Intel Core i9-10900K 10-core Processor and Z490 Chipset Arrive April 2020

[pas008]

1 minute ago, Zando Bob said:

I really don't know how to address this level of just... blinded lunacy? @MageTank Has a good reasoned reply, and if you want a single lineup you need to fix this too:

LGA 1151 CPU vs LGA 2066 CPU. The difference between AM4 and sTR4/sTRX4 is way fuckin bigger, but we'll stick to Intel here. 

Dies:

7980XE

LGA1151 i7s.

You can't physically fit the HEDT Die underneath the LGA1151 IHS, meaning if you want a single lineup, every single CPU needs to be LGA2066, which is a big expensive package. Even Pentiums will need to go under this, driving up their price by a large amount. Will effect mobos too, there's only a single ITX X299 board because the socket is so large, there isn't even an ITX sTR4 board in existence because that package is even bigger. So you can't get mobos as small anymore, half the stuff on your mobos either will or won't work depending on the CPU, and costs for everyone other than HEDT users (it'll stay the same) will go up massively to pay for physically larger and more complicated CPU packages. 

and that's one facet of CPU/socket/chipset differences and why they're split up. Just one, there's a ton more variables. 

Oh, and here's the comparision between sTRX and AM4:

Single TR4 is almost as big as 4 AM4 chips. Oh and no way that die setup fits in the AM4 package:

meaning every single AMD CPU must use the even more expensive, more annoying, less compatible sTR4 package. 

yes i know size difference but its not much

 

and I'm not talking anything amd right now because they did it right

[Zando_]

Just now, pas008 said:

yes i know size difference but its not much

 

and I'm not talking anything amd right now because they did it right

"its not much"

Sir do you have eyes? 

[Drak3]

7 minutes ago, Zando Bob said:

"its not much"

Sir do you have eyes? 

It’s not a huge difference. Basically the same difference between a FourTwo and a Peterbilt.

[MageTank]

7 minutes ago, pas008 said:

 

be easy for intel combine mainstream with their current hedt their damn sockets they pretty damn close in size too

they'd just need to eliminate 115x or lga200 or whatever and go with 2066 or something

then offer different chipsets from there

 

 

its not like amd with am4 and tr sockets

 

fyi you can burn out many boards with 3900x and 9900k

 

 

You're still ignoring the physical layouts of the boards relative to what the CPU's can supply in regards to their memory controller capabilities and PCIe lane allocation. Consumers do not want to pay money for a motherboard and then have features that cannot be used because their CPU simply can't provide them. Again, if you are suggesting that all processors across the entire product stack support the same number of DIMM's and PCIe lanes, that cost has to go somewhere. If you suggest that Intel eats the cost, they will simply stick to their current strategy of market segmentation to save money. If the cost is to be eaten by consumers, they will simply refuse to buy something they don't need, and look elsewhere.

 

If you are suggesting that they simply make everything physically compatible but separate the features via different chipsets, you are defeating the entire purpose of trying to unify the product stack. You will run into issues where a processor physically fits, but can't be used because the chipset itself is refusing the processor. 

 

I am also very aware that you can burn out boards with the 3900X and 9900K. In my post, I gave an example eluding to that using the 9900KS. My point still stands. You cannot unify a platform but exclude board features based on the specific processor installed. This will only lead to anger from consumers, and it will be a nightmare for technical support companies to have to explain and troubleshoot on behalf of their customers.

 

The current segmentation sucks for consumers that want to upgrade to more cores or additional features, but it's still the best option at this moment in time.

[pas008]

6 minutes ago, Zando Bob said:

"its not much"

Sir do you have eyes? 

yes i can see 1/3rd bigger ok but theres plenty of room

geez even cpu coolers holes arent much either

 

 

 

 

[pas008]

2 minutes ago, MageTank said:

You're still ignoring the physical layouts of the boards relative to what the CPU's can supply in regards to their memory controller capabilities and PCIe lane allocation. Consumers do not want to pay money for a motherboard and then have features that cannot be used because their CPU simply can't provide them. Again, if you are suggesting that all processors across the entire product stack support the same number of DIMM's and PCIe lanes, that cost has to go somewhere. If you suggest that Intel eats the cost, they will simply stick to their current strategy of market segmentation to save money. If the cost is to be eaten by consumers, they will simply refuse to buy something they don't need, and look elsewhere.

