ROG Crosshair VI

[Alex@ro]

Interesring, did the same here and it did not worked..., had to plug 8.1 ssd

 

Sent from my SM-N910F using Tapatalk

[flanker]

try the DVD win7 instalation (SATA DVD) or soft tool to create win7 file bootable with USB3 drivers. It worked, ysterday I installed win7 on it

[I.nfraR.ed]

Booting 3200 divider with 2x8GB B-die is easy. You just need to wait for the F9 - shutdown cycle.

The board shuts down 5 times on F9 POST code, then boots.

If 12-12-12-12 doesn't work for your directly, start from 15-15-15-15-36, then bump DRAM voltage if needed and set 12-12-12-12.

 

I guess most reviewers didn't wait for the "training?". Maybe it sets each timing per cycle? Not sure, maybe elmor could answer.

Just playing around, still on stock cooler, which is by the way pretty fine...and that's the weakest Wraith cooler.

 

 

 

Now off to bclk OC.

Edited March 8, 2017 by I.nfraR.ed

[I.nfraR.ed]

@elmor Any words about LLC levels? What I see is big overvolt on any of the levels - about +0.1V load above what is set in bios and +0.2 in idle.

Am I doing something wrong?

[Enigma]

Booting 3200 divider with 2x8GB B-die is easy. You just need to wait for the F9 - shutdown cycle.

The board shuts down 5 times on F9 POST code, then boots.

If 12-12-12-12 doesn't work for your directly, start from 15-15-15-15-36, then bump DRAM voltage if needed and set 12-12-12-12.

 

I guess most reviewers didn't wait for the "training?". Maybe it sets each timing per cycle? Not sure, maybe elmor could answer.

Just playing around, still on stock cooler, which is by the way pretty fine...and that's the weakest Wraith cooler.

 

https://s14.postimg.org/vel1dl19r/screen258.jpg[/i*MG]

 

https://s10.postimg.org/3obwvb3p3/screen259.jpg[/i*MG]

 

Now off to bclk OC.

 

Hi can you or anyone else please do a single core bench in Cinebench R15 with 1800X spec just to see how much higher than ~162 we are getting in Windows 7? I think not only multithreading and SMT are affected in Win10. Thanks in advance and good results man! :celebration:

Edited March 9, 2017 by Enigma

[CornerJack]

A word about my bad experience yesterday.

I prepare my article on the Crosshair and I make a series of benchs at 3600MHz.

I shut down, I make a clear, I reboot and then ... debug led 27

I tried everything: other memories, VGA, PCIe, ... nothing

 

 

After 1 hour, finally, I flashed the bios with USB Bios Flash Back and all seems correct.

What happened ? :confused:

[elmor]

@elmor Any words about LLC levels? What I see is big overvolt on any of the levels - about +0.1V load above what is set in bios and +0.2 in idle.

Am I doing something wrong?

 

Manual or offset mode? Can you confirm with DMM? I believe we have a readout problem.

 

A word about my bad experience yesterday.

I prepare my article on the Crosshair and I make a series of benchs at 3600MHz.

I shut down, I make a clear, I reboot and then ... debug led 27

I tried everything: other memories, VGA, PCIe, ... nothing

 

After 1 hour, finally, I flashed the bios with USB Bios Flash Back and all seems correct.

What happened ? :confused:

 

Don't know, I guess BIOS corruption somehow?

[I.nfraR.ed]

Full manual mode. I always measure with DMM.

Maybe it depends on some other values. I will test more in the evening.

Bios 5803

Edited March 9, 2017 by I.nfraR.ed

[flanker]

Seems, with X370 PRO its better in vcore not fluctuation in load (I think, BIOS there was 0602 or so)

What surprised me, the temps with lowend cooler in AVX stress test were between 60-61 C only! With Ryzen 7 1700. But FCH is hot...

 

Enigma:I will test it today evening (around 10PM CET time :-) )

[flanker]

OK, I did Cinebenchs test in Win7. All at stock, some small difference could be because I have only 2x 4GB memory and at 2133 MHz. The CInebench performance seems similar as in Win10

 

 

 

[elmor]

Updated C6H BIOS 0902, prevents the random BIOS updating bricking. Keep CPU SOC Voltage at max 1.20V for prolonged use.

