• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Help to decide - 3200 CL16 / 3600 CL16?

I highly doubt 3200c16 can do 3600c16 but here you can post results
What would you like as proof?

Here's CPU-Z:
Capture.JPG


The modules are CMK16GX4M2Z3200C16.

I used 1usmus's calculator to get the timings. Haven't had any issues, either on my Asus Prime X470, or on my current MSI X570 Tomahawk.
 
I dont understand you..
The design of the royals is fine and he support with my mobo on QVL so why not? The Taichi have short list of RAMs..
QVL is pretty irrelevant.
 
If you plan on running XMP the difference between 3600cl16 and 3200cl16 is minor at best, if you tweak sometimes a 3200cl16 kit can have better dies that overclock far better, for instance if you get Micron rev E in the 3200 kit while the 3600 kit has Samsung D.
 
Hi,
Yep old thread

Corsair is hit or miss
Mostly miss for myself but corsair sticks have a VER# on each stick if they match they are from the same bin so will work better than if they don't match.

I highly doubt 3200c16 can do 3600c16 but here you can post results

Also fill in your system spec's

What would you like as proof?

Here's CPU-Z:
View attachment 242842

The modules are CMK16GX4M2Z3200C16.

I used 1usmus's calculator to get the timings. Haven't had any issues, either on my Asus Prime X470, or on my current MSI X570 Tomahawk.
Hi,
You removed what to do from my post :laugh:

Here it is again run aida64 and post it on the thread I linked to.
You can get cheap aida64 full versions off ebay for like 3.us just look for guarantied to activate I've used them before.
 
In my experience if you have a IMC or board limitation with 4 dimms, its probably only going to restrict the clock of the ram, not the latency, so if e.g. your 3600 CL 16 ram has to run at 3200mhz, you can probably crank the timing down to the equivalent latency. On my 8600k I used to run my 3200CL14 ram at 3000CL12. (which was actually superior latency, equal would have been CL13).

I would expect 3600CL 16 ram to be better binned than 3200CL 16, so assuming price isnt an issue here, go for the 3600.
 
Hi,
You removed what to do from my post :laugh:
I did no such thing, your post is still there.

@chrcoluk Thanks, yes I think I'll go for the Corsair 3600CL16; that way it can run at more or less stock alongside the OC'd 3200. From what I could see in various tests, its the speed not the CL that counts, especially with 3600 being equal to the IF for my current R53600.
 
@Yearofthegoat
for ram calc: lots stuff isnt what you actually want to use up mto even "completely" wrong/made up (voltage etc),
at least to most with proper understanding of am4/boards/ram/bios

'working" doesnt mean anything, but to make it short:
can you pass +1500% in something like TM5 or HCI (running on windows) with it?
 
@Yearofthegoat
for ram calc: lots stuff isnt what you actually want to use up mto even "completely" wrong/made up (voltage etc),
at least to most with proper understanding of am4/boards/ram/bios

'working" doesnt mean anything, but to make it short:
can you pass +1500% in something like TM5 or HCI (running on windows) with it?
Hi,
You're wasting your time.
 
What would you like as proof?

Here's CPU-Z:
View attachment 242842

The modules are CMK16GX4M2Z3200C16.

I used 1usmus's calculator to get the timings. Haven't had any issues, either on my Asus Prime X470, or on my current MSI X570 Tomahawk.

Some 3200 CL16 kits can OC to DDR4-3600. Most can't. You got lucky, but not everyone who buys a 3200 CL16 kit will.
If you plan on running XMP the difference between 3600cl16 and 3200cl16 is minor at best, if you tweak sometimes a 3200cl16 kit can have better dies that overclock far better, for instance if you get Micron rev E in the 3200 kit while the 3600 kit has Samsung D.
It can, but it probably won't. The overwhelming majority of the time, you'll get better chips on a 3600 CL16 kit than on a 3200 CL16 kit. 3200 CL16 is more likely to be Samsung C-die or Micron 8Gb Rev B, which is usually worse than D-die.
 
