2 sticks is better then 4 sticks?

[Bill_Bright]

This is something TPU could do a review on w/ DDR5, this is a super common question, not alot of data out there.

Too many variables. Unless they did the review by analyzing multiple different motherboards from different makers, multiple different CPUs from different makers and multiple different RAM running in multiple different environments, preferably with both W11 and W10 (maybe Linux too), any review would pretty much be anecdotal.

But this thread here is about AM4 and DDR4.

But is it? The question was, "2 sticks is better then 4 sticks?"

Yes, the OP's example was with an AMD DDR4 board but he did not stipulate the question was exclusively for AMD or DDR4.

In fact, he ended his opening post with, "how much difference is there in % between 2 vs 4 sticks?"

IMO, if the scope of the topic is to be limited/restricted, then it should be specific and the question should become, "With my b550x motherboard and 5600x CPU, what is better, 2 x 32GB or 4 x 16GB?"

Otherwise, we are all just speculating.

 

[phanbuey]

Too many variables. Unless they did the review by analyzing multiple different motherboards from different makers, multiple different CPUs from different makers and multiple different RAM running in multiple different environments, preferably with both W11 and W10 (maybe Linux too), any review would pretty much be anecdotal.


But is it? The question was, "2 sticks is better then 4 sticks?"

Yes, the OP's example was with an AMD DDR4 board but he did not stipulate the question was exclusively for AMD or DDR4.

In fact, he ended his opening post with, "how much difference is there in % between 2 vs 4 sticks?"

IMO, if the scope of the topic is to be limited/restricted, then it should be specific and the question should become, "With my b550x motherboard and 5600x CPU, what is better, 2 x 32GB or 4 x 16GB?"

Otherwise, we are all just speculating.

True but it would still be nice to know setups that work 100%, and how difficult it was to build with them -- there are alot of people that want to build 128GB workstations now that there isn't an HEDT anymore, and those guys are constantly running into issues with this. I see alot of people on youtube and in general running 96GB setups because they're scared the 128GB setup will be to painful to set up.

 

[Bill_Bright]

I am not saying you are wrong but this would be like reviewing a Toyota Camry and trying to say the review applies to all imported sedan cars. Unless one has the same setup, it is just anecdotal.

 

[harm9963]

One thing over look, for example Corsair ,in house tested from Corsair , tested two MB for VENGENCE RT 4600C18 2X16
That's what sold me , cold hard fact , my ASUS DARK HERO is a step up from HERO , I tested 2x16 vs 4x16 , from my testing , tuning both , stable and cool-used Corsair Vengeance Airflow memory cooler, plus day to day use , no crazy OC , got latency at 66 for both , read is the out liner 65800 vs 62500 ,that's it , comes down to what you want spend and capacity .

 

[Bill_Bright]

There is one consistent advantage to using 2 sticks instead of 4. Generally speaking, the energy required to power a RAM stick is roughly the same regardless the capacity of the stick - assuming same architecture.

That is, 1 x 16GB DDR4 takes about the same amount of power to run as 1 x 32GB DDR4.

Likewise, 1 x 64GB DDR5 takes about the same amount of power to run as 1 x 32GB DDR5.

Therefore, running 4 x 16GB sticks of RAM will consume roughly twice the power as running 2 x 32GB sticks. This also means 2 sticks will generation about 1/2 as much heat as 4.

Having said all that, "IF" you are that close to maxing out your PSU that adding 2 more sticks matters, then you need a bigger PSU. And "IF" the added heat from 2 additional sticks RAM significantly affects your case cooling, then your case cooling was totally inadequate to begin with.

 

[harm9963]

If you buy a 4090, without a proper case and PSU , asking for a lot more trouble .

 

[Taraquin]

There is one consistent advantage to using 2 sticks instead of 4. Generally speaking, the energy required to power a RAM stick is roughly the same regardless the capacity of the stick - assuming same architecture.

That is, 1 x 16GB DDR4 takes about the same amount of power to run as 1 x 32GB DDR4.

Likewise, 1 x 64GB DDR5 takes about the same amount of power to run as 1 x 32GB DDR5.

Therefore, running 4 x 16GB sticks of RAM will consume roughly twice the power as running 2 x 32GB sticks. This also means 2 sticks will generation about 1/2 as much heat as 4.

Having said all that, "IF" you are that close to maxing out your PSU that adding 2 more sticks matters, then you need a bigger PSU. And "IF" the added heat from 2 additional sticks RAM significantly affects your case cooling, then your case cooling was totally inadequate to begin with.

Not that it matters much, but many 16gb ddr4-sticks are dual rank with chips on both side. They use more power than single rank. On my setup when stresstesting I got my sr ram up to 5.5W while my DR uses 7.5W.

If you use a 4dimm mb and 2x16 you have an easier time cooling ram due to more space My current setup is dual rank on 2dimm mb. That is hot

 

[AleXXX666]

Practically 2 sticks is always better unless you have a HEDT platform with quad-channel support.

the only correct answer.

4 sticks of RAM had up to 10% better performance than 2 on 5600X. Gamers Nexus did a spot on this a while back.

10% perf for like 30+% in price... F*CK ya, GN

 

[lexluthermiester]

the only correct answer.

Ironically incorrect.

 

[AlB80]

10% perf for like 30+% in price...

+10% of total system performance for +30% of memory price. Good deal.

 

[Noci]

+10% of total system performance for +30% of memory price. Good deal.

Isn't it somehow rule of thumb that every uplift of performance translates in exponential rise of costs one way or another? There will be exceptions of course, but that's not the initial question in this thread.

basic answer im looking for....2x32 is better then 4x16 if theyre all 3600 c18?

how much difference is there in % between 2 vs 4 sticks?

Despite all the discussions, I think there is no right (or wrong) answer here, mainly because the varaible of costs was not clearly defined in this context (at least in the first question).
Gradually the discussion evolved with additional aspects to take in consideration, making it even more complex, with more then one possible right answer depending on which parameter was prevalent.

Best thing coming out of this thread is that hopefully more peeps starting to understand the basics of interaction between single/dual channel, single/dual rank and interleaving (at least I did ).
With a lot of infomation (and misinformation) out there, it is quite confusing to comprehend, especially if knowledgable peeps even go deeper in this complex matter .

I consider myself a noob in this field of technology, but I'm able to relativate and understand that the gains or losses with multiple options only have a marginal impact on performance, power usage and so on.
The only variable that shows significant fluctuations are: costs. So that parameter should be in the basis of a rephrased question, in relation to what one wants to know.

Feel free to disagree, but this is how I see it.

 

[AlB80]

Isn't it somehow rule of thumb that every uplift of performance translates in exponential rise of costs one way or another? There will be exceptions of course, but that's not the initial question in this thread.

+10% of system performance for ~+3% of total price. Good deal.
This shows how important it is for DDR4 to have at least 32 banks per channel (one module 2Rx8 or two modules 1Rx8).
fyi.
Every standard provides a frequency boost, to use a high frequency you need more banks for simultaneous memory requests, and every standard provides this (not true).
SDRAM 1R usually have 2 banks.
DDR 1R = 4 banks.
DDR2 1R = 8 banks.
DDR3 1R = 16 banks.
DDR4 1R = 16 banks. This sucks. But also there are 1Rx16 modules with only 8 banks.
DDR5 1R = 64 banks. Back on the rails.

 

[Noci]

+10% of system performance for ~+3% of total price. Good deal.

I doubt that an upgrade for an AM4 system from 32Gb RAM to 64Gb RAM will result in a 10% overall uplift of system performance (never noticed that myself when going from 32Gb to 64Gb).
But if you're right, it is a good deal.