Overclock.net › Forums › Intel › Intel Motherboards › [Official] MSI X99S/A Motherboard Owners Club
New Posts  All Forums:Forum Nav:

[Official] MSI X99S/A Motherboard Owners Club - Page 217

post #2161 of 3145
Finally getting around to overclocking my 5820K after over a month.

I want to keep voltage as low as possible so I've run Prime95 whilst checking the voltage in aida64 and it hit the 3.6ghz turbo @ 1.0900V (is that high for a 5820K running at stock? I'd read 1.050-1.065).

So I've set CPU Voltage mode to Adaptive + Offset and set that to +0.01V, set the CPU ratio to 36 and stress tested it to validate that speed and voltage which obviously was fine.

Now I plan to see how far I can take it at that +0.01V offset before I get crashes, currently running at 3800.18Mhz and 1.0900V.

It's actually my first time overclocking a CPU, I've never bothered before so if I've made a mistake in the way I've applied the voltages please let me know smile.gif

On a side note I don't plan to change anything else, I'm trying to keep it basic and leaving my RAM at its standard 2400mhz XMP profile.

*edit*
Oh the 1.0900V stable is in primes in-place large FFTs for max heat, when I run it as small FFTs for max FPU stress the voltage (CPU VID in Aida64) goes between 1.0851 and 1.0900V

*edit 2* Oh i use CPU VID for voltages in the overclock page as on the sensors page for "CPU Core" under voltage values it reads 0.208 under load which...err...doesn't look right at all, I thought that should be the actual CPU voltage reading so I'm confused there

*edit 3*
looks like 4Ghz is my limit with +0.001 offset tongue.gif
4.1Ghz gave a BSOD after 10 seconds of prime frown.gif

*edit 4*
With Aida64 system stability test what do you normally choose? CPU and cache? FPU just increases heat which I'm not really interested in at the moment. Or doing memory
Edited by bleets - 2/4/16 at 12:15pm
post #2162 of 3145
Interesting...i found out why my stock voltage was so high. MSI Enhanced Turbo.
With that running all cores turbo up to 3.6Ghz but at what I consider to be a hefty voltage increase to 1.0900 (or more depending on overclock).

Disabling enhanced turbo when stress testing with prime the voltage is between 1.050 and 1.055V however all the cores max out at 3.4Ghz

So Enhanced Turbo gives you better performance for multi core use but quite a bit higher voltage.

Anyways yeah voltage required seems to have an exponential increase after 4ghz, at 4.3 took me to +0.085 to get it booted and +0.130 to get it stable in testing.

Which puts me 4.3Ghz @ 1.2115V

So yeah 4Ghz happily does it at 1.0900 yet 4.3 needs 0.135V extra, ouch!

On a side note auto voltage wants to do 4Ghz @ 1.25V when I have it stable at 1.0900...ok MSI? tongue.gif
Edited by bleets - 2/4/16 at 1:20pm
post #2163 of 3145
Quote:
Originally Posted by om1kron View Post

You make it sound like that was not referenced previously in any of my posts, and also the manual does not explain or tell you to anticipate not being able to use open pci slots for devices or having it randomly assign an occupied slot and the one next to it that's not occupied pci lanes causing a device to lose lanes.


It also shows every configuration possible with the motherboard assuming that you're going to by default shove the first and most powerful graphics card in slot 1 and NEVER use slot 3. It encourages you to put something into 5 at the cost of reducing the speed of an m2 (if you choose to go that route.)

Soooo, after looking at all your recent posts I think we both either have 1) possessed boards or 2) manuals that are incorrectly written. Here is why:

My board got all Oprah too. But isntead of giving away x16 it was Evil Oprah and took them away. Check this out: My board is only configured (thanks to bleets) to only be able to run x16 on slots 1 and 3. Now, if I plug a card into 1 and 5, it trips the switch (listed in the manual albeit incorrectly graphed) to x8. Its set up to think if Ive occupied the slots all the way down to #5 then it figures I am occupying/utlizing the maximum amount of lanes. So, cards in 1 and 5 will force both to run at x8 AND it lists the cards in reverse order. My main card becomes GPU2 (in slot 1) and my slave card becomes GPU1. When I take both out and put just a single card in slot 1 and reboot, it still reads as x8. I have to turn the system completely off for about 30 seconds, take all cards out, reinstall the one in slot 1, turn master power on, power up, and then it reads as x16.

