Intel 9700K Overclocking Question - Overclock.net - An Overclocking Community

Forum Jump: 

Intel 9700K Overclocking Question

Reply
 
Thread Tools
post #1 of 23 (permalink) Old 05-07-2019, 03:07 AM - Thread Starter
Intel User
 
rss013's Avatar
 
Join Date: Nov 2012
Location: The Netherlands
Posts: 331
Rep: 3 (Unique: 3)
Intel 9700K Overclocking Question

Hi all, (Sorry I posted the previous thread in the wrong section)

Yesterday i´ve tried to manual oc my 9700k to 5.0 ghz without any succes.
It keeps failing after a few minutes of p95 on small fft or blend settings, even after increasing my vcore from 1.25 all the way to 1.38 with small steps it gives a BSOD.
At stock speeds the temperatures don´t exceed 25-30c in idle and 50-55c under p95 (blend and small fft´s) full load.

My Specs:

i7 9700KF (Forgot to mention in the title its the KF version)
Corsair H100X
Asus ROG Z390-F Gaming
16 GB DDR4 HyperX Predator 3200MHZ
Nvidia GTX 1060 6GB

I did the following:

- Enabled XMP for RAM

- Set CPU Core Ratio to "Sync All Cores" and I set the multiplier to 50

- AVX Negative Offset 0

- Intel Speedstep Enabled

- Disabled C-States

- Disabled ASUS Multi Core Enhancement

- Disabled CPU SVID

- LLC 6 - CPU Capability 170%

- CPU Core/Cache Current limit set to 9999 (max)

- Internal CPU Power Management set to max

- VCCIO set at 1.2(auto)


I am probably missing something here and I appreciate if someone could give me some insight on this.
Thanks in advance

Coffee Lake R
(13 items)
CPU
Intel Core i7 9700KF
Motherboard
Asus ROG Z390-F Gaming
GPU
Nvidia MSI GTX 1060 6GB Oc
RAM
2x8GB HyperX Predator 3200MHZ
Hard Drive
Crucial BX500 240GB
Power Supply
Seasonic Focus Plus Gold 550
Cooling
Corsair H100X
Case
Sharkoon Night Shark RGB
Operating System
Windows 10 Pro
Monitor
AOC G2590FX
Keyboard
Ducky One DKON1687ST TKL Mx Brown Switches
Mouse
Logitech G303
Mousepad
Steelseries QcK Heavy
▲ hide details ▲
rss013 is offline  
Sponsored Links
Advertisement
 
post #2 of 23 (permalink) Old 05-07-2019, 05:20 PM
New to Overclock.net
 
wingman99's Avatar
 
Join Date: Apr 2012
Posts: 3,232
Rep: 58 (Unique: 43)
My i7 9700k would not run stable at 5.0GHz with 1.350V core voltage. I had to run 4.9GHz

i5 8600K OC 5.0GHz
Noctua NH-U14S
Motherboard Gigabyte Z370-HD3
G.SKILL Ripjaws v 16GB F4-3200C14D-16GVK XMP Speed 3200 14-14-14-34
EVGA GeForce RTX 2070 XC ULTRA GAMING
wingman99 is offline  
post #3 of 23 (permalink) Old 05-08-2019, 03:18 AM - Thread Starter
Intel User
 
rss013's Avatar
 
Join Date: Nov 2012
Location: The Netherlands
Posts: 331
Rep: 3 (Unique: 3)
Quote: Originally Posted by wingman99 View Post
My i7 9700k would not run stable at 5.0GHz with 1.350V core voltage. I had to run 4.9GHz
I think mine has the same

Coffee Lake R
(13 items)
CPU
Intel Core i7 9700KF
Motherboard
Asus ROG Z390-F Gaming
GPU
Nvidia MSI GTX 1060 6GB Oc
RAM
2x8GB HyperX Predator 3200MHZ
Hard Drive
Crucial BX500 240GB
Power Supply
Seasonic Focus Plus Gold 550
Cooling
Corsair H100X
Case
Sharkoon Night Shark RGB
Operating System
Windows 10 Pro
Monitor
AOC G2590FX
Keyboard
Ducky One DKON1687ST TKL Mx Brown Switches
Mouse
Logitech G303
Mousepad
Steelseries QcK Heavy
▲ hide details ▲
rss013 is offline  
Sponsored Links
Advertisement
 
