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 222

post #2211 of 2775
Quote:
Originally Posted by Slink3Slyde View Post

The RealBench stress test also stresses the graphics card, so I think what you are seeing is the CPU cooler sucking in the heated exhaust of the GPU and raising core temps. Apart from lowering what RAM related voltages that you can, which would certainly help, you probably want to look into you case fan setup to try and keep the air going into your CPU cooler as cool as possible.

I have a AIO: Masterliquid Pro 280. The double fans (140mm) are placed at the front. And a 120mm fan at the back. The way I see it, the CPU gets fresh air from out side. De GFX gets air through the AIO fans.
post #2212 of 2775
Quote:
Originally Posted by Xperimental View Post

I am new to RAM tweaking, but would that be enough? Enough temperature drop and no throttling.

If you just go with 1.2 V for RAM and leave every other RAM settings as is, probably won't be stable. Regarding temps, any little helps if you want to stay below 80 degrees.

Quote:
Originally Posted by Xperimental View Post

So I I would summary:
AVX
1) configure AVX offset 2
2) run tests

You missed a step:
0) Lower Vcore a bit, let's say by 0.01 V and if stable, lower it more

Quote:
Originally Posted by Xperimental View Post

Memory :
1) Leave frequency (3000Mhz) and timings.
2) set voltage to 1.2V
3) No boot = loosen timings
(what does this mean?)
4) Boot = Run Hyper Pi
5) unstable: loosen timings, lower frequency (not preferred by me) or raise DRAM voltage.

- when tuning timing, frequency and DRAM voltage: in what order should you start tweaking?
- Increasing DRAM voltage means more heat. Does this also automatically increase the CPU SA and CPU VCCIO? And do they generate heat?


Like I said I am new to RAM tweaking. I have a Corsair Vengeance LPX 32GB 3000Mhz kit.


You wrote you have Ram Speed: 3000Mhz 15-17-17-35 @ 1.3530V. 15-17-17-35 are the timings: CAS, tRCD, tRP, tRAS. You don't have to fully understand them, what's important is lower numbers mean lower latency, faster speed but require more voltage. Bigger numbers, higher latency, lower required voltage.

So:
3) Increase timings to 16-18-18-36 or higher until it boots at least. Won't mean it's stable, but it's at the margin of stability.
5) You need to prefer it if you want to lower temps. Loosening timings has pretty much the same effect as lowering RAM frequency. Not exactly but from a beginners perspective it is, you won't be able to tell a difference and between 1 step in timings or frequency there's a really small, almost unmeasurable difference (in heavy RAM dependent games maybe 1-3 FPS, margin of error).

Here's a good study about the relationship between RAM frequency, timings and voltages required for stability: http://hw-db.com/memory/2768/g-skill-f4-3200c16d-8gvk-review/2. The same principle applies to every RAM just with different values.
Edited by dirtyred - 5/1/17 at 5:24am
    
CPUMotherboardGraphicsRAM
i7 7700 K @ 4.7 GHz ASUS Prime Z270-K Gigabyte GeForce GTX 1060 WINDFORCE OC 6G  G.Skill Ripjaws V 16 GB 
Hard DriveCoolingOSMonitor
Samsung 940 EVO 256 GB Phanteks PH-TC14PE Windows 10 Pro Dell U2312HM 
PowerCaseMouse
Seasonic M12II-620 EVO Fractal Design Define C Razer Taipan 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i7 7700 K @ 4.7 GHz ASUS Prime Z270-K Gigabyte GeForce GTX 1060 WINDFORCE OC 6G  G.Skill Ripjaws V 16 GB 
Hard DriveCoolingOSMonitor
Samsung 940 EVO 256 GB Phanteks PH-TC14PE Windows 10 Pro Dell U2312HM 
PowerCaseMouse
Seasonic M12II-620 EVO Fractal Design Define C Razer Taipan 
  hide details  
Reply
post #2213 of 2775
Quote:
Originally Posted by Xperimental View Post

I have a AIO: Masterliquid Pro 280. The double fans (140mm) are placed at the front. And a 120mm fan at the back. The way I see it, the CPU gets fresh air from out side. De GFX gets air through the AIO fans.

Yup, sorry. The way you worded it made it sound like your temps went up after you installed a new GPU to me.
post #2214 of 2775
Quote:
Originally Posted by Slink3Slyde View Post

Yup, sorry. The way you worded it made it sound like your temps went up after you installed a new GPU to me.