Hold on, this gets better....

After getting master card in slot 1 to read at x16, if I then add the second card into slot 3, 50% (about) of the time it reads at x8. But it reads the gpus correctly. Master card is GPU1, slave card GPU2. If I leave the card in the x16 #3 pcie slot and add a card to #5, thee card in slot 3 (previously slave card) stays at x16 and the card in #5 reads at x8. Thats it fairly normal. And once again, the master GPU is reading as GPU2. This means that slots 1-3 are part of one section and slots 4,5, and 6 are part of another section. This is what causes the "switch" to trip. But...

Lets say I put cards in slots 1,3, and 5. They should read x16, x16, x8...right? Thats what the manual shows.

It doesnt. It knocks them all down to x8. The middle card, in slot 3, becomes GPU1. The card in slot 1 becomes GPU3 and the card in slot 5 becomes GPU2. What?

Now, if I take the card out of slot 5 (the one that is supposedly flipping the switch) this should put me back to slots 1 and 3 at x16 and x16, right? It should. We are just returning to the original configuration that worked for both at x16. It reads as x8 for both cards and they are reading out of order again. I need to power down, break power from the wall, let it sit for 30 seconds, and then reinstall the cards. It took two reboots to get the second card to register as x16

Evil Oprah
post #2164 of 3145
Quote:
Originally Posted by bleets View Post

Interesting...i found out why my stock voltage was so high. MSI Enhanced Turbo.
With that running all cores turbo up to 3.6Ghz but at what I consider to be a hefty voltage increase to 1.0900 (or more depending on overclock).

Disabling enhanced turbo when stress testing with prime the voltage is between 1.050 and 1.055V however all the cores max out at 3.4Ghz

So Enhanced Turbo gives you better performance for multi core use but quite a bit higher voltage.

Anyways yeah voltage required seems to have an exponential increase after 4ghz, at 4.3 took me to +0.085 to get it booted and +0.130 to get it stable in testing.

Which puts me 4.3Ghz @ 1.2115V

So yeah 4Ghz happily does it at 1.0900 yet 4.3 needs 0.135V extra, ouch!

On a side note auto voltage wants to do 4Ghz @ 1.25V when I have it stable at 1.0900...ok MSI? tongue.gif

Yeah, the MSI auto oc things are always way high on voltages. If I use the tame oc switch on the board and use OC genie it puts the chip at 3.9 with more voltage than it took me to hit 4.3 when I was first starting to oc this chip. (before everything stopped working). I think they just do that to make sure that everyones system will run when the use thee auto features

Im still trying to figure out what the OC Genie does to the gpu? In the Command Center it says it will also overclock the gpu. But i, being curious, tried OC Genie and didnt notice any changes to the gpu (unless it changed the pcie freq or bus or something)

BTW, my last 8350 AMD chip did stock voltage all the way from 4ghz up to 4.7ghz. After that it needed ridiculous amounts of voltage for each step up on the freq. It wasnt worth it. I ran it at 4.7 daily and left it alone
post #2165 of 3145
Quote:
Originally Posted by DMatthewStewart View Post

Lets say I put cards in slots 1,3, and 5. They should read x16, x16, x8...right? Thats what the manual shows.

