|
X99 Extreme 6 BIOS |
Post Reply 
|
| Author | |
Spedi 
Newbie 
Joined: 24 Aug 2016 Status: Offline Points: 8 |
 Post Options
 Thanks(0)
 Quote Reply
 Topic: X99 Extreme 6 BIOSPosted: 25 Aug 2016 at 11:02am |
|
Extreme 6 X99 motherboard. Upgrade BIOS to 3.10 or 3.20 and the system fails to boot from my Samsung 950 PRO in the M.2 slot. Version 2.10 of the bios works fine, only thing changed is the version of BIOS. Used instant flash and Windows update. Specifics:
NVME in M.2 slot Samsung 950 PRO boot drive Nvidia GT750 video card with monitor on HDMI Win 7 x64 SP1 Intel i7 5820k processor 32 GB DDR4 RAM 3.20 upgrades from both production and Beta zone. (Those files are identical by the way.) 3.10 upgrade from production Try to boot after upgrade and system shows ASROCK screen. It goes blank and cursor blinking in upper left corner. VERY brief disk activity lights then system is non responsive with the blank screen. Have to hard reset. System will boot USB drive or CD. Any help or thoughts? Jim
|
|
 |
|
|
Spedi
Newbie
Joined: 24 Aug 2016 Status: Offline Points: 8 |
Post Options
Thanks(0)
Quote Reply
Posted: 05 Sep 2016 at 8:01pm |
|
I found my own answer. Here is my configuration:
Asrock X99-Extreme 6
Samsung SSD 950 Pro NVME in the M.2 slot (Boot device)
Samsung XP941 M.2 SSD mounted on a Addonics PCI-E to M.2 adapter
Intel 5820 CPU
Windows 7x64 or Debian Linux (Dual boot using GRUB2)
What happens is the when both the NVME and XP941 are installed and the BIOS is at v3.XX the system refuses to boot. Screen goes blank and you need to hard reset. Remove the XP941 and the system boots just fine. Version 2.10 and below work just fine with everything installed. Is Asrock aware of this and if so is there a fix?
|
|
|
|
|
parsec
Moderator Group 
Joined: 04 May 2015 Location: USA Status: Offline Points: 4996 |
Post Options
Thanks(0)
Quote Reply
Posted: 05 Sep 2016 at 10:10pm |
Booting Windows 7 from an NVMe SSD is unusual, but possible if the Windows 7 installation files are fixed (location of the EFI bootloader folder.) Have you done any configuration of the CSM option in the UEFI? That is normally required when using an NVMe SSD as the OS/boot drive. Did you install Windows 7 with only the 950 Pro in the PC? Sounds like there could be a problem here, I just want some more information. What "own answer" did you find? |
|
|
|
|
|
|
Spedi
Newbie
Joined: 24 Aug 2016 Status: Offline Points: 8 |
Post Options
Thanks(0)
Quote Reply
Posted: 06 Sep 2016 at 6:04am |
|
@parsec I am the only poster in this thread so I answered my original post. The "own answer" is why the black screen when upgrading BIOS from 2.10 to anything 3.xx.
Some answers to your questions: 1. All CSM settings are stock, Legacy only 2. Yes, Win 7 SP1 ISO as downloaded from Microsoft And 3 questions of my own: 1. Why does Linux not boot also? I am using Grub2 and not the Windows boot. I do not get the Grub boot choice screen. Nothing but a black screen and keyboard does not work. 2. Why does everything work fine with BIOS 2.10 and before but break with anything 3.xx or above? Downgrading to 2.10 restores me to a fully working system. I tried re-flashing, battery out reset etc. after going to any v3.xx BIOS. 3. Why does the Extreme 6 have 2 different versions of 3.20 BIOS? The version in the Beta zone and the latest version in the regular download area are both 3.20 yet there are changes between them. I verified that they are different by using Windows CRC32. (edited for spelling)
Edited by Spedi - 06 Sep 2016 at 7:45am |
|
|
|
|
parsec
Moderator Group
Joined: 04 May 2015 Location: USA Status: Offline Points: 4996 |
Post Options
Thanks(0)
Quote Reply
Posted: 06 Sep 2016 at 9:38am |
I just realized something after I had written everything below this paragraph, which I will leave here anyway. You have updated to the Broadwell-E compatible UEFI version (any 3.xx version), but are using a Haswell-E processor. That's Ok, but you must also update the Intel Management Engine software, or you will experience some odd affects. That would be the Intel Management Engine driver ver:11.0.4.1186 on your board's download page, Windows 7 64 bit section: http://www.asrock.com/mb/Intel/X99%20Extreme6/?cat=Download&os=Win764 The basic, incomplete answer why the 3.xx UEFI does not allow you to boot with the XP941, is something is apparently broken in the 3.xx UEFIs, relative to 2.10. As if you did not know that, I know. NO sarcasm