Overclock.net › Forums › Intel › Intel CPUs › Kaby Lake Overclocking Guide [With Statistics]
New Posts  All Forums:Forum Nav:

Kaby Lake Overclocking Guide [With Statistics] - Page 447

post #4461 of 4878
Stock timing for my 3400 mhz kit is 16/18/18 38
I ran hci overnight at 3733 at this timing and no errors
For now cant find a way for the kit to even boot at 3866 or 4000 ....
post #4462 of 4878
Okay REALLY puzzled about what's going on here.

I found my "memory sweet spot" to be 4000, 18, 19, 19, 39, CR-2. (Timings copied from the 4000 kit on G.Skills website).
My own is a a 3600 16, 16, 16, 36 twin kit..

I set DRAM frequency to 4000 18, 19, 19, 39, CR-2.
Then:
Begun lowering primary timings to those copied from slightly slower kits on G.Skill site. I.E. Their 3866Mhz is 18, 18, 18, 38.

I KEPT frequency at 4000 Mhz. Increased voltage and tested for stability. Got to 2265 % coverage error free.

However my memory scores are lower. Not higher. After repeated benchmarks I CANNOT replicate the score I was getting with higher timings.

I got down to as low as 16, 16, 16, 36, CR-2 (4000 MHz at 1.55v). and even at that my DDR4 read/write & copy scores are still lower at lower primary timings. (Despite passing multiple stress tests).

Secondary timings to were to auto.

Anyone any how idea how this could happen: ?



18, 19, 19, 39, CR-2 (4000 MHZ @ 1.5v) <- best despite highest timings!






18, 18, 18, 38, CR-2 (4000 MHZ @ 1.55v)






17, 18, 18, 38, CR-2 (4000 MHZ @ 1.55v)






17, 17, 17, 37, CR-2 (4000 MHZ @ 1.55v)






16, 16, 16, 36, CR-2 (4000 MHZ @ 1.55v) <--worst result despite lowest timings




Update: By upping DRAM voltage to 1.645v on the last screenshot (16, 16, 16, 36) I was able to "approximately" match the performance at 19, 19, 19, 39. But still not beat it. Still doens't make sense. As stress tests were always being passed.
upsidedwnsmiley.gif

Edited by nrpeyton - 10/19/17 at 7:29pm
Nick-PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
I7 7700K ASUS Z270 ROG MAXIMUS IX APEX NVIDIA GeForce GTX 1080Ti G.SKILL Trident Z 3600 C16 
CoolingOSMonitorPower
EK KIT P360 (Custom Water Loop) Microsoft Windows 10 Home, 64 bit ASUS PG278Q 850 Bronze 'be Quiet POWER ZONE' 
Case
Corsair Obsidian 750d Airflow 
  hide details  
Reply
Nick-PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
I7 7700K ASUS Z270 ROG MAXIMUS IX APEX NVIDIA GeForce GTX 1080Ti G.SKILL Trident Z 3600 C16 
CoolingOSMonitorPower
EK KIT P360 (Custom Water Loop) Microsoft Windows 10 Home, 64 bit ASUS PG278Q 850 Bronze 'be Quiet POWER ZONE' 
Case
Corsair Obsidian 750d Airflow 
  hide details  
Reply
post #4463 of 4878
In the first screenshot, try setting your tras to 49, from 39.

