MHz) , is it running really at such speed ? Nope you see this is the Dual Channel Bus technology claimed speed , Dual Bus is a technology made so you can be reading data off both RAMs (Commonly they give out two slots for RAMs in most systems where if more , the claimed bus speed may increase along with channels where it would be 4 channels ~ 8 channels etc) . So by claiming RAMs to be running at 2400 MHz this mean a single stick speed is actually 1200 MHz .

The sad story is there had been a war for years between manufacturers of RAMs and OS makers about what is called "Load Balancing" of data i.e altering the writing to both sticks so upon reading you get the actual Dual Bus speed value , the OS makers believe this should be handled by a smart chip on a RAM stick while manufacturers believe this is OS problem choosing what data goes where as OS is all about control .

The result of this sad story is : We are left with sequential writing i.e the OS keep filling the RAMs based on first free address it finds along the path . Hence one RAM stick needs to be filled first before the other light up .

So guess by now you are starting to gather the plot , if you are actually hardly pushing the use of your RAM capacity to 80% while on 8 GBs , upgrading to 16 GBs might actually get you to use 1 stick the whole time , meaning running at Single Channel Bus (1200 MHz in this example) capacity the whole time which is something about 50% speed loss . So before considering over-clocking your 16 GBs RAMs which would get you marginal MHz speed boost , do consider a down grade to 8 GBs to gain a significant 100% speed boost .

Cheers

Multi-channel memory architecture - Wikipedia

I do not believe this where did you get your info from.

If this were true it would be well known and there would be no such thing as dual channel and especially quad channel, If I had 3600 Mhz quad channel ram does that mean its only running at 900 MHz ---- I think not.

A article I found on dual channel and single channel.

Source:
RAM Performance Benchmark: Single-Channel vs. Dual-Channel - Does It Matter? | GamersNexus - Gaming PC Builds Hardware Benchmarks

A single stick of RAM will operate on a single 64-bit data channel, meaning it can push data down a single pipe that is 64-bits in total width. The channel effectively runs between the memory controller or chipset and the memory socket; in the case of modern architectures, the memory controller is often integrated with the CPU, rather than acting as standalone board component.

By utilizing multi-channel platforms -- something available on every modern build -- we multiply the effective channel width by the count of channels available. "Effective" is key. In the case of dual-channel configurations, we've now got 2x64-bit channels available to the memory. This means we've doubled the data traces running in the memory bus, and now have an effective 128-bit channel, which in turn doubles maximum theoretical bandwidth. This is why I made it a point to say that dual-channel platforms are what exist, not memory -- in the case of dual-channel, the board will host 128 physical traces to handle data communication between the IMC and RAM. This is compared against 64 for single-channel platforms and 256 for quad-channel platforms.

sml156 said:

I do not believe this where did you get your info from.

If this were true it would be well known and there would be no such thing as dual channel and especially quad channel, If I had 3600 Mhz quad channel ram does that mean its only running at 900 MHz ---- I think not.

Let me tell you I am not posting this to start a debate but to rather raise awareness , people on reading fact sheets and tech review posts fail to realize this is all not so different to when is said SATA 3 Bus speed/Transfer rate is "theoretically" 600 MB/s this doesn't make each single drive run at such speed . For RAMs there is a cap to each stick and that's the single channel speed it is marked at in specs . The bus bandwidth incrementally adds bandwidth equal to each stick added speed but that doesn't mean you can squeeze data out of a single stick faster by throwing in more rams but rather that the pool of installed RAMs can rather "Theoretically" operate at such incremented bus speed collectively . Hence on running benchmarks no uniform result can be noted and that is related to how many sticks were utilized during the benchmark .

AS185 said:

That's what the entire world say's, as the answer mentioned by tech's. So, why do all the manufacturer's produce gaming motherboard's with 128GB of random access memory support? Because, we can engage video editing on them? Yeah, right
As noted, I'm running 64GB quad-channel @ 3000Mhz ram. My PC performance is astonishing and I doubt that an average computer with only 16 or 32GB's of random access memory, shall perform the same way. People in the world can debate all they want, why? Because they never possessed an computer that powerful. Maybe their just envious, technologically misinformed & speculating?
For most of the people out there, as to comparison, my PC is one of those however, yes, it's a fact, there's even more impressive collaborated components possessed by others.
The people in the world may think whatever they want but to those who know, this is rather non-sense.

So basically you poses 4 sticks of 16 GBs how about you remove 3 of them to drop to single bus and see if you notice any drop in speed/performance ? :) (Of course this all revolves around how far you utilize this machine , if your ram usage exceeds 25% i.e you are normally using a little above 1 stick then maybe just remove 2 sticks and test)

The point is sometimes the hardware is so advanced that even working on a low performance profile such as single bus speed the performance drop is hardly noticeable because its too fast already , but when it comes to people trying to squeeze out the last drop of an average or low end hardware then this advice of dodging 16+ GBs of RAM if not needed might come handy .

And on a side notice , why would a manufacturer make a 128 GBs capable board , ask yourself this first , what sort of game really utilize such amount of ram ? there are rare games even that broke the 100 GBs barrier of disk space so expecting such games to utilize such amount of memory is like believing a game is loading all game assets for all levels on loading which would be an obsured trend . So why are they making boards capable of handling such capacities ? Because :

1 - Marketing wise people want cutting edge
2 - Who knows what the future may hide , so a consumer buying hardware of high upgrade-ability value render him satisfied
3 - Some consumers are geeky enough and host their own game servers / streaming servers / virtual servers and other blahs that may require who knows what capacity and horse power
4 - Rich spoiled kids who like RAM disk cause its super fast :)

So like you see this is not coming out of necessity but rather out of : Lets give people what they want as long as they are paying for it .

AS185, post your winsat mem score