Seems to be a lot of grumpiness on here. All I wanted to know was whether to go ahead with 3600CL16, given that I have 3200 modules successfully OC'd to 3600.

What does "or ram calc: lots stuff isnt what you actually want to use up mto even "completely" wrong/made up (voltage etc),
at least to most with proper understanding of am4/boards/ram/bios
" mean? I'm serious. Are you saying the Ryzen calculator was made up? It worked for me at least. What is "mto"?

FWIW I checked, and I tested the RAM with MemTest at boot time, as well as HCI MemTest 7.0 (2019) under Windows 10. I remember leaving it running for around a whole day. No errors at the time. To this day, no errors, no WHEA etc. It's fine.
 
fact that you take the calc info as (gold) standard, tells me you want to read up a little more on it.

1 out of 4 times the settings for 3600 xmp are not stable, or at least way too much aggressive/lose, where they shouldnt have to be (as in not producing any gains).
e.g. i have not seen anything that needs 1.45v to run at 3600, unless its something like my b-die doing C14 and some of the lowest sec/tert i have found that are stable,
but those are SOLD as 3600/C16 1.45v kits...

usually you can expect to lower cl by 1 or 2, or you can ago up in speed by 1 or 2 steps, depending on what die is used.
as such, stating (any) kit sold as a certain speed, can reach something better, is blatantly false.
(completely ignoring tha kit does NOT equal what die is used, this can vary on the same kits).

running a prog a day doesnt mean anything, you want test (HCI) to pass 1200-1500%, so even areas used (prior) by win have been covered.
never seen ppl that know there stuff, claim "stable", below those numbers.

unless of course, there isnt any data on your drives that you care to keep, than go ahead..
 
Last edited:
@Fry178 Thanks for posting back. I've since upgraded my CPU to a 5900X, which is working fine with some PBO/Curve Optimizer.

The XMP ('A-XMP' in the MSI BIOS) setting in my board doesn't work with these sticks. I had to set the primary timings myself and after contacting Corsair they recommended upping the voltage to 1.37v. That worried me a bit and backed it off to 1.36v which works fine at stock 3200CL16 (and at 3600CL16).

Re. the calculator, yes, I didn't have any reason to doubt it and it was initially just a way to see if I could get the RAM to run at 3600 - just 3600 at all, not aiming for CL16 or anything other than stability. The reason for that was again advice that Ryzen liked RAM at 1:1 with the infinity fabric. I bought the 3200 RAM because at first that was touted as the 'sweet spot' for an R5 3600, IIRC. But of course the R53600 runs at 3600 (1800), so maybe that was just a mistake.

I haven't bought more RAM yet, as I'm not totally convinced that I really need it. I have 16GB but I read lots of recommendations for 32GB RAM - for audio sample libraries in Cubase/Reaper etc. If I can just add two more 8GB sticks and get them to play nicely (i.e. 3600CL16) with the existing pair, then great, but if it's too difficult to set up I'll wait until I can buy a 'proper' set of 32GB 3600 (or maybe 3800) and sell the original 16GB pair.

Seeing as I've changed the CPU I'll go ahead and re-test the current 16GB with MemTest (HCI). I did a shorter test which was fine but, as you say, it doesn't test all of the RAM since Windows is using some.

I have a program called MemTestHelper which sets the values for HCI (I believe anyway). So I'll go ahead and see what happens. The RAM timings haven't been changed and thus far seem stable (they are a bit looser than those in 1usmus's calculator). The MemBench section in the calculator - after an 'Easy' 6GB test - reports latency of 62.1ns, which I believe is okay?
 
Save the headache and get a Crucial 3600 32GB kit
 
I'll leave this here.
The specs of the kit and so on are within the tabs of CPU-Z and yes, this was a sub I did for validation of it.

This is the exact set I bought and ran for this: G.SKILL TridentZ Series 16GB (2 x 8GB) 288-Pin DDR4 SDRAM DDR4 3600 (PC4 28800) Intel Z170 / Z270 / Z370 / X299 Desktop Memory Model F4-3600C16D-16GTZSW - Newegg.com

RAM entry.jpeg



BTW the board you use matters, sometimes to a good extent.
Must say my X470 Tachi Ultimate is NOT a RAM champ by any means, has difficulty making it to 3600 with these or any others I pop in but the X570 ACE has no issues about it as shown.
 
