DDR4 32Gb Kit 3200 CL16 Vs. 16Gb Kit 3600 CL15

[PIRATA!]

Hi to everyone.

I am interested in how to compare these two kind of solutions as I'd like to know how the total quantity of RAM GB weighs in giving to one of the two kits the 1st place.

 

Motherboard: ASUS ROG STRIX Z370-G GAMING (WI-FI AC)

CPU: i5 8600K

Usage (Work): Heavy Google Chrome browsing (up to n.2 Chrome instances opened together with 30/40 tabs opened total) + Photoshop (both Chrome & Photoshop opened all together)

Also consider Gaming, but for that the computer will be equipped with NVidia 1060 6Gb

 

These are the two kits:

 

Kit 1: 32Gb (2x16Gb)

Name: G.Skill Trident Z 3200 MHz CL16 (16-16-16-36)
Model Number: F4-3200C16D-32GTZA
IC: 8Gb Samsung B-Die
Rank: Dual
Sided: Double
Source: Jedec Info (http://i.imgur.com/z4ReAoo.png)

 

Kit 2: 16Gb (2x8Gb)

Name: G.Skill Trident Z 3600 MHz CL15 (15-15-15-35)
Model Number: F4-3600C15D-16GTZ
IC: 8Gb Samsung B-Die
Rank: Single
Sided: Single
Source: Jedec Info (https://abload.de/img/3600c14thaiphoonr4q8d.jpg)

 

I'd like to compare them at their XMP setting, no extra overclock purpose, so only comparison between their XMP values (3200 16-16-16-36 Vs. 3600 15-15-15-35)

 

Thank you all.

 

 

[Skiiwee29]

Well one is a 32gb kit and the other 16gb. Based on what you have listed, the 32gb kit truthfully will be the better fit.

 

But, Based on the timings and speed, the 3600mhz kit will be faster as it has faster timings and speed, though you likely will not actually notice it in real world scenarios and usage. 

[NelizMastr]

CAS latency and clock speed determine the actual latency. When clock speed goes up and the CAS stays the same, latency goes down thus better performance. When CAS rises with the clock speed, latency stays the same or worsens. Based on that, it's easy to conclude that the 3600 kit at CL15 is much better than the 3200 CL16 kit.

 

As mentioned by @Skiiwee29 this is only theoretical, so whether you'll notice it depends on the workload. Yours isn't that memory speed dependent.

[PIRATA!]

7 minutes ago, NelizMastr said:

CAS latency and clock speed determine the actual latency. When clock speed goes up and the CAS stays the same, latency goes down thus better performance. When CAS rises with the clock speed, latency stays the same or worsens. Based on that, it's easy to conclude that the 3600 kit at CL15 is much better than the 3200 CL16 kit.

 

As mentioned by @Skiiwee29 this is only theoretical, so whether you'll notice it depends on the workload. Yours isn't that memory speed dependent.

Hi and thanks for the reply.

Yes, I know that the 3600 CL15 is better because has higher freq. and lower CAS, but the 3200 are 32Gb and the 3600 16Gb. Does this value of quantity of GB do weight in such "best" determination? When are the GB amount important such to weight enough so to make difference?

 

And about my workload....why am I not memory speed dependent? Isn't the quality and quantity of ram important for my work usage described above?

 

Thank you very much.

[grimreeper132]

go for capacity, although the extra speed might help a bit, the extra capacity will help more, it'll mean you can open 8 chrome tabs, rather than 4 with 16GB  but yea I would get the 32GB because of the extra capacity.

[Sakkura]

22 minutes ago, PIRATA! said:

Hi to everyone.

I am interested in how to compare these two kind of solutions as I'd like to know how the total quantity of RAM GB weighs in giving to one of the two kits the 1st place.

The quantity only matters once you're using it.

 

In this case, kit 2 will be faster until you're using more than 16GB of RAM; then kit 1 will be faster.

[PIRATA!]

