AMD in recent times have used the prefix "X" in naming their High End Enthusiast Chipset. I don't see them using "Z" and we already know that there's a Z390 in the works from intel.

That's a little confusing, because isn't it X390 (and presumably X490) for Threadripper? The 400 series motherboards are already underwhelming as-is. Instead of this Z490, they should've just added the extra lanes to the X470.

I like the "Y". I would like the "Y" to get a little love. Y470.... I like it. It is a bit confusing with the Z.

http://www.guru3d.com/index.php?ct=news&action=file&id=26920 The slide has been doctored. If you simply go up to the header the z490 is listed under Chipset General Purpose PCIe ------>GEN2<------ The Z490 is supposedly gen3 and would have its own slide with its own header not just lumped in with the others as an after thought. Btw Intel is working on a z490 for Coffe Lake so this is where the idea probably came from. Come on you have to do better than this!;) As of May 5th AMD is making a z490 mb but there are no specifics on it's spec's.

i wish PCIE 4.0 would just come out already, what's the hold up?

holler:

i wish PCIE 4.0 would just come out already, what's the hold up?

What's the rush? What do you have/want that would need it? To my knowledge, not even the Titan V saturates all 16 lanes in gen 3.0. With DX12/Vulkan, PCIe bandwidth is reduced even further. Meanwhile when it comes to NVMe drives, hardly any of them offer any noticeable difference on x2 lanes vs x4.

schmidtbag:

What's the rush? What do you have/want that would need it? To my knowledge, not even the Titan V saturates all 16 lanes in gen 3.0. With DX12/Vulkan, PCIe bandwidth is reduced even further. Meanwhile when it comes to NVMe drives, hardly any of them offer any noticeable difference on x2 lanes vs x4.

its not all about bandwidth, pci4.0 introduces Lane margining for one, reduces jitter aka M-GPU stutter [youtube=S_TfX5yYKOU]

holler:

its not all about bandwidth, pci4.0 introduces Lane margining for one, reduces jitter aka M-GPU stutter

This is good, but hardly relevant to AM4 motherboards. This is only especially beneficial when you've got several high-bandwidth expansion cards. The stuttering you encounter from a couple GPUs is not going to be fixed by this. It may be reduced, but the problem is mostly due to timing and synchronization. So, I could see why you may have an interest in PCIe 4.0 with a TR4 motherboard, but for now, AM4 doesn't really need it.

Embra:

I like the "Y". I would like the "Y" to get a little love. Y470.... I like it. It is a bit confusing with the Z.

Then it would look like a bunch of lenovo products (many of their models start with y, such as Y50, or Y500)

Aura89:

Then it would look like a bunch of lenovo products (many of their models start with y, such as Y50, or Y500)

Lenovo uses "Y" so people are not surprised when they start using it and say: "OMG, why?"

schmidtbag:

This is good, but hardly relevant to AM4 motherboards. This is only especially beneficial when you've got several high-bandwidth expansion cards. The stuttering you encounter from a couple GPUs is not going to be fixed by this. It may be reduced, but the problem is mostly due to timing and synchronization. So, I could see why you may have an interest in PCIe 4.0 with a TR4 motherboard, but for now, AM4 doesn't really need it.

I dont understand your replay, what does it even means "doesn't really need it"? PCIe4.0 is not paid feature, so ANY free upgrade is indeed NEEDED, the more the better. "Here, get this totally free Porsche ", no thank you, im ok with my Mazda. Thats how you sounded. P.S. Also, it looks like the PCIe situation on AMD platform is sad, they still use PCIe gen 2. Right now on Intel z370 you can do GPU at x16 + TWO X4 NVMe drives + one SATA M.2 or additional x4 NVME if you disable 2 SATA ports And if you ok with GPU at x8, you can add two more NVMe drives at x4 each using adapter AMD has one GPU at x16 and one x4 NVMe, thats it + 8 useless Gen 2 lanes Going to Gen 4.0 will save 50% PCIe lanes, 2 gen 4 lanes will be as fast as 4 gen 3.0 lanes or 8 gen 2 lanes. Gen 3 is fast enough right now, we dont have devices that saturate the speed on the home market, but lane saving is important, i wont have to disable any SATA ports if i want to use 3 or even 5 full speed NVMe drives at x4 Gen3 or x2 Gen 4 with current PCI express configuration. x8 Gen 4 will be enough for the GPU and we will have 12 more lanes with direct CPU access to use for fast drives instead of going trough the chipset like we do now Dont forget its rare that Intel or AMD adds more lanes to the mainstream products, but it does gets better with generational upgrades, you get 50% "more" lanes if we go by the previous generation

holler:

its not all about bandwidth, pci4.0 introduces Lane margining for one, reduces jitter aka M-GPU stutter

