[Overview] Ryzen CPU & AM4 Mainboard Lineup

[unseenmorbidity]

ShopBlt has 4 msi boards listed now with prices

 

[Mopetar]

The Stilt's hint gives me a discouraged expectation that few of us will be happy to learn what XFR actually is.

Doesn't that just lend some credence to the idea that it's just an auto-OC then? If you manually OC yourself (which will probably allow you to be more aggressive anyways) then it's completely useless. I suppose it's nice for people who don't really understand how in that it gives them to extra performance, but I don't see them buying high-end enthusiast chips to begin with.

The only other thing that makes sense (at least based on the name) is that XFR just means it's a binned chip that has some additional OC headroom relative to others, or perhaps its some combination of both since you'd need good silicon to have the room to do an auto OC.

 

[lolfail9001]

Doesn't that just lend some credence to the idea that it's just an auto-OC then?

No no, you do not quite get my suspicion. Namely it is that non-XFR SKUs will have much more than just better bin and XFR enabled.
I'll express my point this way: Intel CPUs with unlocked multiplier have Extended Frequency Range too.

 

[unseenmorbidity]

No no, you do not quite get my suspicion. Namely it is that non-XFR SKUs will have much more than just better bin and XFR enabled.
I'll express my point this way: Intel CPUs with unlocked multiplier have Extended Frequency Range too.

Except AMD shemselves stated that all zen cpus are unlocked. The xfx might have additional oc features, but they can all be oc'd.

___________________________________________________

MSI Boards

http://www.shopblt.com/search/order_id=805920186&s_max=25&t_all=1&s_all=AMD+AM4+&search=Search

 

[lolfail9001]

Except AMD shemselves stated that all zen cpus are unlocked. The xfx might have additional oc features, but they can all be oc'd.

i7-3820/E5-1620 (or 30?) had unlocked multiplier that was hard capped at 42 multiplier nonetheless.

Anyways, that is my worry, nothing more, and probably false to boot. But it feels kinda weird to have your unlocked 8 cores range in price so widely. To me that just means there's a catch. What is it? Well, then we have this

Admittingly the XFR slide was extremely confusing, but still the slide doesn't say anything about XFR being an "automated overclocking" feature.
XFR stands for Extended Frequency Range and that's exactly what it is, literally.

 

[IEC]

Less than two weeks until NDA expires and more details can be revealed.

 

[unseenmorbidity]

i7-3820/E5-1620 (or 30?) had unlocked multiplier that was hard capped at 42 multiplier nonetheless.

Anyways, that is my worry, nothing more, and probably false to boot. But it feels kinda weird to have your unlocked 8 cores range in price so widely. To me that just means there's a catch. What is it? Well, then we have this

I feel you there. It does seem like we are missing some critical information. Higher binning isn't generally worth a 60% increase in cost. Nearly $200 difference.

 

[lolfail9001]

I feel you there. It does seem like we are missing some critical information. Higher binning isn't generally worth $200.

Well, siliconlottery disagrees, but i feel AMD are not looking to put them out of business either. Eh, we will learn soon indeed. Unless of course AMD only samples reviewers with X SKUs, lol.

 

[Elixer]

Like PCI-E 2.0 lanes.

Um, the Z200 series is physically limited by the 4 PCI-E 3 lanes that make up the DMI 3.0 link between the CPU and PCH, so, I wouldn't exactly say the way AMD is breaking this down is "pretty bad".
AMD could have had them all be PCI-e gen 3 (like intel), but, there isn't enough bandwidth available to service everything (just like intel can't service everything), which means that things slow down, while traffic is serviced.

Overall, it is a design choice limited by the bandwidth they have available.
This is also why some mobo makers could add some PLX chips to increase the number of lanes (great for advertising!), but, they are all still starved for bandwidth (whoops, bad for advertising!) if you have enough devices fighting for it.

 

[lolfail9001]

Overall, it is a design choice limited by the bandwidth they have available.

Probably, though having more lanes off of chipset is way more convincing when they will be choked by bandwidth anyways, don't you think?

I wonder how much x300 boards will cost with them having a fake chipset (so, basically, entire set of controllers outsourced to mobo makers) and all.

 

[itsmydamnation]

Probably, though having more lanes off of chipset is way more convincing when they will be choked by bandwidth anyways, don't you think?

I wonder how much x300 boards will cost with them having a fake chipset (so, basically, entire set of controllers outsourced to mobo makers) and all.

What are you on about?

You have 4x pcie3 dedicated to the CPU for NVME
You have 4x pcie 3 dedicated to the CPU for the chip set
You have 4x usb dedicated to the CPU for usb3.0
You have 16x pcie3 tdeicated to the CPU for GFX

the chip set has 2/6/6 USB , 4 SATA, 2 SATA-E or 1 NVME and 8x pcie2.
What exactly do you need. The Second you have to do anything with your data (bitlocker for example) Zen wins because it can sustain higher bandwidth.

