Compatibility RAM with higher clock speed than CPU and motherboard support

If the processor (i5-6500) supports RAM up to 2133 MHz, as does the motherboard (MSI B150M Bazooka), will 3000 MHz RAM work, but at a lower frequency, or is it not compatible at all?

They will be on 2133MHz, even on boards that support 3000MHz the frequency will be set to 2133MHz, you only need to enable the XMP profile yourself.

They will operate at the maximum frequency supported by the motherboard. Do not worry about the processor memory controller, even the INTEL i7 from what I remember, has a memory controller only 2666 MHz and people put it together and work with RAM with a frequency of 3600 MHz and higher, here is the motherboard and what they support the maximum memory frequencies. If the memories have higher frequencies than the motherboard supports, they will only run at the maximum frequency supported by the motherboard, you won't miss that.

Hello, I'm going to link to the topic as I'm puzzled by this fact. I have an i5 8600K processor @ 4800, no matter what. But it supports a max memory frequency of 2666. I previously had 2 dice in X.M.P set to 3000 and left it that way, and didn't play around. Recently I was putting together a PC for a friend and managed to get DDR4 4000 memory, so I swapped. And they're running X.M.P on that 4000, so getting to the point. Is it better to leave the 4000 or downsize to 2666 and lower the timings as much as possible? Which will be more efficient?

>>21499492 If you have a modern XMP profile, e.g. 2.0 or 3.0, the RAM will be able to run on such a motherboard at higher clock rates than the CPU supports. DDR5 on a processor that does not support it, on the other hand, will not work at all, so called Check the manufacturers offering the capabilities of their motherboards and processors.

On a Z-chipset motherboard and a K-series processor there are no locks, the limiter is the performance of the memory controller in the processor and the quality of the paths on the motherboard. If 4000MT/s was stable on you on this set up then there is no point in limiting your friend's prock.

Well ok, but then why is there this bus of e.g. 2666 MHz in the CPU, if the speed of the memory depends only on the motherboard? I always thought that the board decides how fast the memory runs, i.e. how many operations it can write to itself, e.g. 3200 MHz, but the CPU bus decides how fast then this written data in the RAM will be sent to the CPU. And how is that exactly? For some purpose this bus in the processor is there and has some speed set. If it could run faster, like 3200 MHz memory, then the CPU would support 3200, not 2666 for example. I've read the threads on this forum, but in each one it's the same thing, that is, not fully explained. They wrote that the memory runs as fast as the board allows, but how this then relates to the slower bus in the processor and what it is responsible for, nobody wrote anymore.
I read that it is responsible for communication with the RAM, so logically - what does it matter if the memory is 3200, if it then communicates at 2666? Do I understand this correctly?

fortaigne wrote:

No ok but then why is there this bus of e.g. 2666MHz in the CPU if the speed of the memory depends only on the motherboard

.
For RAM to run faster, you need to have both a board and a processor that supports faster RAM. So the speed doesn't just depend on the board.

[postid:83adb828e1][/postid:83adb828e1]

So as an example: when I put in 3200 MHz memory and the CPU is 2666, the memory, even though the board allows it, how does it end up working? At 3200, which the board shows, or 2666, as the processor can handle?
If the CPU slows them down, why the different benchmark results in the tests with different 2666, 3200 memory settings? I've seen on videos that they were different. If the processor slows them down, shouldn't they be the same?
P.S. After all, I don't think any processor supports 6000 MHz and above, and people put such memory in Ryzen 5 or Core i5/i7. Why then?
I've always been interested in this, but nobody could explain it....

fortaigne wrote:

why in tests at different settings

.
dude, there are dozens of different i5 processors as well as boards, different generations, different chipsets. Each generation and chipset is a separate case. In each case you can check how it is. It is not some kind of secret knowledge.

[postid:490036193f][/postid:490036193f]

Don't sulk this is how people who don't quite know what to answer behave, they immediately get pissed off, others lash out, even that the questioner has no idea or write out how many things it could depend on, only that the question is specific.
Chipsets, boards, procs, different ones have nothing to do with it, in this case that I'm asking, we assume with we have some CPU with frequency X and memory that can run higher than frequency X, and the specific question: the CPU will cut the memory frequency to X (if according to your theory it matters) or it won't, as tests show that with different memory frequencies on the same CPU and board there are different results.
Tests by youtubers who work for these larger shops show clearly:
they have a kit on a specific proc and board (chipset) and they do the tests on the same platform. They only change the memory timing 4800, 5600, 6000 in the BIOS (the CPU has a 4800 bus in their case (i5-14400f)).
Since, according to your theory, the processor matters, it should slow the dice down to 4800 MHz, which the processor supports, and give the same result regardless of the dice setting. In contrast, it is clearly shown in the tests that any increase in the clocking of the dice, even though they exceed the CPU's supported frequency, results in an increase in performance.
Conclusion: either the processor doesn't have an impact after all, and you're thinking, or it does, but its bus frequency doesn't cause the inability of the memory to operate at higher frequencies.
Well, unless it is influenced by something that truncates, but percentage-wise for each bone no less, however, higher frequencies do matter, even if they are higher than the base prock. Unfortunately nobody has explained this anywhere yet, "se check" answers are you know what they are worth....
No one in any topic has yet described how it works, whether it cuts down to the base prock (which as we know from tests is not quite true), or whether it does not cut down and affects something else (which no one has described). Mostly the answers are "it doesn't matter", "it does matter" or, like yours, "it depends on a number of factors", which in this particular case has no effect, because the factors are the same.
Yes, I know, you certainly know more than I do and I am not denying that, but if you can't answer a question because you don't know or you yourself are not sure, then don't make personal excursions, don't sulk or don't speculate, because not knowing everything is no shame.
I know less than you, but before I ask, I look up and read everywhere, that's why out of the 10 questions asked here on the forum 9 have no answer or are ones I have looked up myself, that's how much more I know how to search on the internet. Unfortunately every article they give different pitches, even on the IT portal I found the absolute opposite thesis to yours, that it doesn't matter at all, on the other hand Chat GPT has another one, it says it does and that it will cut down to the same as the base one, which is nonsense, as the tests on YT showed. Also 3 different answers and no one is theoretically right, hence my query here, but it will surely be unresolved.

