PC Drivers & Software

[mockingbird]

Does AMD plan to release a driver for this or are we supposed to use the generic Microsoft driver from 2006?

CPU is Ryzen 2200G.  I downloaded "rv-win10-64bit-whql-radeon-software-17.40.3701-feb12" which is what is offered for it and looked through the driver packages.  There is no AHCI driver in the package at all.  All the chipset drivers on the AMD site only offer southbridge AHCI drivers, and as I understand it, 43B7 is built into the CPU itself.

Thanks

[black_zion]

There is no performance difference between AMDAHCI and MSAHCI drivers.

[plumboby]

Go with the MS drivers more stable for ahci drivers & bother work either way the MS drivers are more upto date.

[mockingbird]

The MS drivers are certainly NOT "more up to date".  They are generic and date from 2006, which would render them more than 10 years old now.

I'm shocked at how no one has brought this up before.

[waltc]

Actually, the correct installation for the new Ryzen chipsets in Windows 10 is as follows:

AMD SATA controller, which is the AMD hardware driver, followed by the Standard SATA AHCI Controller driver, which is the Windows software driver. Both of these drivers should be listed in the Device Manager under the IDE ATA/ATAPI Controllers heading.  (Things do not work in Win10x64 as they did in earlier versions of Windows; that's what has you stumped, I'm sure.) 

[mockingbird]

There most certainly IS a performance difference.

Generic MS driver is unacceptable.  AMD must release an AHCI driver for their product.

[black_zion]

Yea see here's the thing, you're wrong, and I can tell you you're wrong because I have been using Samsung 850 Evo drives for a long time, and those drives push the limits of SATA. Right now I'm using two of them in RAID0 with Microsoft's drivers via Storage Spaces. So halve those figures, they're pushing the practical limits of SATA. No performance loss using Microsoft's drivers

It doesn't even look terrible compared to my 960 Pro 512GB that uses Samsung's drivers.

And there was even a thread on here I started which probably got deleted due to age where I compared AMDAHCI and MSAHCI, and MSAHCI had HIGHER performance than AMDAHCI.

And finally, the drivers are not "10 years old", and I can prove it because the Microsoft Storage Spaces Controller has a date of 2006, which is before it was even conceived of, so it's just Microsoft not changing the driver date, they just change the driver version to whatever version of Windows you are running.

[mockingbird]

black_zion -
I respect your passion (and I like your name).

But it's not all about speed.  How do I know that MS drivers have everything that the AMD ones do?  Unless there's direct collaboration between MS and AMD when it comes to MS writing their own generic AHCI drivers, then it's going to be lacking things.  Drives have TRIM, NCQ, garbage collection, etc...  Only the manufacturer of the chipset can write the driver properly.

AMD must release a driver and let the user choose.

[hardcoregames_]

The stock AHCI driver is all that is needed, the whole thing is standardized for the BIOS and has little to do with the CPU itself

[black_zion]

TRIM is an ATA standard and is supported with Microsoft's drivers, same with NCQ. Garbage collection is handled at the drive level and is transparent to the OS, which allowed for drive performance recovery before TRIM became widely supported as well as currently allows drives which are in a RAID setup to maintain performance.

Seriously you're making more out of it than you need to, you're not losing any performance or features.

[hardcoregames_]

black_zion wrote:

TRIM is an ATA standard and is supported with Microsoft's drivers, same with NCQ. Garbage collection is handled at the drive level and is transparent to the OS, which allowed for drive performance recovery before TRIM became widely supported as well as currently allows drives which are in a RAID setup to maintain performance.

Seriously you're making more out of it than you need to, you're not losing any performance or features.

TRIM is used by SSD drives which are the way the OS tells the drive some sectors are not in use.

NCQ is used by hard disks to stack calls, its similar to the elevator seeking used with Netware long ago

[mockingbird]

black_zion wrote:

TRIM is an ATA standard and is supported with Microsoft's drivers, same with NCQ. Garbage collection is handled at the drive level and is transparent to the OS, which allowed for drive performance recovery before TRIM became widely supported as well as currently allows drives which are in a RAID setup to maintain performance.

Seriously you're making more out of it than you need to, you're not losing any performance or features.

No, I'm not.

What about older SSDs?  An untested, generic driver can be problematic with older chipsets on SSDs.  HOW is NCQ supported with MS' drivers compared to AMD's (for the DEV IDs they do support).  Does MS' drivers handle the same queue depth?  Has the generic MS driver with DEV_ID 43B7 been tested with all iterations of TRIM, including on older SSDs?  So from your point of view, it's OK for AMD to just force people to use an untested, generic driver... I get it.  That's not good enough, PERIOD.  This is all very amateur-ish and unprofessional of AMD, I must say.

Pathetic.