post #4 of 23 (permalink) Old 05-08-2019, 10:41 AM
New to Overclock.net
 
Join Date: Dec 2018
Posts: 30
Rep: 0
Quote: Originally Posted by wingman99 View Post
My i7 9700k would not run stable at 5.0GHz with 1.350V core voltage. I had to run 4.9GHz
Quote: Originally Posted by rss013 View Post
Quote: Originally Posted by wingman99 View Post
My i7 9700k would not run stable at 5.0GHz with 1.350V core voltage. I had to run 4.9GHz
I think mine has the same [IMG class=inlineimg]/forum/images/smilies/frown.gif[/IMG]
So my cpu also looked like it couldnt do 5ghz since the temps were too darn high or i couldnt get anything stable ultimately im running 5.2ghz @ 1.38v VOUT and 1.435 set in bios with extreme llc and i never see higher than 85c-87c on any core. So im coming from ryzen and have only had about a month with my 9700k but i noticed for me setting xmp made my cpu worse and it wouldnt scale right. I know the ram have has extra latency in the xmp profile being the ryzen optimised gskill flareX bdie so i left it at stock 2400 cl16. Turned vccio and vccsa both to 1.2V ( you may need 1.25v or perhaps lower to 1.18 can help, my xmp sets 1.25v both auto) but i disabled cstates also threw 1.22v on dram since i noticed some vdroop, manually set bclk to 100.00(100.01 actually ive heard it helps but not sure tbh) also i use extreme LLC on the cpu internal AC/DC load line but performance mode worked better for 4.9ghz,5ghz then 5.1seemed to like extreme. I noticed the 12v at 11.7v...it still goes to 11.7ish but VRM doesnt overheat anymore. I disabled turbo boost, i would highly recomend you not raising any power or current limit untill you stablized each frequency and see what voltage it wants. So you should have 4.9 stable at 1.35 in bios but look how low VOUT in hwinfo gets...tons of vdroop. I think i hit 5ghz at 1.35v -2 avx offset with medium LLC (aka 5 lower than the highest) gigabyte goes low,medium, high ,turbo, extreme and ultra extreme dont ever use the highest, but 5ghz no avx offset at 1.36v in bios and high LLC. [email protected] turbo LLC -2 AVX, 5.1 no AVX offset i think i needed 1.4v and turbo LLC. [email protected] -2 AVX extreme LLC and finally im now @ 5.2 no offset 1.435v in bios with cache at 4.4ghz. Heres the most important part. Is youll have to use prime95 differently. I use 2 versions one with avx disabled. Just run custom , then change the fft sections with 1344 and again 1344 for the next and then tick run ffts in place. This is a much more realistic stress test. I usually need to wait about 10-15 mins before i see any errors where small ffts was always instant but theres no need to run it small ffts its about 20c too hot to be realistic. Use cinebench r20 for avx disabled testing and r15 for avx enabled, they will follow your offset and dont forget to also do single core as thats usually what really makes the difference in games. also i validate stability with 3dmark at the end and just make sure my scores didnt drop or ive noticed firestrike ultra , extreme and timespy extreme seem to show i needed a tad more vcore and crash or bsod when prime95 and cinebench all showed improvement and passed. look up 9th gen stability testing to see what i mean exactly about the different ways to disable and enable avx in prime95 but those are the settings i use.Oh and im on the gigabyte z390 auros pro and also i am delided with the rockitcool direct die bracket with conductoknaut and needed to upgrade to the EK supremacy waterblock to hit 5.2 as my kraken block just wasnt cutting it but you should take that into account that i probably needed the delid and LM TIM in order to hit this specific speed as temps are almost too hot already. I hope this help. Lookup silicon lottery 9700k and see what they use for voltage its pretty simple but getting that cooled is easier said than done. But yea i think i might need to go back to 5.1 untill i can replace this 280mm rad with a 360 or add another 280 as i have a titan x pascal in the loop. But as for ram im still testing and this is by far my finished setup so just keep that in mind. Good luck. Hope your odds change like mine did once i figured how to properly test. Aida64 works great but i have the pro version so if you got a few extra bucks theres always that.
Exile666 is offline  
post #5 of 23 (permalink) Old 05-08-2019, 11:26 AM
New to Overclock.net
 
Falkentyne's Avatar
 
Join Date: Dec 2007
Location: Riverside
Posts: 5,671
Rep: 370 (Unique: 263)
Quote: Originally Posted by Exile666 View Post
SNIP
Can you *please*, for the love of god, use proper PARAGRAPHS and spacing when you reply?
this was extremely cringeworthy to read and literally made me not want to look at forum posts for an entire day.