1 minute ago, grimreeper132 said:

go for capacity, although the extra speed might help a bit, the extra capacity will help more, it'll mean you can open 8 chrome tabs, rather than 4 with 16GB  but yea I would get the 32GB because of the extra capacity.

 

Just now, Sakkura said:

The quantity only matters once you're using it.

 

In this case, kit 2 will be faster until you're using more than 16GB of RAM; then kit 1 will be faster.

Thank you both...but when will I saturate the 16GB of ram?? Is there even a theoretical way to understand it??

I immagine that in games, both kit will result in no big differences (but still would like to understand which will prevail in gaming).

So my main heavy use will be the Work usage: heavy Google Chrome browsing (up to n.2 Chrome instances opened together using two different Google accounts with 30/40 tabs opened total between both Chrome instances) + Photoshop (both Chrome & Photoshop opened all together).

 

Thank you.

 

[Sakkura]

3 minutes ago, PIRATA! said:

 

Thank you both...but when will I saturate the 16GB of ram?? Is there even a theoretical way to understand it??

I immagine that in games, both kit will result in no big differences (but still would like to understand which will prevail in gaming).

So my main heavy use will be the Work usage: heavy Google Chrome browsing (up to n.2 Chrome instances opened together using two different Google accounts with 30/40 tabs opened total between both Chrome instances) + Photoshop (both Chrome & Photoshop opened all together).

 

Thank you.

 

A game in itself will barely push you to 8GB, let alone 16GB. To benefit from the 32GB kit while gaming, you'd have to leave a lot of Chrome tabs open in the background, and Photoshop as well maybe.

 

It's hard to predict exactly how much RAM you'll be using, the best way to measure is simply to try it out. But your use case doesn't immediately strike me as something that would require 32GB of RAM.

[Jurrunio]

I'd get more RAM than faster RAM. It will be painful if you run out of it (which is likely, consider it's Chrome, the RAM eater)

[PIRATA!]

6 minutes ago, Sakkura said:

A game in itself will barely push you to 8GB, let alone 16GB. To benefit from the 32GB kit while gaming, you'd have to leave a lot of Chrome tabs open in the background, and Photoshop as well maybe.

 

It's hard to predict exactly how much RAM you'll be using, the best way to measure is simply to try it out. But your use case doesn't immediately strike me as something that would require 32GB of RAM.

 

6 minutes ago, Jurrunio said:

I'd get more RAM than faster RAM. It will be painful if you run out of it (which is likely, consider it's Chrome, the RAM eater)

Big dilemma... I'd like to end up with 32Gb as well....but having 3600 CL15 was such a big deal for me..

 

What about getting the 16Gb kit and in future upgrade putting next to it another same 16Gb kit when prices drops? Will they be compatible together? the 16Gb tik is in my motherboard's QVL..

[Jurrunio]

12 minutes ago, PIRATA! said:

 

Big dilemma... I'd like to end up with 32Gb as well....but having 3600 CL15 was such a big deal for me..

 

What about getting the 16Gb kit and in future upgrade putting next to it another same 16Gb kit when prices drops? Will they be compatible together? the 16Gb tik is in my motherboard's QVL..

nah, Coffee Lake isnt that memory intensive. If you're on Ryzen then maybe there's an advantage for the 3600MHz C15 kit that makes it preferred over to 3200MHz kit, but you're not.

 

Also, do you really need an Asus mATX mobo? You've already bought it?

[PIRATA!]

9 minutes ago, Jurrunio said:

nah, Coffee Lake isnt that memory intensive. If you're on Ryzen then maybe there's an advantage for the 3600MHz C15 kit that makes it preferred over to 3200MHz kit, but you're not.

 

Also, do you really need an Asus mATX mobo? You've already bought it?

Yes, I've read that Ryzen cpu's are more ram speed/CL dependent then Intel's, but I really still would like to understand if the capacity is important for responsiveness as well rather then only the amount of stuff that can be done together.