Ah no worries, English isn't my native language wink.gif
post #2215 of 2775
Quote:
Originally Posted by Xperimental View Post

Ah no worries, English isn't my native language wink.gif

Youre doing just fine. thumb.gif
post #2216 of 2775
Quote:
Originally Posted by dirtyred View Post

If you just go with 1.2 V for RAM and leave every other RAM settings as is, probably won't be stable. Regarding temps, any little helps if you want to stay below 80 degrees.

Before I read your reply, I did go for a DRAM voltage drop only. I able to set it to 1.20V. The computer booted and HyperPi finished it's test (32M, priority: normal, processors 8) in 9m33s.
The CPU temperature was 83 degrees at maximum. It was still running at 5GHz without AVX offset.
I am gonna run some new RealBench tests, if there is a temp drop...
Quote:
Originally Posted by dirtyred View Post

You missed a step:
0) Lower Vcore a bit, let's say by 0.01 V and if stable, lower it more
You wrote you have Ram Speed: 3000Mhz 15-17-17-35 @ 1.3530V. 15-17-17-35 are the timings: CAS, tRCD, tRP, tRAS. You don't have to fully understand them, what's important is lower numbers mean lower latency, faster speed but require more voltage. Bigger numbers, higher latency, lower required voltage.

Can you explain when it's the fist step and not the last one?
Quote:
Originally Posted by dirtyred View Post

So:
3) Increase timings to 16-18-18-36 or higher until it boots at least. Won't mean it's stable, but it's at the margin of stability.
5) You need to prefer it if you want to lower temps. Loosening timings has pretty much the same effect as lowering RAM frequency. Not exactly but from a beginners perspective it is, you won't be able to tell a difference and between 1 step in timings or frequency there's a really small, almost unmeasurable difference (in heavy RAM dependent games maybe 1-3 FPS, margin of error).

Here's a good study about the relationship between RAM frequency, timings and voltages required for stability: http://hw-db.com/memory/2768/g-skill-f4-3200c16d-8gvk-review/2. The same principle applies to every RAM just with different values.

Is it save to say that the RAM is stable when Hyper Pi completed the test?
I am gonna dig into that link, anyway smile.gif

Thanks for helping out!
post #2217 of 2775
Quote:
Originally Posted by Darkwizzie View Post

...
To be charted in the main chart, you must fulfill one of the following requirements:
...
OCCT 4.4.1 1 hour

Is any NEWER version of OCCT acceptable? It appears that only 4.5 can be downloaded from OCCT's website. Also, which test mode? (small or large set) (I'm still waiting for them to send me a work-around for non-commercial use on a windows domain, but once they do, I hope to be submitting a 5Ghz @ < 1.3v result.)
    
CPUMotherboardGraphicsRAM
Intel i7-7700K Asus ROG Maximus IX Hero Gigabyte GeForce GTX 1070 G1 Gaming 4x 8GB Vengeance LPX Series 
Hard DriveCoolingCoolingOS
1TB Samsung EVO 960 NVME M.2 EK Supremacy EVO EK SE 360 Radiator Windows 10 Pro x64 
MonitorMonitorKeyboardPower
Asus PB328Q 32" WQHD Asus 25" 16:10 (forgot the model number) Ducky Shine 5 EVGA SuperNOVA 650 G2 
CaseMouseMouse Pad
Corsair Crystal Series 460X RGB Logitech G900 Corsair MM300 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Intel i7-7700K Asus ROG Maximus IX Hero Gigabyte GeForce GTX 1070 G1 Gaming 4x 8GB Vengeance LPX Series 
Hard DriveCoolingCoolingOS
1TB Samsung EVO 960 NVME M.2 EK Supremacy EVO EK SE 360 Radiator Windows 10 Pro x64 
MonitorMonitorKeyboardPower
Asus PB328Q 32" WQHD Asus 25" 16:10 (forgot the model number) Ducky Shine 5 EVGA SuperNOVA 650 G2 
CaseMouseMouse Pad
Corsair Crystal Series 460X RGB Logitech G900 Corsair MM300 
  hide details  
Reply
post #2218 of 2775
Quote:
Originally Posted by Xperimental View Post

Before I read your reply, I did go for a DRAM voltage drop only. I able to set it to 1.20V. The computer booted and HyperPi finished it's test (32M, priority: normal, processors 8) in 9m33s.
The CPU temperature was 83 degrees at maximum. It was still running at 5GHz without AVX offset.
I am gonna run some new RealBench tests, if there is a temp drop...
Can you explain when it's the fist step and not the last one?
Is it save to say that the RAM is stable when Hyper Pi completed the test?
I am gonna dig into that link, anyway smile.gif

Thanks for helping out!