[pokester]

I respect anyone's opinion to do as they choose.

I will say this as an IT pro of over 35 years, I use the MS AHCI drivers currently as they are the most stable and any performance difference is negligible. I would only use another AHCI driver if I had a very specific reason to do so. I have run into nothing but problems using the AMD, NForce or other AHCI (or earlier controller standards) drivers in the past. The only proprietary AHCI drivers that I have seen that are good are the Intel drivers, but they however are pretty much the same as the MS standard driver with added functions on top. I have to totally agree with you black zion on your assesment. The other good thing about the standard MS driver is it just plays nice with everything else too. I have yet on literally thousands of machines I have serviced in our corporation over the years run into a single issue ever using the standard MS driver, not one. The only time I have ever had issue is in the opposite, trying to make one of the 3rd party drivers play nice. My opinion is you stay as close to vanilla, stock settings as possible unless you have real reason to do otherwise.

The exceptions on this all is if you need certain RAID management features or SMART management that only proprietary drivers can do correctly. Even there, my opinion is that is the wrong way to do raid. If you really need raid, and most that think they do probably don't need it, then I would use hardware raid controller not a software raid. I do however think some of what you can do with raid with the new Threadripper chips is pretty neat, but that would really be way beyond my needs.

I also don't think I would jump to a conclusion because a proprietary driver doesn't exist that it is somehow a letdown and they are being lazy. It may simply be that they designed the chip to work perfectly with the standard. It may also be that any other functions are controlled by firmware and do not need special software which is a way better solution than a 3rd party driver could ever be.

Again though, everyone is entitled to their own opinion, and that doesn't make them wrong. Only you can decide what is right for you.

I can tell you that you won't get anything saying something in these forums except a sympathetic ear at best as we are all just users and peers of each other and not AMD employees. If anyone needs to let AMD what they want them to do I suggest you let them know by submitting and Issue Report AMD Issue Reporting Form or Support Request Email Form .  The issue reports are a blind report, they don't respond to you, it's like a suggestion box.

[jsebastian12]

ACTUALLY THOSE WHO "STATE/BELIEVE" THAT THERE IS NO PERFORMANCE DIFFERENCE BETWEEN MS AHCI VS. AMD AHCI DRIVERS ARE MISINFORMED & NOT TRUE!!! I'VE PROVEN THIS MISCONCEPTION FOR OVER A YEAR NOW... HERE'S A SCREENSHOT FROM SAVED CDM BENCHMARKS...

[mockingbird]

Thanks for confirming what we've been insisting on all along.  These screenshots prove that Microsoft's AHCI Driver Performance is significantly slower than branded AHCI drivers (In this case AMD's).  And why wouldn't they be slower?  It's absurd to state that a generic driver should have the same performance as one programmed by the company's own engineering team that know their own product inside-out.

AMD and ASMedia have refused to release a driver (the new chipset is ASMedia eventhough it is branded as AMD).

If you must use an AMD system, I recommend purchasing a motherboard that exposes BOTH the new and old SATA ports.  I was able to switch over to the AMD SATA ports on my board (PRIME B350M-A/CSM).  But most boards only physically have the new chipset SATA ports.

[plumboby]

nope coz Microsoft drivers for ahciis prety standardised. If AMD to make drivers to suit they will run into issues. Most the time the way 10 is build the Microsoft drivers are prety stable. I use them same drivers for old ssds & hard drivers have no preformance hit I actually gain speed as the driver features build into 10 complement the migration. I have many builds with Raid activated using Microsofts drivers without issues, they upgrade them regularly more than the driver date you see. I got a few IT m8s working for Microsoft & in good terms with them, & Microsoft being making drivers to suit Intel & AMD atm as its easier as they can the drivers right per diffrent build of 10 being released. Balck zion is prety right with his postings & a prety onto it contributer to the community from his results they dont vary much from my diffrent builds so I am happy I have stuck with Microsofts drivers. Mockingbird & Hardcoregames your both wrong & worried over nothing. If your running 10 run Microsofts drivers with being older ssds or newer should effect it vice vera on my own testing to what Black zion is done is prety true that the Microsoft drivers arnt as bad as alot abuse them for. They are the original 7 drivers as it  is re written to suit 10 coding. & most new gen builds or builds from 10 years all have bios supporting ahci as it is.

[mockingbird]

Let's face it:

1)  Intel still releases AHCI drivers.  Their latest driver is only two months old and dates from 02/07/2018

2)  AMD has always released AHCI drivers, up to version 1.3.1.277 which dates from 04/15/2015

If MS' standard AHCI drivers were adequate, why does Intel still bother to release its own?  Intel is the gold standard, and AMD is back to their old ways of flying by the seat of their pants.  Why did they bother to release their own AHCI drivers for 10 years and then stop?

