A-Tuning Version 2.0.112.3 For Z97 Extreme 6 OC Tweaker Error

[parsec]

Re: A-Tuning Version 2.0.112.3 For Z97 Extreme 6 OC Tweaker Error

Why am I thinking you have an Extreme 6, you have an Extreme 4! Not that it makes any significant difference, but our UEFIs use different code so we cannot equate their operation.

Your cache (uncore) speed and multiplier/ratio setting is weird stuff, I do not get what is happening. If IXTU and HWiNFO confirm those things, it's a fact, Jack.

The other boards where you saw the cache ratio set to 44 but running at 40 (4.0GHz) were also using an i7-4970K? I'm sure they were, just want to see that in writing.

IMO, if the DC i7-4970K processors were ALL running their cache/uncore at 40 (4.0GHz) with the ratio set above that (44, etc), there would be threads and posts about that all over Intel CPU forums. I'm not saying you are wrong or doing something wrong, just that it would be a major topic if it happened to all i7-4970K users. Maybe I'm missing it... could be... still very weird.

Have you found a strange bug related to some (all?) DC i7-4970K processors? More likely not the processor itself, but the CPU microcode or an internal UEFI/BIOS thing that Intel has a spec for, but is possibly wrong now?

BTW, loose the Auto UEFI settings, and set those ratios to 44 manually and see what happens. You know I must set my CPUs ratios to 44 manually to get them to that speed of course. IMO, this is related to the use of Auto for some reason.

Yes, those VCore and Cache voltage readings in HWiNFO are questionable. The HWiNFO programmer tries very hard to display anything he can correctly, but the info he gets from mobo manufactures is limited or not available. We cannot blame him if those readings are not perfect IMO.

Shoot, gotta go now! Must test that new version of A-Tuning too for ASRock. Keep us (me) updated about this please!

[PowerPie5000]

Re: A-Tuning Version 2.0.112.3 For Z97 Extreme 6 OC Tweaker Error

The other boards where you saw the cache ratio set to 44 but running at 40 (4.0GHz) were also using an i7-4970K? I'm sure they were, just want to see that in writing.

The user in this forum is using a Gigabyte Z97 UD5H along with a 4790K running stock (everything set to auto in the BIOS): Confusing 4790k behavior on Gigabyte Z97 UD5H - [H]ard|Forum.

Here's a HWinfo screenshot from his setup and it shows the same cache/uncore ratio behaviour as mine:



I'll have to try and find the couple of others I saw (I was browsing google image results for HWinfo & 4790k).

Have you found a strange bug related to some (all?) DC i7-4970K processors? More likely not the processor itself, but the CPU microcode or an internal UEFI/BIOS thing that Intel has a spec for, but is possibly wrong now?

It could be possible that not many people have noticed it, especially people who don't OC or even people who jump straight in and manually change the settings.

BTW, loose the Auto UEFI settings, and set those ratios to 44 manually and see what happens. You know I must set my CPUs ratios to 44 manually to get them to that speed of course. IMO, this is related to the use of Auto for some reason.

I know it works if I manually set the cores and cache to x44 and enable the multi core enhancement, but for some reason it makes all my fans run at 100% speed which is fairly loud! I don't know why this happens?

I'll see what happens if I just change the cache from auto to manual x44.

Shoot, gotta go now! Must test that new version of A-Tuning too for ASRock. Keep us (me) updated about this please!

Good luck with that. Hopefully they get the real Vcore readings sorted out . Will keep you updated.

[PowerPie5000]

Re: A-Tuning Version 2.0.112.3 For Z97 Extreme 6 OC Tweaker Error

Ok I just noticed something in the bios that I didn't notice before! With everything set to auto it shows the "Target CPU Turbo Speed" as 4000MHz (should be 4400MHz)!! This is very strange as it still clocks up to 4.4GHZ in Windows when it's under load! The cache also shows a target speed of 4000MHz.





I just changed the cache ratio manually to x44 and it works.. BUT... The cache/uncore will hit 4.4GHz even if the cores are only running at 4.2 or even 4.3GHz turbo states which means the cache is running faster than the core speeds. Very confusing stuff going on here!