 

If you are suggesting that they simply make everything physically compatible but separate the features via different chipsets, you are defeating the entire purpose of trying to unify the product stack. You will run into issues where a processor physically fits, but can't be used because the chipset itself is refusing the processor. 

 

I am also very aware that you can burn out boards with the 3900X and 9900K. In my post, I gave an example eluding to that using the 9900KS. My point still stands. You cannot unify a platform but exclude board features based on the specific processor installed. This will only lead to anger from consumers, and it will be a nightmare for technical support companies to have to explain and troubleshoot on behalf of their customers.

 

The current segmentation sucks for consumers that want to upgrade to more cores or additional features, but it's still the best option at this moment in time.

do you tell someone that buys 8100 to buy z390?

 

do you see prebuilts even doing that?

[Zando_]

3 minutes ago, pas008 said:

yes i can see 1/3rd bigger ok but theres plenty of room

geez even cpu coolers holes arent much either

 

Ya can’t read and your reasoning is... not reasonable, I’m done. 

[pas008]

11 minutes ago, Zando Bob said:

 

Ya can’t read and your reasoning is... not reasonable, I’m done. 

ok guess you are sold on intels reasoning for separate sockets too

 

when you could have just started with a bigger socket in the first place

if you want smaller use mobile

[pas008]

@Zando Bob @MageTank

lga3647 socket should be their hedt now to make sense

and 2066 should be mainstream

similar to how amd is doing their lineup

[MageTank]

1 minute ago, pas008 said:

do you tell someone that buys 8100 to buy z390?

 

do you see prebuilts even doing that?

That is quite the strawman, but I'll answer it nonetheless. If a customer bought an 8100 but wanted to overclock their ram, absolutely. Why? Because overclocking ram is a feature of Z390, and if that feature is desired by a customer, then I'll recommend it. Do I see prebuilts doing that? Yes actually... I have. There was a time where the only chipsets available were Z370 and OEM's were shipping prebuilts with Z370 chipsets paired with plain Core i7 8700 processors that didn't support overclocking. This was also extremely common with mobile laptops that launched for Kaby Lake, as you saw the Clevo P870KM being paired with locked processors due to thermal constraints, despite a board that actually supported overclocking.

 

I still don't see what this has to do with your point, as it actually proves my own. Consumers that need an i3 8100 would not want to spend more money on features they don't need from more expensive chipsets. How can you possibly make every processor across the product stack compatible with each chipset without cannibalizing features? Answer my questions this time Pas, don't just give me a back handed response.

[pas008]

2 minutes ago, MageTank said:

That is quite the strawman, but I'll answer it nonetheless. If a customer bought an 8100 but wanted to overclock their ram, absolutely. Why? Because overclocking ram is a feature of Z390, and if that feature is desired by a customer, then I'll recommend it. Do I see prebuilts doing that? Yes actually... I have. There was a time where the only chipsets available were Z370 and OEM's were shipping prebuilts with Z370 chipsets paired with plain Core i7 8700 processors that didn't support overclocking. This was also extremely common with mobile laptops that launched for Kaby Lake, as you saw the Clevo P870KM being paired with locked processors due to thermal constraints, despite a board that actually supported overclocking.

 

I still don't see what this has to do with your point, as it actually proves my own. Consumers that need an i3 8100 would not want to spend more money on features they don't need from more expensive chipsets. How can you possibly make every processor across the product stack compatible with each chipset without cannibalizing features? Answer my questions this time Pas, don't just give me a back handed response.

hedt should share highend server socket if its going to exist like it use to

mainstream should offer few different chipsets for user configuration of what they want from their system

and then mobile

like I said intels hedt doesnt make sense

didnt linus do a video on this?

