Memory Mhz importance.

[Oldschool297]

I do not play any games and i wonder if mhz is very important? I have 3200 memorys and i do not notice much difference between 2600 and 3200.

 

[P4-630]

I do not play any games and i wonder if mhz is very important? I have 3200 memorys and i do not notice much difference between 2600 and 3200.

You may be able to notice the latency difference, example how fast "my PC" window opens.

Better latency/faster memory = overall snappier feeling.

 

[_JP_]

I do not play any games and i wonder if mhz is very important? I have 3200 memorys and i do not notice much difference between 2600 and 3200.

If not gaming, in any kind of media editing and especially applying compression it makes a difference.
General apps, like office and browsing, not noticably.
But windows updates do run faster, though. TrustedInstaller is a RAM hog since Windows 7 SP1.

So for that last scenario, if you catch a good deal, go for it. Otherwise, depends on your perception of value on performance and its priority.

 

[Bill_Bright]

The amount of RAM is MUCH MORE important than its speed.

In other words, I would MUCH rather have 16GB of 2600MHz RAM than 8GB of 3200MHz RAM. I would MUCH rather have 16GB of 2400MHz RAM than 8GB of 3600MHz RAM!

With all other specs being equal, it is highly unlikely even Superman would notice any difference. So a mere human with our own limited human senses noticing any performance gains if there was, for example, 16GB of 2600MHz RAM instead of 16GB of 3200MHz RAM in a "blind test"? Nope!

"Blind" means if we had no clue what RAM was in there, and all other specs were equal, any differences in performance would not be noticeable. If we knew which RAM was in there, any performance gains we perceived undoubtedly would be our mind playing tricks due to the placebo effect.

It should also be noted if we perform the exact same task 3 times in a row, the time it takes to complete each task will be slightly different each time. Keep that in mind.

Yes, faster RAM can result in certain tasks running faster, but the reality is, you are talking a few (like <5) scant milliseconds at most for the vast majority of even the largest tasks. Smaller tasks will be in the tenths of a millisecond (microseconds). Our eyes and minds just are NOT that sensitive. The average human needs at least 13ms to "see" something.

On paper (benchmark programs)? Sure, you will see better scores. But in playing a game, watching videos, performing office tasks? Nope! And again, that's with all other specs being equal.

Here's a good read on the topic: The Full Guide to RAM Speeds: DDR4 2400 vs 2666 vs 3000 vs 3200 vs 3600 vs 4000 MHz - What in Tech.

Note the following,

Of all the specs, size matters the most.

it’s worth re-iterating: size is way more important than MHz & latency for nearly all users.

Final Word
we’ll reiterate: the most important aspect is size.

CAS Latency matters, but the OP didn't ask about that. So again, with all other specs being equal, size matters most.

The MUCH bigger bottleneck will be the drive - even if a fast SSD. The slowest RAM is many times faster than the access plus read times of the fastest SSD.

I do not play any games and i wonder if mhz is very important?

To answer your question, the answer is, "no". Definitely not "very" important and in the grand scheme of things, not that important at all. Size matters most. Then CAS Latency specs and then speed. Are there exceptions? Of course! But those are relatively rare and generally unique scenarios.

 

[Dr. Dro]

A balance must be struck between capacity, timings and frequency (speed). CAS latency is one of the highest importance timings (which is why people usually use it to describe a kit's latency rating), but it's not the full story. Secondary and tertiary timings all have the potential to greatly impact how fast does a memory chip react to commands sent by the processor, but that, in general, is a more advanced talk that you aren't yet ready for. As Bill says, the single most important factor is capacity. Always prefer a higher capacity kit over anything else if you're in a budget (eg. slower 32 GB kit is heavily preferred over faster 16 GB kit), followed by the timings and then frequency.

Timings and frequency have an interesting relationship, timings will generally increase as frequency rises but a slightly lower frequency with a stricter timing set will generally outperform a kit that goes all-in on bandwidth, for example, you should prefer a 6400 MT/s kit that operates with 30-38-38 over a 7000 MT/s kit that does 40-48-48, the faster timings reduce access latency and permit the memory to complete more refresh cycles within the same time period, resulting in a snappier system.

