F5F Stay Refreshed Hardware Desktop Adjust DDR3 settings and boost I7 3770k performance

Adjust DDR3 settings and boost I7 3770k performance

Adjust DDR3 settings and boost I7 3770k performance

L
LoucoDoPC
Junior Member
44
03-07-2016, 09:49 PM
#1
I received a set of Hynix/Corsair CMI16GX3M2A1866C9R model 16GB with version 5.29 (8x2) running OC@2133 XMP-10-11-11-30-T2 1.6v. I aim to fine-tune the clock speeds but don’t know where to begin. -----------------------------------9-10-9-27-T2-TRC481.5v XMP//// This isn’t the pro version of i7-3770K at 4.2V stock voltage, liquid metal die, etc. I’m trying to push the core frequency higher without fixed voltage. A 4.5GHz is possible. Asus P8Z68 Pro/GEN3 BIOS v3802. I’m looking for ways to improve. Also, I want to set manual RAM timings for better clarity. Template: Primary Timings 2133MHZ Hynix/Corsair cas latency 10 - [ ] RAS to CAS 11 - [ ] RAS pre time 11 - [ ] RAS act time 30 - [ ] Command rate 2 - Secondary Timings__________________ RAS to RAS 7 - [ ] Ref cycle 278 - [ ] Write recovery 10 - [ ] Read to pre 8 - [ ] Four act win 32 - [ ] Write read delay 8 - [ ] CKE minimum pulse width 6 - [ ] CAS write latency 8 - [ ] RTL channel A D0-32_D1-42 - [ ] RTL channel B D0-32_D1-43 - Third Timings______________________ TWRDR(DD) 1 - [ ] TRWDR(DD) 4 - [ ] TRWSR 4 - [ ] TRR(DD) 3 - [ ] TRR(DR) 1 - [ ] TRRSR 4 - [ ] TWW(DD) 3 - [ ] TWW(DR) 3 - [ ] TWWSR 4 - [ ] IM DIVING DEEP Template.txt
L
LoucoDoPC
03-07-2016, 09:49 PM #1

I received a set of Hynix/Corsair CMI16GX3M2A1866C9R model 16GB with version 5.29 (8x2) running OC@2133 XMP-10-11-11-30-T2 1.6v. I aim to fine-tune the clock speeds but don’t know where to begin. -----------------------------------9-10-9-27-T2-TRC481.5v XMP//// This isn’t the pro version of i7-3770K at 4.2V stock voltage, liquid metal die, etc. I’m trying to push the core frequency higher without fixed voltage. A 4.5GHz is possible. Asus P8Z68 Pro/GEN3 BIOS v3802. I’m looking for ways to improve. Also, I want to set manual RAM timings for better clarity. Template: Primary Timings 2133MHZ Hynix/Corsair cas latency 10 - [ ] RAS to CAS 11 - [ ] RAS pre time 11 - [ ] RAS act time 30 - [ ] Command rate 2 - Secondary Timings__________________ RAS to RAS 7 - [ ] Ref cycle 278 - [ ] Write recovery 10 - [ ] Read to pre 8 - [ ] Four act win 32 - [ ] Write read delay 8 - [ ] CKE minimum pulse width 6 - [ ] CAS write latency 8 - [ ] RTL channel A D0-32_D1-42 - [ ] RTL channel B D0-32_D1-43 - Third Timings______________________ TWRDR(DD) 1 - [ ] TRWDR(DD) 4 - [ ] TRWSR 4 - [ ] TRR(DD) 3 - [ ] TRR(DR) 1 - [ ] TRRSR 4 - [ ] TWW(DD) 3 - [ ] TWW(DR) 3 - [ ] TWWSR 4 - [ ] IM DIVING DEEP Template.txt

X
197
03-15-2016, 01:21 AM
#2
For scheduling, lower values are preferable except for tREFI, where higher values are better. Those listed at higher levels should be modified first. why?
X
x_HmOoDy_KsA_x
03-15-2016, 01:21 AM #2

For scheduling, lower values are preferable except for tREFI, where higher values are better. Those listed at higher levels should be modified first. why?

H
HTRSA88
Junior Member
37
04-03-2016, 04:32 PM
#3
if the machine isn't running why increase the power supply and you really cut costs
H
HTRSA88
04-03-2016, 04:32 PM #3

if the machine isn't running why increase the power supply and you really cut costs

Z
zShard
Member
194
04-03-2016, 09:59 PM
#4
You choose these timings for 10-11-11 at 1.60v to match the XMP settings shown in the BIOS (first image). Adjust manually based on what appears there. For example, use 9-10-9-27-36-48 CR 2T. If you find it hard to enter the XMP profile, you don’t need 1.60v unless you aim for speeds above 1000mhz. You might also tweak the Vccio slightly, aiming around 1.1250v but keep it below 1.200v for stability. Higher voltage usage means less savings, since energy is drawn from current rather than just voltage. In reality, the biggest savings come when the load increases—when you actually need power. The difference between 1v and 1.4v at idle is negligible; the real benefit comes during demand. More details follow, but the key is letting the system handle it automatically. For your 8700K, auto mode lets you reach 5ghz with a v-core of 1.360v under load.
Z
zShard
04-03-2016, 09:59 PM #4

You choose these timings for 10-11-11 at 1.60v to match the XMP settings shown in the BIOS (first image). Adjust manually based on what appears there. For example, use 9-10-9-27-36-48 CR 2T. If you find it hard to enter the XMP profile, you don’t need 1.60v unless you aim for speeds above 1000mhz. You might also tweak the Vccio slightly, aiming around 1.1250v but keep it below 1.200v for stability. Higher voltage usage means less savings, since energy is drawn from current rather than just voltage. In reality, the biggest savings come when the load increases—when you actually need power. The difference between 1v and 1.4v at idle is negligible; the real benefit comes during demand. More details follow, but the key is letting the system handle it automatically. For your 8700K, auto mode lets you reach 5ghz with a v-core of 1.360v under load.

