RAM: 4x SR or 2x DR for Ryzen 3000? - Overclock.net - An Overclocking Community

Forum Jump: 

RAM: 4x SR or 2x DR for Ryzen 3000?

 
Thread Tools
post #1 of 10 (permalink) Old 07-16-2019, 03:10 PM - Thread Starter
Bonestorm Professional
 
thrillhaus's Avatar
 
Join Date: Jun 2011
Location: Toronto, ON
Posts: 228
Rep: 10 (Unique: 9)
RAM: 4x SR or 2x DR for Ryzen 3000?

A while a go there was that article where ASRock recommended only using 2 RAM slots with Ryzen 3000 (https://www.tomshardware.com/news/am...ock,39756.html). Does populating all 4 slots with single rank sticks actually reduce the max clock?
thrillhaus is offline  
Sponsored Links
Advertisement
 
post #2 of 10 (permalink) Old 07-16-2019, 04:39 PM
New to Overclock.net
 
Boxman's Avatar
 
Join Date: Jun 2007
Posts: 88
Rep: 3 (Unique: 3)
Only if you intend to go way beyond 3733MHz. Before that, it shouldn't matter that much.

FWIW, I'm on 4x DR and it's working fine* up to 3733C16 - T-Topology board though, although I'm not sure how much that matters at this frequency. If you want to go over that, get a Daisy Chain motherboard and populate only 2 slots.

*no actual stability testing was done
Boxman is offline  
post #3 of 10 (permalink) Old 07-17-2019, 01:21 AM
New to Overclock.net
 
os2wiz's Avatar
 
Join Date: Sep 2012
Location: Brooklyn, New York
Posts: 2,393
Rep: 61 (Unique: 40)
Quote: Originally Posted by Boxman View Post
Only if you intend to go way beyond 3733MHz. Before that, it shouldn't matter that much.

FWIW, I'm on 4x DR and it's working fine* up to 3733C16 - T-Topology board though, although I'm not sure how much that matters at this frequency. If you want to go over that, get a Daisy Chain motherboard and populate only 2 slots.

*no actual stability testing was done
A topology board will excel at using 4 dimm slots, it makes ALL the difference. Most boards use daisy chain technology and using all 4 dimm slots will definitely impoact performance, no matter what their QVL guide says. If using all 4 dimm slots the 4 dimms must be machine matched by the memory manufacturer. They must be bought as a set of 4 dimms. No matter if they all have the same listed speed and timings. It goes beyond that. I have an MSI x570 MEG Ace motherboard. I wanted to run 32GB memory, so I purchaseda G.Skill kit of two 16GB dimms that are dual rank and rated at DDR4 4000 CL19-19-19-39. I have optimized performance at DDR4 3600 CL16-16-17-17-36-59. Going above that may be stable , but produces lower benchmark scores on a consistent basis.

Technology is NOT a liberator . Whom technology serves is a class question.
Technology wielded by the capitalist class benefits only the exploiters.
Technology wielded by the working class will serve ALL Humanity.
Ryzen 9 3900X at 4.2 GHZ, MSI X570 MEG ACE motherboard, AlphaCool Eisbaer 360 cpu AIO cooler
32GB G.Skill TridentZ DDR4 4000mhz CL-19-19-19-39 1.35v Samsung 970 EVO m.2 NVMe 1TB
AMD Radeon VII gpu
os2wiz is offline  
Sponsored Links
Advertisement
 
post #4 of 10 (permalink) Old 07-17-2019, 07:32 AM
New to Overclock.net
 
Boxman's Avatar
 
Join Date: Jun 2007
Posts: 88
Rep: 3 (Unique: 3)
Quote: Originally Posted by os2wiz View Post
A topology board will excel at using 4 dimm slots, it makes ALL the difference. Most boards use daisy chain technology and using all 4 dimm slots will definitely impoact performance, no matter what their QVL guide says.
Afaik it only makes a real difference above 4000MHz-ish. That's what I gathered from the likes of Buildzoid and der8auer. Unless you can point me to another source?

Quote:
If using all 4 dimm slots the 4 dimms must be machine matched by the memory manufacturer. They must be bought as a set of 4 dimms. No matter if they all have the same listed speed and timings. It goes beyond that.
Can you elaborate with a detailed explanation on why? Every set will be binned to at least get the rated speeds/timings, above that it's lottery of how good the chips actually are. There is no physical difference from kit-to-kit. Afaik, they don't perform voodoo-magic to 'match kits' apart from just slapping 4 sticks into 1 blister case. I'm open to being convinced otherwise though, if you can give the references as well.
Boxman is offline  
post #5 of 10 (permalink) Old 07-17-2019, 03:35 PM
New to Overclock.net
 
os2wiz's Avatar
 
Join Date: Sep 2012
Location: Brooklyn, New York
Posts: 2,393
Rep: 61 (Unique: 40)
Quote: Originally Posted by Boxman View Post
Afaik it only makes a real difference above 4000MHz-ish. That's what I gathered from the likes of Buildzoid and der8auer. Unless you can point me to another source?