In case NOT, even if I have intel, Will I might end up with a significant slow computer in my Work usage described above?

If NOT that "significantly slow" but still "slower"..in a matter of "%", how slow would the 3200 16-16-16-36 solution be Vs. the 3600 15-15-15-35 one??

 

And yes, I already have the motherboard as I had the chance to get it with a M.2 sata 500Gb drive in combo with the ASUS Cash Back giveaway of last month at a great discount, so I rather have paid it like some base Intel motherboard.

 

 

[Jurrunio]

43 minutes ago, PIRATA! said:

Yes, I've read that Ryzen cpu's are more ram speed/CL dependent then Intel's, but I really still would like to understand if the capacity is important for responsiveness as well rather then only the amount of stuff that can be done together.

In case NOT, even if I have intel, Will I might end up with a significant slow computer in my Work usage described above?

If NOT that "significantly slow" but still "slower"..in a matter of "%", how slow would the 3200 16-16-16-36 solution be Vs. the 3600 15-15-15-35 one??

With the slower RAM you're looking at most 5-10% speed reduction in memory intensive task (running synthetic benchmarks, especially in Physics score). In 99.99% of other situations, the difference is negligible and you wont feel that without looking at numbers.

 

If you do run out of RAM, you get a stuttering mess, which is horrible.

46 minutes ago, PIRATA! said:

And yes, I already have the motherboard as I had the chance to get it with a M.2 sata 500Gb drive in combo with the ASUS Cash Back giveaway of last month at a great discount, so I rather have paid it like some base Intel motherboard.

Good, because that mobo really isnt impressive. Its components are not worth having the ROG name.

[PIRATA!]

1 minute ago, Jurrunio said:

With the slower RAM you're looking at most 5-10% speed reduction in memory intensive task (running synthetic benchmarks, especially in Physics score). In 99.99% of other situations, the difference is negligible and you wont feel that without looking at numbers.

 

If you do run out of RAM, you get a stuttering mess, which is horrible.

 

Good, because that mobo really isnt impressive. Its components are not worth having the ROG name.

So only in benchmarks, as I really need only "real-life" differences. In games not at all?

 

About he motherboard, I have never read nothing about what you say. Can you point me somewhere to look at?

 

Thank you.

[Biggerboot]

Personally, I always opt for more ram, unless the discussion is 2133/2400 vs 3200.

[Jurrunio]

18 minutes ago, PIRATA! said:

So only in benchmarks, as I really need only "real-life" differences. In games not at all?

 

About he motherboard, I have never read nothing about what you say. Can you point me somewhere to look at?

 

Thank you.

PUBG can also show some performance benefits in terms of minimum frame rates (it's optimized like poop), but that's a tiny benefit compared to what you will get when you run out of memory. Besides, you can overclock the 3200MHz kit further yourself, not like it's locked or something. Samsung B-die should do better than 3200MHz easily with decent timings once you raise the voltage to something like 1.5V (even 1.8V is still safe for DDR4, so 1.5v is easy peasy)

 

It's built like a Prime board, not a traditional ROG board.

[AngryBeaver]

Since you are on an Intel chipset I would go with the higher capacity. The speed isn't as important for intel as it would have been AMD. On top of that the difference would only be seen in synthetic benchmarks and not noticeable in real world usage.

[PIRATA!]

On 7/31/2018 at 5:06 PM, Jurrunio said:

PUBG can also show some performance benefits in terms of minimum frame rates (it's optimized like poop), but that's a tiny benefit compared to what you will get when you run out of memory. Besides, you can overclock the 3200MHz kit further yourself, not like it's locked or something. Samsung B-die should do better than 3200MHz easily with decent timings once you raise the voltage to something like 1.5V (even 1.8V is still safe for DDR4, so 1.5v is easy peasy)

 

It's built like a Prime board, not a traditional ROG board.

Wowowowowow!!!! That review it's amazing!!!!

