[Hurtz0]

Yay! HardStroke. Homie. Still with us? I made a little breakthrough with my VISION solution, wanted to share. I finally got around to disabling BCLK Adaptive Voltage. That setting is evil, and I have exorcised it from my beautiful new BIOS chip.

After turning that off, I was able to go back and knock 85mV off my Vcore, and 15mV off my Vsa. This passed my 30min. P95 test without breaking above 83C at 50x/47x. Then, I went back again and took back these voltage gains, and got 51x/47x past my ballpark testing. Here's the data:

10600k on GIGABYTE Z490 VISION G

Project Notes	Cancelled due to high voltage.	Current project.	For when I'm feelin' frisky.
Speed	50x/47x	50x/47x	51x/47x
Vcore	1.285	1.210	1.272
BCLK Adaptive V	Enabled	Disabled	Disabled
Vio	Auto	1.240	Auto
Vsa	1.360	1.345	1.365
Vaxg		1.210
Vcore realtime max	1.488	1.416	1.476
Vsa r.t. max		1.380
Vaxg r.t. max		1.248

2/5 Table Update- Added notes, realtime Vsa & Vaxg, and my best fixed voltages for the 50x/47x solution. Current project passes 1hr P95 Large FFT concurrent with 2 streaming video apps on 2 monitors.

Rest of my settings haven't changed, except that I fixed my GPU multiplier to 12.5

Is this a normal thing in Windows task manager reporting wrong CPU speed?

Yeah, my last CPU was i7-980X, and Win Task Mgr never got the clock right. But since I've upgraded to 10600K, Win Task Mgr seems to track one of the core clocks, but not the avg/min/max. It gets the high and low end readings right.

You both seem like overclocking newbies, normalising pumping crazy high voltages by chatting to each other about these silly numbers.

Fray_bentos, if it's all the same to you, I'd prefer if we could restrict our conversation to the technology. If you tell me I'm wrong, that's cool. If you tell me that and give me detail or point me to reference, that's appreciated. I thank you when you do that. But if you want to make judgements about my experience (which you know nothing about), my knowledge and my attitude... I don't think the forum wants to read that crap. I don't think forum moderators want that crap choking up their website, either. So let's stick to the tech, OK? Thanks.

 

[fray_bentos]

Is this a normal thing in Windows task manager reporting wrong CPU speed?
it always reports at least 1.1ghz higher than actual. View attachment 2477224

I think it is known to be inaccurate. What load are you running? That said, mine correctly reports 4.9 GHz, matching my clock speed.

 

[fray_bentos]

If you tell me that and give me detail or point me to reference, that's appreciated.

Both myself and sixty9sublime pointed you to buildzoid's videos, which would clear up your misunderstandings. However, your response was to then say "I look forward to seeing credible details"; that isn't the appropriate response based on the technology and science. I've had my say on running chips loaded with excessive Vcore, VCCIO and VCCSA, so I'm bowing out of further interactions on the matter.

 

[HardStroke]

Is this a normal thing in Windows task manager reporting wrong CPU speed?
it always reports at least 1.1ghz higher than actual. View attachment 2477224

yes
task manager is on some mad crap
my old I5 3550 was overclocked to 4.1GHz from 3.7GHz and it was showing 4.7GHz and even 5.6GHz
best apps are HWiNFO64 and CPU-Z
Intel XTU is also pretty good, you can overclock from windows with this app

 

[CoachBombay]

Thanks everyone for your guidance! I figure it's time to share my findings now that I have something stable to report back on. I recently upgraded from an og i7 920 after getting tired of how poorly it ran Win10. I guess ten generations of intel chips was a good enough run lol. I was pumped to find out that I could reuse my H50 Corsair heatsink and 750watt Thermaltake power supply from '09. I decided to upgrade as cheaply as possible since my aim is a conservative, safe overclock and apparently I'll need to set aside over a grand for any shot at an RTX card... so I went with an MSI Z490-A Pro board. I'm very pleased with it other than the fact that it pumps too much wattage and voltage into the chip on default settings and requires some tinkering as others have mentioned. OCCT showed my temps spiking with ~183watts being used, capping at 175w brought that heat down nicely.

I ended up having to do a lot more tinkering than I wanted to because it took a while to stop seeing Kernel Power errors and other sporadic issues from undervolting the chip. These issues were occurring maybe once a day, which made it difficult to realize it was an unstable CPU issue. I nearly replaced my ancient PSU before finally reaching a stable 4.9ghz 47 ring / cache oc at 1.36v adaptive voltage (1.34v max I've seen in OCCT).