@Fry178 Thanks for posting back. I've since upgraded my CPU to a 5900X, which is working fine with some PBO/Curve Optimizer.

The XMP ('A-XMP' in the MSI BIOS) setting in my board doesn't work with these sticks. I had to set the primary timings myself and after contacting Corsair they recommended upping the voltage to 1.37v. That worried me a bit and backed it off to 1.36v which works fine at stock 3200CL16 (and at 3600CL16).

Re. the calculator, yes, I didn't have any reason to doubt it and it was initially just a way to see if I could get the RAM to run at 3600 - just 3600 at all, not aiming for CL16 or anything other than stability. The reason for that was again advice that Ryzen liked RAM at 1:1 with the infinity fabric. I bought the 3200 RAM because at first that was touted as the 'sweet spot' for an R5 3600, IIRC. But of course the R53600 runs at 3600 (1800), so maybe that was just a mistake.

I haven't bought more RAM yet, as I'm not totally convinced that I really need it. I have 16GB but I read lots of recommendations for 32GB RAM - for audio sample libraries in Cubase/Reaper etc. If I can just add two more 8GB sticks and get them to play nicely (i.e. 3600CL16) with the existing pair, then great, but if it's too difficult to set up I'll wait until I can buy a 'proper' set of 32GB 3600 (or maybe 3800) and sell the original 16GB pair.

Seeing as I've changed the CPU I'll go ahead and re-test the current 16GB with MemTest (HCI). I did a shorter test which was fine but, as you say, it doesn't test all of the RAM since Windows is using some.

I have a program called MemTestHelper which sets the values for HCI (I believe anyway). So I'll go ahead and see what happens. The RAM timings haven't been changed and thus far seem stable (they are a bit looser than those in 1usmus's calculator). The MemBench section in the calculator - after an 'Easy' 6GB test - reports latency of 62.1ns, which I believe is okay?
What is the ram dies on the new and old kit according to thaiphoon burner? I would also say eidaraman1s advice of Crucial is best, they are garanteed rev E or B which overclocks like champs and can tune most timings a lot. Easy 15% boost in many games when cpu bound vs running xmp timings.
 
@Yearofthegoat
dont use amp. voltages get boosted, and other stuff might be wacky.
always do it manually..

@Taraquin
with amd i still prefer timings to go as low as it goes and still gains in perf (3600),
had no trouble dropping by 2 on main timings with A0, havent gotten time to test more.
 
I'll leave this here.
The specs of the kit and so on are within the tabs of CPU-Z and yes, this was a sub I did for validation of it.

This is the exact set I bought and ran for this: G.SKILL TridentZ Series 16GB (2 x 8GB) 288-Pin DDR4 SDRAM DDR4 3600 (PC4 28800) Intel Z170 / Z270 / Z370 / X299 Desktop Memory Model F4-3600C16D-16GTZSW - Newegg.com

View attachment 244476


BTW the board you use matters, sometimes to a good extent.
Must say my X470 Tachi Ultimate is NOT a RAM champ by any means, has difficulty making it to 3600 with these or any others I pop in but the X570 ACE has no issues about it as shown.

Nice ram that. mine is this B die 32gb kit-
https://www.overclockers.co.uk/team...00c16-3600mhz-dual-channel-kit-my-002-8p.html
 
What is the ram dies on the new and old kit according to thaiphoon burner? I would also say eidaraman1s advice of Crucial is best, they are garanteed rev E or B which overclocks like champs and can tune most timings a lot. Easy 15% boost in many games when cpu bound vs running xmp timings.
the dies are Hynix CJR revision A0 on the current 2x8GB kit. I don't do any gaming, just looking for stable 1800 clock for Ryzen CPU (although I'm not certain it's worth all the effort). If I buy new RAM, I'll buy 3600, no need for overclocking (apart from the manufacturer rated spec).
@Yearofthegoat
dont use amp. voltages get boosted, and other stuff might be wacky.
always do it manually..