Try to keep TRAS>=Cas=TRCD+tRTP.
SB Rig
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7-2600k Gigabyte P67A-UD5 B3 R9 290X 16GB 
Hard DriveOptical DriveOSMonitor
WD 2 TB, WD 1 TB, 250gb Liteon ihbs212 blu-ray burner/reader XP+W7 Benq XL2720Z 
KeyboardPowerCaseMouse
Ducky Shine 69/Year of the Goat/Shine 5 Seasonic Platinum X1000 Corsair 760T Logitech G502 with r0ach approved ™ sensor 
Mouse Pad
Puretrak Talent 
  hide details  
Reply
SB Rig
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7-2600k Gigabyte P67A-UD5 B3 R9 290X 16GB 
Hard DriveOptical DriveOSMonitor
WD 2 TB, WD 1 TB, 250gb Liteon ihbs212 blu-ray burner/reader XP+W7 Benq XL2720Z 
KeyboardPowerCaseMouse
Ducky Shine 69/Year of the Goat/Shine 5 Seasonic Platinum X1000 Corsair 760T Logitech G502 with r0ach approved ™ sensor 
Mouse Pad
Puretrak Talent 
  hide details  
Reply
post #4464 of 4878
I've stablized my OC at 4.9 GHz with 1.33 VCORE. If I want to jump to 5 then I have to bump the VCORE to 1.39. I was wondering if my motherboard is causing stability issues.
First Rig
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel Coffee Lake 8700k Z370 AORUS Gaming 5 EVGA GTX 1080 Ti SC Black Edition G.Skill TridentZ RGB 3000MHz 
Hard DriveHard DriveHard DriveCooling
Seagate Barracuda 2TB Toshiba OCZ RD400 M.2 500GB Samsung EVO 850 500GB Noctua NH-D15S 
OSMonitorKeyboardPower
Windows 10 Pro 64 bit Dell S2417DG Ducky Shine TKL MX Green Corsair RM750i 
CaseMouseMouse PadAudio
NZXT S340 Elite White Razer Deathadder Elite Cherry G80 Speed Edition Sennheiser GAME ZERO 
Audio
Sound BlasterX AE-5 
  hide details  
Reply
First Rig
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel Coffee Lake 8700k Z370 AORUS Gaming 5 EVGA GTX 1080 Ti SC Black Edition G.Skill TridentZ RGB 3000MHz 
Hard DriveHard DriveHard DriveCooling
Seagate Barracuda 2TB Toshiba OCZ RD400 M.2 500GB Samsung EVO 850 500GB Noctua NH-D15S 
OSMonitorKeyboardPower
Windows 10 Pro 64 bit Dell S2417DG Ducky Shine TKL MX Green Corsair RM750i 
CaseMouseMouse PadAudio
NZXT S340 Elite White Razer Deathadder Elite Cherry G80 Speed Edition Sennheiser GAME ZERO 
Audio
Sound BlasterX AE-5 
  hide details  
Reply
post #4465 of 4878
Quote:
Originally Posted by Falkentyne View Post

In the first screenshot, try setting your tras to 49, from 39.

Try to keep TRAS>=Cas=TRCD+tRTP.

I tried that...

But it begs the question. Why do the XMP profiles of the modules on the G.SKILL website never follow that rule?
Nick-PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
I7 7700K ASUS Z270 ROG MAXIMUS IX APEX NVIDIA GeForce GTX 1080Ti G.SKILL Trident Z 3600 C16 
CoolingOSMonitorPower
EK KIT P360 (Custom Water Loop) Microsoft Windows 10 Home, 64 bit ASUS PG278Q 850 Bronze 'be Quiet POWER ZONE' 
Case
Corsair Obsidian 750d Airflow 
  hide details  
Reply
Nick-PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
I7 7700K ASUS Z270 ROG MAXIMUS IX APEX NVIDIA GeForce GTX 1080Ti G.SKILL Trident Z 3600 C16 
CoolingOSMonitorPower
EK KIT P360 (Custom Water Loop) Microsoft Windows 10 Home, 64 bit ASUS PG278Q 850 Bronze 'be Quiet POWER ZONE' 
Case
Corsair Obsidian 750d Airflow 
  hide details  
Reply
post #4466 of 4878
Well i just de-lidded my CPU using a razor as shown below:

Can u guys see any physical damage (i can't)
I tried to push the blade upwards towards the IHS to avoid scratching the PCB and i think i succeded.
(Unless anyone can see anything different from my pic).

I'm worried i may have slightly warped (bent) the PCB *very slightly". But don't see any damage.

I applied a tiny bit of tape over the bit electrics (4.dots.of metal wbich is part of PCB). I applied it as incase any liquid metal runs off.

Here's pics so far.

I am not planning to reseal the GPU yet. So i'll just:
1. Apply LM
2. Insert CPU into socket.
4. Place IHS ontop.
3. Hold onto IHS while i close retention bracket on MB to hold everything in place.
5. Prey to god the system boots up.




Note: slight discolpuration on one side of DIE is only the photo quality/light.