1
10th_Doctor_
Posting Freak
768
04-04-2016, 03:48 AM
#5
the current clock speeds remain consistent for 2133mhz on this 1866mhz build. i haven’t changed any settings, just using the xmp profile for this kit because it also fine-tunes other parameters besides the main frequency. the 9-10-9-27 setting is meant for 1866mhz and causes instability at 2133mhz even with 1.6 volts. between 1.5 and 1.575 volts also doesn’t work well here. i’m curious how tweaking the 10-11-11-28 timing would help, which secondary clocks to modify and which to leave alone. since no deep ddr3 timing guides have been posted yet, i’ll focus on adjusting only the relevant secondary clocks and avoid overcomplicating things.
1
10th_Doctor_
04-04-2016, 03:48 AM #5

the current clock speeds remain consistent for 2133mhz on this 1866mhz build. i haven’t changed any settings, just using the xmp profile for this kit because it also fine-tunes other parameters besides the main frequency. the 9-10-9-27 setting is meant for 1866mhz and causes instability at 2133mhz even with 1.6 volts. between 1.5 and 1.575 volts also doesn’t work well here. i’m curious how tweaking the 10-11-11-28 timing would help, which secondary clocks to modify and which to leave alone. since no deep ddr3 timing guides have been posted yet, i’ll focus on adjusting only the relevant secondary clocks and avoid overcomplicating things.

N
nickmill48
Member
165
04-12-2016, 09:11 AM
#6
I understand now. These are 2x8 sticks (16GB) and not 4x2s. That explains it. Mmk was interested, but I didn’t examine closely enough. 4x2s would likely keep you at 9-10-9 most of the time. The 8GB sticks put more strain on the memory controller, similar to DDR4 with 16GB sticks. Now that the meal is done, my eyes are functioning better.
N
nickmill48
04-12-2016, 09:11 AM #6

I understand now. These are 2x8 sticks (16GB) and not 4x2s. That explains it. Mmk was interested, but I didn’t examine closely enough. 4x2s would likely keep you at 9-10-9 most of the time. The 8GB sticks put more strain on the memory controller, similar to DDR4 with 16GB sticks. Now that the meal is done, my eyes are functioning better.

R
Royal_X
Junior Member
3
04-12-2016, 10:43 AM
#7
NEW ACHIEVEMENT! 4.5ghz stable on a cryorig m9i liquid metaled under and on top of the I.H.S hopefully i can lower the voltage farther while i wait for the funds to acuire the scythe ninja 5 and direct die cool this sucker since my ihs has a bad spot which is why a single core runs 5 to 10 degree hotter on average
R
Royal_X
04-12-2016, 10:43 AM #7

NEW ACHIEVEMENT! 4.5ghz stable on a cryorig m9i liquid metaled under and on top of the I.H.S hopefully i can lower the voltage farther while i wait for the funds to acuire the scythe ninja 5 and direct die cool this sucker since my ihs has a bad spot which is why a single core runs 5 to 10 degree hotter on average

B
BadMemory
Member
206
04-12-2016, 02:08 PM
#8
Exactly. I can't remember how warm it felt.
B
BadMemory
04-12-2016, 02:08 PM #8

Exactly. I can't remember how warm it felt.

N
nicolaj101
Member
102
04-28-2016, 06:29 AM
#9
You can reach up to 1.7V without problems, but be mindful of scaling effects. My setup caps scaling around 1.68V—higher voltages reduce performance. For timing, primary settings and tREFI had the biggest impact. I also chose 1T because on my board, 1T and 2T behave similarly under those conditions.
N
nicolaj101
04-28-2016, 06:29 AM #9

You can reach up to 1.7V without problems, but be mindful of scaling effects. My setup caps scaling around 1.68V—higher voltages reduce performance. For timing, primary settings and tREFI had the biggest impact. I also chose 1T because on my board, 1T and 2T behave similarly under those conditions.

S
SorannosLP
Junior Member
21
04-29-2016, 05:31 PM
#10
2400rpm 92mm fan single tower cooler, I believe. It seems I hit its limits—this wouldn’t have happened without liquid metal. The RAM kit looks pretty stable at 1T, but I only get a CTDT of 1 to 3 hours while gaming. Maybe the ASUS used a specific memory layout, which is why they need a good overclocking board with just two slots and two channels. Having only two slots would help with RAM overclocking.
S
SorannosLP
04-29-2016, 05:31 PM #10

2400rpm 92mm fan single tower cooler, I believe. It seems I hit its limits—this wouldn’t have happened without liquid metal. The RAM kit looks pretty stable at 1T, but I only get a CTDT of 1 to 3 hours while gaming. Maybe the ASUS used a specific memory layout, which is why they need a good overclocking board with just two slots and two channels. Having only two slots would help with RAM overclocking.