Asrock Fatal1ty X99X Killer SSD SSATA_? Issue
Printed From: ASRock.com
Category: Technical Support
Forum Name: Intel Motherboards
Forum Description: Question about ASRock Intel Motherboards
URL: https://forum.asrock.com/forum_posts.asp?TID=452
Printed Date: 05 May 2024 at 2:22pm
Software Version: Web Wiz Forums 12.04 - http://www.webwizforums.com
Topic: Asrock Fatal1ty X99X Killer SSD SSATA_? Issue
Posted By: TontNZ
Subject: Asrock Fatal1ty X99X Killer SSD SSATA_? Issue
Date Posted: 22 Jul 2015 at 9:18am
|
Hi All I'm having major problems running a Samsung 850 EVO 500GB from the mobos SSATA_x ports. It took 1.5 hours to install Windows 8.1 Update from a 64GB Lexar USB3 flash drive. Copying a 2.5GB file to the drive can take one or two minutes and deleting the file almost as long. I've installed ATTO and Samsung Magician to benchmark the drive and ATTO hangs and Magician never completes the benchmark, it take six or so minutes to get to 8% and then just sits there. Connect the drive to the Intel SATA ports and everything works as it should, the Samsung Magician benchmark completes after 30 seconds or so. I have to connect the boot drive to the SSATA ports rather than the Intel SATA ports as I need to use all six of the Intel ports for RAID. I'll be running 4 x Samsung 850 EVO 500GB in raid 0 and 2 x WD SE 4TB in raid 1. So the system has to boot from one of the SSATA ports. Any ideas please, or please let me know If I can provide more information. Also, a big thank you for taking the time to read this. System: ASRock Fatal1ty X99X Killer LGA 2011-v3 Intel X99 Motherboard. Intel Core i7-5820K Haswell-E 6-Core 3.3GHz cpu. 32GB G.Skill Ripjaws 4 DDR4 2800MHz PC4-22400 CL16 Quad Channel kit (4x8GB). 1 x Samsung EVO 850 500GB (boot) 3 x Samsung EVO 850 500GB (raid 0) 2 x WD SE 4TB (raid 1) 1 x WD SE 4TB
|
Replies:
Posted By: odiebugs
Date Posted: 22 Jul 2015 at 9:54am
|
The windows boot needs to be on Sata3 -0, or in RAID on any of the Sata3 0-5, how does losing 20g make a difference on the RAID 0 which can be the OS partitioned to 40g. You have posted 4 in RAID 0, but below 3 to RAID 0 and the one as boot. So as posted below, that would be the one on Sata3-0 and 3 for 1,2,3, and 2 for RAID 1, on 4,5. 4 drive in RAID 0, with a 40g partition for OS. Why can't this be done and do you think this isn't secure. And throw that non RAID drive anywhere but the Sata3 Intel. That single WD should't even be wasting space on a Sata 2 port.
------------- asrocking |
Posted By: TontNZ
Date Posted: 22 Jul 2015 at 12:54pm
|
Thank you for taking the time to reply, but unfortunately it does not address by problem. I need to be able to boot the system from a Samsung 850 EVO SSD attached to one of the SSATA3_ ports. When I attach the 850 EVO to any of the SSATA3_ ports the system will boot, but performance and reliability is a complete dog. The fact the none of the benchmarks complete when the EVO 850 is attached to one the SSATA3 ports implicitly means that something is fundamentally wrong. This is a business workstation, I need it to be both reliable and stable. I need to know that if I plug an SSD into one of the SSATA3 ports, that device is going to work. I need a large (0.5 > 1TB) system disk and do not want to compromise the raid 0 array which is reserved for a single purpose. I'm planning on using a 1TB m.2 when they become available, hopefully sometime later this year. But what chances would you give on an m.2 SSD working when the same SSATA3 port used by the EVO 850 does not! Yes, you are quite right, raid 0 is inherently risky. That's why critical parts of it will be backed up daily to the raid 1 array (whose sole purpose is fast internal backup), which in turn is backed up to an external raid 5 array. The single WD SE 4TB is attached to one of the SSATA3 ports (seemingly with no issues) as all of the Intel raid ports are in use (4 raid 0, 2 raid 1). I really need a solution to this problem, or sadly I'm going to have to RMA the board. BTW: I have upgraded the BIOS to v1.9, but it's made no difference. Thanks again.