Have i missed anything?
Thx. Nick


P.S.The razor method was.EASY. Took me less than two mins.a lot LESS painsteaking than i thought it would be lol.
Edited by nrpeyton - 10/19/17 at 9:45pm
Nick-PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
I7 7700K ASUS Z270 ROG MAXIMUS IX APEX NVIDIA GeForce GTX 1080Ti G.SKILL Trident Z 3600 C16 
CoolingOSMonitorPower
EK KIT P360 (Custom Water Loop) Microsoft Windows 10 Home, 64 bit ASUS PG278Q 850 Bronze 'be Quiet POWER ZONE' 
Case
Corsair Obsidian 750d Airflow 
  hide details  
Reply
Nick-PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
I7 7700K ASUS Z270 ROG MAXIMUS IX APEX NVIDIA GeForce GTX 1080Ti G.SKILL Trident Z 3600 C16 
CoolingOSMonitorPower
EK KIT P360 (Custom Water Loop) Microsoft Windows 10 Home, 64 bit ASUS PG278Q 850 Bronze 'be Quiet POWER ZONE' 
Case
Corsair Obsidian 750d Airflow 
  hide details  
Reply
post #4467 of 4878
Quote:
Originally Posted by nrpeyton View Post

Okay REALLY puzzled about what's going on here.

I found my "memory sweet spot" to be 4000, 18, 19, 19, 39, CR-2. (Timings copied from the 4000 kit on G.Skills website).
My own is a a 3600 16, 16, 16, 36 twin kit..

I set DRAM frequency to 4000 18, 19, 19, 39, CR-2.
Then:
Begun lowering primary timings to those copied from slightly slower kits on G.Skill site. I.E. Their 3866Mhz is 18, 18, 18, 38.

I KEPT frequency at 4000 Mhz. Increased voltage and tested for stability. Got to 2265 % coverage error free.

However my memory scores are lower. Not higher. After repeated benchmarks I CANNOT replicate the score I was getting with higher timings.

I got down to as low as 16, 16, 16, 36, CR-2 (4000 MHz at 1.55v). and even at that my DDR4 read/write & copy scores are still lower at lower primary timings. (Despite passing multiple stress tests).

Secondary timings to were to auto.

Anyone any how idea how this could happen: ?



18, 19, 19, 39, CR-2 (4000 MHZ @ 1.5v) <- best despite highest timings!


18, 18, 18, 38, CR-2 (4000 MHZ @ 1.55v)


17, 18, 18, 38, CR-2 (4000 MHZ @ 1.55v)


17, 17, 17, 37, CR-2 (4000 MHZ @ 1.55v)


16, 16, 16, 36, CR-2 (4000 MHZ @ 1.55v) <--worst result despite lowest timings


Update: By upping DRAM voltage to 1.645v on the last screenshot (16, 16, 16, 36) I was able to "approximately" match the performance at 19, 19, 19, 39. But still not beat it. Still doens't make sense. As stress tests were always being passed.
upsidedwnsmiley.gif

Stress testing and performance are two different things and not linked to each other.

The reason your scores aren't improving is because the performance does not come from just the primary timings. Your secondary, tertiary and RTL/IOLS have more impact on performance than just the primaries.

Your scores are going backwards, because that indicates that the secondary and tertiary timings are not ideal for the set primaries. The board does not AUTO tune the timings for best performance. You need to do that manually.
Edited by tknight - 10/19/17 at 9:44pm
post #4468 of 4878
Quote:
Originally Posted by nrpeyton View Post

Well i just de-lidded my CPU using a razor as shown below:

Can u guys see any physical damage (i can't)
I tried to push the blade upwards towards the IHS to avoid scratching the PCB and i think i succeded.
(Unless anyone can see anything different from my pic).

I'm worried i may have slightly warped (bent) the PCB *very slightly". But don't see any damage.

I applied a tiny bit of tape over the bit electrics (4.dots.of metal wbich is part of PCB). I applied it as incase any liquid metal runs off.

Here's pics so far.

I am not planning to reseal the GPU yet. So i'll just:
1. Apply LM
2. Insert CPU into socket.
4. Place IHS ontop.
3. Hold onto IHS while i close retention bracket on MB to hold everything in place.
5. Prey to god the system boots up.


Note: slight discolpuration on one side of DIE is only the photo quality/light.


Have i missed anything?
Thx. Nick


P.S.The razor method was.EASY. Took me less than two mins.a lot LESS painsteaking than i thought it would be lol.

The side picture of your cpu is blurry but that bend does not look good if its bent as much as the picture is showing. Because it not being straight, might cause it to not have even pressure applied when locking it down in the socket and not make contact with all the pins and then not boot.

