Memory Speed vs Timings - Overclock.net - An Overclocking Community

Forum Jump: 

Memory Speed vs Timings

 
Thread Tools
post #1 of 8 (permalink) Old 04-27-2019, 03:12 AM - Thread Starter
New to Overclock.net
 
Matt-Matt's Avatar
 
Join Date: Jun 2011
Location: Launceston, Tasmania, Australia
Posts: 5,808
Rep: 224 (Unique: 176)
Memory Speed vs Timings

Hey guys,

I posted a thread a while back about some DDR4 overclocking on my system and I'd like to post and update this thread with my results as well as performance gains/losses.

As we all know; at least in the past with DDR3 Intel systems typically love speed over timings but is this still the case?

My current system is as follows:
CPU: 8700k @ 5.2GHz
RAM: 16GB (2x8GB) G-Skill Trident Z RGB 3000MHz CL16
Board: Gigabyte Gaming 7 z370 (On water)
GPU: GTX1080
Boot Drive: 970 Evo (Soon to be the Evo Plus..)
PSU: EVGA 850w P2

As for the settings I will be using I currently have two setups and am considering a third/fourth based on what people want to see.

All configs will have
CPU: 5.2GHz
Uncore: 5.0GHz
System Agent: 1.050v

Config 1: Pure Speed
Speed: 3900MHz
Timings: 16-16-17-36 1T, tRFC 526
Voltage: 1.38v

Config 2: Mixture
Speed: 3200MHz
Timings: 14-15-15-35 1T, tRFC 401
Voltage: 1.38v

Config 3: Stock (XMP)
Speed: 3000Mhz
Timings: 16-18-18-38 2T rRFC 526
Voltage: 1.38v

Config 4: ???
*** To come ***
I'm welcome to any suggestions for this and to see what I can do.

Benchmarks:
I will do some synthetics to start with, I feel that I need to tweak my mixed timings vs speed overclock a bit more however.
Matt-Matt is offline  
Sponsored Links
Advertisement
 
post #2 of 8 (permalink) Old 04-27-2019, 05:48 AM
professional curmudgeon
 
looniam's Avatar
 
Join Date: Apr 2009
Posts: 9,482
Rep: 776 (Unique: 446)
how low in timings at "stock" 3000?

for something like tightest timings vs fastest speed.

Remember the golden rule of statistics: A personal sample size of one is a sufficient basis upon which to draw universal conclusions.
Upload the computer to Dropbox and provide a link to it so others may download it to examine and give advice for repairs.
loon 3.2
(18 items)
CPU
i7-3770K
Motherboard
Asus P8Z77-V Pro
GPU
EVGA 980TI SC+
RAM
16Gb PNY ddr3 1866
Hard Drive
PNY 1311 240Gb
Hard Drive
1 TB Seagate
Hard Drive
3 TB WD Blue
Optical Drive
DVD DVDRW+/-
Power Supply
EVGA SuperNova 750 G2
Cooling
EKWB P280 kit
Cooling
EK-VGA supremacy
Case
Stryker M [hammered and drilled]
Operating System
Win X
Monitor
LG 24MC57HQ-P
Keyboard
Ducky Zero [blues]
Mouse
corsair M65
Audio
SB Recon3D
Audio
Klipsch ProMedia 2.1
▲ hide details ▲


looniam is offline  
post #3 of 8 (permalink) Old 04-27-2019, 06:13 AM
New to Overclock.net
 
deepor's Avatar
 
Join Date: Feb 2013
Posts: 4,684
Rep: 470 (Unique: 317)
Your "config 1" has faster timings than your "config 2"!

You need to translate the numbers in the timings into real time when comparing. One clock tick at 3900MHz is faster than at 3200MHz.

To do this, divide the timing by the MHz. Then use the results to compare, instead of that "14" or that "16" you've written down.

Here's are your config 1 and 2 timings translated into something that can be compared:

Code:
16-16-17-36 @ 3900MHz:
-> 0.0041, 0.0041, 0.0044, 0.0092

14-15-15-35 @ 3200 MHz:
-> 0.0044, 0.0047, 0.0047, 0.0109
EDIT: I had made a mistake here earlier and copied the wrong set of timings from the first post.

Something else to think about that I don't know how to deal with is, the CPU will never just read a single byte from RAM. It will read a certain length of bytes. This then might make settings that have better (real) timings still end up slower than a different set of settings that has worse timings but more MHz. Reading a length of 64 bytes or something might be done faster because of the MHz despite the whole thing starting slower because of worse timings. I guess it's best to simply benchmark stuff.