 

 

[MageTank]

1 hour ago, pas008 said:

hedt should share highend server socket if its going to exist like it use to

mainstream should offer few different chipsets for user configuration of what they want from their system

and then mobile

like I said intels hedt doesnt make sense

didnt linus do a video on this?

 

 

Once again, you refuse to answer any of my questions. You're simply making statements while refusing to acknowledge anything that is said to you. Explain why HEDT sharing the socket of their server lineup matters at all? Sharing the socket doesn't automatically make them compatible with server boards or vice versa when attempting to use their server processors in consumer boards.

 

Linus ranting about something doesn't automatically make it a fact. Most of his rant was directed to Intel's shady NDA/embargo practices, not their market segmentation as a whole. 

 

If you want to say Intel's HEDT makes no sense from a self-cannibalization stance, then I am sure people will agree with you. However, you can't simply say "Make them all socket compatible and just use different chipsets" as if that is a viable solution. At this point, you are going to need to back your claims up with some evidence, otherwise you will not be taken seriously.

 

EDIT: I went back and watched the video you were likely referring to: 

In this video, his issue is due to Kaby Lake X existing alongside Skylake-X on the same platform. Meaning, the desktop and HEDT CPU's sharing the same board/chipset... That's the very thing you are asking for, is it not? Would this not be evidence against your claim, if you believe what Linus said is the truth? His rant in that video (which I mistook for one of his more recent embargo rants) was directly related to the convergence of the consumer re-branded Kaby Lake chips being used on the exact same motherboard/chipsets. 

 

My point remains. I need to see some evidence to backup your claims in order to understand your point.

 

 

[Belgarathian]

Soooooo i9-10900K 10c/20t v Ryzen 9 3950X 16c/32t

 

Looks DOA to me, would likely need a 10-15% lead in single core performance to be compelling because it's going to be thrashed in multi.

[pas008]

8 minutes ago, MageTank said:

Once again, you refuse to answer any of my questions. You're simply making statements while refusing to acknowledge anything that is said to you. Explain why HEDT sharing the socket of their server lineup matters at all? Sharing the socket doesn't automatically make them compatible with server boards or vice versa when attempting to use their server processors in consumer boards.

 

Linus ranting about something doesn't automatically make it a fact. Most of his rant was directed to Intel's shady NDA/embargo practices, not their market segmentation as a whole. 

 

If you want to say Intel's HEDT makes no sense from a self-cannibalization stance, then I am sure people will agree with you. However, you can't simply say "Make them all socket compatible and just use different chipsets" as if that is a viable solution. At this point, you are going to need to back your claims up with some evidence, otherwise you will not be taken seriously.

we know from the past and how amd is actually doing it now

 

 

[Letgomyleghoe]

4 hours ago, Sauron said:

These are 500$+ chips, they aren't mainstream by any stretch.

well for pc master race i would say mainstream... 

[Drak3]

5 hours ago, Belgarathian said:

Looks DOA to me

Not really. The typical consumer doesn't know what benchmarks or IPC is, the only metric they know about is clockspeed.

 

And Intel advertises their product to typical consumers, AMD doesn't. That alone guarantees that Intel's new line of CPUs are going to sell somewhat well.

[CarlBar]

11 hours ago, pas008 said:

yes i can see 1/3rd bigger ok but theres plenty of room

geez even cpu coolers holes arent much either

 

 

 

 

 

He literally showed you that you physically cannot fit the HEDT CPU die on the desktop socket. And the green substrate thingy it's mounted on, (known as a PCB), is not free.Take a big PCB like the TRx40 one and put a cheapo die like an Athlon on it and the PCB could easily end up costing more than the Die mounted on it. When you get down to the scale of CPU dies you start running into some serious issues with capacitance and other little things that require very exacting standards. All that means you can't just use a cheaper PCB for the cheaper processors and guarantee it will all work fine, (also a lot of the cost is tied to the pure physical dimensions).