This is laziness and incompetance, pure and simple.  Get your act together, AMD.

[elstaci]

I understand you viewpoint. My computer also has MS 2006 AHCI driver. In fact, there are many MS drivers from 2006 being used in my Device Manager. I haven't seen any issues because of those drivers. At one time, I did try to update the MS 2006 drivers and couldn't find any that were correct for my AMD Motherboard or its components.

Exactly what issues are you having that you believe is due to MS 2006 driver not being updated?

[mockingbird]

Some devices use the standard MS drivers (or are renamed by the driver installation to something the manufacturer chooses, but actually use MS files).  But stuff like SMBus or other critical system devices do actually have custom drivers that are installed during a chipset driver installation.

I am having no issues, because I moved the HDD to a port which is DEV_7901 instead of 43B7.  The point is I don't want to have to find out​ about issues in the future.  I have _never_ used MS generic AHCI drivers.  In fact, it is the first driver I update because I want proper SMART and AHCI interaction between the HDD/SSD and the system before doing anything else.

[plumboby]

The microsoft drivers allow for it in there current state. I run smart interaction between my ssd/hdd no issues on MS drivers & are the best drivers for the interaction. Wouldn't trust intel chipset drivers usually they release patches as they are flawed. I run MS drivers for ssd & hdd as I know they always are compatible & run no issues smooth as.

[mockingbird]

I'll cease installing chipset drivers altogether and stop using official AHCI drivers when pigs fly.

[hardcoregames_]

AHCI is handled in Windows/Linux etc with no driver needed.

It goes back before UEFI but they updated it as part of the UEFI spec

[mockingbird]

Comparing Linux to Windows here is unfair.  Linux uses an open source generic AHCI module.

The Windows generic AHCI driver is closed source.

Since Windows is closed source, the manufacturer must maintain their own driver for the OS.

43B7 is almost certainly a Silicon Images device rebranded with an AMD device ID.  AMD and Silicon Images need to get their act together and release a Windows driver, period.  No ifs, ands, or buts.

[Matt_AMD]

Our ACHI driver is bundled in box with Windows 10 OS, that's why we don't include it in the Chipset/driver package. You can safely ignore the date that says the driver is from 2006 in this instance.

[mockingbird]

amdmatt -

Windows 10 is using generic MS drivers for 43B7 which is actually a rebadged ASMedia AHCI controller.

I recommend people not use these ports and physically move their devices to the DEV_7902 ports, which are actual AMD AHCI ports and have official AMD drivers available for them.

And if anyone doubts that there is no official AMD driver, and that the one Windows 10 uses is the generic Microsoft driver, then one needs not look any further than the Microsoft Update Catalog:

Microsoft Update Catalog

If you follow that link, you will clearly see that AMD has not submitted a driver for DEV_43B7 to Microsoft, otherwise it would have shown up in their database.  Windows 10 isn't doing anything special for this device, it is just installing the generic AHCI driver for it.  I must reiterate that this is very poor practice on AMD's part, for the very simple reason that this is unprecedented.  43B7 will be the first AHCI device ever, from any vendor, not to have an official driver released for it.  There is no one to turn to or speak with if glitches arise in the future with this device/driver combination, and there is no accountability, which is very unprofessional on the part of AMD.

[Matt_AMD]

I have passed your feedback onto engineering and will update this thread once i hear back from them.

[mockingbird]

Thank you.

I am looking forward to their response.

[Matt_AMD]

We do not provide a Sata ACHI driver for Ryzen or Raven CPUs. We utilize and do our testing using the in the box driver provided by Microsoft, the same applies to Linux.

Can you double check the DEV_7902 port, as this is not related to the AMD Sata Controller, which is DEV_7901 port?

[hardcoregames_]

DEV_7902 is the FCH SATA Controller [RAID mode] for AMD

no driver is needed as this is handled by the operating system as are all SATA and IDE devices

the spec goes back to the early days of IDE and the 80386 when ISA paddle board with a 40 pin header on them came to the market

over time the 40-pin changed to the serial format used today but the same ATA command set is still used with some extensions

so in answer to the OP, the entire disk controller stack is handled by the OS with no driver needed

[mockingbird]

You're muddying the waters of the topic at hand here with an analogy that appeals to ignorance.

SATA is a far more sophisticated interface than the serial port of olde, and the differences with regard to proprietry between the SATA controllers per each manufacturer necessitates a unique driver release for each controller.

[hardcoregames_]

mockingbird wrote:

You're muddying the waters of the topic at hand here with an analogy that appeals to ignorance.