I didn't have any luck tinkering with the LLC settings, I saw crashes and Kernel errors so I've left that on auto. I also have my 32gb 3600mhz patriot viper ram on the stock XMP 1 setting... since one of the A Pro's shortcomings is a lack of bios reset button and XMP 2 is greyed out. I'm not sure that I'd see much of a real world difference moving away from XMP 1 anyway?

I'm pretty confident my chip could do 5.0ghz stable if I was willing to add voltage and run it hotter but I don't see any reason to currently when 4.9ghz absolutely crushes any game I throw at it 1440p 144hz and the chip runs at an icy cold max of 72c, worlds different from what I was seeing on the i7 920.

Here's the conservative overclock I've ended up with in pictures... hopefully it acts as a good starting template for any future 10600k ocers.

 

[Nexon1338]

i can do 5.1 200% stable but not more whit ht on atleast can hit 5.4 ht off i 5.2 i can game bit i like it to 200%stable no mather wath;D is failing cbr23 instant whit an error someone know anything about this issu?

 

[Nexon1338]

i run my g skillz tritend z rgb 3200cl 16-18-18-38 at 3600 18-20-20-36 630 clocks 2T
10600k 5.1 allcore vcore idel 1.394 1.332 under load LLC 4 migth be some more adjusting on this but it run super smoth at the time so;D

 

[Nexon1338]

Thanks everyone for your guidance! I figure it's time to share my findings now that I have something stable to report back on. I recently upgraded from an og i7 920 after getting tired of how poorly it ran Win10. I guess ten generations of intel chips was a good enough run lol. I was pumped to find out that I could reuse my H50 Corsair heatsink and 750watt Thermaltake power supply from '09. I decided to upgrade as cheaply as possible since my aim is a conservative, safe overclock and apparently I'll need to set aside over a grand for any shot at an RTX card... so I went with an MSI Z490-A Pro board. I'm very pleased with it other than the fact that it pumps too much wattage and voltage into the chip on default settings and requires some tinkering as others have mentioned. OCCT showed my temps spiking with ~183watts being used, capping at 175w brought that heat down nicely.

I ended up having to do a lot more tinkering than I wanted to because it took a while to stop seeing Kernel Power errors and other sporadic issues from undervolting the chip. These issues were occurring maybe once a day, which made it difficult to realize it was an unstable CPU issue. I nearly replaced my ancient PSU before finally reaching a stable 4.9ghz 47 ring / cache oc at 1.36v adaptive voltage (1.34v max I've seen in OCCT).

I didn't have any luck tinkering with the LLC settings, I saw crashes and Kernel errors so I've left that on auto. I also have my 32gb 3600mhz patriot viper ram on the stock XMP 1 setting... since one of the A Pro's shortcomings is a lack of bios reset button and XMP 2 is greyed out. I'm not sure that I'd see much of a real world difference moving away from XMP 1 anyway?

I'm pretty confident my chip could do 5.0ghz stable if I was willing to add voltage and run it hotter but I don't see any reason to currently when 4.9ghz absolutely crushes any game I throw at it 1440p 144hz and the chip runs at an icy cold max of 72c, worlds different from what I was seeing on the i7 920.

Here's the conservative overclock I've ended up with in pictures... hopefully it acts as a good starting template for any future 10600k ocers.

View attachment 2478626 View attachment 2478627 View attachment 2478628

wath ram kit are u using? gskillz trident z rgb maby? try timings 18-20-20-36 insted maby not a big difrent but sligthly faster if it post i run my gskillz trident z rgb 3200 at this timings and 3600mhz non xmp i recomend settingas mutch as u can manual in bios tvb shuld be disabled also oratleast try disable it wen u do manual overclocking and se wath the difrence is i see so menny things on this biospic that isent as it shuld long duration 25 sec max out power limits dude^^

 

[acrvr]

Turned out that cooling is what's preventing me to hit 5.1 all core.

I had been running at 5GHz for months and couldn't really get 5.1 all core stable. Then I changed the cooler from an EK AIO 360 to a NZXT X63 because I was moving to a Cooler Master NR200 case. Previous case was Lian Li 011D Mini and before that Phanteks P500A. I may have had a not so good copy of the EK AIO 360 because contrary to reviews, I was getting cooler temps with the Kraken X63.

With the new X63 AIO, I gave 5.1GHz another run and this time it's fully stable through many runs of OCCT. Voltage was 1.37V under load LLC7. A 30 minutes run OCCT AVX2 Large Threads I was getting average temps of high 70C on all individual cores, Peak temps of 85-90C on individual cores. I don't really need 5.1GHz daily but it's nice to finally be able to hit that number.