As far as capacity goes, currently the sweet spot for Windows-based computers is 32 GB, whether you are a gamer or not. Avoid purchasing memory kits or computers that have less than 32 GB of RAM if it can be helped, RAM is inexpensive and it is always best to have it and not need it, than not have it and need it. It's also important to note that you should populate each memory channel available on your system with at least one stick. So avoid purchasing one big memory stick, most computers have a dual channel memory architecture and as such you should opt for a kit that comes with two of them, each containing half the capacity.

 

[ShrimpBrime]

I do not play any games and i wonder if mhz is very important? I have 3200 memorys and i do not notice much difference between 2600 and 3200.

To directly answer the question.

Yes - Mhz are very important.

If Mhz where NOT important, we could still be using DDR1 speeds like 200mhz CL2 memory kits and DDR2 would never need to be invented.

The Frequency goes hand in hand with bandwidth. We move up generations of DDR to accommodate the bandwidth needed to move larger amounts of data.

Since you're using the same type of DDR (4), frequency differences don't make a lot of difference in normal use.

But if you ran a 5800X on DDR1, I'm sure the performance would be horrible.

Amount of memory is on a need be basis. Some of these guys could utilize 128gb. Others would not. Storage capacity will have no effect on thr bandwidth capability of the the generation of DDR in use.

And never mind the word "feel" others may use to describe how memory works because it is all pretty measurable, you can straight see the difference by the numbers produced from simple benchmarks.

At one time, benchmarks like PiMod and PiFast where popular because you could measure between two systems which settings produce a lower time, which is desired tlfor your system to ""feel" "snappy"..

OK, so then we can simply measure and prove 3200mhz is "better/faster" than 2600mhz no matter the capacity.

GL

 

[Bill_Bright]

If Mhz where NOT important, we could still be using DDR1 speeds like 200mhz CL2 memory kits and DDR2 would never need to be invented.

Nah! First, nobody said MHz where "NOT" important so that's just obfuscating the issue. Yes, it is important, just not the top priority.

Second, sorry but you don't understand the difference between DDR (DDR1) and DDR2. Actually the clock speeds were essentially the same. It was
the transfer rates that were faster due to the enhanced input/output bus signal which had nothing to do with MHz capability of the RAM. DDR2 had a 4-bit prefetch, twice that of DDR. DDR2 could also reach data rates of 533 to 800MT/s.

DDR2 memory also supported "dual channel mode". DDR1 did not support that.

More significantly, later versions supported higher density chips for MORE RAM because again, more is better than faster. Later versions also consumed less power.

So there where lots of reasons DDR1 was replaced. Speed really was not a top priority.

 

[68Olds]

DDR2 memory also supported "dual channel mode". DDR1 did not support that.

Sorry, but that is incorrect. DDR1 has support for dual channel mode. Here's an example of it...

 

[_JP_]

DDR1 did not support that.

It did, in from Northwood and ClawHammer onwards. One just wouldn't enable it because your FSB would already be 1:1 with the RAM's bus speed and for those chips (depending on the uArch), you were looking to compensate the process pipeline, so latencies were all the rage.
You'd still want 1:1 while FSB was a thing, though, so if you had a better CPU (especially a dual-core later on), so that is where Dual-Channel would come to be relevant. Intel dual-cores also coincided with DDR2 becoming mainstream, which is why I fell you have that impression.

 

[Dr. Dro]

It did, in from Northwood and ClawHammer onwards. One just wouldn't enable it because your FSB would already be 1:1 with the RAM's bus speed and for those chips (depending on the uArch), you were looking to compensate the process pipeline, so latencies were all the rage.
You'd still want 1:1 while FSB was a thing, though, so if you had a better CPU (especially a dual-core later on), so that is where Dual-Channel would come to be relevant. Intel dual-cores also coincided with DDR2 becoming mainstream, which is why I fell you have that impression.

Back in those days the memory controller was on the chipset instead of integrated onto the processor (this only became a thing with AMD's K8 CPUs afaik, and Intel adopted it much later only with Nehalem - Core 2 still has memory controller on the northbridge), so you could get dual channel on the earlier Willamette chips as well. Much of this confusion back in the day likely originates from the Rambus systems that operated at 16-bit per channel. Glad that thing didn't take off... especially the requirement for terminator/continuity modules... yikes.

