ASRock.com Homepage
Forum Home Forum Home > Technical Support > AMD Motherboards
  New Posts New Posts RSS Feed - 970M Pro3 VRM components
  FAQ FAQ  Forum Search Search  Events   Register Register  Login Login

970M Pro3 VRM components

 Post Reply Post Reply Page  123>
Author
Message
PetrolHead View Drop Down
Groupie
Groupie


Joined: 07 Oct 2015
Status: Offline
Points: 403
Post Options Post Options   Thanks (0) Thanks(0)   Quote PetrolHead Quote  Post ReplyReply Direct Link To This Post Topic: 970M Pro3 VRM components
    Posted: 17 Jan 2016 at 7:58am
I came across a review of this motherbord (http://cxzoid.blogspot.fi/2015/06/asrock-970m-pro3-review.html) and the reviewer had found out that the MOSFETs are unlabeled. Since the manufacturer likely has an impact on quality, I'm interested in knowing where those MOSFETs have come from. Furthermore, it would be nice to know what current they are rated for.

I know this is a long shot, but does anyone have any information on this?
Ryzen 5 1500X, ASRock AB350M Pro4, 2x8 GB G.Skill Trident Z 3466CL16, Sapphire Pulse RX Vega56 8G HBM2, Corsair RM550x, Samsung 960 EVO SSD (NVMe) 250GB, Samsung 850 EVO SSD 500 GB, Windows 10 64-bit
Back to Top
wardog View Drop Down
Moderator Group
Moderator Group


Joined: 15 Jul 2015
Status: Offline
Points: 6447
Post Options Post Options   Thanks (0) Thanks(0)   Quote wardog Quote  Post ReplyReply Direct Link To This Post Posted: 17 Jan 2016 at 8:22am
Looks to me like LF Pak, NXP Semi, LowRDS
Back to Top
PetrolHead View Drop Down
Groupie
Groupie


Joined: 07 Oct 2015
Status: Offline
Points: 403
Post Options Post Options   Thanks (0) Thanks(0)   Quote PetrolHead Quote  Post ReplyReply Direct Link To This Post Posted: 19 Jan 2016 at 7:31pm
Thank you for the info, wardog. I tried to figure out whether that was good or bad, but only managed to determine (using Sin's outdated VRM list: http://sinhardware.com/images/vrmlist.png) that the parts were probably just acceptable and not suitable for overclocking.

Over at overclock.net, The Stilt had given this verdict on the VRM (http://www.overclock.net/t/946407/amd-motherboards-vrm-info-database/1470#post_24802727):

ASRock 970M Pro3 = 4+1 true phase analog, <95W continuous (estimated)- NOTE: (don't use with any FX-8K series)

When I asked what this estimation was based on, I received the following answer:

It is pretty easy to tell the VRM capabilities just from the looks of it. At least after you have some experience on the platform and after you have designed couple of motherboard VRMs yourself biggrin.gif

970M Pro3 has 4+1 phase analog VRM with Renesas K03J2 (LS) and K03J3 (HS). While these are relatively high quality fets the fact that there are just 4 phases for VDDCR, combined with the usual issues typical to ASRock (low copper, low layer PCB, low quality high-loss inductors) makes the board certainly not recommended for anything higher than 95W. Nearly identical construction is used on several ASRock AMD boards (970 Performance, Extreme series, etc). with the only real difference being the advertized phase count. These boards usually have the very same VRM structure, but with doubled component count ("8+2" phase, in fake phase configuration). Despite ASRock advertized them as "220W TDP, FX-9K series" compatible they commonly have issues in running slightly overclocked FX-8K series CPU without the VRM cooking off / throttling.

So no, I would not recommend this board for anything higher than 95W 32nm CPU biggrin.gif

The interesting thing is that in a comment previous to this, The Stilt commented on my worries about having the Phenom II X6 overclocked and showing over 100W power draw in HWMonitor nearly constantly:

The VRM requirements for 45nm 10h parts and 32nm 15h parts are not really comparable. For example the Phenom II X4 965 (HDZ965FBK4DGI) is rated for 140W TDP and 91.8A IDDMax. Meanwhile the FX-8150 which has TDP of 125W, has significantly higher IDDMax of 145A. The component requirements for a VRM which is able to supply 100A @ 2.0V are significantly easier to meet than the requirements for a VRM which is able to supply 200A @ 1.0V.

So the 95 W recommendation is meant for 32 nm CPUs and 45 nm CPUs may be able to get away with a larger continuous power draw, depending on the max load current.

In any case, ASRock is basically making promises the hardware shouldn't be expected to keep. Again. A stock FX-6300 should be in the clear, but anything above that is in the danger zone.
Ryzen 5 1500X, ASRock AB350M Pro4, 2x8 GB G.Skill Trident Z 3466CL16, Sapphire Pulse RX Vega56 8G HBM2, Corsair RM550x, Samsung 960 EVO SSD (NVMe) 250GB, Samsung 850 EVO SSD 500 GB, Windows 10 64-bit
Back to Top
parsec View Drop Down
Moderator Group
Moderator Group
Avatar

Joined: 04 May 2015
Location: USA
Status: Offline
Points: 4996
Post Options Post Options   Thanks (0) Thanks(0)   Quote parsec Quote  Post ReplyReply Direct Link To This Post Posted: 19 Jan 2016 at 11:39pm
I'm glad people are paying attention to the components and number of phases used in the processor's VRM design in mother boards.

Then we have a typically critical analysis of some AMD boards like the one included above.

Reality Check: The board in question costs $65 at a well known E-tailer in the USA.

What can be expected from a board that is sold for $65?! Confused

How can any board at this price be considered for over clocking? To compare it with boards that are at least three times its price for component and build quality does not make sense.

We don't get what we don't pay for, sorry to say. Geek
Back to Top
PetrolHead View Drop Down
Groupie
Groupie


Joined: 07 Oct 2015
Status: Offline
Points: 403
Post Options Post Options   Thanks (0) Thanks(0)   Quote PetrolHead Quote  Post ReplyReply Direct Link To This Post Posted: 20 Jan 2016 at 2:31am
1. Several things can affect the price at a retailer. I personally paid ~100 e for the board, and while that's still relatively cheap, it wasn't the cheapest board available. It was, however, the only available mATX board with a 970 chipset. In any case, the price is besides the point.

2. OC capability is not the issue and I think you're seriously missing the point here. The point is that ASRock claims this board can handle CPUs that the VRM can't really be expected to handle. Support for up to 140 W CPUs? Compatibility with FX-8xxx CPUs? I wouldn't take the risk knowing what I now know. It doesn't matter if the board costs 200 $ or 50 $, it should be able to deliver what the manufacturer says it will.
Ryzen 5 1500X, ASRock AB350M Pro4, 2x8 GB G.Skill Trident Z 3466CL16, Sapphire Pulse RX Vega56 8G HBM2, Corsair RM550x, Samsung 960 EVO SSD (NVMe) 250GB, Samsung 850 EVO SSD 500 GB, Windows 10 64-bit
Back to Top
PetrolHead View Drop Down
Groupie
Groupie


Joined: 07 Oct 2015
Status: Offline
Points: 403
Post Options Post Options   Thanks (0) Thanks(0)   Quote PetrolHead Quote  Post ReplyReply Direct Link To This Post Posted: 20 Jan 2016 at 6:06am
By the way, I thought it might be a good thing to share some things about The Stilt. After all, if you haven't heard of him, why would you trust him over manufacturer's specifications? This is a very good question. Maybe you shouldn't. But consider these facts:

-The Stilt is a very experienced overclocker. For example, he has pushed an A10-6800K to 8.2 Ghz, an FX-8370 to 8.7 GHz and an FX-9370 to 8.5 GHz.  http://hwbot.org/user/the_stilt/#Hardware_Library

-The Stilt is the one that found and fixed a BIOS bug that was holding AMD CPUs back in SuperPi 32M.

Finnish overclocker, The Stilt, figured out how to considerably improve performance by going through the BIOS developers guides. The exact same guides available to the BIOS R&D teams of motherboard vendors, a surprising fact considering a single man managed to outdo an entire industry.

http://www.techpowerup.com/186056/amd-super-pi-history-to-be-rewritten-courtesy-the-stilt.html


As far as AMD hardware goes - possibly Intel as well - I think it's pretty safe to assume The Stilt is one of the most knowledgeable people you can run into on internet forums.
Ryzen 5 1500X, ASRock AB350M Pro4, 2x8 GB G.Skill Trident Z 3466CL16, Sapphire Pulse RX Vega56 8G HBM2, Corsair RM550x, Samsung 960 EVO SSD (NVMe) 250GB, Samsung 850 EVO SSD 500 GB, Windows 10 64-bit
Back to Top
wardog View Drop Down
Moderator Group
Moderator Group


Joined: 15 Jul 2015
Status: Offline
Points: 6447
Post Options Post Options   Thanks (0) Thanks(0)   Quote wardog Quote  Post ReplyReply Direct Link To This Post Posted: 20 Jan 2016 at 6:21am
Originally posted by PetrolHead PetrolHead wrote:

By the way, I thought it might be a good thing to share some things about The Stilt. After all, if you haven't heard of him, why would you trust him over manufacturer's specifications?


While not addressing previous posts above, nothing beats experience.

We live and we learn along the way.

Sin's and Stilt's engineering backgrounds I'm jealous of, yet even w/o their degrees, experience is how I came by it. I maybe can't decipher a boards schematic but I can now determine when a boards manufacturer is pushing hard against the limits of believably.


Edited by wardog - 20 Jan 2016 at 6:22am
Back to Top
PetrolHead View Drop Down
Groupie
Groupie


Joined: 07 Oct 2015
Status: Offline
Points: 403
Post Options Post Options   Thanks (0) Thanks(0)   Quote PetrolHead Quote  Post ReplyReply Direct Link To This Post Posted: 22 Jan 2016 at 9:34am
While contemplating where to take my OC next and what my VRM can or can't take, I've noticed that there's no mention of over current protection (OCP) regarding the 970M Pro3 anywhere. There's some sort of surge protection, which is meant to deal with power spikes from the PSU, but as far as I can tell, the VRM is not protected in any way. The BIOS does limit the maximum Vcore value to 1.55 V, which may protect the board from the user to some extent. However, as The Stilt pointed out, voltage is not the only variable that affects the durability of the VRM. The load current value of the CPU is another important part of the equation, and this value depends on the CPU. Sadly, it seems very hard to find out what this value is, so it's hard to say for which CPUs that 1.55 V could be considered acceptable, if any. At least there's good reason to believe that it's way too much for FX CPUs on this board...

In any case, back to the lack of OCP. What I'm wondering is will the lack of OCP mean that the VRM will not throttle no matter what and instead just go up in flames when it's had enough? For example, at the moment I'm knowingly taking a risk in light of wardog's advice in another thread, which means that I've crossed the 1.4 V border with the CPU and set it to 1.4375 V in an attempt to see if it's enough for 3.9 GHz on all six cores. HWmonitor does show Vcore values lower than the one that I've set, and most of the time the value is below 1.38 V, so maybe this is still "safe". Maybe not. The point is I've been running Prime95 now for about 45 minutes and there has been no throttling. The cores are under 100% load and the clock speed has been 3900 MHz with only the sort of ~30 MHz occasional fluctuation that is present even at stock speeds. Max temperatures have been 69 C socket and 49 C for the cores, so while the socket is at the limit of what is considered safe, neither value should be an issue. So, everything seems okay and Prime95 has reported no issues. But maybe this stability is just an illusion. Maybe the VRM is just waiting to pop. There's no temperature sensor on the VRM and there's no OCP, so there's no way of knowing how big a risk I'm taking and there is no safety net. At least that's what it seems like at the moment.

I think I'm going to buy a handheld infrared thermometer at some point to monitor the VRM temperatures. Then again, what would the VRM be rated for? 85 C, 105 C or something else?
Ryzen 5 1500X, ASRock AB350M Pro4, 2x8 GB G.Skill Trident Z 3466CL16, Sapphire Pulse RX Vega56 8G HBM2, Corsair RM550x, Samsung 960 EVO SSD (NVMe) 250GB, Samsung 850 EVO SSD 500 GB, Windows 10 64-bit
Back to Top
parsec View Drop Down
Moderator Group
Moderator Group
Avatar

Joined: 04 May 2015
Location: USA
Status: Offline
Points: 4996
Post Options Post Options   Thanks (0) Thanks(0)   Quote parsec Quote  Post ReplyReply Direct Link To This Post Posted: 22 Jan 2016 at 12:39pm
Originally posted by PetrolHead PetrolHead wrote:

1. Several things can affect the price at a retailer. I personally paid ~100 e for the board, and while that's still relatively cheap, it wasn't the cheapest board available. It was, however, the only available mATX board with a 970 chipset. In any case, the price is besides the point.

2. OC capability is not the issue and I think you're seriously missing the point here. The point is that ASRock claims this board can handle CPUs that the VRM can't really be expected to handle. Support for up to 140 W CPUs? Compatibility with FX-8xxx CPUs? I wouldn't take the risk knowing what I now know. It doesn't matter if the board costs 200 $ or 50 $, it should be able to deliver what the manufacturer says it will.


You are the one that posted Stilt's quote about over clocking with boards like this. Why wouldn't someone assume that is what you are referring to?

I only see one 140W TDP processor in the CPU support list, a Phenom II x4 processor. Should you use 140W FX-8000 series processors in this board? No, they are not in the support list.

Yes it states support for 140W processors in the specs. But when you read the fine print (CPU Support list), you find only one, and no 140W FX-8000 processors.

So no OC with this board? Fine, the only mention of over clocking in its description is the X-Boost feature, which is for unlocking cores.

Personally, after being burned many times by Marketing "specs" like "ASMedia or Marvell SATA III chips", which qualify as SATA III only because they can surpass SATA II speeds somewhat, we must look closer at the true specifications of a board to see what is really the case. Does "AM3+ 8-Core Processor Support" mean that all AM3+ 8-Core processors are supported? Many people will read that as such, but does it really say that? IMO, no it doesn't.

Do I agree a board should be able to perform to the level claimed for it? Of course. But if the argument here is it does not support 140W FX-8000 when it claims to do so, the specs are vague enough and the CPU Support list detailed enough to show that argument does not apply.

Frankly, I understand your frustration, and I don't want you to have a board you don't like. IMO, the AMD FX-8000 and FX-9000 series processor and mother board compatibility situation is terrible. It bothers me more than you know or may believe. IMO, AMD shares a good part of the fault by releasing these processors without some control over how they are used. While Intel is hated for the amount of control they force upon mother board manufactures, only allowing processors to work or be over clocked on specific platforms, AMD could learn something from them in this case.
Back to Top
PetrolHead View Drop Down
Groupie
Groupie


Joined: 07 Oct 2015
Status: Offline
Points: 403
Post Options Post Options   Thanks (0) Thanks(0)   Quote PetrolHead Quote  Post ReplyReply Direct Link To This Post Posted: 22 Jan 2016 at 7:38pm
Originally posted by parsec parsec wrote:

You are the one that posted Stilt's quote about over clocking with boards like this. Why wouldn't someone assume that is what you are referring to?


The quotes from The Stilt are not about overclocking with boards like this. He only mentions overclocking once as an example of how overly optimistic ASRock's promises are: If a board is supposed to be able to handle an FX-9xxx, it should be able to handle a slightly overclocked FX-8xxx without issues (even knowing that the FX-9xxx probably requires less voltage to achieve its rated speed than an FX-8xxx would). What The Stilt's post is about is this: The 970M Pro3 should be used with FX-8xxx processors, and isn't recommended for 32 nm processors with a TDP of over 95 W, which in practice means the FX-6350.

Quote I only see one 140W TDP processor in the CPU support list, a Phenom II x4 processor. Should you use 140W FX-8000 series processors in this board? No, they are not in the support list.


I meant those to be two separate, unrelated questions. However, since it's stated in the specs as if there were no differences between similar TDP values of different CPUs, it would be easy to assume that even with FX-8xxx CPUs the user has ~15 W of headroom for overclocking. In any case it's now easy to be sceptical of that value, no matter what CPU it might refer to.

Btw, I'm a bit surprised to see that TDP value. I wonder why it's so high compared to other Deneb CPUs?

Quote So no OC with this board? Fine, the only mention of over clocking in its description is the X-Boost feature, which is for unlocking cores.


The manual also dedicates several pages to overclocking, as the board comes with AXTU. Also the overclocking options in the BIOS are covered. While the board isn't advertised as a good board for overclocking, ASRock doesn't really advise against it either.

But again, the ability to OC wasn't really the point. The point is the CPU support list is optimistic when it shouldn't be. The ability to OC would follow from the CPU support list, however, assuming it could be trusted. For example: if the FX-8370 is supported (as it is), there should be some headroom to overclock an FX-6300. After all, even a stock FX-8370 can go well beyond 125 W TDP depending on load and CPU temperature.

Quote But if the argument here is it does not support 140W FX-8000 when it claims to do so, the specs are vague enough and the CPU Support list detailed enough to show that argument does not apply.


The argument is that the board doesn't really support FX-8xxx CPUs, and that even an FX-6350 is a risk. And by "support" I mean that those CPUs could be run reliably at any load (at stock settings). Of course if the CPU never sees use that takes advantage of what it can offer, maybe one can get away with running one for a longer period of time.

Quote Frankly, I understand your frustration, and I don't want you to have a board you don't like.


Even though it may seem that I complain a lot about this board, I actually like it at the moment. It has so far been able to handle what I've thrown at it, and frankly, there doesn't seem to be anything better on the market in the mATX size. What I am mostly frustrated about is ASRock and the promises they make that make it pretty much impossible for a consumer to make a reasonably informed decision on what to buy. If I had gone for an FX-6350, I would probably have an issue with the board as well, but I got lucky in the CPU department. The other thing I'm frustrated about is that ASRock apparently didn't think making this board any better was worth it, because I see that as a missed opportunity. They already had better VRM designs on other boards, the mATX size doesn't limit the VRM size and the price hike due to the components would not have made this board expensive, since it's so cheap to begin with. This board could have been... wait for it... legendary. ;)

Quote IMO, the AMD FX-8000 and FX-9000 series processor and mother board compatibility situation is terrible. It bothers me more than you know or may believe. IMO, AMD shares a good part of the fault by releasing these processors without some control over how they are used. While Intel is hated for the amount of control they force upon mother board manufactures, only allowing processors to work or be over clocked on specific platforms, AMD could learn something from them in this case.


I believe you and I agree.


Edited by PetrolHead - 22 Jan 2016 at 8:34pm
Ryzen 5 1500X, ASRock AB350M Pro4, 2x8 GB G.Skill Trident Z 3466CL16, Sapphire Pulse RX Vega56 8G HBM2, Corsair RM550x, Samsung 960 EVO SSD (NVMe) 250GB, Samsung 850 EVO SSD 500 GB, Windows 10 64-bit
Back to Top
 Post Reply Post Reply Page  123>
  Share Topic   

Forum Jump Forum Permissions View Drop Down

Forum Software by Web Wiz Forums® version 12.04
Copyright ©2001-2021 Web Wiz Ltd.

This page was generated in 0.344 seconds.