[Johan45]

I figured this is a better place to ask. Taken from OCN

 

Most boards are not affected as previously stated, you would probably be OK. Then again there's still a slight risk that your board will brick when using those bioses with increased SOC Voltage. Up to you if you want to take the risk. From our internal tests 5803 is the best performance wise, we haven't had time to evaluate that on 0902 yet.

 

Is there any easy way to tell which ones are affected and if this is actually fixable through BIOS to the extent that higher voltages while benching cold aren't going to kill my next board?

[elmor]

I figured this is a better place to ask. Taken from OCN

 

 

 

Is there any easy way to tell which ones are affected and if this is actually fixable through BIOS to the extent that higher voltages while benching cold aren't going to kill my next board?

 

There's no way to tell until it happens unfortunately, I recommend all to upgrade to 0902. Up to 1.25V should be OK on SOC Voltage, which is more than enough even during LN2 benching. Actually the new BIOS prevents the BIOS updating brick even at higher voltages. You won't brick the board but might still cause different issues if you go above that limit.

[Johan45]

Thanks elmor, I appreciate the candor and all the work you guys have put into this issue to get it resolved ASAP.

[chispy]

Updated C6H BIOS 0902, prevents the random BIOS updating bricking. Keep CPU SOC Voltage at max 1.20V for prolonged use.

 

Thank you for the PM Jon , appreciate it a lot :celebration:. With this new Bios now i feel confident using my boards again :ws:

[chew*]

flanker test wprime for me. I'm seeing preety large variance between win7 and 10 not sure why.

 

just 32m needed for now.

[Daxten]

after having some time testing, I don't think my problem is anywhere near solved

I'm running everything on AUTO, have all energy stuff enabled, balanced energy mode in windows and most of the time the cpu is between 1.2v - 1.4v in idle, very randomly it goes down to 0.844v for 1 tick.

temps at ~44c in idle.

cpu monitor shows 4-5% utilization

 

just updated to latest bios version posted here and it didnt change anything, also multiple times reseted, tried default, restarted, tried different settings.

 

I saw someone post 24c idle here, what would be an acceptable idle temp with an 1800x?

Also with 0902 FanXpert seems to have no control over the fans

Edited March 10, 2017 by Daxten

[I.nfraR.ed]

@elmor, thanks!

 

So for bios flashing we have a SPI header to recover from this brick situation.

I'm missing 2 things on this board:

1) dual EEPROM bios chips, preferably socketed with hw switch

2) voltage read points in the form of connectors with supplied voltage cables

 

Perhaps 1) is somehow limited by the platform.

For 2) I don't know why noone adds these anymore, they are much more convenient than the current measurement points.

 

Can we have the pinout of the header?

[elmor]

after having some time testing, I don't think my problem is anywhere near solved

I'm running everything on AUTO, have all energy stuff enabled, balanced energy mode in windows and most of the time the cpu is between 1.2v - 1.4v in idle, very randomly it goes down to 0.844v for 1 tick.

temps at ~44c in idle.

cpu monitor shows 4-5% utilization

 

just updated to latest bios version posted here and it didnt change anything, also multiple times reseted, tried default, restarted, tried different settings.

 

I saw someone post 24c idle here, what would be an acceptable idle temp with an 1800x?

Also with 0902 FanXpert seems to have no control over the fans

 

40-50*C idle is normal. You do seem to have something wrong with your OS, 4-5% utilization is way too much in idle.

 

@elmor, thanks!

 

So for bios flashing we have a SPI header to recover from this brick situation.

I'm missing 2 things on this board:

1) dual EEPROM bios chips, preferably socketed with hw switch

2) voltage read points in the form of connectors with supplied voltage cables

 

Perhaps 1) is somehow limited by the platform.

For 2) I don't know why noone adds these anymore, they are much more convenient than the current measurement points.

 

Can we have the pinout of the header?

 

It's not BIOS flashing, that can be done through USB BIOS Flashback. It's updating the EC firmware you need to do.

 

You only get dual BIOS chips on Extreme/Apex, and it wouldn't have helped this situation. USB BIOS Flashback can solve any situation where dual BIOS would help, since it's damn near impossible to kill the BIOS chip.

 

Regarding connectors it would add cost and very few people use it. Additionally there are space constraints on the board. And the Hero is not targeting extreme overclocking specifically.

 

There's unfortunately no header on the board for flashing the EC, what you need is an EEPROM writer plus SOIC-8 clamp that can fit on top of this IC. Space is a bit limited as well.