 

And thats before we get into the issue that if you put a dual channel CPU in a quad channel board there is no physical way to avoid having some of the RAM slots disabled with the dual channel mode as the CPU chipset has nothing to do with the functioning of that. Ditto for many of the PCI-E slots, (EPYC doesn't even have a motherboard chipset from what i understand, the motherboard is just a set of electrical connections and mounting slots for everything).

 

You can have all the different Motherboard chipsets you want. It isn't going to change the fact that some CPU's will have most of their features disabled in some boards. And with the memory controller at least it may well be the case that a quad channel memory controller doesn't function well in a dual channel configuration, (and capacitance issues might create problems for dual channel in quad channel for that matter).

 

And all this is ignoring the VRM differences or the issues for small form factor that a big socket creates.

 

In the end everything would be just as segmented as now because what boards a given CPU would work in would still have to be restricted, just all using the same socket and with everything but the HEDT CPU's costing more.There's no advantage to it if all it does is make things cost more and make small form factors harder or impossible. Why would anyone ever want either of those things?

[Quadriplegic]

Is 10900k going to be the first in a while "mainstream" CPU to hit 300w power draw? 

9900k already can reach 250w when overclocked and I doubt binning can help Intel in this case

[pas008]

4 hours ago, CarlBar said:

 

He literally showed you that you physically cannot fit the HEDT CPU die on the desktop socket. And the green substrate thingy it's mounted on, (known as a PCB), is not free.Take a big PCB like the TRx40 one and put a cheapo die like an Athlon on it and the PCB could easily end up costing more than the Die mounted on it. When you get down to the scale of CPU dies you start running into some serious issues with capacitance and other little things that require very exacting standards. All that means you can't just use a cheaper PCB for the cheaper processors and guarantee it will all work fine, (also a lot of the cost is tied to the pure physical dimensions).

 

And thats before we get into the issue that if you put a dual channel CPU in a quad channel board there is no physical way to avoid having some of the RAM slots disabled with the dual channel mode as the CPU chipset has nothing to do with the functioning of that. Ditto for many of the PCI-E slots, (EPYC doesn't even have a motherboard chipset from what i understand, the motherboard is just a set of electrical connections and mounting slots for everything).

 

You can have all the different Motherboard chipsets you want. It isn't going to change the fact that some CPU's will have most of their features disabled in some boards. And with the memory controller at least it may well be the case that a quad channel memory controller doesn't function well in a dual channel configuration, (and capacitance issues might create problems for dual channel in quad channel for that matter).

 

And all this is ignoring the VRM differences or the issues for small form factor that a big socket creates.

 

In the end everything would be just as segmented as now because what boards a given CPU would work in would still have to be restricted, just all using the same socket and with everything but the HEDT CPU's costing more.There's no advantage to it if all it does is make things cost more and make small form factors harder or impossible. Why would anyone ever want either of those things?

oh wow

i have said numerous times amd is doing it right for their hedt

didnt intel do this with x38 and x48?

 

2066 should be the mainstream socket with b365 for dual channel and i3 to i7 cpus

x299/z390 for any one could be for dual channel one for quad

other features come and go in between

 

and your memory controller wont function well comment what?

many people use dual channel on x99/x299 as is they went hedt for the cpu/pcie lanes

[MageTank]

14 hours ago, pas008 said:

we know from the past and how amd is actually doing it now

 

 

Ah yes, AMD's past. The one that stifled innovation for the sake of keeping older AM3 processors compatible,severely limiting what they could do with newer processors, and ultimately delivering one of the worst CPU designs we've seen in modern hardware. Not exactly the example I would use to prove one of my points, but I suppose desperate times call for desperate measures. You still put so much significance into the sockets themselves, so tell me... What benefits do consumers get from Threadripper sharing EPYC's socket pin count? It's not like they are getting a wider variety of boards or features of the higher EPYC processors as they are being limited by a single ID pin and BIOS microcode, preventing you from using a threadripper CPU in an SP4 motherboard, or EPYC CPU in a TR4 board. You keep touting this as a genius move, but you've yet to make a case for a single benefit of doing this.

