Who's getting/got a 8086k?

[porina]

3 minutes ago, Christophe Corazza said:

So that means it's over and out in terms of increasing the clock speed? 

I've been concentrating on bagging easy runs to submit to hwbot for now. I need to play more with fixed voltage so I can really work out what's going on, and if I have more headroom. So I wouldn't rule out 5.3 but it looks like 5.2 is the easy limit for my sample after delid.

[Christophe Corazza]

5 minutes ago, porina said:

I've been concentrating on bagging easy runs to submit to hwbot for now. I need to play more with fixed voltage so I can really work out what's going on, and if I have more headroom. So I wouldn't rule out 5.3 but it looks like 5.2 is the easy limit for my sample after delid.

 

So your next step will be...?? 

[porina]

Just now, Christophe Corazza said:

So your next step will be...?? 

Going to bed, as it is getting late in my time zone. Tomorrow after work I dunno, I might start working on fixed voltages, or I might continue with the 5.2 bench submissions. Will see how I feel at the time. Also for now I'm running my ram at XMP defaults. I know I can squeeze more out of it, but I didn't want to complicate stability by doing it at the moment.

[Christophe Corazza]

1 minute ago, porina said:

Going to bed, as it is getting late in my time zone.

 

I'm next to your time zone... 00:22 

 

2 minutes ago, porina said:

Tomorrow after work I dunno, I might start working on fixed voltages, or I might continue with the 5.2 bench submissions. Will see how I feel at the time. Also for now I'm running my ram at XMP defaults. I know I can squeeze more out of it, but I didn't want to complicate stability by doing it at the moment.

 

Keep the updates coming 

[DarkSmith2]

28 minutes ago, porina said:

I did the delid and LM application. Before and after, Prime95 12x64k FFT core temps went from 76-79 to around mid 60's. So... around 10C+ drop there.

 

5.2 seems bench stable at +150mV offset in XTU. I didn't get a crash during the bench session under that condition. I still need to work out what actual voltage the mobo is supplying and fix it, as voltage monitoring software seems erratic. I see reported peaks of 1.48v which is on the high side, but under load it settles to around 1.36v. I've done a small selection of benches on hwbot, but my XTU score sucks compared to others (~2100 at 5.2) so I haven't uploaded it yet. Can't figure out why...

 

Can I push clock further?... not really. 5.3 was not stable enough to complete Cinebench R15 even with +250mV offset.

Oh hmn, i can.

 

But well, i also have a much different board and do overclocking via manual voltage. I think our CPUs are pretty similar from the things you wrote.

If i would have to guess i would say the CPU isnt the limiting factor, leaning more towards your Motherboard.

[porina]

5 minutes ago, DarkSmith2 said:

But well, i also have a much different board and do overclocking via manual voltage. I think our CPUs are pretty similar from the things you wrote.

What cooling - AIO according to your profile? I got up to 1728 at 5.2, so 1750 at 5.3 seems slightly low but within run to run variations and your ram is faster than I'm currently running. Do you repeat runs to try to get a better score?

[DarkSmith2]

10 minutes ago, porina said:

What cooling - AIO according to your profile? I got up to 1728 at 5.2, so 1750 at 5.3 seems slightly low but within run to run variations and your ram is faster than I'm currently running. Do you repeat runs to try to get a better score?

Yes i did repeat that run until it actually crashed because that wasnt completely stable either with only 1.35volts. But the outcome wasnt much different, i ran it like 5-6 times. The difference from 5.2 to 5.3 in cinebench score was only about ~25points, was also pretty disappointed with the score, i guess you would have to crank up RAM and Cache just to not get bandwidth limited to get higher scores at this size of frequency. 

 

 

Edit: Im using a 360mm AiO. Alphacool Eisbaer LT360, with Noctua Black.swap fans.

[Kalm_Traveler]

Mine finally arrives tomorrow, hoping that the radiator shows up too so that I can finish the build and do some benchmarks with the Titan V. 

 

Anyone else have a custom loop build for one of these? 

[porina]

Spent some time looking at voltage on my mobo, the Asrock Z370 Pro4. This is a low end board, bought at the time as it was the cheapest one that wasn't MSI, I'm not going there again after bad experience with their Z170 boards.

 

I started by repeating what I already did, and monitored the voltage in CPU-Z, hwinfo64, and A-tune (Asrock's own hardware utility). They all reported the same voltage. hwinfo64 reacted fastest, CPU-Z took a little longer, and A-tune looks like they put some averaging in the software, presumably to make it not jump around as much, but it took forever to get to the voltage the other two were already at.

 

Using Cinebench R15 as the load, I got around 1.2v with CPU at stock. Using XTU to add +0.10 offset, it wasn't a surprise to get around 1.31v. Overclocking to 5000, it was now showing 1.36v, so about 0.05v extra from the extra clock. Any predictions on what happens if I set +0.15 offset? Wrong, it went down to 1.34v. Going up to 5.2 still at +0.15 offset, it stayed at 1.34v.

 

Fixing voltage in bios to 1.35v, I had a choice of load line settings. It defaulted to level 1 when I set the fixed voltage. The little graphic suggested this was the highest correction. Back in Windows, stock showed 1.344v idle, 1.312v under load. Back in bios, set load line level 3, back to Windows. Same idle, but load dropped to 1.248. Ok, graphic was correct, level 1 was highest and I set that back. While there, I set the voltage to 1.395. Why not 1.400? It showed that in red. Must be bad

 

Back to Windows. Idle voltage was static at 1.392v regardless of CPU clock. Load at stock clock (4.3 all core) was 1.360. At 5.0 it was 1.344, as it was at 5.2. Still at 5.2, I applied an extra +0.05 offset, but this didn't change the reported voltage.