SATA is a far more sophisticated interface than the serial port of olde, and the differences with regard to proprietry between the SATA controllers per each manufacturer necessitates a unique driver release for each controller.

drivers are generally not needed for most SATA controllers and USB devices as standards are very clear with reference designs etrc

SATA came to be for an entirely new reason, the connector is designed to be plugged into a backplane for servers and RAID boxes. SATA uses one channel and SAS uses an extra 7 pins to add another channel for more bandwidth.

SATA uses the ATA version 6 command set while SAS uses the SCSI command set. SCSI is more expensive as it requires more logic on the device such as bus mastering etc.

USB is host and peripheral so its architecture is simple, while Firewire, like SCSI needs logic to handle bus mastering etc

[mockingbird]

Ironic that you mention Firewire and compare it to SCSI, when in fact Firewire is the only interface out of all the ones you mention that actually almost always has drivers built into Windows for it.

[mockingbird]

The Linux driver is open source, the Microsoft drivers are not.

What exactly do you do when you encounter problems with the Microsoft driver?  Relying on Microsoft to write a driver for your hardware is unprofessional, not to mention poor practice.

Yes, it may have been DEV_7901.  I no longer have the systems I already took them to the customer.  I installed the pure AHCI drivers on those ports (1.3.1.277, amdsata.sys and amdxata.sys).

You really do need to get in touch with your partner ASMedia and have them release a public driver for 43B7.

[Matt_AMD]

If you experience an issue with the in the box driver you can post about it here  and we will investigate, if we can reproduce the issue we will try to resolve it.

The driver you have installed (1.3.1.277) is for Windows 7 and was last updated in 2015. We do not test or recommend using this driver with Windows 10, so if you experience issues we recommend uninstalling that driver and using the in the box driver.

[mockingbird]

1.3.1.277 is not for Windows 7.  It is for Windows 8-10.

The last driver for Windows 7 was 1.2.1.402.

You are misreperesenting AMD by saying you do not recommend 1.3.1.277.

Ok, I'm going to be selling another Ryzen system soon and will run my own performance tests on 7901 vs 43B7.

[hardcoregames_]

I have a motherboard or 2 with ASmedia logic on it, works fine. I have a PCI Express card with ASmedia on it too.

ASmedia also has USB 3.0 in addition to SATA ports which again are standardized.

ASmedia builds to the same design as Intel and AMD use with their logic the functions are standardized long ago.

[mockingbird]

Except ASMedia releases drivers for their own SATA chips.

Their latest version is 3.2.3.0 from November of 2017.

AMD could easily ask ASMEdia to release a driver for the 43B7 pots.  Maybe they're hiding something...  Does it have something to do with the promontory exploits?

[plumboby]

I think mockingbird is an Intel Fanboy the ways hes stating it to amdmatt..Mockingbird unfortunately Intel does the same thing with there drivers. Motherboard venders may release chipset drivers but the achii driver from Microsoft will always supersede any new release AMD or Intel as will clash does happen. Not bad practise its a thing Microsoft does for both Intel/AMD.  There nothing wrong with Microsofts achi drivers they always update them with the newer 10 updates on the 10 Insider program updates There been a lot of Ryzen rework for 10s next update whys been delayed so long. Know a couple testing the engenering chip & found a Microsofts achi  drivers no issues with the new 470 boards. Mockingbird stop scare mongering its your personal view but not every1 even alot of IT pros here that just run Microsofts achi drivers.

[mockingbird]

Why am I planning my future builds on Ryzen 2200GE if I'm an Intel fanboy?  I'm a price for value fanboy.  Right now, Ryzen 2200GE really appeals to me for office builds (I found the 2200G runs a bit too hot - because AMD is using TIM instead of solder with these).  Nothing Intel offers can compare with it ATM, and Intel's Vega CPUs only seem to be for their higher-priced parts.

And no, Plumboy, Intel doesn't do, nor have they ever done this sort of thing.  Intel has official AHCI drivers from their ICH6R chipset all the way up to their most modern AHCI implementation.

Not that I care.  If Intel has a better product, I sell that product, and vice versa.  I'm complaining about this ​because​ I like the product and I want to be happy with it, not because I'm against the product.

[angryuser1]

I came here to ask which .inf file in chipset driver installer was the sata driver, because my hdds are slower than they used to be on intel, and found out that there's actually NO driver? Really? Do I understand it correctly? I expected a mistake on my part or bug in the installer, but that is just very, very weird. I literally get ~40-60MB/s copy speed between drives for larger files (above 1GB), when on z97 intel I had consistent 120-130MB/s speeds on the very same drives(It'd actually start at 300-400MB/s for a few seconds thanks to the disks cache being used, and then dropping to 100+).

I guess I wasted some money on a x370 motherboard then. I'll still recheck if it's asmedia or AMD controller that handles my drives, because I'm unsure about that, but it's still dissapointing it's not hassle free a year after release.