All you can do is try it and hope it boots up. Also you need to check the top of the pcb as that is where any damage would have occurred from the blade. All it takes is the tiniest nick and the cpu will not boot.
post #4469 of 4878
Quote:
Originally Posted by tknight View Post

The side picture of your cpu is blurry but that bend does not look good if its bent as much as the picture is showing. Because it not being straight, might cause it to not have even pressure applied when locking it down in the socket and not make contact with all the pins and then not boot.

All you can do is try it and hope it boots up. Also you need to check the top of the pcb as that is where any damage would have occurred from the blade. All it takes is the tiniest nick and the cpu will not boot.

Thank goodness!!!

It worked out OKAY!

I am not as impressed with the temperature drop as I'd like to of seen. (although I did drop -- just not as far as I'd have liked to).

I used Conductonaut between DIE & IHS and Kryonaut between IHS and EK Supremacy Block.

What do you guys think of my new temps?

First of all Settings were:

CPU: 4900 Mhz
v.core: (1.35v adaptive).
(all power and LLC settings left to default/auto).

Cache: 4600 Mhz

DDR4: 4000 Mhz
18, 19, 19, 29, 2T. 1.5v
All other timings on auto.


New delidded temps!

Prime95 (AVX enabled):

Averages:
Core #0 - difference of: 15c
Core #1 -difference of: 11c
Core #2 -difference of: 12c
Core #3 - difference of: 18c

After a 15 min run 'core max' was 17c cooler
and 'package max' was 12c cooler



Realbench 2.56:


Average 'CPU Package Temp' at 10 mins (on 2nd successful Hash Check) was: 53c
= -18c difference

Max CPU Package Temp (reported in RealBench): 60c
= -18c difference




x264, 16 thread loop

After resetting hwinf64 at 2 mins.

Then taking a reading at 5 mins;

'Average CPU package' temp was: 52c (=18c cooler)
&
'Core Max': 54c (= 20c cooler)


Summary

Most insignificant improvement was prime95 core #1 only seen a 11c drop.

Most promising improvement was core #3 seen a 18c improvement.

And best improvements show 18-20c improvements at best on x264 and Realbench (averages and max cores).

And finally, even large FFT (maximum Heat on AVX enabled Prime95 barely ever exceed 50c. With a maximum core average of only 47c

Some cores even running in the 30's!! smile.gif






Next Step:

See if she can pass all stability tests now at 5GHz 1.4.V. (never possible before).
Edited by nrpeyton - 10/20/17 at 1:06am
Nick-PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
I7 7700K ASUS Z270 ROG MAXIMUS IX APEX NVIDIA GeForce GTX 1080Ti G.SKILL Trident Z 3600 C16 
CoolingOSMonitorPower
EK KIT P360 (Custom Water Loop) Microsoft Windows 10 Home, 64 bit ASUS PG278Q 850 Bronze 'be Quiet POWER ZONE' 
Case
Corsair Obsidian 750d Airflow 
  hide details  
Reply
Nick-PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
I7 7700K ASUS Z270 ROG MAXIMUS IX APEX NVIDIA GeForce GTX 1080Ti G.SKILL Trident Z 3600 C16 
CoolingOSMonitorPower
EK KIT P360 (Custom Water Loop) Microsoft Windows 10 Home, 64 bit ASUS PG278Q 850 Bronze 'be Quiet POWER ZONE' 
Case
Corsair Obsidian 750d Airflow 
  hide details  
Reply
post #4470 of 4878
Quote:
Originally Posted by nrpeyton View Post


Did you applied LM on the dye only or dye and underneath of ihs ?
And just to confirm you are using the float method ? haven't glued the ihs back yet....
The stuff..
(8 items)
 
  
CPUMotherboardGraphicsRAM
i7 7700k Maximus Viii Impact Gainward GeForce GTX 260 896MB GDDR3 G.Skill 3200 C15 2x16 
Hard DriveCoolingOSPower
Intel 750 800 NH D14 Windows 10 XFX P1-550B-BEFX 650w XTR Gold Fully Modular 
  hide details  
Reply
The stuff..
(8 items)
 
  
CPUMotherboardGraphicsRAM
i7 7700k Maximus Viii Impact Gainward GeForce GTX 260 896MB GDDR3 G.Skill 3200 C15 2x16 
Hard DriveCoolingOSPower
Intel 750 800 NH D14 Windows 10 XFX P1-550B-BEFX 650w XTR Gold Fully Modular 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › Kaby Lake Overclocking Guide [With Statistics]