Can you elaborate with a detailed explanation on why? Every set will be binned to at least get the rated speeds/timings, above that it's lottery of how good the chips actually are. There is no physical difference from kit-to-kit. Afaik, they don't perform voodoo-magic to 'match kits' apart from just slapping 4 sticks into 1 blister case. I'm open to being convinced otherwise though, if you can give the references as well.
iF you get 2 memory kits each with 2 dimms of the same sku, they will not always work well together. That is because they are not perfectly binned to match each other. I know this from my own experience . I have 2 kits G.Skill DDR4- 4266 CL19-19-19-39. Same sku each with Samsung b-die chips according to Thaiphoon. I use each kit solo and overclock the memory in bios. One kit at 3600mhz will allow me 16-16-16-36 the other 16-17-17-37 with stability testing confirmed. So they show intrinsic differences although both binned to boot in a particular motherboard with a particular model cpu with XMP. Now if I put these 4 dimms to occupy all 4 dimm slots on my daisy-chained board it will define the maximum speed and settings by the dimms with the slowest timings, the least common denominator. This is based on logic as well as empirical observation. So no it is NOT voodoo but there are still differences between dimms within the same model and sku number. I do not worship Gurus. I will show respect to those who have a level of expertise but they are not gods. When you stop thinking and behave solely by rote thinking you are underselling your capabilities. Always follow your gut instinct and accumulated knowledge and realize there is somebody else out there who may know a bit more whether recognized or acclaimed or just quietly acquiring knowledge by experience.

Technology is NOT a liberator . Whom technology serves is a class question.
Technology wielded by the capitalist class benefits only the exploiters.
Technology wielded by the working class will serve ALL Humanity.
Ryzen 9 3900X at 4.2 GHZ, MSI X570 MEG ACE motherboard, AlphaCool Eisbaer 360 cpu AIO cooler
32GB G.Skill TridentZ DDR4 4000mhz CL-19-19-19-39 1.35v Samsung 970 EVO m.2 NVMe 1TB
AMD Radeon VII gpu

Last edited by os2wiz; 07-17-2019 at 04:04 PM.
os2wiz is offline  
post #6 of 10 (permalink) Old 07-17-2019, 03:40 PM
Realist
 
chakku's Avatar
 
Join Date: Mar 2016
Posts: 360
Rep: 3 (Unique: 3)
4x8 SR kits seem to be rated for higher speeds on QVL lists and from what I've seen people have achieved better timings/lower voltages at 3733 with these than me, even on daisychain boards. I have 2x16 DR B-Die and 3733CL16/3600CL14 is the best I've got 100% stable. (20K%+ in RAMTest)

PC
(19 items)
CPU
AMD Ryzen R7 3700X
Motherboard
ASUS ROG Crosshair VII Hero (Wi-Fi)
GPU
ASUS ROG Strix GTX 1080 Ti OC
RAM
G.Skill Ripjaws V F4-3200C14D-32GVK
Hard Drive
Samsung 970 Pro 512GB NVMe
Hard Drive
Samsung 860 EVO 1TB
Hard Drive
STOREMI (Intel 730 Series 240GB + Hitachi Ultrastar 7K4000 4TB)
Power Supply
EVGA SuperNOVA 850 P2
Cooling
Noctua NH D15 SE-AM4
Case
Phanteks Enthoo Evolv X Satin Black
Operating System
Windows 10 Pro 1809 64-bit
Monitor
Dell P2715Q
Monitor
Dell P2715Q
Keyboard
Topre Realforce 87U 55g Uniform (Norbaforce/BT-500 Bluetooth)
Mouse
Logitech G Pro Wireless
Mousepad
Artisan Hayate Kou
Audio
Benchmark DAC2 HGC
Audio
JBL LSR305
Audio
Sennheiser HD6XX
▲ hide details ▲
chakku is offline  
post #7 of 10 (permalink) Old 07-17-2019, 04:31 PM
New to Overclock.net
 