 

[I.nfraR.ed]

 

It's not BIOS flashing, that can be done through USB BIOS Flashback. It's updating the EC firmware you need to do.

 

You only get dual BIOS chips on Extreme/Apex, and it wouldn't have helped this situation. USB BIOS Flashback can solve any situation where dual BIOS would help, since it's damn near impossible to kill the BIOS chip.

 

Regarding connectors it would add cost and very few people use it. Additionally there are space constraints on the board. And the Hero is not targeting extreme overclocking specifically.

 

There's unfortunately no header on the board for flashing the EC, what you need is an EEPROM writer plus SOIC-8 clamp that can fit on top of this IC. Space is a bit limited as well.

 

 

Ok, I understand the reasoning behind not adding the connectors. At least we have reading points. My problem is all my hardware is retail from the store and if I solder connectors to the reading points, it will void the warranty.

Usually I don't care about warranty when there's some solid reason of voiding it, but in this case I would like to keep things at stock due to the high risk something goes bad and it's not really a cheap board for my country standards.

 

Am I safe to assume that this bricking situation only happens with auto update over internet?

I don't even have an internet connection on my bench systems.

 

I will try the new bios, but 5803 seems fine in terms of performance, so I might stick with it for a while.

 

Thanks again for all the work you and ROG team are doing for the community.

I expect some more extreme board in the near future though, so this is one of the reasons I don't want to modify the Hero in any way, which will help me sell it.

 

The AM3 coolers support idea is brilliant though and a big selling point. Thumbs up.

 

PS: LLC seems to be working now, not sure what was the problem before. It seems LLC was stuck at lvl5 which overshoots with about 100mV. Level 1 and 2 seem to be the best for me.

Edited March 11, 2017 by I.nfraR.ed

[Masterchief79]

Anyone else getting complete bonkers temperatures with 0902 Bios? It's showing 16°C in Idle but it's at least 22°C in my room, so yeah.. I didn't realize my watercooling ghetto mounting solution was THAT effective

[Ross Allan]

Anyone else getting complete bonkers temperatures with 0902 Bios? It's showing 16°C in Idle but it's at least 22°C in my room, so yeah.. I didn't realize my watercooling ghetto mounting solution was THAT effective

 

Similarly stupid temps here. 18c idle, 50c stress testing full load at 20C ambient.

[I.nfraR.ed]

flanker test wprime for me. I'm seeing preety large variance between win7 and 10 not sure why.

 

just 32m needed for now.

 

chew*, here's your wprime on Win7 x64 SP1, full auto in bios, CPU running at 3.7GHz on load. No tweaks.

 

Edited March 11, 2017 by I.nfraR.ed

[chew*]

chew*, here's your wprime on Win7 x64 SP1, full auto in bios, CPU running at 3.7GHz on load. No tweaks.

 

 

Ok thx i'm not crazy. UD5 performance in 7 is garbage...

[elmor]

Ok, I understand the reasoning behind not adding the connectors. At least we have reading points. My problem is all my hardware is retail from the store and if I solder connectors to the reading points, it will void the warranty.

Usually I don't care about warranty when there's some solid reason of voiding it, but in this case I would like to keep things at stock due to the high risk something goes bad and it's not really a cheap board for my country standards.

 

Am I safe to assume that this bricking situation only happens with auto update over internet?

I don't even have an internet connection on my bench systems.

 

I will try the new bios, but 5803 seems fine in terms of performance, so I might stick with it for a while.

 

Thanks again for all the work you and ROG team are doing for the community.

I expect some more extreme board in the near future though, so this is one of the reasons I don't want to modify the Hero in any way, which will help me sell it.

 

The AM3 coolers support idea is brilliant though and a big selling point. Thumbs up.

 

PS: LLC seems to be working now, not sure what was the problem before. It seems LLC was stuck at lvl5 which overshoots with about 100mV. Level 1 and 2 seem to be the best for me.

 

No it can happen randomly at reboot, I really suggest you update to 0902. We'll have future updates to make it as good or better than 5803 in terms of overclocking

 

Anyone else getting complete bonkers temperatures with 0902 Bios? It's showing 16°C in Idle but it's at least 22°C in my room, so yeah.. I didn't realize my watercooling ghetto mounting solution was THAT effective

 

We went a little bit too far with the tweaking, probably future BIOSes will go back to 5803 behavior. You can manually restore this by setting the Sense Mi skew to 272 under Tweaker's paradise.