Well....now my question is: enabling XMP and setting Ram to their XMP values (so no extra overclock rather than the XMP values)...will it trigger the heat problem explained in the video?? ...or there is no risk if setting any Ram to their XMP freq./latency??

I'm interested in having 16GB or 32GB of Ram at XMP 3200 values or XMP 3600. Not interested in overclock of cpu, neither Ram rather then their XMP.

Will it be better to stick with a 3200 kit, or could I push it up and so get a 3600 kit?

 

Thank you.

[Jurrunio]

7 minutes ago, PIRATA! said:

Wowowowowow!!!! That review it's amazing!!!!

Well....now my question is: enabling XMP and setting Ram to their XMP values (so no extra overclock rather than the XMP values)...will it trigger the heat problem explained in the video?? ...or there is no risk if setting any Ram to their XMP freq./latency??

I'm interested in having 16GB or 32GB of Ram at XMP 3200 values or XMP 3600. Not interested in overclock of cpu, neither Ram rather then their XMP.

Will it be better to stick with a 3200 kit, or could I push it up and so get a 3600 kit?

 

Thank you.

No, overclocking memory doesnt produce any noticeable amount of heat.

 

just stick with 3200 kit. the performance gain from 3200 to 3600 is soooo small, especially when the CPU's running at stock.

though in terms of leaving CPU at stock clocks, might as well get the i7-8700 instead for hyperthreading. https://www.pugetsystems.com/labs/articles/Photoshop-CC-2018-CPU-Performance-AMD-Ryzen-2-vs-Intel-8th-Gen-1136/

[Srnewbee]

On ‎2018‎년 ‎7‎월 ‎31‎일 at 10:05 PM, PIRATA! said:

 

Thank you both...but when will I saturate the 16GB of ram?? Is there even a theoretical way to understand it??

I immagine that in games, both kit will result in no big differences (but still would like to understand which will prevail in gaming).

So my main heavy use will be the Work usage: heavy Google Chrome browsing (up to n.2 Chrome instances opened together using two different Google accounts with 30/40 tabs opened total between both Chrome instances) + Photoshop (both Chrome & Photoshop opened all together).

 

Thank you.

 

go with 32GB. If you have Chrome and Photoshop opened together, you are gonna need a lot more than 16gb. Especially with large photoshop files, its gonna really tax your memory.

[PIRATA!]

1 hour ago, Jurrunio said:

No, overclocking memory doesnt produce any noticeable amount of heat.

 

just stick with 3200 kit. the performance gain from 3200 to 3600 is soooo small, especially when the CPU's running at stock.

though in terms of leaving CPU at stock clocks, might as well get the i7-8700 instead for hyperthreading. https://www.pugetsystems.com/labs/articles/Photoshop-CC-2018-CPU-Performance-AMD-Ryzen-2-vs-Intel-8th-Gen-1136/

Oh no!!

 

Intel® Core™ i5-8600K Processor

Intel® Hyper-Threading Technology ^‡  No

 

I haven't thought to check on that before getting the 8600K!! Well...I'll comfort my self with the fact that I bought it used at less than $200, and that the guy that has it before had used it for less then 10h as it was on its second machine, plus it was able to run it stable at stock under-volting it a bit. I hope it'll be a good cpu with me as well!  

 

About the lack of hyper-threading...will it be such a big loss? With hyper-threading would I have noticed such big differences?? As I see here, it's not that bad:

Intel Core i7-8700K vs Core i5-8600K
Userbenchmark Effective CPU Speed: 107% vs 102%  
Intel Core i7-8700K  - 8th / 1118  
Intel Core i5-8600K  - 11th / 1118

 

....and about the ram freq. you say that it's better to save money on them and best invest money in ram Gb?? So better keep some money in my pockets not investing in the extra 400MHz between the 3600 and 3200, and put it in the extra 16Gb of total amount of Ram??

And about the timings?? If getting a 3200, even if 32Gb, high timings are more relevant then the once in the 3600 Ram kits.