Boxman's Avatar
 
Join Date: Jun 2007
Posts: 88
Rep: 3 (Unique: 3)
Quote: Originally Posted by os2wiz View Post
iF you get 2 memory kits each with 2 dimms of the same sku, they will not always work well together. That is because they are not perfectly binned to match each other. I know this from my own experience . I have 2 kits G.Skill DDR4- 4266 CL19-19-19-39. Same sku each with Samsung b-die chips according to Thaiphoon. I use each kit solo and overclock the memory in bios. One kit at 3600mhz will allow me 16-16-16-36 the other 16-17-17-37 with stability testing confirmed. So they show intrinsic differences although both binned to boot in a particular motherboard with a particular model cpu with XMP. Now if I put these 4 dimms to occupy all 4 dimm slots on my daisy-chained board it will define the maximum speed and settings by the dimms with the slowest timings, the least common denominator. This is based on logic as well as empirical observation. So no it is NOT voodoo but there are still differences between dimms within the same model and sku number. I do not worship Gurus. I will show respect to those who have a level of expertise but they are not gods. When you stop thinking and behave solely by rote thinking you are underselling your capabilities. Always follow your gut instinct and accumulated knowledge and realize there is somebody else out there who may know a bit more whether recognized or acclaimed or just quietly acquiring knowledge by experience.
Quote:
If using all 4 dimm slots the 4 dimms must be machine matched by the memory manufacturer. They must be bought as a set of 4 dimms.
I'm just picking your brain because I was puzzled by your remark on why you should never combine two kits. See, by saying "they will not always work well together", you're implying that there will be inherent issues with running two separate kits together. But that's not the case, of course - that's what I meant with the 'voodoo' remark.

In reality, if 4-dimm kits are as carefully binned as you claim, all you're achieving is that your sticks all OC roughly the same. BUT! That might as well be worse than the 2 separate kits, there's no guarantee your 4-dimm kit is a better bin than the 2-dimm kits. It's still simply a lottery. Indeed the logic is simple. This 4-stick kit might as well only do the 16-17-17-37 timings, and you'd never have known a different kit could've been faster. At the same time, 2 separate kits where one does 16-16-16-36 and the other does 16-17-17-37, will be limited by the 16-17-17-37 kit but as a whole perform exactly the same as the 16-17-17-37 4-stick kit. And you'd be none the wiser, the result is identical. Or the 2 separate kits might both, by sheer luck, be able to do 16-16-16-36, and outperform the 4-stick kit when put together. You can't know. It's a lottery.

But there's no inherent reason not to put two kits together. And nothing states that a 4-dimm kit will always be better binned than two 2-dimm kits - might as well be worse.. From a technological standpoint, two separate kits will work together completely fine.

On another note, have you ever separately tested the dimms from a 4-stick kit to see if one was faster than the other, or did you assume they would be 100% identical in performance anyway?
Boxman is offline  
post #8 of 10 (permalink) Old 07-17-2019, 05:02 PM
New to Overclock.net
 
os2wiz's Avatar
 
Join Date: Sep 2012
Location: Brooklyn, New York
Posts: 2,393
Rep: 61 (Unique: 40)
Quote: Originally Posted by Boxman View Post
I'm just picking your brain because I was puzzled by your remark on why you should never combine two kits. See, by saying "they will not always work well together", you're implying that there will be inherent issues with running two separate kits together. But that's not the case, of course - that's what I meant with the 'voodoo' remark.

In reality, if 4-dimm kits are as carefully binned as you claim, all you're achieving is that your sticks all OC roughly the same. BUT! That might as well be worse than the 2 separate kits, there's no guarantee your 4-dimm kit is a better bin than the 2-dimm kits. It's still simply a lottery. Indeed the logic is simple. This 4-stick kit might as well only do the 16-17-17-37 timings, and you'd never have known a different kit could've been faster. At the same time, 2 separate kits where one does 16-16-16-36 and the other does 16-17-17-37, will be limited by the 16-17-17-37 kit but as a whole perform exactly the same as the 16-17-17-37 4-stick kit. And you'd be none the wiser, the result is identical. Or the 2 separate kits might both, by sheer luck, be able to do 16-16-16-36, and outperform the 4-stick kit when put together. You can't know. It's a lottery.

But there's no inherent reason not to put two kits together. And nothing states that a 4-dimm kit will always be better binned than two 2-dimm kits - might as well be worse.. From a technological standpoint, two separate kits will work together completely fine.

On another note, have you ever separately tested the dimms from a 4-stick kit to see if one was faster than the other, or did you assume they would be 100% identical in performance anyway?
No because it is impossible to get a 4 dimm kit above 4000mhz. I would rather use two 16GB dual rank dimms, I believe the performance is better with dual rank dimms and timings can be tighter . Now you can NOT get a dual rank kit for 32GB above 400mhz either. But I will much rather take my chances on the dual rank than the 4 dimm kit. That is my comfort level. I see no reason with Ryzen 3000 to run memory faster than 3600mhz. You get reduced performance even at 3733mhz compared to 3600mh. I do not try to do things for bragging rights alone. If it gives me better performance that is what gives me satisfaction. Otherwise let it be damned. So all those seeing their memory boot at 4500 or 4600mhz great. It is of zero use and in many cases it is NOT stable. Realize that derbauer and others claiming stability at that level have cherry picked cpu's, cherry picked memory and cherry picked motherboards. The odds we could accomplish the same are next to none and for what end? May be when zen3 comes along the infinity fabric can handle better than 3733mhz before the ratio collapsing from 2:1 to 1:1 destroys performance. Until then it is a geekish exercise with no practical use.