here, I simply don't have a complete answer for you. My theory is, and only a theory, is in the 2.10 UEFI, the UEFI configuration settings required for using an NVMe SSD as the OS/boot drive, are being set automatically, behind the scenes, when an PCIe NVMe SSD is detected. That means, minimally, the CSM option Launch Storage OpROM Policy is being set to UEFI Only. This has always been required for a PCIe NVMe drive used as the OS drive. How is that related to your XP941 that is not an OS drive? Did you install Windows and/or Linux with both the 950 Pro and XP941 installed in the PC? The Windows installer has a annoying quirk of putting the MBR/GPT partition on another drive besides the OS drive, if one is available during the installation process. Does the Linux installer also do that? I'm not sure. What bothers me is this does not really explain your situation. Plus I have not seen other X99 Extreme6 owners complaining about the problem you are having. If your XP941 has a Windows "System" (MBR or GPT boot) partition on it, then that would confirm what I asked about your Windows 7 installation. Again, I cannot understand how an unmodified Windows 7 installation file structure could use the EFI bootloader. That is a known bug in the Windows 7 installation file/folder layout. But only Windows 7, not newer versions. One thing I will suggest to you is, in the Boot screen, find the CSM option at the very bottom of that screen, scroll down if necessary. CSM should be Enabled if I follow what you've said so far. Open the CSM option, and set the Launch Storage OpROM Policy to UEFI Only. This causes the EFI Bootloader to be used, which is necessary for an NVMe PCIe SSD. Windows requires an OS drive using the EFI Bootloader to be GPT partitioned. Not sure about Linux. A full UEFI booting OS installation (CSM Disabled) requires the video source to be GOP compatible, a UEFI requirement. I know that all EVGA 700 series video cards are GOP compatible, but that does not mean other Nvidia based 700 series products will be. Is your video card an Nvidia product, or made by someone else? If a video card is not GOP compatible, it will not work at all with a full UEFI booting OS installation, and you will see nothing on your screen. I'm not sure if this is your situation. If the 3.20 Beta and 3.20 Production UEFI versions are identical, or different, then that is just a sloppy file naming mistake. The 3.20 Beta has a newer date, and seems to be different in download size. The descriptions are not identical. You won't notice any difference in the UEFI between the 3.20 Beta and 3.20 Production UEFI versions if you are not using a Broadwell-E processor. But I can assure you they are different, despite what the CRC32 check tells you. The 3.20 Beta version adds the Broadwell-E Turbo Boost Max 3.0 option, that does not apply to Haswell-E processors. |
|
|
|
|
|
|
|
Spedi
Newbie
Joined: 24 Aug 2016 Status: Offline Points: 8 |
Post Options
Thanks(0)
Quote Reply
Posted: 07 Sep 2016 at 9:03am |
|
parsec,
The 941 does not have a boot partition or boot sector on it. (Verified that by Linux DD.) I never used it as a boot device and Windows was installed before the device was. Digging further I let the system sit after trying to boot from 3.20 and the 941 installed, and all of a sudden, about 5 minutes after turn on, I got a message from Grub saying: (paraphrasing here) Device 8ffe5368-... was not found. Grub Rescue> When I did a ls it hung up after only showing 2 disks. (I have 7 in the system) The 2 it showed me it could not read with Unknown file system errors. These may be the 950 and my first real hard drive. Eventually it timed out and came back to Grub Rescue>. The Device listed above is my Linux root partition's UUID which means it could not find the Linux disk, which is a separate SSD. Later today I am going to try and install a newer version of grub to see if that helps. Both the 941 and the 950 are formatted to MBR as they are only 256 and 512 MB respectively. Do you think one or the other should be GPT? Changing the CSM had no effect. My video card is GOP compatible.
Edited by Spedi - 07 Sep 2016 at 9:04am |
|
|
|
|
Spedi
Newbie
Joined: 24 Aug 2016 Status: Offline Points: 8 |
Post Options
Thanks(0)
Quote Reply
Posted: 17 Sep 2016 at 7:21am |
|
I guess parsec gave up trying to help me. Oh well, go and fix those other X99 P3.20 BIOS problems. Maybe you will discover a fix for this one too.