 

[QuietBob]

It did, in from Northwood and ClawHammer onwards

Actually, the Athlon XP/Duron family also had dual channel support with specific versions of the nForce chipset. But IIRC the additional memory bandwidth didn't translate into real-life gains.

 

[Bill_Bright]

Sorry, but that is incorrect. DDR1 has support for dual channel mode. Here's an example of it...

Hmmm, odd because when creating that post, my memory was agreeing with you. But I like to verify my facts before posting because I learned long ago that my memory is not perfect - especially with technologies that are constantly evolving.

So I did some digging and found a couple links, including this from Crucial. I made another mistake by "assuming". I should never do that. I assumed Crucial, one of the biggest names in memory, should be considered an authority on this. Maybe not.

Note it says under DDR vs. DDR2,

DDR2 memory can be installed in pairs to run in "dual channel mode," which can increase memory throughput even more.

I assumed that meant this was something new to DDR2. Apparently that is incorrect.

So I have now gone back and looked at other claims (latencies, bus rates, I/O, prefetch, efficiencies) from multiple sources and believe they are correct.

Thanks simply pointing out the errors and not personally attacking the messenger!

 

[ShrimpBrime]

Nah! First, nobody said MHz where "NOT" important so that's just obfuscating the issue. Yes, it is important, just not the top priority.

Second, sorry but you don't understand the difference between DDR (DDR1) and DDR2. Actually the clock speeds were essentially the same. It was
the transfer rates that were faster due to the enhanced input/output bus signal which had nothing to do with MHz capability of the RAM. DDR2 had a 4-bit prefetch, twice that of DDR. DDR2 could also reach data rates of 533 to 800MT/s.

DDR2 memory also supported "dual channel mode". DDR1 did not support that.

More significantly, later versions supported higher density chips for MORE RAM because again, more is better than faster. Later versions also consumed less power.

So there where lots of reasons DDR1 was replaced. Speed really was not a top priority.

Priority?
Difference from.ddr to ddr2?
You explained the difference. Ddr2 is 50% higher bandwidth at double the frequency.

More memory in a time when 32 bit OS was mostly common and 3.5gb was overkill in this time period? Like when running 2x 512mb modules was normal?

There's a lot to it. We can agree to disagree, but I didn't ask to be quoted.

 

[_JP_]

so you could get dual channel on the earlier Willamette chips as well

True, for the ones that were rebranded for socket 478, not for the 423. But for Willamate's low FSB it was still negligible, only if you went for 200MHz DDR.

Actually, the Athlon XP/Duron family also had dual channel support with specific versions of the nForce chipset. But IIRC the additional memory bandwidth didn't translate into real-life gains.

True and I did have it on the NF-7, but yes gains were none most of the time, so I think I benchmarked once. Thank you, I stand corrected.

 

[Dr. Dro]

So I did some digging and found a couple links, including this from Crucial. I made another mistake by "assuming". I should never do that. I assumed Crucial, one of the biggest names in memory, should be considered an authority on this. Maybe not.

Probably written by an underpaid trainee years ago

True and I did have it on the NF-7, but yes gains were none most of the time, so I think I benchmarked once. Thank you, I stand corrected.

I wonder if that would hold true with heavier workloads available today, although you probably can't get anything newer than XP running on that...

 

[ShrimpBrime]

Sorry, but that is incorrect. DDR1 has support for dual channel mode. Here's an example of it...


View attachment 353384

Cheers

 

[chrcoluk]

I have no idea, as I even when I tried never been able to get memory to run at high mhz, I always have hit a clock wall around 3000-3200mhz, I could only get above when changing it to gear 2, my only ddr5 system is a n100 NUC.

I could beat the clock wall with gear 2, but didnt bother doing anything other than stability testing as it will be inferior to lower clocked gear 1.

I have seen a couple of games react positively to very tight latency though, (stutters going away or becoming micro stutters), and logically more bandwidth will reduce i/o wait so improve CPU efficiency.

I have done some testing on linux software between 2133mhz and 2866mhz and there was noticeable CPU gains.

 

[iameatingjam]

It depends on the task. I mean my laptop still uses ddr3 and feels snappy. But all it does is watch youtube and browse the web. Still overall, I think its less important than people think it is, especially after a certain point (like 6000). Doesn't stop me from obsessing over it at times though.