Technology is NOT a liberator . Whom technology serves is a class question.
Technology wielded by the capitalist class benefits only the exploiters.
Technology wielded by the working class will serve ALL Humanity.
Ryzen 9 3900X at 4.2 GHZ, MSI X570 MEG ACE motherboard, AlphaCool Eisbaer 360 cpu AIO cooler
32GB G.Skill TridentZ DDR4 4000mhz CL-19-19-19-39 1.35v Samsung 970 EVO m.2 NVMe 1TB
AMD Radeon VII gpu
os2wiz is offline  
post #9 of 10 (permalink) Old 07-17-2019, 05:16 PM
New to Overclock.net
 
Boxman's Avatar
 
Join Date: Jun 2007
Posts: 88
Rep: 3 (Unique: 3)
I'm afraild you've lost me You stated to never mix two identical sku 2-dimm kits but instead one must always get a uniquely packed 4-dimm kit or things 'will not work properly', but now you seem to have gone on a tangent.

I'm just asking, if you have ever verified whether all dimms in a 4-dimm kit actually perform exactly the same when tested separately, or if you're just assuming this to be true?
Boxman is offline  
post #10 of 10 (permalink) Old 07-17-2019, 06:05 PM
New to Overclock.net
 
AllenG's Avatar
 
Join Date: Mar 2009
Location: OC, CA
Posts: 566
Rep: 25 (Unique: 22)
How ya liking OS/2 running on that 3900x?

CPU
i7 990x @ 4.69GHz
Motherboard
Asus Rampage III Extreme
GPU
EVGA GTX780 FTW ACX 2.0 3GB
RAM
12 GB Samsung DDR3L @ 2100Mhz
Hard Drive
Samsung 840 Pro 256gb Raid0 On LSI 9207
Hard Drive
Samsung 840 Pro 512gb
Hard Drive
PSN across 10Gb Fiber
Optical Drive
Sony Optiarc DVDRW
Power Supply
Corsair 850HX
Cooling
DTek Fuzion V2 CPU Block w/ Quad Insert
Cooling
Laing D5 pump {2MT}
Cooling
2 x Black Ice Pro 240 Radiators {2MT}
Cooling
EK Full Cover RIIIE Nickel/Acetal Block
Case
Lian Li PC-A10B
Operating System
Windows 8.1 Enterprise X64
Monitor
Monoprice 27in WQHD IPS
Keyboard
Das Keyboard Model S
Audio
Creative X-Fi Fatal1ty Platinum
Other
LSI 9207-8i SAS6G Raid Controller
Other
Intel XF SR 10Gb Fiber Adapter
CPU
Pentium G3258 @ 4.5GHZ
Motherboard
MSI PC Mate Z97
GPU
EVGA GTX 970 SC ACX2.0 4GB
RAM
8GB Mushkin Blackline DDR3 2133
Hard Drive
256 GB Plextor M6 SSD
Hard Drive
WD Black 640GB Raid0
Optical Drive
Sony Optiarc DVDRW
Power Supply
Seasonic S12II 620W
Cooling
Corsair H50
Case
Old Antec
Operating System
Windows 10 Enterprise x64
Monitor
Changhong 42in 1080p TV
Keyboard
HP Wireless Keyboard & Thumb tracker
Mouse
HP Wireless Keyboard & Thumb tracker
Audio
Creative Sound Blaster X-Fi Fatality
CPU
Intel Xeon x5687 @ 3.6GHZ
CPU
Intel Xeon x5687 @ 3.6GHZ
Motherboard
Supermicro X8DT6-F
GPU
EVGA GT630 2GB
GPU
EVGA GT630 1GB
RAM
24GB Hynix DDR3 1333 2R-X4 ECC
Hard Drive
Samsung 850 Evo 500GB
Power Supply
Seasonic S12II 620W
Cooling
Intel LGA1366 HDX Cooler for Extreme Edition CPU
Cooling
Intel LGA1366 HDX Cooler for Extreme Edition CPU
Case
CM Stacker
Operating System
Windows 7 Pro x64
Monitor
Qnix QX2410R DP
▲ hide details ▲
AllenG is offline  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off