Were you using offset (DVID)/Auto voltages or fixed (manual) override voltages?

If you were using override voltages, the (Internal VR Settings), AC/DC loadline values (presets in CPU Internal AC/DC Load Line for these) have absolutely ZERO effect on your voltage!
Why did you say that the "Extreme" settings for CPU Internal AC/DC loadline helped you? The AC/DC loadlines are *only* used when you are either using Auto voltages, DVID offsets, or SVID offset (Svid offset is really weird; it adjusts a voltage offset via SVID based on the type of load being delivered; SVID offset = Enabled + AC loadline = 1.0 mOhms (100 in the bios) works well at 5 ghz on some CPU's).

Anyway, regardless. What you said about extreme CPU internal AC/DC loadline preset working better at 5.1 and 5.2 ghz is wrong. Manual voltages ignore the AC/DC values totally. The CPU VID is influenced by this, but not the CPU Vcore/VR VOUT or IOUT/POUT etc. What happened when you used a *manual* voltage at 5.1 ghz or 5.2 ghz, and you set CPU Internal AC/DC loadline to power saving? There should have been zero difference in stability, as long as your "Loadline Calibration" and "CPU Vcore" were set to the exact same values, as they were when you set CPU Internal AC/DC loadline to "Extreme".

The presets for CPU Internal Load Line change the AC/DC loadline settings in internal VR settings to certain values, if they are set to 0 (auto) there:

Extreme: 2.1 mOhms/2.1 mOhms (ACLL=210, DCLL=210)
Turbo: 1.6 mOhms / 1.6 mOhms (ACLL=160, DCLL=160)
Performance: 1.0 mOhms / 1.3 mOhms (ACLL=100, DCLL=130)
Power Saving: 0.4 mOhms / 1.3 mOhms (ACLL=40, DCLL=130).

These only affect CPU Vcore (VR VOUT) on auto/SVID offset/Normal voltages only.

On Auto voltages, or if DVID is +0.00v, CPU VID will equal CPU VR VOUT if the DC loadline value is equal to the VRM (CPU Vcore Loadline Calibration) mOhms value. DC loadline does not affect voltages--it's for power measurements only (VID and CPU Package Power).

VRM Loadline Calibration presets (aka CPU Vcore Loadline Calibration):
Auto/Normal/Standard: 1.6 mOhms
Low: 1.3 mOhms
Medium: 1.0 mOhms
High: 0.8 mOhms
Turbo: 0.4 mOhms
Extreme: 0.2 mOhms
Ultra Extreme: 0.01 mOhms.

Total CPU Vdroop= Amps * Resistance (mOhms). Bios target voltage (converted to millivolts)- vdroop (mv)=VR VOUT (convert back to volts from millivolts)

Higher levels of Loadline Calibration greatly decrease transient voltage response (stability) at heavier amps loads.

[email protected] ghz, RX Vega 64, 32GB DDR4, Gigabyte Aorus Master, Seasonic Platinum 1000W, Corsair 760T
Alt: MSI GT73VR Throttlebook with 7820HK @ 4.7 ghz, GTX 1070 MXM TDP mod to 230W, 32 GB RAM

Last edited by Falkentyne; 05-08-2019 at 11:31 AM.
Falkentyne is offline  
post #6 of 23 (permalink) Old 05-08-2019, 06:41 PM
New to Overclock.net
 
Join Date: Dec 2018
Posts: 30
Rep: 0
Quote: Originally Posted by Falkentyne View Post
Quote: Originally Posted by Exile666 View Post
SNIP
Can you *please*, for the love of god, use proper PARAGRAPHS and spacing when you reply?
this was extremely cringeworthy to read and literally made me not want to look at forum posts for an entire day.

Were you using offset (DVID)/Auto voltages or fixed (manual) override voltages?

If you were using override voltages, the (Internal VR Settings), AC/DC loadline values (presets in CPU Internal AC/DC Load Line for these) have absolutely ZERO effect on your voltage!
Why did you say that the "Extreme" settings for CPU Internal AC/DC loadline helped you? The AC/DC loadlines are *only* used when you are either using Auto voltages, DVID offsets, or SVID offset (Svid offset is really weird; it adjusts a voltage offset via SVID based on the type of load being delivered; SVID offset = Enabled + AC loadline = 1.0 mOhms (100 in the bios) works well at 5 ghz on some CPU's).