 

[Outback Bronze]

Sorry, but that is incorrect. DDR1 has support for dual channel mode. Here's an example of it...

Yeah, was gona say, pretty sure this is where it first started no? I used to run a 2.4B 533 FSB Northwood then the 3.0C 800 FSB Northwood which supported dual channel and HT. I was running DDR400 in dual channel on P865's then later P875 mobo's

 

[Bill_Bright]

There's a lot to it.

??? As I noted right off the bat in my first post here. In fact I noted that by saying no less than 3 times,

With all other specs being equal

I also noted other factors when I mentioned,

CAS Latency matters

the drive

Are there exceptions? Of course!

Priority?

Come on, Dude! Stop taking things out of context then pretend they support your claim. You've done that twice now and it is deceptive at best.

What I said both times I used the word "priority", was that speed was not the "top priority".

We can agree to disagree

If you are going to continue to claim for most people speed is more important than the amount of RAM, then we definitely disagree. But note you are essentially saying you are right and the rest of the world is wrong. I provided supporting evidence. You provided none.

Here's more.
Faster RAM vs More RAM - Which is better for your Workload? (cgdirector.com)

Professional users should almost always opt for “more RAM”, though.

Making that RAM fast can be good as well, but the biggest performance improvement will be present in making sure that you never run your RAM to full capacity outside of the projects meant to push the limits of your system.

What Is More Important RAM Speed Or Size? - Computer noobs

Generally speaking, RAM size is more important than RAM speed, at least up to a point. In fact, the conventional user will not notice any difference when having a higher RAM speed or frequency. On the other hand, RAM size can make a substantial difference.

What Is More Important RAM Speed Or Size? - Computer noobs

Having enough RAM capacity is even more important than focusing on high speeds. If your PC doesn’t have enough RAM, the lack of it will become a major bottleneck,

So let me be clear, for most users more RAM is more important and offers greater performance gains than faster RAM. Are there exceptions? Of course! When the computer is already loaded with lots of RAM, more than enough to meet the user's demands, then faster RAM may provide a noticeable performance gain. But so might upgrading from a hard drive to a SSD. Or upgrading the graphics solution or CPU, or both. Or even swapping the RAM with the same size RAM that has better timings/latency specs.

I NEVER claimed "more RAM is ALWAYS better!" In fact, I specifically said in my first post, "faster RAM can result in certain tasks running faster."

but I didn't ask to be quoted.

Then don't post anything.

Quoting is how other readers know who you are replying to.

It sure is interesting, if not a bit hypocritical, to have an aversion to being quoted yet have no problem quoting others, like @Oldschool297, @68Olds, or me.

Oh well. I believe the OP has his answer. Speed is important, just not the "top" priority - as he witnessed himself with his own comparison between 2600 and 3200MHz RAM.

Time to move on.

 

[ShrimpBrime]

??? As I noted right off the bat in my first post here. In fact I noted that by saying no less than 3 times,

I also noted other factors when I mentioned,



Come on, Dude! Stop taking things out of context then pretend they support your claim. You've done that twice now and it is deceptive at best.

What I said both times I used the word "priority", was that speed was not the "top priority".


If you are going to continue to claim for most people speed is more important than the amount of RAM, then we definitely disagree. But note you are essentially saying you are right and the rest of the world is wrong. I provided supporting evidence. You provided none.

Here's more.
Faster RAM vs More RAM - Which is better for your Workload? (cgdirector.com)


What Is More Important RAM Speed Or Size? - Computer noobs
What Is More Important RAM Speed Or Size? - Computer noobs


So let me be clear, for most users more RAM is more important and offers greater performance gains than faster RAM. Are there exceptions? Of course! When the computer is already loaded with lots of RAM, more than enough to meet the user's demands, then faster RAM may provide a noticeable performance gain. But so might upgrading from a hard drive to a SSD. Or upgrading the graphics solution or CPU, or both. Or even swapping the RAM with the same size RAM that has better timings/latency specs.

I NEVER claimed "more RAM is ALWAYS better!" In fact, I specifically said in my first post, "faster RAM can result in certain tasks running faster."


Then don't post anything.

Quoting is how other readers know who you are replying to.