The radiator is mounted on the side in the NR200, and this frees up the top panel for some Arctic P12 intake fans. With the fans blowing cool air directly at the RAM, I was also getting better RAM OC of 4300MHz C17 1.47V and 4400MHz C17 1.52V stable through anta777 TM5 runs.

Didn't expect this case to perform so well but it just happened to work out quite nicely.

 

[Martin v r]

locked on all to
5GHZ at 1.355vcore rest is on auto
have not tried more in now, come tendons, cooler with an RX360 XSPC / xspc waterblock
is not quite done with my pc coming a pci-e riser cable today .
PC MSI MAG Z490 TOMAHAWKIntel Core i5-10600K Processor HyperX Fury RGB DDR4 3600MHz 16GB cl 17 1000W Sinlet Pro ROG STRIX-GTX1070-O8G-GAMING

 

[Hurtz0]

Cool, looks like we have some more good 50x solutions. I have not reached my final testing stage yet, but I have 2 profiles saved for CPU/Ring/AVXO/GPU = 50x/47x/0x/14.0x. One is in Override mode, the other in Adaptive Single Offset, and both operate under 1350mV Loaded Vcore. I'm pretty sure I've confirmed 50x on my CPU though.

GIGABYTE LLCs and HWiNFO's Voltage Readings
I wanted to see what load voltages I got under the 7 different LLC settings with all else equal for this board. But I've had to stop using the Vcore reading in HWiNFO, because it only has 12mV resolution and that's not good enough for fine-tuning.

Other folks suggest that VR Out is a better HWiNFO reading for CPU Vcore. According to HWiNFO, Vcore comes from an ITE IT8688E chip, and the VR Out comes from the ISL69269 chip. I looked these up, and the ITE chip is a "SuperIO" chip which provides thermal, fan, voltage sensors; while the ISL chip is the actual VRM controller chip. Therefore I think VR Out is a more direct measure of CPU Vcore than Vcore (HWiNFO reading). Could VRM controller be reporting die-sense and SuperIO chip reporting socket-sense?

So I did my LLC test with all 3 voltage readings in HWiNFO: VID, Vcore, and VR Out. I got these results (pardon the crude notes... too lazy to build a spreadsheet):

The chart shows the 7 rows of test data to the left. Y-axis runs from 1340-1440mV, and the horizontal lines are 25mV apart. X-axis is LLC setting. + = Vcore; o = VR Out; * = VID.

Test Procedure: Each result obtained by running P95 Large FFT AVX/on for 15 min. HWiNFO counters are reset right after P95 starts, and readings taken just before P95 is stopped. The maximum values are used for all voltage readings (same with Pkg Temp T and Pkg Power P).

So I verified that VROut tracks Vcore linearly, and I was amazed at how closely 6 of the 7 VR Out values lined up perfectly. It was also cool to note that while VID rises only a few mV across all 7 LLCs; Vcore (and VROut) rise approx. 50 mV.

Also interesting is that VROut = VID (what the CPU suggests from it's internal V/Load curve) at LLC = 3. LLC 3 was recommended by one of the clocking video authors which HardStroke shared earlier, who was also on GIGABYTE. But then LLC2 is attractive for being the setting which provides just 5mV over VID. I must investigate these 2 LLC settings further.

 

[Redlurkeraite]

10600K
5.3GHz (5.4GHz boots but instantly crashes in OCCT with voltage up to 1.5v)
AVX Offset -1
Ring Ratio 4.7GHz (No amount of voltage allowed for 4.8GHz )
LLC 3 1.4V

The CPU was lapped briefly for about 30 minutes.
Liquid metal applied, watercooled with a XD3 pump & XPX Block on a 360mm (25mm) push pull config.

OCCT ~ High 60 degrees Celsius
Gaming ~ Mid 50s degrees Celsius
Idle ~ Low 30s degrees Celsius

Ram, running in dual channel 2x16gb 3200cl14
4400Mhz 16-17-17-35
~ 37.8ns

Mobo - Z490 MPG Gaming PLUS

Any suggestions on how to achieve a higher ring ratio?

 

[fray_bentos]

Ram, running in dual channel 2x16gb 3200cl14
4400Mhz 16-17-17-35
~ 37.8ns

What are your VCCSA and VCCIO voltages?

 

[Redlurkeraite]

What are your VCCSA and VCCIO voltages?

1.3V both, I'm going to see if I can lower them and still have it run stable.

 

[fray_bentos]

1.3V both, I'm going to see if I can lower them and still have it run stable.

Yeah, I'm curious how low you can go whilst keeping those nice memory speeds.