 

[lolfail9001]

What are you on about?

On the chipset itself. I mean, sure, Zen is a SoC, and this "chipset" only serves same purpose ASMedia's USB3.1 controllers served on Z170 motherboards + general management purpose, but nonetheless, it does not make pricing justice.

the chip set has 2/6/6 USB , 4 SATA, 2 SATA-E or 1 NVME and 8x pcie2.

Actually, only pci-e 2 and USB, if this has any weight: http://media.gamersnexus.net/images/media/2017/CES/amd/AM4-block-diagram-gn_1.png

And since Crosshair VI has 1 less M.2 than Maximus IX [Hero versions of both], i conclude that this chart is correct.

 

[krumme]

A 350b board that can oc is 79usd. 79usd. Man. Its what 99.99% users need. Its cheap. You are just looking for some bad things to say about zen. "Pricing justice".
Post after post. Same agenda.

Think the glass is full not half empty. The negativity is not even the slightest rational.

So you can have a 320usd cpu on a 79usd mb for a total cost of 400 usd and get 6900 like perf from it. Its a third of the cost. The world is so unfair. Thats not "justice" ?

Well go buy intel if you find zen a bad deal because its a third of the cost. Pls stop the complaining. I am sure you can find a sata port or pci lane as a reason and enjoy your prime number crunsching.

 

[itsmydamnation]

Actually, only pci-e 2 and USB, if this has any weight: http://media.gamersnexus.net/images/media/2017/CES/amd/AM4-block-diagram-gn_1.png

And since Crosshair VI has 1 less M.2 than Maximus IX [Hero versions of both], i conclude that this chart is correct.

Your interpretation is wrong and your conclusion is wrong, here are the chips sets.
http://hothardware.com/ContentImage...summary.jpg.ashx?maxwidth=1170&maxheight=1170


its disk IO is 4 SATA off the chipset + 4PCIE gen 3 broken up into whatever the MB maker wants it to be.
Majord covered it here:

http://i.imgur.com/Luc2HEW.png

 

[lolfail9001]

You are just looking for some bad things to say about zen.

I am looking to say bad things about mobo makers. Mostly because i see no cool itx boards in sight and that annoys me.

Post after post. Same agenda.

You have me on ignore i presume, because you would know my agenda by now i'd think: Ryzen is great, but we do not know the catch of "x" letter.

A 350b board that can oc is 79usd. 79usd.

https://mdata.yandex.net/i?path=b0424104327_img_id1355643266809802025.jpeg
Here is it's Skylake brother. Would you want to OC on such board?

Your interpretation is wrong and your conclusion is wrong, here are the chips sets.

My facts are dead on though: no 2xNVMe on the flagship board. So probably no 2xNVMe at all, so majord's interpretation may just be wrong as well. Especially since it copies itself.

EDIT: On second look though even if it is correct, there's some weird stuff going on with lanes.

 

[krumme]

The vrm specs for zen is tight from amd side. No mb crap this time as stilt said. So no i would have no trouble oc on a cheap board.

 

[krumme]

The lack of itx mb at launch is a real world problem.

Lack of avx2, prime bm crap, 2 sata ports or a few pci lanes is all stuff no consumer needs and will need. Reviewers and Intel fans will talk about it but it have no real impact for 99.99% of the users. Heck there is even a cf and sli option to make sure the last 0.05% segment is covered. So its all sidetracking it. Its not relevant in consumer market.

 

[EliteRetard]

I wish there was an option even higher than the X370, but I'm pretty sure that'd require a different CPU/socket at this point.
Basically I was hoping with the higher core count (8+) CPUs we would have an option like Intels HEDT with 40x PCIe 3.0. It looks like there's roughly 24 lanes on the AMD CPUs. If they had made the socket capable of say 40 lanes, but limited the available lanes and chipsets for the lower core/lower price CPUs (like Intel splits their HEDT line). I wonder how much extra it would cost. I imagine it would be very minimal...since you wouldn't even have to put the extra physical pins in the socket. Just some extra plastic and a physically larger socket.

Example: CPU/socket has 1500 pins, only 1000 are needed for everything X370 and lower. If you plan to use one of those chipsets on your MOBO you simply exclude the extra pins and don't bother with their traces etc. Any 1500 pin CPU you choose to place into the socket will be limited (clearly labeled and marketed). There would be say an X400 chip that utilizes the extra lanes with a bigger link to the chipset (say x8) and x32 lanes direct from the CPU for graphics NVMe drives etc. So you could have x16 x8 GPU and 2 full speed x4 NVMe drives direct to CPU, or x16 x8 x8 for triple GPU setups (and still have good bandwidth for an x4 NVMe drive and USB3.1 Gen 2/thunderbolt 3 at the same time through the chipset). They already have an XFR designation, so it could've been something like XFR-400 (match the chipset name) for the higher end CPUs with the extra lanes. And if you stuck a lower end CPU into the higher end socket it would also limit you (disable certain features/slots).