 

8 minutes ago, pas008 said:

oh wow

i have said numerous times amd is doing it right for their hedt

didnt intel do this with x38 and x48?

 

2066 should be the mainstream socket with b365 for dual channel and i3 to i7 cpus

x299/z390 for any one could be for dual channel one for quad

other features come and go in between

 

and your memory controller wont function well comment what?

many people use dual channel on x99/x299 as is they went hedt for the cpu/pcie lanes

Oh, you've absolutely said it dozens of times, you've just yet to actually offer any substance that proves your point. So far, you've only proven that you want to further obfuscate the product stack and confuse more consumers. Start answering my questions pas.

[pas008]

1 hour ago, MageTank said:

Ah yes, AMD's past. The one that stifled innovation for the sake of keeping older AM3 processors compatible,severely limiting what they could do with newer processors, and ultimately delivering one of the worst CPU designs we've seen in modern hardware. Not exactly the example I would use to prove one of my points, but I suppose desperate times call for desperate measures. You still put so much significance into the sockets themselves, so tell me... What benefits do consumers get from Threadripper sharing EPYC's socket pin count? It's not like they are getting a wider variety of boards or features of the higher EPYC processors as they are being limited by a single ID pin and BIOS microcode, preventing you from using a threadripper CPU in an SP4 motherboard, or EPYC CPU in a TR4 board. You keep touting this as a genius move, but you've yet to make a case for a single benefit of doing this.

 

Oh, you've absolutely said it dozens of times, you've just yet to actually offer any substance that proves your point. So far, you've only proven that you want to further obfuscate the product stack and confuse more consumers. Start answering my questions pas.

for starters many extra pins for longer lifetime and possible future compatibility

no need refresh for few damn pins

amd is going have 4 gens on same socket

 

if they did like amd hedt  i could understand, because they can easily move eypc to tr like they just did with 32 cores like a yr ago

or if they did like x38 and x48 days where hedt cpus was mixed with mainstream

 

not to mention intels chipset shortages should never be an issue

 

[MageTank]

4 minutes ago, pas008 said:

for starters many extra pins for longer lifetime and possible future compatibility

no need refresh for few damn pins

amd is going have 4 gens on same socket

 

if they did like amd hedt  i could understand, because they can easily move eypc to tr like they just did with 32 cores like a yr ago

or if they did like x38 and x48 days where hedt cpus was mixed with mainstream

 

 

 

With this being the entire foundation of your point, I am going to need you to explain this to me: https://en.wikipedia.org/wiki/Socket_SP3

 

Threadripper's TR4 platform and TRX40 have the exact same pin count, same physical socket, and yet TR4 only supported 2 generations of Threadripper. 

 

The socket doesn't guarantee a longer lifetime, nor does it guarantee future compatibility. It's not a magical seer or oracle that can predict what future technologies look like. Sometimes, you end up in a situation where a new piece of tech is released and your current design simply can't facilitate it, so you have to switch it up. Can it be used nefariously for forced obsoletion? Absolutely, but that's why you trust in a competitive market and go elsewhere if you feel you are being mislead.

 

I am going to need you to step back from your blind faith for AMD and take an objective look at things. The very company you are using as your source is contradicting your point. It's time to come to an understanding that what you are requesting, while somewhat understandable, isn't best for consumers right now.

[leadeater]

1 hour ago, pas008 said:

i have said numerous times amd is doing it right for their hedt

didnt intel do this with x38 and x48?

Putting Xeons in to HEDT motherboards was extremely niche back then and today there is even less reason to do it as Xeons today clock much lower than HEDT CPUs and are locked. Xeon chipsets are also locked. So you gain exactly nothing by designing these CPUs lines with cross compatibility and only offer worse performance more expensive options for the non server users, pay more for less? Pass.

 

AMD may be doing a lot of good things for HEDT but it's got nothing to do with the socket or chipset and everything to do with the fact PCIe lanes are not chopped off on lower SKUs and there is on die HEDT features like 10Gb networking. Their success has everything to do with acceptable pricing and feature parity across the entire product line, something you cannot say about Intel until 10th Gen HEDT.