|
Posted By: Xaltar
Date Posted: 22 Jul 2015 at 3:34pm
|
What port do you have the evo boot drive connected to? Try to avoid connecting it to shared ports on the Asmedia controller (SSATA_x). Port 2 is shared with the M2 slot so it could cause problems connecting a boot drive there as the M.2 slot will have preference to resources which while unlikely could cause the system to bug out even in the instance where the M.2 slot is unpopulated. The Asmedia controller that provides the SSATA_x ports is really intended for optical/mechanical drives and is not ideally suited to high performance SSDs like the Evo series. Oddly the M.2 slot is also provided by the Asmedia controller so obviously the performance potential is there or what would be the point of the slot. This is the first time I have seen this problem occur so I am going to have to do some digging to see what can be done about it. If you have any other brand SSDs laying about you could test on the SSATA ports that would be most helpful as it will allow us to eliminate firmware conflicts with the Evo/Asmedia combo. As you said, the mechanical drives seem to work as intended on the same controller. If you have not already I suggest you put in a tech support request and get the support guys on this. The Evo is a very popular drive series so this kind of thing is exactly what they will be looking to address. They may be able to provide you with a bios update that corrects the problem with the asmedia SATA ports. Good luck
|
Posted By: TontNZ
Date Posted: 22 Jul 2015 at 7:40pm
|
Xaltar, thank you for your reply. I have tried the EVO 850 500GB on SSATA3_0 and SSATA3_1 but not the shared ports SSATA3_2 and SSATA3_3. Given that the Asmedia chip supports m.2, an EVO 850 should present no problem. I can rule out the other SSD's and HDD's as the single EVO 850 was the only SSD present when I did the 90 minute Windows 8.1 install! I have done as you suggest and logged a technical support query with Asrock, although the email I received from them suggest I contact my dealer. I'll do this anyway in case I need to RMA the mobo. If I have time tomorrow I will try the EVO on SSATA3_2 or _3, just to see if anything changes. The major problem I have is that this is a workstation for my business, I'm also self employed. So all this downtime messing with this issue means I'm not earning any money. So I can sustain only so much pain before I have to dump this mobo. BTW: All four EVO 850 500GB's are running the same EMT01B6Q firmware. The Samsung Magician software reports that this is the latest firmware revision. Thank you again for your time, I really do appreciate it.
|
Posted By: odiebugs
Date Posted: 22 Jul 2015 at 8:02pm
|
There is no info for this ASmedia chip, so I don't know what firmware it has, actually I can't find anything about it. This is what you need to do or find out. 1- What is the devise ID of the controller 2- What is the driver that is running it. 3- What settings in the BIOS / UEFI are there for this storage controller, are they even there. If you know how to get the devise ID of the hardware so we can see what it is, we might be able to deal with it. If you can check the BIOS and if you find settings for it can you set them to AHCI. ------------- asrocking |
Posted By: TontNZ
Date Posted: 23 Jul 2015 at 10:40am
|
I have tried the EVO 850 on SSATA3_3 and the problem only got worse. Write speed is abysmal, it is 2.5% (no - that number is correct) of the same drive connected to one of the Intel SATA ports. I also looked through Device Manager and i can't find an Asmedia driver for the EVO when connected to the SSATA3 ports. It seems to be using standard Microsoft drivers: disk.sys, edevmon.sys, EhStorClass.sys and partmgr.sys. For those of you that are familiar with the ATTO benchmark (default settings) at a transfer size of 64KB when connected to Intel port the write figure is 266,197, when connected to SSATA3 it is 6798. The only interesting part is that on the ATTO benchmark (run on default settings, which failed again at 64 and did not complete) the read speed is considerably better than when the drive is connected to the Intel port. Samsung Magician is showing AHCI active when connected to a SSATA3 port, but not when connected to an Intel port (I've had to turn RAID on in the bios) through the Intel Rapid Storage Technology (RST) drivers. The ATTO 64K benchmark figures for reads are Intel: 279,920 and SSATA3_3: 533,963. Thanks to everyone for reading this. |
Posted By: odiebugs
Date Posted: 23 Jul 2015 at 11:26am
|
Most ASmedia Sata 3 controllers have around a Sata 2 speed, but yours are not operating correctly. If anyone has one of these boards can you post the devise ID of the ASmedia controller so we know what it has. Kinda helps when you know what series the controller is and what it's specs are. P.S. Some hardware actually functions better off the generic Microsoft drivers than it's own MFG driver. But knowing the drivers are loaded and your speed and bench not finishing is a sign something is wrong. Also, at 64k the evo should be at around 400MB and not 200MB on the Intel, can you try another drive, sounds like the drive has a problem. ------------- asrocking |
Posted By: Xaltar
Date Posted: 23 Jul 2015 at 9:40pm
|
It has come to my attention that I was wrong. The X99X Killer does not use an Asmedia controller for any of the SATA ports, all ports are provided by intel's chipset with only the SATA3_x ports supporting RAID functionality. Please forgive my confusion, the socket 1150 boards are the ones that usually have ports provided by the Asmedia controller. There is a http://forum.asrock.com/forum_posts.asp?TID=456&title=asrock-x99x-killer-sata-ports" rel="nofollow - second post about problems with drive performance on the SSATA ports on this same board so whatever the problem is it is likely the same for both you and the other poster.
|
Posted By: odiebugs
Date Posted: 23 Jul 2015 at 9:47pm
Who would have thought, up until the X99, Intel only allowed 6 Sata 3 for chipsets. Now looking at the chipset we can see how they have 10. Like you said about the other poster, they have a problem with speed on the S Sata, I guess we'll be looking for tests on these ports to see if there is a difference between the 6 that have RAID and the 4 that say no RAID support. It's not your fault Xaltar, the manual should have this info very accessible as they only provided 6 before. I only see one bench where someone went and benched all of the drive ports, and there was no difference between them. So either there is a UEFI setting, or an actual UEFI problem with Sata. Either someone or TECH needs to test them and see if any UEFI settings cause a problem with the S Sata ports.