Yeah, that would be great! If only you'd have enough PCIe lanes for mGPU (especially with mGPU that PCIe 2.0 is hurting as it's not even 2x16 but only 16/8), and if only those devs would put their work into properly supporting it, especially beyond dx11 / in the future. That's a lot of ifs to worry about PCIe 4.0 and mGPU systems, in my opinion. Good that AMD keeps on improving it's platform if this leak is true.

My assumption has been that we will see PCIE 4.0 on AMD chipsets that come out with the 7nm Zen 2.

Whatever the truth, the issues are simply danced around to confuse folks to the max imo. Its not a big deal to add lanes in a switched environment like a chipset. Intel have been offering multiple chipset pcie3 nvme ports for ages. Nowhere do they make clear tho, that in reality, the whole chipset must share the 4GB/s link to the main system (equal to about one nvme) - adding lanes simply shares this 4GB/s among more resources. If this report is true, as it may well be, then it is not about magically finding 4 more lanes from the am4/ryzen platforms current user total of 24, it is just a chipset mod which allows you to hog the entire chipset bandwidth with one device, which doesnt have to be a bad thing, but they dont make this clear. To use raid 0 on 2 such nvmeS would be nonsense e.g., but intel give the impression you can, and provide the means for triple raid nvme even on z170 afaik.

MegaFalloutFan:

I dont understand your replay, what does it even means "doesn't really need it"? PCIe4.0 is not paid feature, so ANY free upgrade is indeed NEEDED, the more the better.

I understand that... but the point is holler wants something that will have little to no impact on performance. I'm not advocating against PCIe 4.0, I'm just saying there's no point in wishing for something that won't improve anything (for now). By the time it reaches consumer hardware, I'm sure we'll see something that can actually take advantage of it.

"Here, get this totally free Porsche ", no thank you, im ok with my Mazda. Thats how you sounded.

Even though I've already clarified how that's not a good analogy, I would actually gladly take the Mazda over the Porsche (unless it was an RX-7). Not only are Mazdas cheaper to maintain and repair, but a lot of them are more practical. Unless you take the Porsche on a track (which I probably wouldn't bother with), I won't be able to take advantage of the performance upgrade (just like I wouldn't be able to with PCIe 4.0).

P.S. Also, it looks like the PCIe situation on AMD platform is sad, they still use PCIe gen 2. Right now on Intel z370 you can do GPU at x16 + TWO X4 NVMe drives + one SATA M.2 or additional x4 NVME if you disable 2 SATA ports

Uh... no? AMD offers x20 3.0 lanes (vs Intel's x24) - x16 are primarily meant for GPUs, while the x4 lanes for M.2. The remaining PCIe 2.0 lanes are used for the x1 slots, which people tend to use for stuff like wifi, and is plenty sufficient for such needs. In other words, AMD motherboards don't use PCIe 2.0 slots for anything bandwidth-intensive like GPUs or SSDs.

And if you ok with GPU at x8, you can add two more NVMe drives at x4 each using adapter AMD has one GPU at x16 and one x4 NVMe, thats it + 8 useless Gen 2 lanes

...you do realize that x8 2.0 lanes has equivalent performance to x4 3.0 lanes, right? I suggest you do some actual research on how much bandwidth you actually need. I think you're gravely overestimating how much bandwidth you think you need.

Going to Gen 4.0 will save 50% PCIe lanes, 2 gen 4 lanes will be as fast as 4 gen 3.0 lanes or 8 gen 2 lanes.