I am a firm believer that the manual was drafted for bios 1.0 which throws those lane switches and assignments right out of the window when flashed to 1.1 and even different when flashed to 1.2. I almost want to try a beta bios to see if that helps with the lane assignments but I have no idea how to flash back a version. The motherboard prevented me from reverting back to 1.1 after installing 1.2.

mad.gif
Quote:
Now, if I take the card out of slot 5 (the one that is supposedly flipping the switch) this should put me back to slots 1 and 3 at x16 and x16, right? It should. We are just returning to the original configuration that worked for both at x16. It reads as x8 for both cards and they are reading out of order again. I need to power down, break power from the wall, let it sit for 30 seconds, and then reinstall the cards. It took two reboots to get the second card to register as x16

What else is plugged into your system? sata drives and what ports do you have the sata drives in? Don't sata drives share pci lanes? I know if you plug anything into the 5th slot period you're going to have a bad time. If your sata drives are plugged into the sata ports that get disabled when running an m2, then I imagine you're running into the issue with lane 5 if anything is plugged into the sata ports. Try disconnecting all of your hard drives and see what the bios says with only your memory and cards added in.
post #2166 of 3145
Well that's annoying, after I had it stable @ 4Ghz @ +0.001V I went to 4.3 @ +0.135V which I wasn't happy with so took it back down to 4Ghz @ +0.001 and now that won't work with a BSOD 0x00000124 :/ Ho hum.

It's annoying that it was fine before and makes my paranoid self wonder if the extra voltage to 4.3Ghz has in some way affected the CPU.
post #2167 of 3145
Quote:
Originally Posted by bleets View Post

Well that's annoying, after I had it stable @ 4Ghz @ +0.001V I went to 4.3 @ +0.135V which I wasn't happy with so took it back down to 4Ghz @ +0.001 and now that won't work with a BSOD 0x00000124 :/ Ho hum.

It's annoying that it was fine before and makes my paranoid self wonder if the extra voltage to 4.3Ghz has in some way affected the CPU.

No your chip is fine, you didn't hurt anything.
Panty Dropper
(16 items)
 
I like Boobs
(20 items)
 
VagDestroyer
(19 items)
 
CPUMotherboardGraphicsRAM
1920x Asus Zenith x399 EVGA 1080TI SC Corsair LED 3200 
Hard DriveHard DriveHard DriveCooling
Samsung 960 evo NVMe 500 gig Samsung 960 evo NVMe 500 gig Samsung 960 evo NVMe 500 gig Full custom Wc 
OSMonitorKeyboardPower
Win 10 pro Acer x34 predator Corsair k65 LUX EVGA 1000 G3 
CaseMouseMouse PadAudio
Thermaltake X9 Roccat Leadr Roccat aliminac Creative ZxR 
CPUMotherboardGraphicsGraphics
5960x MSI X99 Xpower AC EVGA 980TI SC EVGA 980TI SC 
RAMHard DriveHard DriveHard Drive
Corsair LPX DDR4 2800 Crucial M550 512gb Samsung M.2 XP941 256gb Samsung 840 pro 512gb 
Hard DriveHard DriveOptical DriveCooling
Kingston hyper X 240gb WD Red 2TB x 2 LG blu-ray burner Custom Water cooling 
OSMonitorKeyboardPower
Windows 8.1 Acer x34 Predator Roccat Ryos Pro TKL Corsaix AX1200 
CaseMouseMouse PadAudio
Case Labs STH10 Roccat Kone XTD Roccat Alumic double sided Creative ZxR 
CPUMotherboardGraphicsRAM
Intel 6850K @ 4.5Mhz MSI x99A Carbon Pro EVGA 1080 SC Corsair Vengeance LED (red) 3200 - 32GB 
Hard DriveHard DriveHard DriveCooling
Samsung 960 256GB Evo M2 Samsung 950 evo 500GB x 2, Steam and Origin drives Muskin Reactor 1TB x 4 EK Predator 360 with EK GPU full block 
OSMonitorKeyboardPower
Win 10 64x Acer Predator 34x 100hz Roccat TKL pro EVGA G3 1000 
CaseMouseMouse PadAudio
Be Quiet! 900 pro Roccat Tyon Roccat Aluminac Creative ZxR 
AudioAudioOther
Monoprice 8" studio monitors Senn HD800 pros HTC Vive VR 
  hide details  
Reply
Panty Dropper
(16 items)
 
