ASRock Z87 Extreme6 OC issues (UEFI 2.10)

[scorpian007]

Hey all!

I'm new to this forum so be gentle :)

I've been tearing my hair out the last few days trying to work this out and I have no idea what is going on. I'm by no means an expert overclocker but I wouldn't call myself a novice either. I've managed to overclock my 4770K to a modest 4.3GHz at a fixed voltage of 1.19v and temperatures under load in OCCT are acceptable (just under 80 degrees on my aging H50 cooler). When the CPU downclocks to 800MHz in Windows, the voltage doesn't drop, just stays at 1.19v so I've enabled adaptive mode in the UEFI hoping to use the power saving features that Haswell offers, and while it does lower the voltage when idle to around 0.72v, the voltage jumps up to a whopping 1.27v in OCCT which causes the CPU to overheat very quickly and the test to fail. I've tried manually setting up an adaptive voltage but no matter what I set it to, it makes no difference.

Has anyone had any issues with this? I'm wondering if I'm doing anything wrong or if the UEFI is just broken?

Thanks all

[Shiari]

Re: ASRock Z87 Extreme6 OC issues (UEFI 2.10)

You may not have the best overclocker by the sound of it, though it could be related to your cooler as well. Haswell really does get hot when pushed (why I got one of the best air coolers possible).

In my experience, for adaptive voltage you want to set the base voltage to about 1v, then the offset to 190mv if 1.19v is where you like to end up at. However, keep in mind that what you have set is not necessarily what the CPU will get, benchmarks and stress tests particularly may demand and get a higher voltage. The idle voltage of 0.76v does sound about right though. For me, benchmarks can push it to about 1.325v ... with 1.257 set as the total adaptive voltage, so much higher than I've asked for.

All that said, don't worry too much about benchmarks ... they're useful for helping to get to a "stable" overclock, but unless you're running benchmarks all day long you won't be seeing those temperatures in day to day use. Best test stability with software you'll actually use. I've found some Handbrake encoding of a full length movie to be a pretty good stability test, along with playing games.

You should also do a search on this forum, we've had a lot of discussion about Haswell overclocking on the Extreme 6 already.

[scorpian007]

Re: ASRock Z87 Extreme6 OC issues (UEFI 2.10)

I had a look at some of the previous threads but didn't really see anything that matched my issue.

My CPU doesn't OC too well, I could probably get 4.5GHz @ 1.27v but it overheats under OCCT really quickly. I think my cooler just sucks though, might have to invest in a H100i.

It seems like the adaptive voltage pushes it way too high even when I set a manual voltage. It's almost as if it's completely ignoring the voltage that I set in the adaptive settings and giving the CPU what it 'thinks' it needs. Would be nice if you could set a maximum in the adaptive voltage and it stayed under it :)

[profJim]

Re: ASRock Z87 Extreme6 OC issues (UEFI 2.10)

Check out 3 Step Guide to Overclock Your i7 / i5 Haswell Platform | Overclockers in the Adaptive Voltage section:

Adaptive Voltage

Adaptive voltage is a lot smarter than your average offset voltage. What it does is raise the part of the voltage curve that you need under load, but leaves the lower end of the curve (when the CPU is operating at 800 MHz) alone. Thus, the higher voltage you need is there when the CPU calls for it, but you have no increase in idle voltage at all. This is going to be the best choice for letting your CPU breathe, as it were, when idle. You set the maximum voltage you want the CPU to get and go on your merry way knowing the CPU can draw the voltage it needs when it needs it.

A picture is worth a thousand words, and this slide outlines the voltage override modes quite well.
Various Integrated Voltage Regulator Configurations

[Shiari]

Re: ASRock Z87 Extreme6 OC issues (UEFI 2.10)

I'm no fan of the AIO water coolers. Not personally used any, but I'm reading a lot about them elsewhere. They're quite noisy compared to an air cooler with good fans, and when either of their two moving parts (fans, but particularly the pump) fails you have a serious problem and your CPU will kill itself. You also have a small risk of leaks. That said, air coolers have some issues too: not all ram fits well under them, and some are huge, which is a problem if you have a smaller case. But your CPU won't fry immediately if the fan(s) fail. Performance wise a good air cooler will pretty much match the performance of an AIO.

And again, are you going to be running OCCT all day? Sure, my 4770K @ 4.6 hits 85-95 degrees on stress tests as well, but it's stable and won't throttle (which ruins performance) which is really what matters. The chip can take that for a while for sure. Under normal load it might hit mid 70's or so.

