Skylake 6700K memory tests, speed vs latency - Overclock.net - An Overclocking Community

Forum Jump: 

Skylake 6700K memory tests, speed vs latency

 
Thread Tools
post #1 of 4 (permalink) Old 08-19-2015, 11:52 AM - Thread Starter
New to Overclock.net
 
qwwwizx's Avatar
 
Join Date: Jun 2011
Location: Denmark
Posts: 521
Rep: 23 (Unique: 22)

What is faster - 3000C15 or 2400C12 or 3200C16 or 2666C13 ?????

 

Conclusion so far:

 

1) 4 sticks are a lot faster than 2. (click 2 vs 4 sticks in the sheet below)

2) You need high frequency sticks, go for 3200+ even if they are CL16 or worse.

 

The reason for doing this and for you to read it

When ordering my 6700k, I tried to find some information on what kind of memory modules to get for optimal performance. I couldn't really find any usable information, so I decided to make some myself and share it. 

 

Most high performance sticks out there suffers from lousy timings. High bandwidth on memory can be hard for an overclocked CPU so it was about to find spot with decent bandwidth and good latency or the other way around. In the end I want a fast 24/7 stable system. 

I ended up with some Kingston 2666CL13 sticks. Not fast, but with ok latency and not too expensive in case of poor performance. 

The good thing is that these sticks had a lot of unused potential, so I ended up overclocking and trying a lot of different timings. Now that I done all that, you can see my results and make your own conclusions.

 

Test

My tests are only done in XTU 6.0.2.2. All tests are done by changing bandwitdth and timing in bios, load windows, test 3 times in a row, back to bios and change settings etc. until I had the whole data-set. 





 

Now you know what to expect. Enjoy and get some good sticks. You can also get my XTU profile from the red-line directly from hwbot using this link.


H2O - 6770K (XTU 1697)
[H2O] 6700K + MSI Z170 Titanitum 4.9 GHz (turbo boost, eist) - EK HF Supreme
[H2O] EVGA SC 980 - EK FC
[AIR] Samsung 950 Pro 512GB, Samsung SM951 512GB
[AIR] Ripjaws V (4x16GB) 3200 @ 14-14-14-35 2T
qwwwizx is offline  
Sponsored Links
Advertisement
 
post #2 of 4 (permalink) Old 08-28-2015, 11:31 AM
New to Overclock.net
 
ldsh's Avatar
 
Join Date: Jun 2015
Posts: 19
Rep: 1 (Unique: 1)
Could the difference in speed be linked to the memory capacity? (I don't know How XTU is doing things).

I find it strange that the increase due to the usage of 4 stick is not a fixed percentage, the higher the frequency, the higher it benefit from the 4 sticks (instead of 2):
Freq Increase when using 4 sticks instead of 2
2666 => +6.67%
3000 => +8.37%
3200 => +9.04%

I also don't understand the big difference between the
3200 16 16 40 2 12000 1652
Of the first page test and the
3200 16 16 39 2 13000 1294
3200 16 16 39 2 13000 1411
What makes the first better than the two others?

ldsh is offline  
post #3 of 4 (permalink) Old 08-29-2015, 02:13 PM - Thread Starter
New to Overclock.net
 
qwwwizx's Avatar
 
Join Date: Jun 2011
Location: Denmark
Posts: 521
Rep: 23 (Unique: 22)
Quote:
Originally Posted by ldsh View Post

Could the difference in speed be linked to the memory capacity? (I don't know How XTU is doing things).

I find it strange that the increase due to the usage of 4 stick is not a fixed percentage, the higher the frequency, the higher it benefit from the 4 sticks (instead of 2):
Freq Increase when using 4 sticks instead of 2
2666 => +6.67%
3000 => +8.37%
3200 => +9.04%

I also don't understand the big difference between the
3200 16 16 40 2 12000 1652
Of the first page test and the
3200 16 16 39 2 13000 1294
3200 16 16 39 2 13000 1411
What makes the first better than the two others?

 

I cant tell you the reason. I just did the tests and took notes. I ran each test 3 times in a row, no reboots, no new applications on system. Just testing, testing, testing, reboot, adjust, and redo. Everything in bios like speedstep etc. is on auto settings. 

 

"

I also don't understand the big difference between the
3200 16 16 40 2 12000 1652

"

 

That was with 4 memorysticks, CPU OC at 4x 4.9Ghz, Cache at 4.7Ghz. The CPU is overclocked a lot, to be sure I not to saturate the CPU to early.


"

Of the first page test and the
3200 16 16 39 2 13000 1294
3200 16 16 39 2 13000 1411

"

 

Thats on Skylake default clocks (CPU not overclocked), only playing with memory 2 sticks vs 4 sticks. This is what you should expect with normal sticks like g.skill 3200CL16 and no CPU overclock, if you get a pack with 2 sticks vs 4.
 

Page 1 was just to show how memory speed and latency should be prioritized. 

 

Page 2 was to see if it made any difference using 2 or 4 sticks. 

 

My conclusion was something like, get 4 cheap 3200 CL16 sticks. Then you dont need to tweak it further to get good efficiency. 


H2O - 6770K (XTU 1697)
[H2O] 6700K + MSI Z170 Titanitum 4.9 GHz (turbo boost, eist) - EK HF Supreme
[H2O] EVGA SC 980 - EK FC
[AIR] Samsung 950 Pro 512GB, Samsung SM951 512GB
[AIR] Ripjaws V (4x16GB) 3200 @ 14-14-14-35 2T
qwwwizx is offline  
Sponsored Links
Advertisement
 
post #4 of 4 (permalink) Old 08-31-2015, 12:05 AM
New to Overclock.net
 
ldsh's Avatar
 
Join Date: Jun 2015
Posts: 19
Rep: 1 (Unique: 1)
I firstly thought there was a specific XTU benchmark for ram, but it is a global score for the machine performance.
So the difference in CPU overclock makes logically a difference between the results of the two tables.

It may be interesting to also have some benchmark memory specifics (results like theses: http://www.maxxpi.net/pages/downloads/maxxmemsup2---preview.php ). What really change between 2 or 4 sticks may be clearer.

I would also been very interested on CFD benchmarks but the servers where the STARS Euler3D CFD one should be seems down.

ldsh 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