fortaigne wrote:

I mean, as an example: when I put in 3200 MHz memory and the CPU has 2666, the memory, even though the board allows it, how does it end up working? At 3200, which the board shows, or 2666, as the processor can handle?

.
With Intel there are a lot of variables (read - artificial locks) that can artificially limit the max speed of supported memory to varying degrees. Limits to 2133 or later 2666 were as artificial as possible, locked in software at the processor and/or board level and had nothing to do with the actual performance of the memory controller in the processor or the stability of the tracks in the motherboard.

fortaigne wrote:

If the processor slows them down, why the different benchmark results in the tests at different 2666 memory settings, 3200? I've seen in the videos that they were different. If the processor slows them down, shouldn't they be the same?

.
Intel's K-series processors along with Z-series chipset boards (and all AMD processors) do not have the above locks, in them there will be no slowing down of RAM, and the limiter will be the physical performance/stability of the components.

fortaigne wrote:

P.S. After all, I don't think any processor supports 6000 MHz or more, and people are putting that kind of memory into Ryzen 5 or Core i5/i7. Why then?
I've always been interested in this, but nobody could explain it...

.
Because you put them in platforms that don't have locks and don't slow down the RAM. In the case of Ryzen on the AM5 platform, even AMD itself recommended 6000CL30 as the best price/performance trade-off, even though officially, according to JEDEC certification (the RAM development and standards body) the procs only support 5200. Official certification and rejoinder is one thing, actual performance is another.

fortaigne wrote:

Don't sulk

.
if you spent 10% of what time you waste writing mini novellas on the forum on googling or at least asking AI to gain knowledge on the topic, you wouldn't be asking stupid questions.

fortaigne wrote:

So as an example: when I put in 3200 MHz memory and the CPU has 2666, the memory even though the board allows it, how does it end up working? At 3200, which the board shows, or 2666, as the processor can handle?
If the processor slows them down, why the different benchmark results in the tests with different 2666, 3200 memory settings?

.

To explain this, you'd probably have to write a book, and ideally you'd have to discuss the design of the PC in turn and all the stages of its development since 1980 .
The problem comes from the fact that people often confuse the clock frequency of memory, and forget about the access time of memory when writing and reading.
Hence interesting situations come out, like for example a computer clocked at 4000 MHz is slower than 3000 MHz, but people think that after all 4000 is more than 3000 and are ready to choose a slightly more expensive product, which is actually slower .
Note that a PC is a collection of different parts from different manufacturers.
Generally, when it comes to RAM, the most important thing is access time, not clocking.
To read a RAM cell, you need to send to it at different times usually the 2 most important parameters, i.e. the column address and the row address, but the time between these signals has some maximum parameter for a given memory and you are unlikely to cheat this.
I'll give you a purely theoretical example, I'm taking the parameters from memory, so treat this as a theoretical example.
You have a controller and a RAM, you set it to 200 MHz and in order to read the RAM, you have to issue half of the address and communicate that with the RAS signal, which has to last a minimum of 10 cycles, then issue the other half and the CAS signal. Now by waiting 20 cycles, we are sure that the data to be read is ready and we can retrieve it.
Now, let's say we clock the same memory at 400 MHz and what happens? Well, we do the same, but after issuing RAS we wait 20 cycles, then we issue CAS and after 40 cycles we have ready data .
To quote a classic .
Someone will think:
- After all, it's the same thing.
- But how it sounds!

gregor124 wrote:

Generally, when it comes to RAM, the most important thing is access time, not clocking.

.
Both are important, with similar access times the higher clocked dice will fill the CPU with data better.

turkuc11 wrote:

It is important to have both, with similar access times the higher clocked dice will fill the CPU better with data.

.
I'll admit that I don't understand much of this.
What clocking are you writing about? Can you write more clearly?
Generally RAM is a big matrix of capacitors, one per bit.
In order to charge a capacitor to a suitable (safe) level, you need to spend some time, which is expressed by the formula
Charging time = R * C * ln (1 - V/V₀).
And you change the frequency without changing any of these parameters.
In addition, in memory, in order to get to a particular bit, you have to switch some number of transistors, which no matter what you clock with, is the same and introduces exactly the same delay, if we are talking about the same memory.
But, of course, the world is not perfect and two of the same chip can be minimally different, so manufacturers for given frequencies give the access characteristics of the module (not the individual chip) in constant memory and usually do so for the worst case.
The memory controller reads this parameter and sets itself accordingly.
Of course the manufacturer of the motherboard can change this parameter even programmatically, so with millions of RAM access cycles per second there may be differences even on the same chipset.
But as I wrote, this is a river subject...