[SOLVED] I recently bought new ram, is the latency good?

Luca_21

Reputable
Aug 1, 2017
68
3
4,535
0
I recently bought Vengeance DDR4 3600mhz "C19"
I put C19 in quotations because the amazon page says C19, but when i search the product tag it shows me a C18 version.

My question is, will this ram last me for the next few years? Is it low latency enough?
I ask this because another person told me "Aslong as ifs C14-16 itll last for the next few years while DDR5 is lowering in cost"

So id like to know if this ram kit is future proof for atleast 3~ years.
Thank you in advance!
 

DimkaTsv

Proper
Nov 7, 2021
156
23
115
15
Is it low latency enough?
Ngl, question sound kinda stupid, but i will try to answer why.

Short answer: You will not see any difference, but C18 is better than C19. And yes, it would last few years for sure, why it shouldn't. Not that CPU cann't take more than it supposed to (this one more relates to frequency), so chasing for better RAM can be not as productive as you think. After 3200MT C16 there is drop for increasing efficiency even for Ryzen's which are more dependant on memory than Intel

Long answer:
Difference between RAM can be extrapolated in 2 main roundups.
  1. Frequency - higher = better, lower = worse. But you may not be able to run more than 3600MT frequency at all even on new CPU, because of CPU I/O silicon quality or motherboard limitations. For older CPU it is even more harsh, as for most of them even 3200 can be a bit of struggle and 50/50 type of deal (like for Ryzen 1XXX and 2XXX).
  2. Latency - higher = worse, lower = better. BUT! All these latencies should be looked only in same frequency category. Because latency as number means "amount of cycles", which have no specific value. Value for 1T comes from memory frequency.
For example: 3600 MT memory with C18 and same, but with C16 and 3200 MT with C16 and C14 respectively.
First! We must know 1T value.
For 3600 MT memory it will be equal to 1/(3600/2) or 1/1800 of second. Why we dividing 3600 by 2? Because DDR already means Double Data Rate, which meaning is that manufacturer frequency is just double from actual frequency of RAM stick.
Anyways.
1/1800 = 0.0005(5) of secons = 0.5(5) ns.
1/1600 (for 3200 MT ones) = 0.000625 = 0.625 ns.
Then we take latencies in account
3600 C18 = CAS latency of 10.00 ns [С19 will be equal to 10.5(5) ns]
3600 C16 = CAS latency of 8.8(8) ns
3200 C16 = CAS latency of 10.00 ns
3200 C14 = CAS latency of 8.75 ns.

What that means on practice. Actually for you it will probably not mean much. Except part that C18 is better than C19. If you are asking that question then my answer is "You will most likely not notice that difference unless you try really hard on benchmarks, but not on real usage, because difference will be that miniscule".

If you saying that you looked for 3600 CL19 memory, it means that you already looked for more of a budget solutions. Because people who actually care have pretty expensive systems, or enthusiasts in memory OC. When people look for more performative memory (IF they actually think this will be worth for them), they look for 3600 CL16 or 3200 CL14, and better of all for memories when timings are X / X / X / 2X+N (N may be equal to 0 there), and not X / X+Z / X+Z / 2X+Z+N, because that will be sign of wanted by them Samsung B-die which is most effective atm chip for overclocking, lowering timings, and overall much better temperature for voltage curves.
But you aren't close to both categories, so... No any difference for you. And yes, it will last whatever long you want it to stay (until you decide to buy new PC with DDR5), no problems. Just better check that RAM is working correctly.

P.S. CAS latency is not only thing that affects performance. Secondary and thertiary timings may sometimes have even bigger effect on RAM overall effectiveness, and some manufacturers sometimes actually make them terrible from what i saw. And when you can see secondary timings and these depends mostly on chip manufacturer/technology/binning (so you can rarely see unnecessary things there) thertiary are much more annoying to deal with, because you cannot see them until you buy RAM and place it in. So... No one is safe.

Anyways. RAM is like rabbit hole, more you dig - deeper it goes, so you and i probably will be better to stop at this point, as answer was given.
 
Last edited:
Reactions: ROOT#

DimkaTsv

Proper
Nov 7, 2021
156
23
115
15
Is it low latency enough?
Ngl, question sound kinda stupid, but i will try to answer why.

Short answer: You will not see any difference, but C18 is better than C19. And yes, it would last few years for sure, why it shouldn't. Not that CPU cann't take more than it supposed to (this one more relates to frequency), so chasing for better RAM can be not as productive as you think. After 3200MT C16 there is drop for increasing efficiency even for Ryzen's which are more dependant on memory than Intel

Long answer:
Difference between RAM can be extrapolated in 2 main roundups.
  1. Frequency - higher = better, lower = worse. But you may not be able to run more than 3600MT frequency at all even on new CPU, because of CPU I/O silicon quality or motherboard limitations. For older CPU it is even more harsh, as for most of them even 3200 can be a bit of struggle and 50/50 type of deal (like for Ryzen 1XXX and 2XXX).
  2. Latency - higher = worse, lower = better. BUT! All these latencies should be looked only in same frequency category. Because latency as number means "amount of cycles", which have no specific value. Value for 1T comes from memory frequency.
For example: 3600 MT memory with C18 and same, but with C16 and 3200 MT with C16 and C14 respectively.
First! We must know 1T value.
For 3600 MT memory it will be equal to 1/(3600/2) or 1/1800 of second. Why we dividing 3600 by 2? Because DDR already means Double Data Rate, which meaning is that manufacturer frequency is just double from actual frequency of RAM stick.
Anyways.
1/1800 = 0.0005(5) of secons = 0.5(5) ns.
1/1600 (for 3200 MT ones) = 0.000625 = 0.625 ns.
Then we take latencies in account
3600 C18 = CAS latency of 10.00 ns [С19 will be equal to 10.5(5) ns]
3600 C16 = CAS latency of 8.8(8) ns
3200 C16 = CAS latency of 10.00 ns
3200 C14 = CAS latency of 8.75 ns.