Last edited by deepor; 04-27-2019 at 07:07 AM.
deepor is offline  
Sponsored Links
Advertisement
 
post #4 of 8 (permalink) Old 04-27-2019, 06:23 AM - Thread Starter
New to Overclock.net
 
Matt-Matt's Avatar
 
Join Date: Jun 2011
Location: Launceston, Tasmania, Australia
Posts: 5,808
Rep: 224 (Unique: 176)
Quote: Originally Posted by looniam View Post
how low in timings at "stock" 3000?

for something like tightest timings vs fastest speed.
Yeah that is something that I am going to do next, it'll probably be a few days before I get any testing done on this however.
I'll also go ahead and update my BIOS as well, it seems to be having issues with a few things.

Quote: Originally Posted by deepor View Post
Your "config 1" has faster timings than your "config 2"!

You need to translate the numbers in the timings into real time when comparing. One clock tick at 3900MHz is faster than at 3200MHz.

To do this, divide the timing by the MHz. Then use the results to compare, instead of that "14" or that "16" you've written down.

Here's are your config 1 and 2 timings translated into something that can be compared:

Code:
16-16-17-36 @ 3900MHz:
    0.0041, 0.0041, 0.0044, 0.0092

16-18-18-38 @ 3200 MHz:
    0.0044, 0.0047, 0.0047, 0.0109
Of course, that is how timings work. But we can see that lower timings vs higher speed is more the comparison.

So you're saying if i use the formula; it will tell me which is better?

As for the memory timings; 14-15-15-35 1T tRFC 401 is tighter then 16-16-17-36 1T tRFC 401 as in the numbers are lower.

I wasn't able to get the board to post any lower then the above at 3200MHz, but my BIOS is also playing up so I will flash that tomorrow.
Matt-Matt is offline  
post #5 of 8 (permalink) Old 04-27-2019, 07:02 AM
New to Overclock.net
 
deepor's Avatar
 
Join Date: Feb 2013
Posts: 4,684
Rep: 470 (Unique: 317)
Quote: Originally Posted by Matt-Matt View Post
As for the memory timings; 14-15-15-35 1T tRFC 401 is tighter then 16-16-17-36 1T tRFC 401 as in the numbers are lower.
You can only easily compare if it's the same MHz. If the MHz are different, you only know what's tighter after you do some calculations.

The memory timing numbers are the amount of clock ticks it takes before the RAM starts to answer a request from the CPU. If the clock runs at a different MHz, then the timings mean a different real time. So what you need to compare is not:

Code:
14 vs. 16
What you need to compare is:

Code:
14/3200 vs. 16/3900
Another way to calculate stuff is this way here:

Code:
16 / 3900 * 3200 = 13.1
That "16" timing at 3900MHz is the same real time as what a "13.1" timing would be at 3200MHz. But of course you can't set a timing like "13.1", you can only configure either "13" or "14". Using 13 at 3200MHz would be a tiny bit tighter than the 16 at 3900MHz, and using 14 at 3200MHz would be quite a bit away from the 16 at 3900MHz.
deepor is offline  
post #6 of 8 (permalink) Old 04-27-2019, 07:11 AM
New to Overclock.net
 
deepor's Avatar
 
Join Date: Feb 2013
Posts: 4,684
Rep: 470 (Unique: 317)
@Matt-Matt :

I had made a mistake in my post earlier. I had copy+pasted the wrong timings from your first post, the results of the calculations should have looked like this:

Code:
16-16-17-36 @ 3900MHz:
-> 0.0041, 0.0041, 0.0044, 0.0092

14-15-15-35 @ 3200 MHz:
-> 0.0044, 0.0047, 0.0047, 0.0109
deepor is offline  
post #7 of 8 (permalink) Old 04-28-2019, 01:02 AM - Thread Starter
New to Overclock.net
 
Matt-Matt's Avatar
 
Join Date: Jun 2011
Location: Launceston, Tasmania, Australia
Posts: 5,808
Rep: 224 (Unique: 176)
Update: My board is now playing up with the RAM speeds, it will POST at 3900MHz and pass memtest, but CPU-Z is reporting that it's only at 3400MHz despite it being set to 3900MHz in the BIOS?

I updated the BIOS and set the clocks the same and getting the same behavior, are there are other programs I can test this with?

I'm still not sure if this is the board changing it back or CPU-Z being broken? Thinking its the first..

Oddly enough if I set it to 3400MHz and these same timings it works in CPU-Z?

Aida64 is also reporting the same speed.. darn
Matt-Matt is offline  
post #8 of 8 (permalink) Old 05-05-2019, 03:50 AM
New to Overclock.net
 
MrHarris's Avatar
 
Join Date: Jan 2008
Posts: 597
Rep: 9 (Unique: 9)
Hwinfo64 ..... or memtweakit from asus...
MrHarris 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