[PowerPie5000]

Re: A-Tuning Version 2.0.112.3 For Z97 Extreme 6 OC Tweaker Error

More strange findings....

Setting the CPU ratio to "per core" mode automatically sets the CPU turbo and Cache target speeds to 4400MHz (cache still only reaches 4000MHz in Windows though), but also at the same time it increases both Vcore and Cache voltages to 1.272V and changes the LLC from 5 (lowest) to 1 (highest). HWinfo also shows each core VID at a constant 1.249V no matter what speed the cores are running at. Manually changing the CPU settings in the BIOS also makes the fans run at 100% full speed and I can't seem to change this for some reason.











I can't say I find Asrocks EUFI behaviour very amusing. Is it possible I could have a faulty board or even a faulty CPU? I should have some spare time tomorrow so i'll install the new chip if I manage to find my small IC extractor tool.

[parsec]

Re: A-Tuning Version 2.0.112.3 For Z97 Extreme 6 OC Tweaker Error

A quick comment about the CPU fan speeds. I've seen similar behavior with other ASRock boards when the CPU multipliers are set at or above the max Turbo speed of the CPU being used. The CPU fan speed option is automatically set to Full On, no doubt as a precaution for users that don't know better and may be using the stock Intel cooler. For those of us that do know better, that can be annoying, but the fix which has always worked for me is something we should know IMO.

I use the Customize setting, which takes some configuration but once done is remembered by the UEFI if you change settings. In my experience the Customize setting is not reset to Full On when the CPU ratio is changed beyond the CPU's Turbo speed.

Of course, I use PWM fans on my CPU coolers so the Customize setting is really meant for those fans.

Thanks for posting all the info, I'll check it out.

[parsec]

Re: A-Tuning Version 2.0.112.3 For Z97 Extreme 6 OC Tweaker Error

[ATTACH]7338[/ATTACH]
However, we provide few ways for you.
1. Please re-install A-Tuning for try.
2. We provide the latest A-Tuning version(2.0.214) for you to verify.
Please refer below link to download it for try. https://mega.co.nz/#!e59ljD5b!xaSGzl...fTulYPrMEjEFak
If the symptom still exist, please provide what value you have set in BIOS to us for checking.
Thank you.

Hi Davis,

Thanks to you and ASRock for looking into what I talked about in this thread. I will try the new version of A-Tuning that you provided, thank you for that.

The settings in the UEFI you described above did not cause the issue in A-Tuning ver:2.0.112.3, in the OC Tweaker screen. Note that this is not the new version.

But, the UEFI settings in the OC Tweaker screen (not A-Tuning) that I used are not the same as the ones ASRock used to test OC Tweaker in A-Tuning.

I should have described the settings I was using in the UEFI in better detail than I did in my first post, I apologize for that. I will do that now.

These are the OC Tweaker screen settings I used when I saw the issue in A-Tuning I described:

[ATTACH]7340[/ATTACH]

I also used settings in Advanced > CPU Configuration that are important. I have all the CPU C State options set to Enabled. Package C State support is set to C7.

I was using Adaptive Voltage for the CPU cores, and the core speeds will drop to 800MHz when the CPU is idle when using these settings.

With these settings in the UEFI, and the CPU slowing to 800MHz at idle, I then started A-Tuning and found the CPU Ratio set to 16. If you have any questions about other UEFI settings, please let me know.

I will try the new version of A-Tuning today with the UEFI settings I described above, and post the results.

I installed A-Tuning version 2.0.214 provided by ASRock to test the issue I described above and in the first post of this thread.

To review, the UEFI settings in the OC Tweaker screen that ASRock used to test the previous version of A-Tuning provided for this board, 2.0.112.3 were not the same as those I used, and they did not have the issue I described in this thread. The UEFI settings in the OC Tweaker screen that I used did cause the issue I found, and I used those settings again to test A-Tuning version 2.0.214.