I might have forgotten to tell you not to use XMP profiles. If you did set it, do a BIOS reset and set your RAM manually. XMP is basically auto OC so I don't recommend it.
If you apply AVX offset, you can combine it with a Vcore reduction. If you lower one there's sense in lowering the other one as well.

Hyper Pi is not 100% but a fast way to detect instability. If you want to be sure, you run MemTest86 as it bypasses the OS. It takes time however so do it overnight.

Btw I found out the hard way after many RB crashes (LuxMark GPU test) that the cause was MSI Afterburner. After I disabled to autostart with Windows, I could lower my Vcore more and still pass a 30 min test. Before that it was crashing randomly between 3-30 minutes. The explanation is that it makes some API calls and that can cause instability.
Edited by dirtyred - 5/1/17 at 11:06am
    
CPUMotherboardGraphicsRAM
i7 7700 K @ 4.7 GHz ASUS Prime Z270-K Gigabyte GeForce GTX 1060 WINDFORCE OC 6G  G.Skill Ripjaws V 16 GB 
Hard DriveCoolingOSMonitor
Samsung 940 EVO 256 GB Phanteks PH-TC14PE Windows 10 Pro Dell U2312HM 
PowerCaseMouse
Seasonic M12II-620 EVO Fractal Design Define C Razer Taipan 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i7 7700 K @ 4.7 GHz ASUS Prime Z270-K Gigabyte GeForce GTX 1060 WINDFORCE OC 6G  G.Skill Ripjaws V 16 GB 
Hard DriveCoolingOSMonitor
Samsung 940 EVO 256 GB Phanteks PH-TC14PE Windows 10 Pro Dell U2312HM 
PowerCaseMouse
Seasonic M12II-620 EVO Fractal Design Define C Razer Taipan 
  hide details  
Reply
post #2219 of 2775
Quote:
Originally Posted by dirtyred View Post

Quote:
Originally Posted by Xperimental View Post

Before I read your reply, I did go for a DRAM voltage drop only. I able to set it to 1.20V. The computer booted and HyperPi finished it's test (32M, priority: normal, processors 8) in 9m33s.
The CPU temperature was 83 degrees at maximum. It was still running at 5GHz without AVX offset.
I am gonna run some new RealBench tests, if there is a temp drop...
Can you explain when it's the fist step and not the last one?
Is it save to say that the RAM is stable when Hyper Pi completed the test?
I am gonna dig into that link, anyway smile.gif

Thanks for helping out![/quote

I might have forgotten to tell you not to use XMP profiles. If you did set it, do a BIOS reset and set your RAM manually. XMP is basically auto OC so I don't recommend it.
If you apply AVX offset, you can combine it with a Vcore reduction. If you lower one there's sense in lowering the other one as well.

Hyper Pi is not 100% but a fast way to detect instability. If you want to be sure, you run MemTest86 as it bypasses the OS. It takes time however so do it overnight.

Btw I found out the hard way after many RB crashes (LuxMark GPU test) that the cause was MSI Afterburner. After I disabled to autostart with Windows, I could lower my Vcore more and still pass a 30 min test. Before that it was crashing randomly between 3-30 minutes. The explanation is that it makes some API calls and that can cause instability.



Not trying to be an ass by correcting you, just that Memtest86 actually isnt very good for detecting a bad RAM OC, it is very good at testing if the RAM is faulty. If you check out the Asus guide, its recommended to use HCI Memtest for OC stability. You need to run as many instances as you have threads on the CPU with 90% of your RAM divided between them.

As Ive understood it, the fact that Windows is is reserving some of the RAM doesnt matter, running HCI Memtest to 500% should mean your good to go. I have previously used 32M Hyper Pi and before that Memtest86 myself until I heard about HCI Memtest, and its worked very well for me this time around. smile.gif
post #2220 of 2775
Quote:
Originally Posted by dlss View Post

Well, you didn't mention you have i5-7600K so I wrongly assumed you have an i7-7700K.
Yeah, 38x is the stock cache speed for that CPU. smile.gif
What is stock cache for i7 6700k?
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]