 

For fast calculus of best performance, I do these:

 

1. MHz / CL = X (higher is better) - Bandwidth (is it? need confirmation!)
2. ( CL / MHz ) * 1000 = Y (lower is better) - Latency

...and the fastest are the timings, so less is the differences in performance between a 3200 kit vs a 3600 one (assuming to have same timings in both 3200 and 3600 kits), because the ratio regarding the latency will converge.

 

So regarding using a 32Gb kit, I think I'll have to go for the 3200 because of a matter of total price plus motherboard compatibility:  all 32Gb 3600MHz kits compatible with my motherboard are 4x8Gb and have also terrible timings....plus they cost too much $$$!!!

 

So regarding considering 32Gb 3200 MHz kits, as I have found both CL 15-15-15-35 and CL 16-16-16-36 kits....for understanding if its worth going for the fastest ram or not... >>>> how much should cost in real $$$ that "1" CL of difference?? Really speaking....giving it a matter of money value...I shall consider in going for the CL15 if that "1" CL costs no more then [$ .... ,00] (fill the dots  ).

[Srnewbee]

the difference between  CL 15-15-15-35 and CL 16-16-16-36 is so negligible that it is virtually non-existent. In fact, CL 16 could potentially overclock better. So to answer your question, the answer is $5 dollars at max.

[Jurrunio]

43 minutes ago, PIRATA! said:

I haven't thought to check on that before getting the 8600K!! Well...I'll comfort my self with the fact that I bought it used at less than $200, and that the guy that has it before had used it for less then 10h as it was on its second machine, plus it was able to run it stable at stock under-volting it a bit. I hope it'll be a good cpu with me as well!  

 

About the lack of hyper-threading...will it be such a big loss? With hyper-threading would I have noticed such big differences?? As I see here, it's not that bad:

Intel Core i7-8700K vs Core i5-8600K
Userbenchmark Effective CPU Speed: 107% vs 102%  
Intel Core i7-8700K  - 8th / 1118  
Intel Core i5-8600K  - 11th / 1118

depends on the workload you're doing, 8600k (stock) can perform the same to 10% worse than 8700. Difference in games with a 1060 6gb is virtually none.  For that price, I'd say you're good.

 

45 minutes ago, PIRATA! said:

....and about the ram freq. you say that it's better to save money on them and best invest money in ram Gb?? So better keep some money in my pockets not investing in the extra 400MHz between the 3600 and 3200, and put it in the extra 16Gb of total amount of Ram??

And about the timings?? If getting a 3200, even if 32Gb, high timings are more relevant then the once in the 3600 Ram kits.

 

For fast calculus of best performance, I do these:

 

1. MHz / CL = X (higher is better) - Bandwidth (is it? need confirmation!)
2. ( CL / MHz ) * 1000 = Y (lower is better) - Latency

...and the fastest are the timings, so less is the differences in performance between a 3200 kit vs a 3600 one (assuming to have same timings in both 3200 and 3600 kits), because the ratio regarding the latency will converge.

 

So regarding using a 32Gb kit, I think I'll have to go for the 3200 because of a matter of total price plus motherboard compatibility:  all 32Gb 3600MHz kits compatible with my motherboard are 4x8Gb and have also terrible timings....plus they cost too much $$$!!!

 

So regarding considering 32Gb 3200 MHz kits, as I have found both CL 15-15-15-35 and CL 16-16-16-36 kits....for understanding if its worth going for the fastest ram or not... >>>> how much should cost in real $$$ that "1" CL of difference?? Really speaking....giving it a matter of money value...I shall consider in going for the CL15 if that "1" CL costs no more then [$ .... ,00] (fill the dots  ).

If you ask me I would go as far as 2400MHz 32GB memory, since memory frequency doesnt help performance in any noticeable way with a 1060 6gb or photoshop. Memory transfer rate and latency isnt as important as you seemed to think on Coffee Lake.