The result was this:



The CPU Ratio in the A-Tuning OC Tweaker tool is now 38 (instead of 16) which is the correct value given the stock, Per Core, CPU Ratio setting I used in the UEFI.

IMO, this issue is fixed in A-Tuning 2.0.214, thanks for that ASRock.

[PowerPie5000]

Re: A-Tuning Version 2.0.112.3 For Z97 Extreme 6 OC Tweaker Error

I've installed the new BIOS chip sent from Asrock and... IT'S EXACTLY THE SAME! The same ME firmware version, same BIOS version and all the same weird little glitches that come along with it. I also discovered I didn't need my IC extractor tool as these chips are very easily pulled out by hand.

Another note... The VRM circuitry runs extremely hot with this board (same VRM setup as the Z97 Extreme6). It's gets so hot that it actually heats up the connectors and I/O shield on the back of the PC. The backside of the motherboard around the CPU socket is red hot and it heats up the motherboard tray and even the side panel on the case!! That can't be good, but i suppose Asrocks trademark thin PCB could be contributing to this? I'm actually tempted to give up and try my luck with a Gigabyte or MSI Z97 board (fully digital). Or should I hang on to this board for now? I don't want it to start degrading my CPU with silly voltages at stock, and I can't put up with 100% fans when I change anything to do with the CPU in the BIOS!

[parsec]

Re: A-Tuning Version 2.0.112.3 For Z97 Extreme 6 OC Tweaker Error

I've installed the new BIOS chip sent from Asrock and... IT'S EXACTLY THE SAME! The same ME firmware version, same BIOS version and all the same weird little glitches that come along with it. I also discovered I didn't need my IC extractor tool as these chips are very easily pulled out by hand.

Another note... The VRM circuitry runs extremely hot with this board (same VRM setup as the Z97 Extreme6). It's gets so hot that it actually heats up the connectors and I/O shield on the back of the PC. The backside of the motherboard around the CPU socket is red hot and it heats up the motherboard tray and even the side panel on the case!! That can't be good, but i suppose Asrocks trademark thin PCB could be contributing to this? I'm actually tempted to give up and try my luck with a Gigabyte or MSI Z97 board (fully digital). Or should I hang on to this board for now? I don't want it to start degrading my CPU with silly voltages at stock, and I can't put up with 100% fans when I change anything to do with the CPU in the BIOS!

Seriously, the IO shield is hot and the mother board tray?!?! IF so, that is not good.

Of course I'm not using my Z97 EX 6 board PC now, but I will check that out later. I have an IR thermometer so I can check temperatures anywhere. I'll do that later, won't happen until tonight.

About the CMOS/BIOS chip, I'm not surprised the ME firmware is the same, I don't understand why you thought it would be different? ASRock probably assumed your CMOS chip was bad and the ME firmware on that chip was corrupted, or the UEFI was corrupted, or both.

Since that does not seem to be the case, now you can move on from there. Don't forget that I'm not having the cache ratio issue that you have.

Don't forget that the stock voltage your CPU is using is being requested by the CPU itself via the VCore VID, and applied by the internal Haswell processor voltage regulators, not by the VRMs on your board. That is unique to Haswell processors and different than all previous Intel processors.

I agree that the stock voltage with Adaptive and Auto is extreme, and is most likely the same on any Haswell board. You can fix that by not using Auto voltage, I set a specific voltage and then use a negative offset to reduce the voltage further. Sometimes we must do things ourselves, and this is a great example.

[PowerPie5000]

Re: A-Tuning Version 2.0.112.3 For Z97 Extreme 6 OC Tweaker Error

Seriously, the IO shield is hot and the mother board tray?!?! IF so, that is not good.

Yeah they get pretty hot, but i'm also now thinking the GTX 780 could be contributing too as it uses a blower type cooler (exhausts hot air through the rear of the case). I know my stock VID & Vcore are a bit high, and this probably makes the VRM circuitry work a bit harder. Asrock claim it's a 12 phase power design but in reality it's only a 6 phase design with doublers.

