CL16-16-16-36 vs CL16-19-19-39 worth 55% higher price? (B-Die or maybe E-Die)

[Shou Miko]

I been looking 2x16GB 3600MHz kit because my two mixed Geil Dragon kits won't properly be able to run 3600MHz running together and I do not want to go down to 16GB ram since I can fill that up with no issue even not gaming.

These are my 2 memory kits running at 3000MHz CL16-17-17-35
GWW416GB3000C15DC this is SK Hynix M-die CL15-17-17-35 (Read with Thaiphoon Burner)
GWW416GB4000C19DC this is Samsung B-die CL19-25-25-45 (Read with Thaiphoon Burner)

I could properly tune them to do CL15-17-17-35 using manually timings but I am not sure how much of a boost that will give me over what they are currently running plus a decent 3600MHz kit will properly out run them any day of the week.

So on my searching I looked at the G.Skill's Trident Z Neo F4-3600C16D-32GTZN and I found that these are properly the best 3600MHz kit on the market using Samsung B-die chips at CL16-16-16-36 for about £242 but that's like 55% higher than G.Skill's F4-360016D-32GTZNC which is SK Hynix CJR running CL16-19-19-39 and cost about £156.

Yes I know that Samsung B-die is great and properly the best ram chips for overclocking but when I just want to run XMP all day long and a Infinity Fabric 1:1 on my AMD Ryzen 3900X shouldn't kit with SK Hynix CJR be fine?

Because in all fairness I would properly not feel the difference in daily use and gaming and spending the 55% more when I won't overclock that would hurt my pocket more then what I would properly gain.


I know owning a Samsung B-die kit would be awesome but when I would decide to sell them on most users won't know the advantage of B-die anyway and won't pay the extra cost even they can do more with manually turning.

With these things in mind I look into Micron E-die for a good 3600MHz kit like the BL2K16G36C16U4BL runs CL16-18-18-38 at about £191 which is cheaper then the F4-3600C16D-32GTZN by £51 but I want them in white and that bump up the price with about 31% to £250 for the BL2K16G36C16U4WL which is the same kit just white

I know for daily and gaming I wouldn't notice the difference between running Samsung B-die 16-16-16-36 or SK Hynix CJR 16-19-19-39 or Micron E-die 16-18-18-38 but in the benchmarks I sometimes do I will but I don't think 31% or 55% more will be great value for how many procent difference there is?

At last can anyone tell me why white coating cost so much on the ram heatsinks vs black because it's the BL2K16G36C16U4BL and BL2K16G36C16U4WL is the same PCB, RGB, E-die chips and heatsink design only difference is the colour coating black vs white

Sorry about this rambling I hope it's to figure out since english ain't my native language and it's shown a lot in different scenarios.

 

[Tatty_Two]

I play around with my memory constantly and TBH I can't "feel" and difference between 4000mhz CL16-16-16-32 than I can if I run the sticks at stock which is 4133mhz CL19-17-17-34, obviously there are some differences in latency etc in synthetic benchmarks but in everyday use I cannot tell the difference personally.

If I were you I would possibly go for the GSkill Trident Z's 32GB CL17 for less than £150 and it's likely with a small voltage bump you would get your timings anyway. Obviously it's upto you but if you are happy with what you already have I probably wouldn't bother.

 

[Shou Miko]

@Tatty_One are you talking about the F4-3600C17D-32GTZR with CL17-19-19-39?

They are black only and the price difference from the F4-360016D-32GTZNC is only about £15.10 and then they are black and white so that I can live with.

I have a feeling that the F4-3600C17D-32GTZR is also SK Hynix CJR I am not sure but with those timings.

 

[Crashman]

I quantified things like this a while ago (I don't know which TPU article compares)

Ryzen Above: Best Memory Settings for AMD's 3000 CPUs, Tested

Is Ryzen 3000 optimized for DDR4-3200? What if we want more? We examine everything from frequency to rank count to nail down optimal settings.

www.tomshardware.com

Basically you don't get that much from the lower timings, and the biggest benefit you can find is from having two ranks of memory per channel (a double sided or two single sided DIMMs per channel).

 

[tabascosauz]

@Tatty_One are you talking about the F4-3600C17D-32GTZR with CL17-19-19-39?

They are black only and the price difference from the F4-360016D-32GTZNC is only about £15.10 and then they are black and white so that I can live with.

I have a feeling that the F4-3600C17D-32GTZR is also SK Hynix CJR I am not sure but with those timings.

I have the RGB 3600 C17 32GB kit. It's DJR. Generally, newer kits at 17-19-19 and 16-19-19 3600 will be DJR as it is essentially a newer revision of CJR. They take the same suggestions on timings, just that DJR is sometimes reputed for lower tRFC and lower VDRAM at the same speed than CJR. If the price difference is negligible, take the C16 kit for less of a silicon binning lottery.

If you're doing just as you say, running XMP, you will not notice a performance difference to B-die, period. I'd say you stand to gain about as much from learning to reach 3733 or 3800 @ 16-19-19 and learning about your subtimings, than you would moving from 3600/16 CJR XMP to 3600/16 B-die XMP.