 

1. More yes than no

2. More yes than no

Why not just conclude that into MHz/CL, the higher the better?

 

Just in terms of bandwidth, the 3600MHz CL15 kit is worth the price. However, actual performance difference isnt worth that cost. If you create RAM disk with it then maybe worth buying, but otherwise a big NO.

 

 

[PIRATA!]

On 8/4/2018 at 6:15 AM, Jurrunio said:

depends on the workload you're doing, 8600k (stock) can perform the same to 10% worse than 8700. Difference in games with a 1060 6gb is virtually none.  For that price, I'd say you're good.

 

If you ask me I would go as far as 2400MHz 32GB memory, since memory frequency doesnt help performance in any noticeable way with a 1060 6gb or photoshop. Memory transfer rate and latency isnt as important as you seemed to think on Coffee Lake.

 

1. More yes than no

2. More yes than no

Why not just conclude that into MHz/CL, the higher the better?

 

Just in terms of bandwidth, the 3600MHz CL15 kit is worth the price. However, actual performance difference isnt worth that cost. If you create RAM disk with it then maybe worth buying, but otherwise a big NO.

 

 

For the cpu, well I'm happy that I went for the 8600 as I have these 3 good reasons that plays for me: 1) I've paid it not too much, 2) "only" 10% difference from 8700, and 3) I'll use at least a 1060 6gb NVidia graphic card. The bad thing (probably..) is that if just one of these 3 things have not happened, I would have probably ended up in not doing the right thing, as all of them practically weigh a lot in the whole  equation.

 

For the ram, there is no way to find a 32gb kit 3600 with all CL15 that is compatible with my motherboard. Not even all CL16, and not 2x16Gb but all 4x8Gb. Plus to that, all of the 32gb 3600 are soooo expensive.

 

So the two roads are to get: 1) the good 16Gb kit 2x8Gb 3600 all CL15 and start a journey in following the market for price drops on another kit like that for placing it in compo so to have then 32Gb in my system......or 2) go directly for a 32Gb kit, probably 3200, (CL timings decision would come subsequently) that in a matter of price is practically very similar to the 16gb kit 3600 CL15.

 

I have to consider then what you've told me about the freq., as probably even having a 2400, or lets say a 2666 (as its my stock Jedec max freq. compatible with my cpu) I'll end up with a system where I'd probably not notice that much in performance differences if having a 3200 kit.

 

In my particular case, as I have gone for a "K" cpu, I'd like to use it with " > 2600 " rams...and as I was considering in spending money in a 16gb kit 3600 CL15, well...probably I will have chance to get a 32Gb 3200 for the same price.

 

As I'm checking now, strangely the 32gb 3200 all CL16 compatible for my motherboard have much more higher prices than the all CL15 brothers!!!

I've also found a 32gb 3200 all CL14 kit that has interesting price!!

Making a list in order by price, I have 1) 16Gb 3600 all CL15 kit that is about $310, then 2) with $30 more I can get a 32Gb 3200 all CL15 kit, and 3) with extra $50 more ($80 from the 3600) I can get a 32Gb 3200 all CL14 kit!!!

 

Maybe the 3) is not that worth...but speaking of those equations I've exposes before, I can notice a difference of about half of a ns in latency between the 3600 all Cl15 and the 3200 all CL14, and about only 1ns in latency between the 3600 all CL and the 3200 all CL 15.

Speaking about bandwidth, the 3600 all CL15 is about 5/6% bigger than the 3200 all CL14, and 11/12% bigger than the 3200 all CL15.

Remember still that the 3600 is only 16Gb, while the others are 32Gb kits.

 

So...which kit would you chose?? The n.2) 3200 all CL15, or what else? 

 

Thanks!

[Jurrunio]

7 hours ago, PIRATA! said:

So...which kit would you chose?? The n.2) 3200 all CL15, or what else? 

the 3200MHz CL15 32gb kit. If you're lucky you can lower the timings to that of the CL14 kit, without paying for it.