Anyway, regardless. What you said about extreme CPU internal AC/DC loadline preset working better at 5.1 and 5.2 ghz is wrong. Manual voltages ignore the AC/DC values totally. The CPU VID is influenced by this, but not the CPU Vcore/VR VOUT or IOUT/POUT etc. What happened when you used a *manual* voltage at 5.1 ghz or 5.2 ghz, and you set CPU Internal AC/DC loadline to power saving? There should have been zero difference in stability, as long as your "Loadline Calibration" and "CPU Vcore" were set to the exact same values, as they were when you set CPU Internal AC/DC loadline to "Extreme".

The presets for CPU Internal Load Line change the AC/DC loadline settings in internal VR settings to certain values, if they are set to 0 (auto) there:

Extreme: 2.1 mOhms/2.1 mOhms (ACLL=210, DCLL=210)
Turbo: 1.6 mOhms / 1.6 mOhms (ACLL=160, DCLL=160)
Performance: 1.0 mOhms / 1.3 mOhms (ACLL=100, DCLL=130)
Power Saving: 0.4 mOhms / 1.3 mOhms (ACLL=40, DCLL=130).

These only affect CPU Vcore (VR VOUT) on auto/SVID offset/Normal voltages only.

On Auto voltages, or if DVID is +0.00v, CPU VID will equal CPU VR VOUT if the DC loadline value is equal to the VRM (CPU Vcore Loadline Calibration) mOhms value. DC loadline does not affect voltages--it's for power measurements only (VID and CPU Package Power).

VRM Loadline Calibration presets (aka CPU Vcore Loadline Calibration):
Auto/Normal/Standard: 1.6 mOhms
Low: 1.3 mOhms
Medium: 1.0 mOhms
High: 0.8 mOhms
Turbo: 0.4 mOhms
Extreme: 0.2 mOhms
Ultra Extreme: 0.01 mOhms.

Total CPU Vdroop= Amps * Resistance (mOhms). Bios target voltage (converted to millivolts)- vdroop (mv)=VR VOUT (convert back to volts from millivolts)

Higher levels of Loadline Calibration greatly decrease transient voltage response (stability) at heavier amps loads.
0.o dang sorry about the "chaotic" organization I have a way of rambling on to try and make my points as clear as I can plus doing everything on android doesn't help, still sorry mate.

I didnt get very far with dynamic vcore so ive been using manual voltage and kind of hovering around LLC. You are absolutely right about the internal LLC not doing anything for stability as I just set it back to auto and at first I noticed only VID changed but I'm well aware VID is simply a request from the controller.

That got me thinking about when I actually did see a difference and that was when I had to uninstall Intel XTU as it was setting dynamic vcore on boot and i was wondering why it wasn't setting what I put in as manual voltage or core clock and I just continued to experiment with different combinations until something worked.

Thank you very much for clearing up that confusion on my part as I thought it was the preconversion from 12v, 5v and 3.3v that was effected by that setting but I will keep testing as my manual voltages seemed around the right ballpark.

So after I read your response I turned the internal CPU LLC to auto and there was one more thing I noticed was my power consumption was closer to 180w-190w (not good) could having those LLC settings in place raise my power reading to make it seem like I was pulling less power thus fooling my results/readings in hwinfo64?

I did think my power consumption was still pushing it but it was better as I only remember seeing 160w with occasional spikes.

If it helps my next step was putting that -2 avx offset and then cinebench r15 didn't make it a single run that's with extreme LLC on CPU only, tried with -1 but same result. Perhaps I'm hitting over current protection. Back to 5ghz and will soon be messing with dynamic voltage perhaps that now i understand it better.

Thank you again for clearing my confusion and tolerating the cringiness but I myself dont usually throw my 2 sense in but I was experiencing what looked like the same problem and stopped what I was doing to try and give as accurate as possible info and hey you ended up improving my knowledge. Didn't think it was that difficult to read. Seems a bit much not to want to reply at all. But hey I overwhelm myself sometimes too.