On a side note, I've done a fair bit of testing with my Trident kit, and with default XMP timings and auto subtimings, 16-19-19, 17-19-19 or 18-19-19 all perform similarly in time-limited benchmarks such as Membench. But when you really put in the work on subtimings, that's what makes all the difference.

Yes, B-die is also capable of hitting 3600/14 or 3800/14, but it's all dependent on the kit you receive, and XMP-rated 3600/16 B-die is pretty mundane and run-of-the-mill for B-die.

As for Rev.E, actual performance is similar to CJR, but it overclocks much better on frequency. Not that you'll be doing so with a 3900X and your stated use case.

 

[Ripcord]

NO

 

[Shou Miko]

@tabascosauz still in the learing phase of AMD Ryzen OC but so far the max all-core stable is 4250MHz at 1.3V doing 4300MHz all-core was a no go even for cinebench Windows just locked up running like 15mins doing nothing.

What is for sure is that AMD Ryzen like 3600MHz ram because the Inifinity Fabric like 1:1 to get the best performance with no penalty.

 

[Tatty_Two]

@Tatty_One are you talking about the F4-3600C17D-32GTZR with CL17-19-19-39?

They are black only and the price difference from the F4-360016D-32GTZNC is only about £15.10 and then they are black and white so that I can live with.

I have a feeling that the F4-3600C17D-32GTZR is also SK Hynix CJR I am not sure but with those timings.

I meant these ……………………….

G.Skill Trident Z RGB 32GB (2x 16GB) 3600MHz DDR4 - F4-3600C17D-32GTZR

G.Skill branded powered by Intel XMP profile support, ample 32GB DDR4 stick kit, low voltage requirement (1.35V) RAM and with CL17 latency timing, Mfg Code: F4-3600C17D-32GTZR

www.cclonline.com

 

[tabascosauz]

@tabascosauz still in the learing phase of AMD Ryzen OC but so far the max all-core stable is 4250MHz at 1.3V doing 4300MHz all-core was a no go even for cinebench Windows just locked up running like 15mins doing nothing.

What is for sure is that AMD Ryzen like 3600MHz ram because the Inifinity Fabric like 1:1 to get the best performance with no penalty.

For Rev.E overclocking I meant MT/s on the RAM, not frequency on the CPU. Overclocking a 3900X too much on all core torpedoes its efficiency, anyways.

I played with a fair bit of 3733 and 3800 on my 3600/17 Trident kit, but at the end of the day, GPU temperature and frequency as well as GPU drivers matter much, much more in games and daily usage. I don't know about "lucky", because I can only hit CJR-range tRFC while the purported DJR ranges are out of reach for my kit, but I've been running 3600 16-19-19-37 for almost a year now with no problems. I use about 1.37V to be completely safe on stability, but 1.35V is probably just fine. One thing is for sure - 1T with no GearDown is impossible to make completely stable at 3600/16 without 1.45V+. Leave Geardown on, or just set 2T command rate without Geardown, like XMP will probably be.

CJR and DJR are otherwise identical other than DJR maybe/kinda/sometimes being a little better on some timings or freq. Treat them the same.



3733 was a walk in the park for my kit at 1.38V, but the problem was the IF on my 3700X logged occasional hardware interconnect errors that didn't affect usage after I dialed in VDDP/VDDG, but I wasn't comfortable with the hardware errors. 3800 was doable for all benchmarks at 1.44V, but again, didn't want to push the already-uncomfortable IF even harder.

Technically, you could luck out on getting the Trident 3600/17 instead of the Trident 3600/16 kits, but the latter are inevitably of a slightly higher bin. Safer to go with the CL16 if price difference is negligible.

 

[EarthDog]

To answer the tbread title... nope. Not by a Longshot. MAYBE if you're trying to get the most out of any integrated graphics (which you dont have in the first place).

 

[moproblems99]

Mine do 3800 16-19-19 so I stopped there and moved on.

 

[ThrashZone]

Hi,
C19 at 3600 is not a very good set
But on amd 3600 is about it's max so not a lot of oc'ing left

Anything without tight timings 16-16-16/ 15-15-15/.... I'd make sure it was at least on the qvl of the amd board or the set said optimized for ryzen/ amd seeing it is pretty picky about memory.

 

[Shou Miko]

Hi,
C19 at 3600 is not a very good set
But on amd 3600 is about it's max so not a lot of oc'ing left

Anything without tight timings 16-16-16/ 15-15-15/.... I'd make sure it was at least on the qvl of the amd board or the set said optimized for ryzen/ amd seeing it is pretty picky about memory.

Current sticks ain't an many QVL lists and they do fine and weirdly I am unable at this time to get the QVL list of my board because MSI's weird website just keeps going on in a infinite loop spend over 10mins trying to get it in different browsers.