For the adaptive voltage you really need to keep that base voltage high (around the 1v mark) or your voltage may shoot up way higher than you'd expect. Don't know why, but it's what we have noticed ... we've discussed that here before as well.

[scorpian007]

Re: ASRock Z87 Extreme6 OC issues (UEFI 2.10)

Thanks for the response profJim, looks like there is nothing I can do about it. I guess I wasn't doing anything wrong after all :)

I was actually looking at a Noctua NH-D14 because of it's amazing performance but there is no way it would fit with the Dominator Platinum RAM modules I have so I'm stuck with AIO solutions for the time being.

I guess I'm just being paranoid about worst case scenario. When I run Handbrake the voltage stabilises at 1.2v at 4.3GHz which is fine and temps hover around 60-65 degrees. I doubt games will push the CPU as hard as Handbrake or OCCT does so I don't think I need to worry in that case.

I'll muck around a little more with the adaptive voltage settings a little more. What would you suggest as a base and offset value? Should I leave the cache at auto? And should I just leave the input voltage at 1.9v?

Thanks for your help :)

[Shiari]

Re: ASRock Z87 Extreme6 OC issues (UEFI 2.10)

Since you should be using slots 2 and 4 anyway I think they reach just beyond the heatsink so the clearing shouldn't be a problem ... I can check that when I get home later. You would have to lift the fan, but that shouldn't be an issue I suppose ... the 650D should be plenty wide at least. There's also the option of removing the heatspreaders (or perhaps part of it, if possible), which doesn't affect the ram really.

As to the adaptive voltage settings, if you want 1.2v then try 1.0 + 0.2 for the settings. I haven't got cache at auto I believe, I've tweaked to go lower while remaining stable. Was some effort though.

Yeah, input voltage should remain at 1.9V ... it HAS to be at least vcore + 0.4V, and having it a little higher doesn't matter for your temps really.

[Shiari]

Re: ASRock Z87 Extreme6 OC issues (UEFI 2.10)

You can find my latest Intel XTU submitted 4.6 settings here: Shiari`s XTU score: 1073 marks with a Core i7 4770K (click on XTU settings) ... 1.09V / 136mV I used actually.

Oh and a little side story ... I was playing with settings to see whether I could get 4.7 stable without insane temperatures. I think everything was fine for several days, no crashes or anything, even when running Handbrake and benchmarks. Until a few minutes into a game of Civ 5 ... my 4.7 profiles are on hwbot.org as well. :)

[scorpian007]

Re: ASRock Z87 Extreme6 OC issues (UEFI 2.10)

I'll try it out tomorrow. My cache voltage is at 1.16v and it runs very stable, don't think changing it will improve things at all so I'll leave it as is.

Thanks for the help, I'll post back tomorrow to let you know how it goes! Bed time now :)

[Shiari]

Re: ASRock Z87 Extreme6 OC issues (UEFI 2.10)

Just a few quick snapshots, can't do much better than this without pulling the computer out entirely from under the desk. The NH-D14 is a beast, as you're no doubt aware:



With the front fan removed, you can just about see my RAM only just fits under the heatsink. I believe it would be possible to have the RAM in slot 2 stick out beyond the heatsink, would have to be forced slightly when it's taller than my RAM but that shouldn't really be a problem. My heatsink is also not entirely straight on it would seem, which doesn't help. :)



The NH-U14S could be an alternative if you're worried about the available space. Here's a review: Noctua NH-U12S and NH-U14S review - Article

[�code]

Re: ASRock Z87 Extreme6 OC issues (UEFI 2.10)

Check out 3 Step Guide to Overclock Your i7 / i5 Haswell Platform | Overclockers in the Adaptive Voltage section:

Hey thanks ProfJim, Adaptive Voltage looks very much like "additional turbo voltage" in the older chipsets.

[profJim]

Re: ASRock Z87 Extreme6 OC issues (UEFI 2.10)

You're welcome.
I wonder if using Adaptive Voltage and delidding a Haswell cpu plus Coollaboratory Liquid Pro (or Ultra) will still increase the maximum cpu voltage to ~1.36 volts while stress testing?

Coollaboratory*Liquid Pro - the first heat conduction paste of 100% metall
Coollaboratory*Liquid Ultra - top performance for High-End systems!

Deliding a 4770K (Haswell). Improving temperatures and maximizing overclockablity.

Google video search: delid haswell

[parsec]

Re: ASRock Z87 Extreme6 OC issues (UEFI 2.10)

I had a look at some of the previous threads but didn't really see anything that matched my issue.

My CPU doesn't OC too well, I could probably get 4.5GHz @ 1.27v but it overheats under OCCT really quickly. I think my cooler just sucks though, might have to invest in a H100i.