Edit . so let me get this straight you didn't want to visit the whole forum(yet you replied pretty quick) because you were triggered by a single post that you were very capable of helping with(or ignoring) but decided to comment on someone else's thread anyway. That's pretty petty to claim I made you not want to be involved with the whole forum for a whole day of almost an hour. needless drama no? So thanks but no thanks for the help if it needs to come with some down talking, preachy intelligence quota about grammer and your issues. Thats your problem not mine. If this was YOUR thread and it frustrated you that would be a different story and id feel bad but come on man why spread the cancer attitude when we're all here to learn. I can appreciate someone correcting me as you know your facts so once again thanks for that but that little tangent was unneccesary and pretty dramatic. Not trying to be disrespectful but no one came to see someone flaunt a god complex

Last edited by Exile666; 05-09-2019 at 01:54 AM.
Exile666 is offline  
post #7 of 23 (permalink) Old 05-08-2019, 07:10 PM
New to Overclock.net
 
Falkentyne's Avatar
 
Join Date: Dec 2007
Location: Riverside
Posts: 5,671
Rep: 370 (Unique: 263)
Quote: Originally Posted by Exile666 View Post
0.o dang sorry about the "chaotic" organization I have a way of rambling on to try and make my points as clear as I can plus doing everything on android doesn't help, still sorry mate.

I didnt get very far with dynamic vcore so ive been using manual voltage and kind of hovering around LLC. You are absolutely right about the internal LLC not doing anything for stability as I just set it back to auto and at first I noticed only VID changed but I'm well aware VID is simply a request from the controller.

That got me thinking about when I actually did see a difference and that was when I had to uninstall Intel XTU as it was setting dynamic vcore on boot and i was wondering why it wasn't setting what I put in as manual voltage or core clock and I just continued to experiment with different combinations until something worked.

Thank you very much for clearing up that confusion on my part as I thought it was the preconversion from 12v, 5v and 3.3v that was effected by that setting but I will keep testing as my manual voltages seemed around the right ballpark.

So after I read your response I turned the internal CPU LLC to auto and there was one more thing I noticed was my power consumption was closer to 180w-190w (not good) could having those LLC settings in place raise my power reading to make it seem like I was pulling less power thus fooling my results/readings in hwinfo64?

I did think my power consumption was still pushing it but it was better as I only remember seeing 160w with occasional spikes.

If it helps my next step was putting that -2 avx offset and then cinebench r15 didn't make it a single run that's with extreme LLC on CPU only, tried with -1 but same result. Perhaps I'm hitting over current protection. Back to 5ghz and will soon be messing with dynamic voltage perhaps that now i understand it better.

Thank you again for clearing my confusion and tolerating the cringiness but I myself dont usually throw my 2 sense in but I was experiencing what looked like the same problem and stopped what I was doing to try and give as accurate as possible info and hey you ended up improving my knowledge. Didn't think it was that difficult to read. Seems a bit much not to want to reply at all. But hey I overwhelm myself sometimes too.
I explained in the last reply that CPU Package Power is a direct effect of CPU VID (VID * Amps), and VID is influenced by the Internal AC/DC Loadlines.
Current (IOUT) shows a more accurate power consumption, directly from the vrm.

[email protected] ghz, RX Vega 64, 32GB DDR4, Gigabyte Aorus Master, Seasonic Platinum 1000W, Corsair 760T
Alt: MSI GT73VR Throttlebook with 7820HK @ 4.7 ghz, GTX 1070 MXM TDP mod to 230W, 32 GB RAM
Falkentyne is offline  
post #8 of 23 (permalink) Old 05-09-2019, 03:31 AM - Thread Starter
Intel User
 