It sure is interesting, if not a bit hypocritical, to have an aversion to being quoted yet have no problem quoting others, like @Oldschool297, @68Olds, or me.

Oh well. I believe the OP has his answer. Speed is important, just not the "top" priority - as he witnessed himself with his own comparison between 2600 and 3200MHz RAM.

Time to move on.

See, you turn the shit into who is right and who is wrong.

Correct, speed is very important.

It's why people enable XMP and not use system defaults.

XMP is a frequency memory profile.

Competitive benchmarking has shown for decades the importance of memory speed.

You lower memory latency with simply overclocking a cpu.

You've proven nothing to me. I overclock for speed. Because it's important..

No matter your opinion, the fact frequency matters will go unchanged.

I agree to say quantity of RAM is important to a single user, ram speed is important to all users.

 

[Bill_Bright]

No matter your opinion, the fact frequency matters will go unchanged.

And there you go twisting facts and context again. That's dishonest.

You just quoted my post and NO WHERE in that did I say frequency doesn't matter. Yet here you are suggesting I did. That deception is pathetic.

As I noted, the OP has his answer. I'm done here.

 

[A Computer Guy]

I do not play any games and i wonder if mhz is very important? I have 3200 memorys and i do not notice much difference between 2600 and 3200.

What kind of CPU do you have?

From the best I know of with AM4 you get better CPU utilization as you approach DDR4-3800 but the gains above DDR4-3200 are minimal unless you can blow past the 1:1 ratio by a significant margin by becoming an expert RAM over clocker. Will jumping from DDR4-2600 to DDR4-3200 save you 30 minutes of time (and time is money) in any particular workload that spans 1hr - probably not for normal user workloads.

I'm going to address a different angle than other posts here...

Where speed is always important is related to the cost (at a needed capacity) you are willing to pay for it. In the AM4 example:

• Is DDR4-4800 worth getting at top dollar - probably not if it's not going to save you time or improve your performance in a meaningful way. (you need to research your use case)
• Is DDR4-2666 worth getting at bottom dollar - probably not when DDR4-3200/DDR4-3600 are so cheap and it utilizes the CPU more effectively.

 

[ShrimpBrime]

And there you go twisting facts and context again. That's dishonest.

You just quoted my post and NO WHERE in that did I say frequency doesn't matter. Yet here you are suggesting I did. That deception is pathetic.

As I noted, the OP has his answer. I'm done here.

No need for name calling here Bill.

You say quantity is most important. More Important. Both??

Not really more important no. Because users purchase the quantity they need at the frequency they want and seldom use stock board default memory speed and timings.

That being said, the general concensus of the public opinion, purchasing XMP kits, frequency is more important.

Gamers want speed which = fps and should always buy fast kits according to their system specs regardless of the amount. Which is Not as important to people while most standard kits these days are reasonably priced for 32gb.

I don't believe people actually think quantity is a problem. This is something you are trying to prove. It's almost not even factual. 128gb desktop for niche users is the same as 16gb DDR5 would be. I actually have a 16gb ddr5 kit, and you might say it's not enough, but hey, that's not your business.

I just want to know....

Memory Mhz Importance.
Would be the same question as
CPU Mhz importance.

 

[InVasMani]

Comes down to you're overall hardware in part and usage. Some usage requires a lot more memory capacity some requires a bit more bandwidth and others more memory latency sensitive. The other hardware is a factor as well since it might expect more memory and faster memory to operate it's full potential. The software itself is a huge factor because it can very significantly change expectations on capacity and performance by being more demanding or less demanding of memory.

The biggest importance for games is dominantly the GPU. You can bottleneck the hell out a RTX4090 in like a old LGA775 DDR2 system and it'll still beat a lot of other GPU's that are weaker even on slightly better CPU's. It's not ideal, but you can still push a ton of FPS simply due to the GPU hardware and importance of it. In fact newer GPU hardware like that example case include more VRAM than you could even populate with system memory on top of it being faster performing once it's in VRAM though transferring it into VRAM takes some loading or buffering period not too terrible though on SSD's.

I wouldn't recommend that scenario, but it's not as terrible as many present it though not advisable to pair a GPU that's over $1000's with a total system that's maybe $50-$100'S today or very possibly free from someone that no longer is going to even use that level of system hardware.