@Taraquin
with amd i still prefer timings to go as low as it goes and still gains in perf (3600),
had no trouble dropping by 2 on main timings with A0, havent gotten time to test more.
I don't use A-XMP (it wasn't stable at stock 3200, never mind 3600). All the settings were manual from 1usmus' calculator, and for memory only apart from IF clock set 1:1. I have, however, now got PBO dialled in, plus Curve Optimizer for the 5900X. (Runs really nicely. Cinebench 23 reports 1633 single core @4.93GHz and 23099 multicore @4.58GHz. Temps max 76C).

I ran HCI memory test up to 1600% yesterday with no errors. I read, though, that it's not a great test as it runs under Windows and can't ever test all the RAM (is that actually correct?). So I also ran MemTest 86 off a bootable CD. Again no errors.

I think I'll get the Corsair 3600CL16 2x8GB pair. If they work with my OCd 3200 pair, then great. If they don't - but work on their own of course - then I guess I'll sell the 3200s and just get another 2x8GB 3600CL16 pair.
 
If you plan on running XMP the difference between 3600cl16 and 3200cl16 is minor at best, if you tweak sometimes a 3200cl16 kit can have better dies that overclock far better, for instance if you get Micron rev E in the 3200 kit while the 3600 kit has Samsung D.
I agree my Crucial Ballistix 3200 CL16 does 3600 CL14 easily with Rev. E
C9BJZ 3600 CL14 Stable 2 2 12 6.jpg
 
If you care about overclocking, get the 3600 kit. Its higher rated factory speed might mean it's better quality too.

Though personally, I wouldn't bother, as RAM overclocking only gives you a few unnoticeable percents of extra performance at best.
 
the dies are Hynix CJR revision A0 on the current 2x8GB kit. I don't do any gaming, just looking for stable 1800 clock for Ryzen CPU (although I'm not certain it's worth all the effort). If I buy new RAM, I'll buy 3600, no need for overclocking (apart from the manufacturer rated spec).

I don't use A-XMP (it wasn't stable at stock 3200, never mind 3600). All the settings were manual from 1usmus' calculator, and for memory only apart from IF clock set 1:1. I have, however, now got PBO dialled in, plus Curve Optimizer for the 5900X. (Runs really nicely. Cinebench 23 reports 1633 single core @4.93GHz and 23099 multicore @4.58GHz. Temps max 76C).

I ran HCI memory test up to 1600% yesterday with no errors. I read, though, that it's not a great test as it runs under Windows and can't ever test all the RAM (is that actually correct?). So I also ran MemTest 86 off a bootable CD. Again no errors.

I think I'll get the Corsair 3600CL16 2x8GB pair. If they work with my OCd 3200 pair, then great. If they don't - but work on their own of course - then I guess I'll sell the 3200s and just get another 2x8GB 3600CL16 pair.
Corsair with AM4 is just asking for trouble, esp if getting a set from the Vengeance line.
Too many issues and troubleshooting posts about it to even count between a set of Vengeance sticks and Ryzen, it's best to grab a set of G.Skills instead and don't worry about it.
 
Hi,
Buying a 2x8gb kit these days is not wise.
 
Corsair with AM4 is just asking for trouble, esp if getting a set from the Vengeance line.
Too many issues and troubleshooting posts about it to even count between a set of Vengeance sticks and Ryzen, it's best to grab a set of G.Skills instead and don't worry about it.
Yeah, I've read that a lot. Mine were fine at stock after a bump in the voltage and avoiding A-XMP. I bought a set that were specifically labelled as 'optimized for Ryzen', don't know if that's a factor, but they've been okay since, although at 3600 they needed tweaking.

Maybe I'm reading your post wrong, but would a set of G.Skills work with the existing Corsairs? Everything I've read says not to mix brands. I'm trying to save ca. £100 by adding RAM rather than buying a new set of 32GB, so that would mean more Corsairs.
 
Back
Top