rss013's Avatar
 
Join Date: Nov 2012
Location: The Netherlands
Posts: 331
Rep: 3 (Unique: 3)
Quote: Originally Posted by Exile666 View Post
So my cpu also looked like it couldnt do 5ghz since the temps were too darn high or i couldnt get anything stable ultimately im running 5.2ghz @ 1.38v VOUT and 1.435 set in bios with extreme llc and i never see higher than 85c-87c on any core. So im coming from ryzen and have only had about a month with my 9700k but i noticed for me setting xmp made my cpu worse and it wouldnt scale right. I know the ram have has extra latency in the xmp profile being the ryzen optimised gskill flareX bdie so i left it at stock 2400 cl16. Turned vccio and vccsa both to 1.2V ( you may need 1.25v or perhaps lower to 1.18 can help, my xmp sets 1.25v both auto) but i disabled cstates also threw 1.22v on dram since i noticed some vdroop, manually set bclk to 100.00(100.01 actually ive heard it helps but not sure tbh) also i use extreme LLC on the cpu internal AC/DC load line but performance mode worked better for 4.9ghz,5ghz then 5.1seemed to like extreme. I noticed the 12v at 11.7v...it still goes to 11.7ish but VRM doesnt overheat anymore. I disabled turbo boost, i would highly recomend you not raising any power or current limit untill you stablized each frequency and see what voltage it wants. So you should have 4.9 stable at 1.35 in bios but look how low VOUT in hwinfo gets...tons of vdroop. I think i hit 5ghz at 1.35v -2 avx offset with medium LLC (aka 5 lower than the highest) gigabyte goes low,medium, high ,turbo, extreme and ultra extreme dont ever use the highest, but 5ghz no avx offset at 1.36v in bios and high LLC. [email protected] turbo LLC -2 AVX, 5.1 no AVX offset i think i needed 1.4v and turbo LLC. [email protected] -2 AVX extreme LLC and finally im now @ 5.2 no offset 1.435v in bios with cache at 4.4ghz. Heres the most important part. Is youll have to use prime95 differently. I use 2 versions one with avx disabled. Just run custom , then change the fft sections with 1344 and again 1344 for the next and then tick run ffts in place. This is a much more realistic stress test. I usually need to wait about 10-15 mins before i see any errors where small ffts was always instant but theres no need to run it small ffts its about 20c too hot to be realistic. Use cinebench r20 for avx disabled testing and r15 for avx enabled, they will follow your offset and dont forget to also do single core as thats usually what really makes the difference in games. also i validate stability with 3dmark at the end and just make sure my scores didnt drop or ive noticed firestrike ultra , extreme and timespy extreme seem to show i needed a tad more vcore and crash or bsod when prime95 and cinebench all showed improvement and passed. look up 9th gen stability testing to see what i mean exactly about the different ways to disable and enable avx in prime95 but those are the settings i use.Oh and im on the gigabyte z390 auros pro and also i am delided with the rockitcool direct die bracket with conductoknaut and needed to upgrade to the EK supremacy waterblock to hit 5.2 as my kraken block just wasnt cutting it but you should take that into account that i probably needed the delid and LM TIM in order to hit this specific speed as temps are almost too hot already. I hope this help. Lookup silicon lottery 9700k and see what they use for voltage its pretty simple but getting that cooled is easier said than done. But yea i think i might need to go back to 5.1 untill i can replace this 280mm rad with a 360 or add another 280 as i have a titan x pascal in the loop. But as for ram im still testing and this is by far my finished setup so just keep that in mind. Good luck. Hope your odds change like mine did once i figured how to properly test. Aida64 works great but i have the pro version so if you got a few extra bucks theres always that.
Thanks for your reply,
I personally don't want to go above 1.39v and even that is too high for my personal taste.
Recently i've lowered the VCCIO and VCSSA to 0.95 and 1.10 while they where on 1.328 before and it made a positive difference in temperature.
If my CPU can't get 5ghz at a reasonable vcore, i give up hope and won't overclock it at all but still thanks for your help.

Coffee Lake R
(13 items)
CPU
Intel Core i7 9700KF
Motherboard
Asus ROG Z390-F Gaming
GPU
Nvidia MSI GTX 1060 6GB Oc
RAM
2x8GB HyperX Predator 3200MHZ
Hard Drive
Crucial BX500 240GB
Power Supply
Seasonic Focus Plus Gold 550
Cooling
Corsair H100X
Case
Sharkoon Night Shark RGB
Operating System
Windows 10 Pro
Monitor
AOC G2590FX
Keyboard
Ducky One DKON1687ST TKL Mx Brown Switches
Mouse
Logitech G303
Mousepad
Steelseries QcK Heavy
▲ hide details ▲
rss013 is offline  
post #9 of 23 (permalink) Old 05-09-2019, 10:16 AM
New to Overclock.net
 