As much as the split in Intels HEDT lineup annoys me (because they're all supposed to be the highest end choices) it could be a smart choice for a full range lineup that goes top to bottom...and if they kept the socket around for a decade (like AM3) it would offer incentive and a lot of expansion capability to buy a higher end MOBO even if you start with a lower end CPU. And as tech advanced, you'd have the option of dropping off old low end chipsets and making more lanes available to new low end options. That'd also make it easier to phase in a new socket/high end platform for the next decade. All the CPUs/MOBOs at the end of cycle could be using previously extra lanes to include the newest relevant features (unlike AM3 which kinda had to hack on extras) so the platform wouldn't seem so desperately dated after a decade.

 

[itsmydamnation]

There are 32 pci-e 3 lanes on each Zepplin die, So on the 32 core parts they do get upto 128lanes of PCI-E add to that 8 channel memory and thats why SP3 socket is 4k pins.

 

[bjt2]

that's what we thought, but someone here, presumably with inside info, said we were making bad assumptions.

I wrote an article on B&C, having access to the papers that AMD wrote in conferences and journals on previous technologies up to Brisrtol Ridge. XFR is already present in a limited form on Bristol Ridge and is called shadow p-states. For limited i mean that the clocks applied to the CPU are limited to the known p-states. XFR is just shadow p-states made free to go to higher clocks in 25MHz increments... The article is here: http://www.bitsandchips.it/52-english-news/8058-xfr-and-its-anchestors

 

[bjt2]

Less than two weeks until NDA expires and more details can be revealed.

Actually if you have access to the papers, you can have a description of the inner working up to Bristol Ridge. As soon as the ISSCC papers are available on ieeexplore, you can donwload and read it...

 

[Elixer]

And since Crosshair VI has 1 less M.2 than Maximus IX [Hero versions of both], i conclude that this chart is correct.

It is, again, a design choice. Once both those M.2 slots are both enabled, you lose SATA ports.

Could they have done the same on the Crosshair? Sure, as for why, they obviously don't think that is a major selling point.

 

[Mopetar]

No no, you do not quite get my suspicion. Namely it is that non-XFR SKUs will have much more than just better bin and XFR enabled.

I'm not sure I understand what you mean there. Did you mistype something, because you make it sound as though the non-X chips (no idea if the X denotes XFR or not) have a better bin and XFR enabled. But that doesn't make sense logically because non-XFR chips wouldn't have XFR enabled by definition.

I'll express my point this way: Intel CPUs with unlocked multiplier have Extended Frequency Range too.

Still doesn't make sense as AMD have said all the chips are unlocked (it just requires a capable mobo) so XFR wouldn't just be top binning and capable of hitting higher clocks. I'm guessing that like the articles have suggested it is the ability for the chip to detect that its sitting at low temperatures and give itself a bit of an OC beyond what the turbo is by default. So basically if you bought an XFR chip and put it under water or LN2 it would boost higher automatically than it would if on a shoddy air cooler. Intel chips won't do that to my knowledge.

Of course I don't think it matters much if the XFR chips have the best binning because most people who buy one are going to manually OC it and I suspect that the XFR automatic OC isn't anywhere near as aggressive as a manual one. It's a nice thought for AMD, but it really seems like something that would be better if it were enabled across the line. The people who could likely benefit most aren't buying the chips with the feature enabled.

 

[lolfail9001]

I'm not sure I understand what you mean there. Did you mistype something, because you make it sound as though the non-X chips (no idea if the X denotes XFR or not) have a better bin and XFR enabled. But that doesn't make sense logically because non-XFR chips wouldn't have XFR enabled by definition.

Yeah, i mistyped, my analogy puts it more aptly.

Still doesn't make sense as AMD have said all the chips are unlocked (it just requires a capable mobo) so XFR wouldn't just be top binning and capable of hitting higher clocks. I'm guessing that like the articles have suggested it is the ability for the chip to detect that its sitting at low temperatures and give itself a bit of an OC beyond what the turbo is by default. So basically if you bought an XFR chip and put it under water or LN2 it would boost higher automatically than it would if on a shoddy air cooler. Intel chips won't do that to my knowledge.

Of course I don't think it matters much if the XFR chips have the best binning because most people who buy one are going to manually OC it and I suspect that the XFR automatic OC isn't anywhere near as aggressive as a manual one. It's a nice thought for AMD, but it really seems like something that would be better if it were enabled across the line. The people who could likely benefit most aren't buying the chips with the feature enabled.

No, it makes perfect sense, because unlocked multiplier can still have a range it can fall into. Anyways, we will learn soon.

 

[Mopetar]

No, it makes perfect sense, because unlocked multiplier can still have a range it can fall into. Anyways, we will learn soon.

I'm still not sure exactly what you're trying to say. If it just means can clock higher, than almost all Ryzen chips will have XFR as they're all unlocked assuming you're using a B or X model board.

What exactly is it that you think XFR means or does?