That makes absolutely no difference if the cards you slot into them aren't also gen 4.0. A gen 3.0 M.2 device will not run faster in a gen 4.0 slot (assuming the amount of lanes are the same).

but lane saving is important, i wont have to disable any SATA ports if i want to use 3 or even 5 full speed NVMe drives at x4 Gen3 or x2 Gen 4 with current PCI express configuration.

That's not how it works... Gen 4.0 doesn't give you more lanes, it just gives you more bandwidth per lane. The reason the SATA ports need to be disabled is because the lanes used for the SATA controller are instead dropped in favor of the M.2 slots. So unless the chipset involves more lanes, you're still going to have to disable SATA ports. However... Gen 4.0 might yield 6 SATA ports instead of 4.

x8 Gen 4 will be enough for the GPU and we will have 12 more lanes with direct CPU access to use for fast drives instead of going trough the chipset like we do now Dont forget its rare that Intel or AMD adds more lanes to the mainstream products, but it does gets better with generational upgrades, you get 50% "more" lanes if we go by the previous generation

Again, that's only true if the expansion cards are also gen 4. For argument's sake, let's say you have a gen 3.0 GPU that bottlenecks at x8 3.0 lanes. If you run that GPU in a gen 4.0 x8 slot, you're still going to be bottlenecked.

not 100% accurate seeing as AMD x300 and x400 chipsets USE PCI-e 3.0 and do have as well PCI-e 2.0, so to claim z490 will use 3.0 because x300/400 only use PCI-e 2.0 is not at all accurate more lanes would be good though for most people are absolutely not necessary...think If Threadripper (two Ryzen dies) has 64, and EPYC (four Ryzen dies) has 128, so therefore 1 die (Ryzen) is 32 total lanes (at least on a fully enabled chipset such as the high end x370/470 vs x350/x450 type deal, or specific cpu models such as the APU which cut off 8 lanes provided to m.2 or whatever) (AM4 allows for 24. Eight of those are dedicated to the chipset and an M.2 slot, meaning you effectively have 16 to mess around with in terms of expansion cards (that are dynamically shared or wired electrically by vendors as they see fit) x370/470 adds another 8 on top of this at 2.0 speeds)

schmidtbag:

What's the rush? What do you have/want that would need it? To my knowledge, not even the Titan V saturates all 16 lanes in gen 3.0. With DX12/Vulkan, PCIe bandwidth is reduced even further. Meanwhile when it comes to NVMe drives, hardly any of them offer any noticeable difference on x2 lanes vs x4.

The difference between a Samsung 960 evo on x2 and a Samsung 960 pro on x4 is like comparing night and day.

The Goose:

The difference between a Samsung 960 evo on x2 and a Samsung 960 pro on x4 is like comparing night and day.

Key word here was "noticeable". There is absolutely a measurable difference, but in the most (but probably not all) cases, you wouldn't be able to tell the difference.

kalub:

When I read Z490, I thought about the Z390 from Intel. It becomes painful this naming story, it is a total confusion between the two. Let's hope it's wrong ... If that's true, Intel will be upset the day they release a Z490 (successor to the Z390), and I would not be surprised if they call it Z499 instead.

I'm pretty sure this is intentional, from AMD. Their naming scheme is starkly similar to Intel's, and it never really made sense to me why they started the chipset at 300 instead of 100. That is, until you realize that AMD was probably deliberately trying to one-up Intel's chipsets, to confuse the market. I figure AMD was trying to get Intel shoppers looking for a 200 series chipset and see there's a "newer" 300 series (from AMD). I'm not exactly fond of this, but at least it's not that difficult to distinguish the products just simply based on which brand you're referring to.

schmidtbag:

I'm pretty sure this is intentional, from AMD. Their naming scheme is starkly similar to Intel's, and it never really made sense to me why they started the chipset at 300 instead of 100. That is, until you realize that AMD was probably deliberately trying to one-up Intel's chipsets, to confuse the market. I figure AMD was trying to get Intel shoppers looking for a 200 series chipset and see there's a "newer" 300 series (from AMD). I'm not exactly fond of this, but at least it's not that difficult to distinguish the products just simply based on which brand you're referring to.

Trying to use such techniques on customers to confuse them? Hmm sounds logical, although you can probably never prove it's true 😀