------------- asrocking |
Xaltar wrote:Posted By: DevillEars
Date Posted: 23 Jul 2015 at 11:15pm
|
Hi, I'm the "other poster with X99X Killer SATA challenges"... Opening Intel's X99 chipset webpage shows the following from the block diagram: a) CPU to Discrete Graphics: "Up to 40 lanes" (only 28 lanes for Core i7-5820K) b) X99 to PICe 2.0: Up to 5Gb/s x1 bi-directional c) X99 to 10 xSATA 3.0 Ports: Up to 6Gb/s (asterisk note states "all SATA ports capable of 6Gb/s") NOTE THE USAGE OF THE WORDS "UP TO" IN THE SPECS... Under "Features and Benefits" Intel's site contradicts the above as follows: Feature: Serial ATA (SATA) 6 Gb/s Benefit: Next generation high-speed storage interface supporting up tp 6Gb/s transfer rates with up to six SATA ports (Note reference 6) Note 6: Actual number of ports available may vary by system configuration. Please consult your system vendor for more information. This last Note 6 comment implies that ports 7-10 not offered as standard by Intel but implemented by mobo manufacturer... Intel's website is not an easy site from which to extract detail, so checked out Wikipedia which has the following to say regarding SATA implementation in the X99 chipset:
|
Posted By: Xaltar
Date Posted: 23 Jul 2015 at 11:18pm
| Thanks for the info Dave, that is very useful to know. |
Posted By: odiebugs
Date Posted: 23 Jul 2015 at 11:22pm
|
Thanks Dave: Glad you popped over into this thread and posted this info, real nice of you. ------------- asrocking |
Posted By: DevillEars
Date Posted: 23 Jul 2015 at 11:43pm
|
One thought... The X99 capabilities need to be considered in conjunction with the partnering CPU: Scenario 1: X99 + Core i7-5820K (limitation of only 28 PCIe 3.0 lanes) Scenario 2: X99 + Core i7-59xxK (support for 40 PCIe 3.0 lanes) There may well be some - unpublished - constraints with regard to X99 capabilities when used with the 28-lane i7-5820K. The constraint may well lie in the DMI interface between CPU and X99 and dictated by Intel's CPU implementation... Guessing time...  ------------- If music be the love of food, eat on |
Posted By: DevillEars
Date Posted: 24 Jul 2015 at 12:11am
Some further "web-digging" elicited the following from an Intel Haswell-E review by AnandTech:
If the reviewer at AnandTech is correct, this would explain the problems experienced with SSD usage on these ports as the hub will, effectively, function as a "multiplexer" to extend the number of connector ports while keeping the interface lane-count constant. Dividing the fixed lane bandwidth by more connector ports will see bandwidth degradation - particularly when data is being moved between multiple drives attached to the same SATA hub. Also, contention may also be the reason for some of the SSD tests failing due to timing issues? My suggestion to the OP: If you want to run your disks as explained, ignore the S_SATA ports and install an Intel 8-port 8x PCIe 3.0 RAID card which will provide the full bandwidth capability of a PCIe 3.0 lane for each port (The more I think about this, the more convinced I become that it is the solution for me as well...) Dave
------------- If music be the love of food, eat on |
Posted By: parsec
Date Posted: 24 Jul 2015 at 1:04am
|
I'll need to check the datasheet of the X99 chipset to learn how the four secondary (my terminology) SATA III ports are implemented. Intel has always provisioned the PCIe lanes used for the SATA ports from those provided by the board's chipset. Those have always been PCIe 2.0 lanes. The PCIe 3.0 lanes are used exclusively for the video card slots, and now for PCIe SSDs. I'll be testing the SSATA ports on a SSD to check the performance, and will report back soon. |
Posted By: odiebugs
Date Posted: 24 Jul 2015 at 1:29am
|
Even though the chipset has the 10 sata , from what I see, these four posts have the same bandwidth as the 6 intel with RAID. Honestly I can't figure if the 28 lanes were maxed out, if this has any affect on the 2.0 on the chip to become burdened. I also noticed that Intel has removed the Enterprise RAID from the X99 and dosn't have ROM switch like the X79 had. Now there's only RST and no RSTe. Not sure if I can find if the UEFI has injected another AHCI in it for the other ports or if because there's still RAID on the six if the UEFI has a separate software firmware. It's taking the X79 and X99 UEFI and seeing the changes to the UEFI. So it could be added code in the original one for the extra four. There are massive changes to the UEFI files inside the X99, I can't even find the Sata Driver or the CMScore. I wonder if Parsec can use the RST AHCI driver to switch over the four posts to show as storage devices to see what devise ID they have or what they have for it with windows driver.
------------- asrocking |