Join Date: Dec 2018
Posts: 30
Rep: 0
Quote: Originally Posted by rss013 View Post
Quote: Originally Posted by Exile666 View Post
So my cpu also looked like it couldnt do 5ghz since the temps were too darn high or i couldnt get anything stable ultimately im running 5.2ghz @ 1.38v VOUT and 1.435 set in bios with extreme llc and i never see higher than 85c-87c on any core. So im coming from ryzen and have only had about a month with my 9700k but i noticed for me setting xmp made my cpu worse and it wouldnt scale right. I know the ram have has extra latency in the xmp profile being the ryzen optimised gskill flareX bdie so i left it at stock 2400 cl16. Turned vccio and vccsa both to 1.2V ( you may need 1.25v or perhaps lower to 1.18 can help, my xmp sets 1.25v both auto) but i disabled cstates also threw 1.22v on dram since i noticed some vdroop, manually set bclk to 100.00(100.01 actually ive heard it helps but not sure tbh) also i use extreme LLC on the cpu internal AC/DC load line but performance mode worked better for 4.9ghz,5ghz then 5.1seemed to like extreme. I noticed the 12v at 11.7v...it still goes to 11.7ish but VRM doesnt overheat anymore. I disabled turbo boost, i would highly recomend you not raising any power or current limit untill you stablized each frequency and see what voltage it wants. So you should have 4.9 stable at 1.35 in bios but look how low VOUT in hwinfo gets...tons of vdroop. I think i hit 5ghz at 1.35v -2 avx offset with medium LLC (aka 5 lower than the highest) gigabyte goes low,medium, high ,turbo, extreme and ultra extreme dont ever use the highest, but 5ghz no avx offset at 1.36v in bios and high LLC. [email protected] turbo LLC -2 AVX, 5.1 no AVX offset i think i needed 1.4v and turbo LLC. [email protected] -2 AVX extreme LLC and finally im now @ 5.2 no offset 1.435v in bios with cache at 4.4ghz. Heres the most important part. Is youll have to use prime95 differently. I use 2 versions one with avx disabled. Just run custom , then change the fft sections with 1344 and again 1344 for the next and then tick run ffts in place. This is a much more realistic stress test. I usually need to wait about 10-15 mins before i see any errors where small ffts was always instant but theres no need to run it small ffts its about 20c too hot to be realistic. Use cinebench r20 for avx disabled testing and r15 for avx enabled, they will follow your offset and dont forget to also do single core as thats usually what really makes the difference in games. also i validate stability with 3dmark at the end and just make sure my scores didnt drop or ive noticed firestrike ultra , extreme and timespy extreme seem to show i needed a tad more vcore and crash or bsod when prime95 and cinebench all showed improvement and passed. look up 9th gen stability testing to see what i mean exactly about the different ways to disable and enable avx in prime95 but those are the settings i use.Oh and im on the gigabyte z390 auros pro and also i am delided with the rockitcool direct die bracket with conductoknaut and needed to upgrade to the EK supremacy waterblock to hit 5.2 as my kraken block just wasnt cutting it but you should take that into account that i probably needed the delid and LM TIM in order to hit this specific speed as temps are almost too hot already. I hope this help. Lookup silicon lottery 9700k and see what they use for voltage its pretty simple but getting that cooled is easier said than done. But yea i think i might need to go back to 5.1 untill i can replace this 280mm rad with a 360 or add another 280 as i have a titan x pascal in the loop. But as for ram im still testing and this is by far my finished setup so just keep that in mind. Good luck. Hope your odds change like mine did once i figured how to properly test. Aida64 works great but i have the pro version so if you got a few extra bucks theres always that.
Thanks for your reply,
I personally don't want to go above 1.39v and even that is too high for my personal taste.
Recently i've lowered the VCCIO and VCSSA to 0.95 and 1.10 while they where on 1.328 before and it made a positive difference in temperature.
If my CPU can't get 5ghz at a reasonable vcore, i give up hope and won't overclock it at all but still thanks for your help.
Have you tried to OC the CPU before setting your ram XMP? That was the major issue I was having as it jacks up vccio and vccsa once you set XMP. But 1.32 sounds needlessly high for some reason I know 0.95v and 1.1v work at stock. Also I would say it's a must to revert bios to default especially after enabling XMP and disabling it.

you should be able to get 5ghz at much better temps than you think if you use prime95 custom tesy like I said with 1344 set as both min and max fft size and run ffts in place checked? Look how low your temps are compared to small ffts. I can't run small ffts past 4.9 myself.
You can enable AVX by adding "CpuSupportsAVX=1" no quotes to the end of local.text in prime95 folder and disable by changing to =0

I think you'll be pleasant surprised how low temps are and I definitely needed some CPU LLC for 5ghz.
If I dont use LLC I need to set something like 1.4v and there's just too much vdroop still it ends up lower than 1.2 VOUT. I think that's when I saw my VRM over 70c when it usually stays around 60c.