Of course I'm not using my Z97 EX 6 board PC now, but I will check that out later. I have an IR thermometer so I can check temperatures anywhere. I'll do that later, won't happen until tonight.

Thanks, it'll be interesting to see how hot they get. I only discovered how hot it was after removing the side panel which had a very warm patch.

About the CMOS/BIOS chip, I'm not surprised the ME firmware is the same, I don't understand why you thought it would be different? ASRock probably assumed your CMOS chip was bad and the ME firmware on that chip was corrupted, or the UEFI was corrupted, or both.

I just assumed they might have flashed a later ME firmware as they said they want to send a new chip to make sure it's not an ME problem, otherwise all I had to do was flick the BIOS switch on the board to use the secondary chip.

I agree that the stock voltage with Adaptive and Auto is extreme, and is most likely the same on any Haswell board. You can fix that by not using Auto voltage, I set a specific voltage and then use a negative offset to reduce the voltage further. Sometimes we must do things ourselves, and this is a great example.

I thought setting a negative offset would also affect the idle voltage? I wouldn't want the CPU stalling if the Vcore gets too low. I'll have another go at playing with the voltages later, but so far it seems anything I change makes the fans run at 100% speed which is bit annoying. I'll check again later though. I'll keep the LLC as it is (level 5).

[PowerPie5000]

Re: A-Tuning Version 2.0.112.3 For Z97 Extreme 6 OC Tweaker Error

Ok I can tweak the voltages without it messing with the fans. I started of with a mild -0.040V Vcore offset and it works. Max VID is now 1.224V (was 1.263V) and the max Vcore is now 1.240V (was 1.280V)... The only problem is it affects the lower CPU states too! Meaning the VID is now around 0.680V when idle and Vcore is around 0.740V when idle (still backwards). Now I need to try and configure the offset to affect only the Turbo state voltages... I know this can be done with some boards using adaptive voltage as the adaptive voltage is only supposed to affect the higher clocks (I could be wrong, but i'm sure that's what I read).

EDIT: Changing the adaptive Vcore on it's own without adding an offset doesn't appear to do anything. Is the value entered in to the Adaptive Vcore box the amount added to the original Vcore under load (turbo states), or would I need to enter the max desired voltage such as 1.22V for example? The Vcore at 4.0Ghz without the turbo is around 1.100V, so I entered 0.120V in to the adaptive Vcore box and it did nothing (assuming the max would then be 1.220V when running at 4.4GHz). Haswell is definitely a head scratcher compared to other generations. I can't wait for Skylake when Intel remove the FIVR from the CPU (they probably now realise it was a mistake).

[parsec]

Re: A-Tuning Version 2.0.112.3 For Z97 Extreme 6 OC Tweaker Error

I have never had a problem with the VID at very low values, mine can go to a bit below 0.660V, and the PC runs fine. I would not worry about that.

I have a UEFI profile with stock clocks (38 Turbo), Adaptive voltage with CPU VCore set to 1.200V, and the offset set to -0.100V. Max VID is 1.030V.

Adaptive voltage for the cache at stock Turbo (38) set to 1.200V, and the offset set to -0.115V. Max VID (per HWiNFO) is 1.016V.

All C-States enabled and Package C-State set to C7. CPU Input voltage set at just under 1.6V.

Runs fine, drops to 800MHz at idle at VID of 0.654V minimum. Since that is a VID, the actual voltage is lower, I've read about other Z97 board users that apparently have a VCore reading, that say it can go under 0.300V.

That is what Haswell is really all about IMO, low power usage and low voltage stability at low clocks. But Intel will release several 'K' Haswell processor versions with TIM instead of solder under the IHS, and call them over clocking CPUs.

IMO, the last truly OC worthy mainstream processors from Intel are the i7-2600K, i7-2700K, i5-2500K, and i5-2550K.

You may be right about the FIVR, I saw that too, but I would not be surprised if it survives in some form on mobile processors.

I'll be checking my Z97 EX 6's temperatures tonight, never done that with my IR thermometer yet, should be interesting...