I like Boobs
(20 items)
 
VagDestroyer
(19 items)
 
CPUMotherboardGraphicsRAM
1920x Asus Zenith x399 EVGA 1080TI SC Corsair LED 3200 
Hard DriveHard DriveHard DriveCooling
Samsung 960 evo NVMe 500 gig Samsung 960 evo NVMe 500 gig Samsung 960 evo NVMe 500 gig Full custom Wc 
OSMonitorKeyboardPower
Win 10 pro Acer x34 predator Corsair k65 LUX EVGA 1000 G3 
CaseMouseMouse PadAudio
Thermaltake X9 Roccat Leadr Roccat aliminac Creative ZxR 
CPUMotherboardGraphicsGraphics
5960x MSI X99 Xpower AC EVGA 980TI SC EVGA 980TI SC 
RAMHard DriveHard DriveHard Drive
Corsair LPX DDR4 2800 Crucial M550 512gb Samsung M.2 XP941 256gb Samsung 840 pro 512gb 
Hard DriveHard DriveOptical DriveCooling
Kingston hyper X 240gb WD Red 2TB x 2 LG blu-ray burner Custom Water cooling 
OSMonitorKeyboardPower
Windows 8.1 Acer x34 Predator Roccat Ryos Pro TKL Corsaix AX1200 
CaseMouseMouse PadAudio
Case Labs STH10 Roccat Kone XTD Roccat Alumic double sided Creative ZxR 
CPUMotherboardGraphicsRAM
Intel 6850K @ 4.5Mhz MSI x99A Carbon Pro EVGA 1080 SC Corsair Vengeance LED (red) 3200 - 32GB 
Hard DriveHard DriveHard DriveCooling
Samsung 960 256GB Evo M2 Samsung 950 evo 500GB x 2, Steam and Origin drives Muskin Reactor 1TB x 4 EK Predator 360 with EK GPU full block 
OSMonitorKeyboardPower
Win 10 64x Acer Predator 34x 100hz Roccat TKL pro EVGA G3 1000 
CaseMouseMouse PadAudio
Be Quiet! 900 pro Roccat Tyon Roccat Aluminac Creative ZxR 
AudioAudioOther
Monoprice 8" studio monitors Senn HD800 pros HTC Vive VR 
  hide details  
Reply
post #2168 of 3145
Quote:
Originally Posted by TheGovernment View Post

No your chip is fine, you didn't hurt anything.

Confusing then since it was very happy running at 4Ghz @ +0.001 for 6 hours on aida64 and prime but after taking it higher with more voltage it starts acting up :/
post #2169 of 3145
Hmm...
It looks like when it was failing @ 4ghz the CPU VID was running at 1.8510V for some reason.

I defaulted and reflashed my BIOS and it's back stress testing @ 1.0900V again

So right Aida64 stability test, what are the best options to test an overclock?
I've typically only ever used prime smile.gif

*edit*
OK the reason it seemed to be stable with 4Ghz @ +0.001V was because of Aida64.
Doing the stability test there...caused no issues. It's when I switched to Prime95 (an old version - 27.9) that it started getting BSODs.

To get it stable with prime95 I need +0.020V frown.gif
Edited by bleets - 2/4/16 at 5:30pm
post #2170 of 3145
Question: Can an unstable overclock cause issues with time keeping/clocks on a system?
While my 4Ghz @ +0.20V was fine stress testing it gave me headaches trying to get it working with 3D Mark Firestrike as it kept reporting an invalid result due to time mismatch since 3D Mark compares time from 3 sources...doesn't give any details on what the mismatch actually is though.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel Motherboards
Overclock.net › Forums › Intel › Intel Motherboards › [Official] MSI X99S/A Motherboard Owners Club