Aim for 4.9ghz @1.35v and try that prime95 trick and test with and without AVX. If you crash or show errors raise LLC. You won't need to set LLC too high by any means.

Mine only needed 1.36v set and one more notch up on LLC to get 5ghz at amazing temps.

But the ram was my 1st issue. I did set 1.2 for vccio and vccsa before core clock to try to keep things consistent as one is for uncore ratio I believe. Not 100% which one tbh. But I haven't gotten further myself.

Good luck

Last edited by Exile666; 05-09-2019 at 11:03 AM.
Exile666 is offline  
post #10 of 23 (permalink) Old 05-09-2019, 11:58 PM - Thread Starter
Intel User
 
rss013's Avatar
 
Join Date: Nov 2012
Location: The Netherlands
Posts: 331
Rep: 3 (Unique: 3)
Quote: Originally Posted by Exile666 View Post
Have you tried to OC the CPU before setting your ram XMP? That was the major issue I was having as it jacks up vccio and vccsa once you set XMP. But 1.32 sounds needlessly high for some reason I know 0.95v and 1.1v work at stock. Also I would say it's a must to revert bios to default especially after enabling XMP and disabling it.

you should be able to get 5ghz at much better temps than you think if you use prime95 custom tesy like I said with 1344 set as both min and max fft size and run ffts in place checked? Look how low your temps are compared to small ffts. I can't run small ffts past 4.9 myself.
You can enable AVX by adding "CpuSupportsAVX=1" no quotes to the end of local.text in prime95 folder and disable by changing to =0

I think you'll be pleasant surprised how low temps are and I definitely needed some CPU LLC for 5ghz.
If I dont use LLC I need to set something like 1.4v and there's just too much vdroop still it ends up lower than 1.2 VOUT. I think that's when I saw my VRM over 70c when it usually stays around 60c.

Aim for 4.9ghz @1.35v and try that prime95 trick and test with and without AVX. If you crash or show errors raise LLC. You won't need to set LLC too high by any means.

Mine only needed 1.36v set and one more notch up on LLC to get 5ghz at amazing temps.

But the ram was my 1st issue. I did set 1.2 for vccio and vccsa before core clock to try to keep things consistent as one is for uncore ratio I believe. Not 100% which one tbh. But I haven't gotten further myself.

Good luck
Thanks!, I haven't tried to OC the cpu without enabling XMP first, but i might gonna try that later today.
And yeah, the VCCIO and VCCSA on ASUS ROG Z390 boards seem ridicilous high on auto settings when you enable XMP, lowering that made a big difference in temps for me.
Yesterday i did try to run 5ghz/4.5 cache manual on 1.35 fixed with LLC 6, AVX offset 1 and after 2 minutes of prime the temps spike to 95c and it bsod's shortly after.
I think this chip just is just having a hard time maintaining 5ghz,in that case i might keep it on stock settings for a longer while and might OC it later to 4.8 / 4.9 when i need the extra little bit of performance.

I'm currently running 23-25c idle and 50-550c p95 full load after 1 hour, not oc'ed or anything but im not sure if i wanna sacrifice these great temps for a 35-40 idle with 80-90 max load.
I think my previous 3570K spoiled me too much with winning the silicon lottery as it run 4.5 ghz @ 1.19 fixed for over 6 years stable.
Good old Ivy

Coffee Lake R
(13 items)
CPU
Intel Core i7 9700KF
Motherboard
Asus ROG Z390-F Gaming
GPU
Nvidia MSI GTX 1060 6GB Oc
RAM
2x8GB HyperX Predator 3200MHZ
Hard Drive
Crucial BX500 240GB
Power Supply
Seasonic Focus Plus Gold 550
Cooling
Corsair H100X
Case
Sharkoon Night Shark RGB
Operating System
Windows 10 Pro
Monitor
AOC G2590FX
Keyboard
Ducky One DKON1687ST TKL Mx Brown Switches
Mouse
Logitech G303
Mousepad
Steelseries QcK Heavy
▲ hide details ▲

Last edited by rss013; 05-10-2019 at 12:01 AM.
rss013 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