It seems like the adaptive voltage pushes it way too high even when I set a manual voltage. It's almost as if it's completely ignoring the voltage that I set in the adaptive settings and giving the CPU what it 'thinks' it needs. Would be nice if you could set a maximum in the adaptive voltage and it stayed under it :)

Welcome to Haswell OCing! 4.5GHz at 1.27V is not bad for Haswell, but it sounds like you're throttling ("... it overheats under OCCT really quickly."), by hitting 100C on one or more cores. I know the voltage seems high, but it really isn't relative to Haswell.

Does OCCT use AVX2 instructions for its test? If so, that is what pushes the CPU into throttling.

AIDA64 allows you to select what is stressed in its benchmark. I can run my i5-4670K with compute, memory, and graphics testing simultaneously, without throttling. But add AVX2 testing to any one of the others at 4.4GHz, and throttling in under one minute. Stock clocks, no throttling.

I use an H60, with dual fans push-pull, with shrouds on my Haswell system. Same setup as my Sandy Bridge and Ivy Bridge (actually single fans on the SB and IVB), and I've never throttled on either of those, at 4.8GHz on the SB, and 4.6 on IVB. They'll BSOD or lock up at high OCs, but never throttle. I'm not a hardcore OC person, but the first time I've ever seen throttling is on my Haswell PC.

If you'll be judging your Haswell OC by the usual benchmark methods, Prime95, IBT, OCCT, AIDA64, and go with high stress testing with the CPU coolers we considered adequate or great in the past, your CPU/core temps will be near or at the throttling temperature. All the OC enthusiasts at OCN delid their Haswell CPUs, use liquid metal TIM (CLU, CLP) and custom water cooling to keep throttling from happening. Which is why I've stopped trying for an OC higher than ~4.4GHz on my Haswell processor, it's just not worth it for my usage.

For my standard usage, with all CPU power saving options enabled, I can use Adaptive voltage settings and set the OC to 4.4GHz. I'll see jumps in CPU voltage to 1.3V+ but that does not last long.

Another major issue is monitoring CPU core voltage on Haswell PCs, which is more difficult with ASRock boards like ours, IMO. All monitoring programs, including A-Tuning and IXTU, show the VID as the core voltage. The VID and Vcore readings for my SB and IVB PCs are not identical, and the "Vcore" reading on any monitoring program that is not the core VID, is the CPU Input Voltage, nominally 1.800V.

I've had discussions with the author of the HWiNFO monitoring program, and he assures me that the true Haswell core voltage is NOT available on most if not all ASRock Z87 boards. The best he can do is provide the core VIDs, and the "Vcore" reading is the CPU Input Voltage (which I helped him get right, check the HWiNFO forums.) Gigabyte Z87 boards use two monitoring chips in order to provide the (apparently) true core voltage. Using two monitoring chips is unheard of on mother boards, until now.

Another thing I found (apparently) about our ASRock Z87 board, is a discrepancy between the Adaptive Voltage setting in the UEFI, and the actual value. This is in the HWiNFO main information screen, not the monitoring or summary screen, in the Central Processor section. It shows the base CPU core voltage setting from the UEFI, not dynamically unfortunately, but it is read from the UEFI setting. Using Adaptive mode, the display shows a consistent 19mv (0.019V) value beyond the UEFI setting. For example, if the Adaptive voltage is set to 1.000V in the UEFI, the value on the HWiNFO display is 1.019V. The same is true for the iGPU voltage, and Ring voltage. I discussed this with the HWiNFO author, and he believes his reading is correct. Just FYI, 19mv may not seem like much, but if you're worried about high CPU core voltage, then this would matter, no?

[scorpian007]

Re: ASRock Z87 Extreme6 OC issues (UEFI 2.10)

Thanks for the info everyone, I tried a base voltage of 1v and an offset of 0.2v but the voltages seem to spike up to 1.4v under load using HWInfo64! If I set the base to 0.8 and the offset to 0.01 then it's at a more reasonable 1.2v under load. However, if I OC to 4.4GHz, the voltage remains at 1.2v and crashes after about 90seconds under load, changing the base or offset causes the voltage to spike up to 1.36-1.4v :(

[Shiari]

Re: ASRock Z87 Extreme6 OC issues (UEFI 2.10)

Strange ... if I set the base to somewhere under 1v (don't know the exact value, 0.8v does for sure) my and raise the offset accordingly my load voltage shoots up. I had to set the base value higher rather than lower (and the offset lower) to keep the voltage down within reason.

Edit: 1.2V under load likely isn't enough for 4.4GHz btw.