Edited by Spedi - 17 Sep 2016 at 7:24am |
|
|
|
|
parsec
Moderator Group
Joined: 04 May 2015 Location: USA Status: Offline Points: 4996 |
Post Options
Thanks(0)
Quote Reply
Posted: 17 Sep 2016 at 9:00am |
I had nothing more to tell you for several reasons: I don't know Linux well enough to give you any advise that would be helpful. NVMe SSDs need to use the EFI boot loader provided by the OS, rather than the legacy boot loader. This is generally called UEFI booting. Windows 7 has a known bug in the location of the EFI boot loader file, which must be fixed before it can be used to be installed in UEFI booting mode. I asked you about this, and your response was you used a standard Windows 7 ISO. Which seems to mean it was unmodified, but not specifically answered either way. I'll call this Part 1. Next, you said you use the default CSM option setting, Enabled, with all the CSM sub-options set to their defaults, Legacy Only. Your 950 Pro is formatted as MBR. Both of those things are just the opposite of how UEFI booting works, as I have known it to work in the past. The GPT formatting of the OS drive has nothing to do with its capacity, that is a UEFI booting/EFI boot loader requirement, or has always been so in the past. I'll call this Part 2. The Samsung 950 Pro requires an NVMe driver in order to function. Windows 7 does not have an NVMe driver, at least the ISO installation file doesn't have one. I've heard of a Windows update that apparently adds an NVMe driver to Windows 7, but that does nothing for the Windows 7 installation media. There is a hacked F6 version of the Samsung NVMe driver which can be installed during a Windows installation, but you never mentioned anything about an NVMe driver. The XP941 is an AHCI PCIe SSD, which should have no affect on an NVMe SSD. Add Linux into this regarding an NVMe driver, which is unknown to me. That is Part 3. Put Part 1, Part 2, and Part 3 together, and it all makes no sense, given my experience with UEFI booting and installing an OS on an NVMe SSD. It contradicts the most basic facts about UEFI booting that I have managed to discover via research and experience. In short, it means from my viewpoint, your configuration simply should not work. Yet you claim it does, or did with UEFI version 2.10 and earlier versions. That is a wall or barrier that prevents me from providing any further advice, with one exception. That exception was explained in my second post in this thread. That is, it seems the UEFI is able to detect an NVMe SSD, and manages to force the use of the EFI boot loader, even if the Windows installation was not done in UEFI booting mode, and the CSM option was not configured correctly. But that is only a guess on my part, since I cannot find any information that confirms it. I then said the UEFI versions beyond 2.10 seem to no longer provide the automatic use of the EFI boot loader. The fact that you could not fix the booting issue by setting the CSM option as I suggested, with an unmodified Windows 7 installation, seems to confirm my theory. I also told you about the need to use the new version of the Intel IME software with any of the 3.xx UEFI versions, that adds support for Broadwell-E processors. You made no comments about trying that or not. What more can I do when you seem to ignore almost everything I have told you? I don't program the UEFIs for these boards, and don't have access to the programmer(s) that work on them. I don't have a magic fix that requires nothing from the user, which seems to be the only help you have any interest in. |
|
|
|
|
|
|
|
Spedi
Newbie
Joined: 24 Aug 2016 Status: Offline Points: 8 |
Post Options
Thanks(0)
Quote Reply
Posted: 18 Sep 2016 at 2:27am |
|
@parsec
You are correct Putting part 1, 2 and 3 together makes no sense but what happens regardless of BIOS settings is the boot order gets changed so that ATA, which is the 941, tries to boot first. Linux or windows does not matter. Nothing changes this as far as I have seen. I had no choice but to upgrade the ME firmware when I installed the 3.20 bios. When I went to the Intel website and downloaded and installed the latest IME and installed a WINDOWS driver. So how does a driver in windows (upgrading the IME driver) effect the drive selection before windows is even started? You got me on this if you can answer this question. Second I must be running the Samsung driver if I can even access this device. Thirdly I don't expect you to write the code to boot UEFI vs MBR but I do expect you, as an agent of Asrock, like the rest of the people on this forum expect you too, is to have some knowledge that can help me. If you can't or you don't want to please tell me who to contact at Asrock. Oh, I did try the CSM options like you said. NOTHING DIFFERENT TO REPORT. Sorry for not expressly reporting nothing to report. I have tried everything you suggested. Nothing works. Sorry but that is the way it is. Have your tier 2 folks build a system similar and they will see no matter what the ATA drive tries to boot first. Jim Edited by Spedi - 18 Sep 2016 at 2:31am |
|
|
|
|
Post Reply
|
|
Tweet
|
| Forum Jump | Forum Permissions You cannot post new topics in this forum You cannot reply to topics in this forum You cannot delete your posts in this forum You cannot edit your posts in this forum You cannot create polls in this forum You cannot vote in polls in this forum |
Topic Options
Spedi wrote: