AMD Bristol/Stoney Ridge Thread

[The Stilt]

Hmm, what about FIVR... Does that make boards vrm any simpler ? I would guess so, looking at the haswell boards.

Yes.
Because FIVR input voltage is much higher (1.8V nominal) than on any even remotely modern CPU, the motherboard VRM can be built to much lower standards than usually since the currents are significantly lower.
While the FIVR itself decreases the overall efficiency by quite a lot, the motherboard VRM on these boards is slightly more efficient than usual. Higher output voltage means the duty cycle is higher and the lower currents will reduce the losses through the VRM and the conduction path.
But still despite that, the overall efficiency of FIVR is smaller than on conventional configurations. FIVR operates at extremely high frequencies (due the physical inductor size, in core), around 450x faster (~140MHz) than convertional the VRMs found on motherboards and graphics cards.
Intel has not published the efficiency of newer FIVR implementations, however Haswell's FIVR had around 78% efficiency according to Intel.

The total efficiency of Haswell VRMs (motherboard + FIVR) is around 69% (78% FIVR, 88% motherboard), whereas a high-end conventional VRM on a motherboard can hit around 85% total efficiency on average.

 

[majord]

R9 Nano, 175w , 150+A , 4+1phase is another good practical example of what can be achieved.. , and can handle upwards of 250w

The point that Stilt makes is the key though - Cost of implementing such a VRM on motherboards were competition is fierce is impossible.. we're talking about entire motherboards that retail at as low as $50. We've even seen the same issues on AIB cards, like powercolor's RX 480.

Probably also worth mentioning - the challenge here boils down to putting an 8 core CPU on mainstream socket. For x TDP, the higher the core count, the higher current. Intel don't have this problem currently, with their mainstream socket limited to Quad cores only, so there VRM spec can be more lax.

If it means comparable entry level AMD boards need to cost ~10$ more than Intel counterpart to get a reasonable VRM then i'm all for it. Whether marketing has the same opinion is another matter

 

[hojnikb]

Yes.
Because FIVR input voltage is much higher (1.8V nominal) than on any even remotely modern CPU, the motherboard VRM can be built to much lower standards than usually since the currents are significantly lower.
While the FIVR itself decreases the overall efficiency by quite a lot, the motherboard VRM on these boards is slightly more efficient than usual. Higher output voltage means the duty cycle is higher and the lower currents will reduce the losses through the VRM and the conduction path.
But still despite that, the overall efficiency of FIVR is smaller than on conventional configurations. FIVR operates at extremely high frequencies (due the physical inductor size, in core), around 450x faster (~140MHz) than convertional the VRMs found on motherboards and graphics cards.
Intel has not published the efficiency of newer FIVR implementations, however Haswell's FIVR had around 78% efficiency according to Intel.

The total efficiency of Haswell VRMs (motherboard + FIVR) is around 69% (78% FIVR, 88% motherboard), whereas a high-end conventional VRM on a motherboard can hit around 85% total efficiency on average.

That makes a lot of sense. Also explains why 1150 boards are cheaper. Having cheap 2+1 vrm setup (from what i can gather) can definitely push the pricing down.
Although one does ask himself why they bothered with FIVR in the first place given the efficiency of such setup and the ever need for more efficient CPUs.

 

[inf64]

Edit: wrong thread

 

[cbn]

Comparing Athlon x 4 845 and Athlon x 4 950 (two Excavator quad cores @ 3.5GHz/3.8 Ghz), what effect do people think the memory controller (DDR3 2133 vs. DDR4 2400) difference will have?

Athlon x4 950 a bit slower? About the same? or a bit faster?

 

[ET]

Any rumour regarding release date of DIY Bristol Ridge? I've been waiting for it (im)patiently since year's start, I'm stuck with a 4TB drive that my HTPC's nForce drivers can't use, and I'm starting to feel really sorry that I didn't just go for an A10-7860K instead of waiting for the pie in the sky that is Bristol Ridge AM4.

 

[dark zero]

And no news about Stoney Ridge.

 

[Insert_Nickname]

Any rumour regarding release date of DIY Bristol Ridge?

"Soon". Sometime in this month perhaps. February at the latest.

I've been waiting for it (im)patiently since year's start, I'm stuck with a 4TB drive that my HTPC's nForce drivers can't use, and I'm starting to feel really sorry that I didn't just go for an A10-7860K instead of waiting for the pie in the sky that is Bristol Ridge AM4.

Technically, the reason you can't use a 4TB drive on an nForce chipset is the BIOS. Not the drivers.

As for upgrading, under the circumstances, I'd wait until BR/KBL anyhow. Reason being native HDMI 2.0 support.

 

[ET]

Technically, the reason you can't use a 4TB drive on an nForce chipset is the BIOS. Not the drivers.

As for upgrading, under the circumstances, I'd wait until BR/KBL anyhow. Reason being native HDMI 2.0 support.

From what I've read, technically the reason is the drivers. In fact the BIOS screen reports it as 4TB. I've read that Linux can work with 4TB drives on nForce, though haven't tried it.

As for upgrading, I'm not sure what to do. I really need that HDD upgrade.

Edit: I found a post explaining how to disable the nVIDIA driver for a specific controller and use the default Windows driver, which should solve the problem. I will try this later. If that works, it should tide me over until BR is released.

 

[Insert_Nickname]

From what I've read, technically the reason is the drivers. In fact the BIOS screen reports it as 4TB. I've read that Linux can work with 4TB drives on nForce, though haven't tried it.

Sorry, meant as a boot drive. Booting from a 2TB+ disk requires using GPT, and using UEFI mode for Windows. I don't recall seeing any nForce boards use UEFI. Legacy BIOS can only boot MBR drives, and those are limited to 2TB.

As a data disk such a drive should be pretty much plug and play on 7 and newer. You might encounter a few issues getting the drive initialized and partitioned however, I'd recommend using a computer with UEFI for the initial setup. Saves hassle and the need to use 3rd party software.

As for upgrading, I'm not sure what to do. I really need that HDD upgrade.

Edit: I found a post explaining how to disable the nVIDIA driver for a specific controller and use the default Windows driver, which should solve the problem. I will try this later. If that works, it should tide me over until BR is released.

For 7 and newer I'd just use the Windows standard AHCI driver. If you're running in AHCI mode. If you're running in IDE mode then the standard IDE driver. They're actually quite good. There are a few compatibility issues surrounding nForce SATA controllers however, so YMMV.

If all else fails, a cheap SATA add-on card will fix at least accessing the drive.

 

[ET]

Setting the IDE driver worked nicely. So need for BR lessened for now.

 

[LTC8K6]

Comparing Athlon x 4 845 and Athlon x 4 950 (two Excavator quad cores @ 3.5GHz/3.8 Ghz), what effect do people think the memory controller (DDR3 2133 vs. DDR4 2400) difference will have?

Athlon x4 950 a bit slower? About the same? or a bit faster?

I think it would be hard to tell the difference.

Which board can use the fastest ram with the chip?
That's probably the decider?

 

[Abwx]

1-Liter-Mini-PC mit leistungsstarker 35-Watt-APU AMD Pro A12-9800E

https://www.computerbase.de/2016-10/lenovo-thinkpad-thinkcentre-tiny-amd-pro/

 

[KTE]

1-Liter-Mini-PC mit leistungsstarker 35-Watt-APU AMD Pro A12-9800E

https://www.computerbase.de/2016-10/lenovo-thinkpad-thinkcentre-tiny-amd-pro/

I really need to test this or see a test of it ASAP. Then I can recommend at work, since Lenovo is our partner

Sent from HTC 10
(Opinions are own)

 

[ET]

The press release for that has some interesting figures.

 

[Erenhardt]

1-Liter-Mini-PC mit leistungsstarker 35-Watt-APU AMD Pro A12-9800E

https://www.computerbase.de/2016-10/lenovo-thinkpad-thinkcentre-tiny-amd-pro/

Quite a large pencil

(Inappropriate graphics for joke, but still )


Anyway. I would rather see DIY Market parts.

 

[DrMrLordX]

Anyway. I would rather see DIY Market parts.

Agreed! I am quite disappointed that AMD hasn't put anything new in the DIY channel. No B350 boards, no A12s, no nothing. It was supposed to be available Oct. 2016, and now we are a week away from month's end. We are now looking at a Nov/Dec release. What gives?

 

[ET]

It was supposed to be available Oct. 2016

It was supposed to be available in March, then June, ... Hard to base buying decisions on rumours.

I still prefer seeing OEM announcements to seeing nothing at all, but yeah, let me buy DIY Bristol Ridge AM4 already!

 

[Justinbaileyman]

Yeah we all have been waiting!! Whats the deal here, where are they?Jeez AMD release something and let us give you our money already..

 

[Abwx]

It was supposed to be available in March, then June, ... Hard to base buying decisions on rumours.

I still prefer seeing OEM announcements to seeing nothing at all, but yeah, let me buy DIY Bristol Ridge AM4 already!

But it was available in march for the mobile segment, the confidential RM leaked by Benchlife info in march 2015 has been quite accurate overall since it mention the DT version launch as projected for july 2016, wich is right on schedule since OEMs got the chip in early august at the latest to release products in september..

Surely that the reason for no DIY launch has to do with OEMs being enough to absorb the whole current
production, and while they re at it they benefit from this demand to get rid of previous inventories in the DIY market, it s clearly visible that previous chips prices are progressively reduced since a few months.

 

[VirtualLarry]

I would love to talk a friend of mine into a Bristol Ridge desktop. He watches a lot of internet-based video, and has an Athlon II X4 currently, with a GT610.

 

[AtenRa]

The press release for that has some interesting figures.

Link is broken i cannot open it.

 

[ET]

Link is broken i cannot open it.

This is the press release at AMD's site, but it doesn't contain the footnotes, which are of some interest, and do appear on that link (which I also find inaccessible at the moment).

Anyway, the spec sheet is interesting. (Link taken from the Tech Report.)

 

[AtenRa]

This is the press release at AMD's site, but it doesn't contain the footnotes, which are of some interest, and do appear on that link (which I also find inaccessible at the moment).

Anyway, the spec sheet is interesting. (Link taken from the Tech Report.)

thanks

 

[DrMrLordX]

It was supposed to be available in March, then June, ... Hard to base buying decisions on rumours.

I still prefer seeing OEM announcements to seeing nothing at all, but yeah, let me buy DIY Bristol Ridge AM4 already!

Well it's like this, Bristol Ridge is available from OEMs, so all the rumour junk is basically over. Bristol Ridge is here, AM4 is here, just . . . not for DIY buyers.

Yeah all the people predicting availability as far back as March were looking for signs and/or reading tea leaves. In all probability AMD could have prioritized availability of Bristol Ridge and started shipping back then. If you want to get technical Abwx is right but it's not worth it to really argue that point.

Point is AMD is acting like they have no desire/incentive to update their consumer APU lineup for DIY buyers. It must be that they think their market share in that segment is so low that it may not even be worth trying.

I for one would probably get one, but at this rate I guess I'm also looking at Summit Ridge? I really don't want to ride this A88x-Pro until whenever Raven Ridge finally launches. I've had it since late 2014! It was a hell of a deal at the time but I'm ready to move forward. I just don't know how comfortable I am shifting to a system that relies on a dGPU for all graphics duties. Having an iGPU is really handy.