What that means on practice. Actually for you it will probably not mean much. Except part that C18 is better than C19. If you are asking that question then my answer is "You will most likely not notice that difference unless you try really hard on benchmarks, but not on real usage, because difference will be that miniscule".

If you saying that you looked for 3600 CL19 memory, it means that you already looked for more of a budget solutions. Because people who actually care have pretty expensive systems, or enthusiasts in memory OC. When people look for more performative memory (IF they actually think this will be worth for them), they look for 3600 CL16 or 3200 CL14, and better of all for memories when timings are X / X / X / 2X+N (N may be equal to 0 there), and not X / X+Z / X+Z / 2X+Z+N, because that will be sign of wanted by them Samsung B-die which is most effective atm chip for overclocking, lowering timings, and overall much better temperature for voltage curves.
But you aren't close to both categories, so... No any difference for you. And yes, it will last whatever long you want it to stay (until you decide to buy new PC with DDR5), no problems. Just better check that RAM is working correctly.

P.S. CAS latency is not only thing that affects performance. Secondary and thertiary timings may sometimes have even bigger effect on RAM overall effectiveness, and some manufacturers sometimes actually make them terrible from what i saw. And when you can see secondary timings and these depends mostly on chip manufacturer/technology/binning (so you can rarely see unnecessary things there) thertiary are much more annoying to deal with, because you cannot see them until you buy RAM and place it in. So... No one is safe.

Anyways. RAM is like rabbit hole, more you dig - deeper it goes, so you and i probably will be better to stop at this point, as answer was given.
 
Last edited:
Reactions: ROOT#

Luca_21

Reputable
Aug 1, 2017
68
3
4,535
0
Ngl, question sound kinda stupid, but i will try to answer why.

Short answer: You will not see any difference, but C18 is better than C19. And yes, it would last few years for sure, why it shouldn't. Not that CPU cann't take more than it supposed to (this one more relates to frequency), so chasing for better RAM can be not as productive as you think. After 3200MT C16 there is drop for increasing efficiency even for Ryzen's which are more dependant on memory than Intel

Long answer:
Difference between RAM can be extrapolated in 2 main roundups.
  1. Frequency - higher = better, lower = worse. But you may not be able to run more than 3600MT frequency at all even on new CPU, because of CPU I/O silicon quality or motherboard limitations. For older CPU it is even more harsh, as for most of them even 3200 can be a bit of struggle and 50/50 type of deal (like for Ryzen 1XXX and 2XXX).
  2. Latency - higher = worse, lower = better. BUT! All these latencies should be looked only in same frequency category. Because latency as number means "amount of cycles", which have no specific value. Value for 1T comes from memory frequency.
For example: 3600 MT memory with C18 and same, but with C16 and 3200 MT with C16 and C14 respectively.
First! We must know 1T value.
For 3600 MT memory it will be equal to 1/(3600/2) or 1/1800 of second. Why we dividing 3600 by 2? Because DDR already means Double Data Rate, which meaning is that manufacturer frequency is just double from actual frequency of RAM stick.
Anyways.
1/1800 = 0.0005(5) of secons = 0.5(5) ns.
1/1600 (for 3200 MT ones) = 0.000625 = 0.625 ns.
Then we take latencies in account
3600 C18 = CAS latency of 10.00 ns [С19 will be equal to 10.5(5) ns]
3600 C16 = CAS latency of 8.8(8) ns
3200 C16 = CAS latency of 10.00 ns
3200 C14 = CAS latency of 8.75 ns.

What that means on practice. Actually for you it will probably not mean much. Except part that C18 is better than C19. If you are asking that question then my answer is "You will most likely not notice that difference unless you try really hard on benchmarks, but not on real usage, because difference will be that miniscule".

If you saying that you looked for 3600 CL19 memory, it means that you already looked for more of a budget solutions. Because people who actually care have pretty expensive systems, or enthusiasts in memory OC. When people look for more performative memory (IF they actually think this will be worth for them), they look for 3600 CL16 or 3200 CL14, and better of all for memories when timings are X / X / X / 2X+N (N may be equal to 0 there), and not X / X+Z / X+Z / 2X+Z+N, because that will be sign of wanted by them Samsung B-die which is most effective atm chip for overclocking, lowering timings, and overall much better temperature for voltage curves.
But you aren't close to both categories, so... No any difference for you. And yes, it will last whatever long you want it to stay (until you decide to buy new PC with DDR5), no problems. Just better check that RAM is working correctly.

P.S. CAS latency is not only thing that affects performance. Secondary and thertiary timings may sometimes have even bigger effect on RAM overall effectiveness, and some manufacturers sometimes actually make them terrible from what i saw. And when you can see secondary timings and these depends mostly on chip manufacturer/technology/binning (so you can rarely see unnecessary things there) thertiary are much more annoying to deal with, because you cannot see them until you buy RAM and place it in. So... No one is safe.

Anyways. RAM is like rabbit hole, more you dig - deeper it goes, so you and i probably will be better to stop at this point, as answer was given.
Thank you for the detailed answer, i think what i should get out of it is.. its complicated, and i should just focus on finding something reliable and affordable within my budget instead of really worrying about latency all that much.
 

ASK THE COMMUNITY