 

The question now was, is it actually changing the voltage, or is there some limit going on? hwinfo64 also reports the package power, maybe that could be used to indirectly see what's happening. With the 1.395v set in bios, it drew 157W in Cinebench. Using XTU to override the voltage to 1.40, I got 164W, and at 1.42v I got 166W. Power is going up with voltage, although not necessarily linearly. If you look at Anandtech's OC, they also had a similar pattern of reported voltage not going up, yet power was.

 

Fine, let's do a suicide run. I reset XTU to use bios voltage, and set +0.1v, giving a nominal level of 1.495v. I set 5.3, started Cinebench and... BSOD. I guess it isn't happening.

 

I'm calling it a night here. 5.3 still seems to be a step too far, and I have a known working 5.2 setting. Back to benching on the weekend.

[DarkSmith2]

33 minutes ago, porina said:

Fine, let's do a suicide run. I reset XTU to use bios voltage, and set +0.1v, giving a nominal level of 1.495v. I set 5.3, started Cinebench and... BSOD. I guess it isn't happening.

yea, i also think it isnt happening with this board. Do you consider upgrading it after Z390 release?

[porina]

12 minutes ago, DarkSmith2 said:

yea, i also think it isnt happening with this board. Do you consider upgrading it after Z390 release?

I have a spare 8350k which I originally bought with this mobo. Getting another mobo to allow both to be used is an option, but I'm not in a rush. I could just get low end for the 8350k, or go higher end and see if that helps with the OC. I also noticed in bios there weren't any voltage control options that I have on my Z170 VIII Hero. All sorts of things I don't really understand like switching frequencies and some other stuff... so if I got a similar level board for Z3xx that could be interesting.

 

Thing is, if I wait long enough for Z390 and the long rumoured 8-core, I'd probably will be playing with that CPU then too.

[Christophe Corazza]

1 minute ago, porina said:

Thing is, if I wait long enough for Z390 and the long rumoured 8-core, I'd probably will be playing with that CPU then too.

 

You're planning to benchmark/OC that too?

[porina]

Just now, Christophe Corazza said:

You're planning to benchmark that too?

I'll see how rich I feel at the time

 

I keep saying I need to sell older systems to make space but I never actually do it... still got an assortment of Intel dual core systems, older quad cores, even 6-core Ryzen seems too slow now but I'll keep the 8 until I replace that. On the dual cores, I might have waited a bit too late to shift them... who's going to buy it now for any amount of money actually worth the time to do? Maybe I should give them away to console peasant friends who at most PC game on non-gaming laptops.

[DarkSmith2]

13 minutes ago, porina said:

Thing is, if I wait long enough for Z390 and the long rumoured 8-core, I'd probably will be playing with that CPU then too.

Sounds like a decent plan Having opportunities is always a good thing. I would definitely go with something not limiting, if it goes in the same manner with z390 as with z370 a taichi or hero atleast. Apex if you really take it serious ..

[Christophe Corazza]

1 hour ago, porina said:

I keep saying I need to sell older systems to make space but I never actually do it...

 

Although I don't have heaps of CPUs like you, I've never sold any of my hardware either.

I keep older computers/laptops/hardware components around  just so I can tinker with it 

(and also a bit for nostalgia's sake )

[Christophe Corazza]

10 minutes ago, porina said:

Maybe I should give them away to console peasant friends who at most PC game on non-gaming laptops.

 

Luring them to the magnificent world of desktop gaming 

[DarkSmith2]

1 minute ago, Christophe Corazza said:

 

Although I don't have heaps of CPUs like your, I've never sold any of my hardware either.

I keep older computers/laptops/hardware components around  just so I can tinker with it 

(and also a bit for nostalgia's sake )

everytime i upgrade i pass on my old hardware to someone needy.

[Christophe Corazza]

5 minutes ago, DarkSmith2 said:

everytime i upgrade i pass on my old hardware to someone needy.

 

How often do you upgrade?

[Kalm_Traveler]

well I was excited to start benchmarks myself this weekend since everything arrived.

 

However.... Aquatuning sent me a reservoir instead of the radiator on the invoice they included in the box. Going to order from a company that can hopefully ship out the correct item and if I'm lucky maybe they'll overnight it so I can get it Saturday at least.

[Rambo]

 

Just got mine the other day. Sitting on the desk right now waiting for a new motherboard. 

For some reason my 3930k system has started BSODing, maybe it feels it's gonna be replaced soon

[DarkSmith2]

9 hours ago, Christophe Corazza said:

 

How often do you upgrade?

admittedly not very often , but seeing the 8086k Box on the pic above this post... hard to withstand ... no, no, i need to safe up for GPU.. *WUSA*

 

[porina]

Just had a look on hwbot, and more entries are starting to appear for the 8086k now. There's some water cooled submissions above 5.3 GHz... don't make me look to upgrade cooling... maybe the mobo as well, but I'd need a better PSU... and so it goes on  

[Christophe Corazza]

5 hours ago, Rambo said:

For some reason my 3930k system has started BSODing, maybe it feels it's gonna be replaced soon

 

It knows mate, it knows... 

[Christophe Corazza]

1 hour ago, DarkSmith2 said:

admittedly not very often , but seeing the 8086k Box on the pic above this post... hard to withstand ... no, no, i need to safe up for GPU.. *WUSA*

 

Overclocking CPUs seems to be a very expensive hobby 

I mean... look at @porina... simply to overclock only 1 CPU 

 

1 hour ago, porina said:

don't make me look to upgrade cooling... maybe the mobo as well, but I'd need a better PSU

 

[DarkSmith2]

1 minute ago, Christophe Corazza said:

Overclocking CPUs seems to be a very expensive hobby 

I mean... look at @porina... simply to overclock only 1 CPU 

and he also forgot potentially a new RAM kit that clocks higher