 

[ThrashZone]

Current sticks ain't an many QVL lists and they do fine and weirdly I am unable at this time to get the QVL list of my board because MSI's weird website just keeps going on in a infinite loop spend over 10mins trying to get it in different browsers.

Hi,
G-skill has it's own qvl list just look up the memory on their website see if any amd boards are listed doesn't have to be your board just one would be good enough.

 

[oxrufiioxo]

Kinda depends really..... WIth a 2080 ti or faster at 1080p I see a decent difference with my 3900X going from 3600 CL16-16-16-32 vs 3800 CL16-16-16-32 but at 1440p there is no gain beyond 3200 CL 14-14-14-28.... All other sub timings identical. For programs there isn't any difference in anything beyond margin of error. I see you have a 5700XT and a 4k monitor so honestly any memory kit likely above 3000 is gonna be the same I actually highly doubt you will gain anything over running your current kits at 3000 with tighten timings at 4k.

I do have a kit that does 3800 CL13-13-13-28 but haven't tested it on my AMD system don't even know if my 3900X memory controller can handle it so don't know how that scales but its a $500 USD kit for 32GB so it makes zero sense unless you really love tinkering.

Not sure what you're actually doing with this PC but gaming at 4k this seems like a bigger numbers must mean more better scenario.

Now if you plan on getting a 3090 or Big Navi isn't garbage like the last couple supposedly high end AMD cards (vega 64/Radeon 7) Then get the lower latency stuff......

I could properly tune them to do CL15-17-17-35 using manually timings but I am not sure how much of a boost that will give me over what they are currently running plus a decent 3600MHz kit will properly out run them any day of the week.

And no a Properly tunned 3000 kit even at 1080p/medium can match or exceed a 3600 non tunned CL16 in most games

Testing 3rd-Gen Ryzen DDR4 Memory Performance and Scaling

When we reviewed Ryzen's latest iteration we briefly checked out different DDR4 memory speeds but now that things have settled we were put on a mission to...

www.techspot.com

 

[lexluthermiester]

CL16-16-16-36 vs CL16-19-19-39 worth 55% higher price? (B-Die or maybe E-Die)

Short answer, no. Not worth it unless you absolutely need the tighter timings.

 

[Caring1]

Current sticks ain't an many QVL lists and they do fine and weirdly I am unable at this time to get the QVL list of my board because MSI's weird website just keeps going on in a infinite loop spend over 10mins trying to get it in different browsers.

Yep, MSI want people to register on their site before they will divulge any info now, scumbags.

 

[9-5 sid]

hi i got my my gskill 3600 16-19-19-39 run at 3800 and its fast enough..maybe not a must ,to buy the expencive mem

 

[freeagent]

The Higher M/T's are more important then timings imo. Running 1900/1900 14-14-14-34 feels the same as 1900/1900 16-16-16-36. They bench fairly close too. You see it a little in certain benchmarks but.. I dunno.. this feels wrong to say but you would probably be just better off running 3600-3800 C16 with a dual rank setup.. Tighter timings do get you a little more, but I don't know if I would say its worth it. If you have OCD with your computer it is I suppose. It just doesn't feel as rewarding as it should.. but it could just be my motherboard.. I am considering trying another.

 

[oxrufiioxo]

The Higher M/T's are more important then timings imo. Running 1900/1900 14-14-14-34 feels the same as 1900/1900 16-16-16-36. They bench fairly close too. You see it a little in certain benchmarks but.. I dunno.. this feels wrong to say but you would probably be just better off running 3600-3800 C16 with a dual rank setup.. Tighter timings do get you a little more, but I don't know if I would say its worth it. If you have OCD with your computer it is I suppose. It just doesn't feel as rewarding as it should.. but it could just be my motherboard.. I am considering trying another.

I agree, you would likely need a 3090 and be gaming at 720p to notice a difference...... With my 2080 ti and 4x8GB there is no difference going from 3200 CL 14-14-14-28 to 3800 CL 16-16-16-32 even though the latency goes from 68ns to 63ns in the aida64

I would still recommend people grab a 3200 CL14 kit though as it can tweak a lot better than the CJR/DJR/micron based 3600 kits

 

[freeagent]

I agree, you would likely need a 3090 and be gaming at 720p to notice a difference...... With my 2080 ti and 4x8GB there is no difference going from 3200 CL 14-14-14-28 to 3800 CL 16-16-16-32 even though the latency goes from 68ns to 63ns in the aida64

I would still recommend people grab a 3200 CL14 kit though as it can tweak a lot better than the CJR/DJR/micron based 3600 kits

Yup 3200 C14 is what I bought. These are the best sticks of ram that I have ever owned. They are so flexible, I have never played with memory so in depth as I have with these. I've had them at 3600 C12 up to 4400 C16. Crazy stuff in my eyes lol.

Not going to lie but I kind of wish I had an Intel setup to exploit these further..

 

[toilet pepper]

Does it really have to be gskill? I got the 32Gb Patriot Viper 3600Mhz 17-19-19-39-68 1.35V. It might not look as good as the Gskill but it is considerably cheaper and is a Bdie.