 

1 hour ago, pas008 said:

and your memory controller wont function well comment what?

many people use dual channel on x99/x299 as is they went hedt for the cpu/pcie lanes

There are dedicated pins on the CPUs for each memory channel and PCIe lane, these are directly traced across the motherboard to the slots which means if you put a 2 channel CPU in a 4 channel design board then you will have memory slots that will not connect to any CPU pins so those slots cannot be used, they go nowhere. This is the exact problem of Kaby Lake-X on X299, memory slots and PCIe slots don't work as those CPUs do not have the pins to connect them.

 

Kaby Lake-X was discontinued because of these problems and nobody was buying them, pay twice as much for an HEDT motherboard that half the board end up disabled if you put in a Kaby Lake-X CPU? That's just silly. The CPUs weren't cheap either so nobody was going to stepping stone from those to a proper HEDT CPU, you could nearly build an entire system for the money wasted doing it, which would be a better option than buying Kaby Lake-X.....

 

Multiple CPU and archiecture generations on the same socket and chipset is nice on the consumer product line so long as it does not inhibit progress.

[pas008]

3 minutes ago, MageTank said:

With this being the entire foundation of your point, I am going to need you to explain this to me: https://en.wikipedia.org/wiki/Socket_SP3

 

Threadripper's TR4 platform and TRX40 have the exact same pin count, same physical socket, and yet TR4 only supported 2 generations of Threadripper. 

 

The socket doesn't guarantee a longer lifetime, nor does it guarantee future compatibility. It's not a magical seer or oracle that can predict what future technologies look like. Sometimes, you end up in a situation where a new piece of tech is released and your current design simply can't facilitate it, so you have to switch it up. Can it be used nefariously for forced obsoletion? Absolutely, but that's why you trust in a competitive market and go elsewhere if you feel you are being mislead.

 

I am going to need you to step back from your blind faith for AMD and take an objective look at things. The very company you are using as your source is contradicting your point. It's time to come to an understanding that what you are requesting, while somewhat understandable, isn't best for consumers right now.

blind faith in amd 

no

i'm looking at fact intel didnt limit themselves in the past on moving high core count cpu from their server lineup to their hedt like they are now

amd sockets might not work but they can move their eypc cpus to threadrippers like they showed us with 32 core

you think they cant move the 64 core one to tr?

 

1 minute ago, leadeater said:

Putting Xeons in to HEDT motherboards was extremely niche back then and today there is even less reason to do it as Xeons today clock much lower than HEDT CPUs and are locked. Xeon chipsets are also locked. So you gain exactly nothing by designing these CPUs lines with cross compatibility and only offer worse performance more expensive options for the non server users, pay more for less? Pass.

 

AMD may be doing a lot of good things for HEDT but it's got nothing to do with the socket or chipset and everything to do with the fact PCIe lanes are not chopped off on lower SKUs and there is on die HEDT features like 10Gb networking. Their success has everything to do with acceptable pricing and feature parity across the entire product line, something you cannot say about Intel until 10th Gen HEDT.

 

There are dedicated pins on the CPUs for each memory channel and PCIe lane, these are directly traced across the motherboard to the slots which means if you put a 2 channel CPU in a 4 channel design board then you will have memory slots that will not connect to any CPU pins so those slots cannot be used, they go nowhere. This is the exact problem of Kaby Lake-X on X299, memory slots and PCIe slots don't work as those CPUs do not have the pins to connect them.

 

Kaby Lake-X was discontinued because of these problems and nobody was buying them, pay twice as much for an HEDT motherboard that half the board end up disabled if you put in a Kaby Lake-X CPU? That's just silly. The CPUs weren't cheap either so nobody was going to stepping stone from those to a proper HEDT CPU, you could nearly build an entire system for the money wasted doing it, which would be a better option than buying Kaby Lake-X.....

 

Multiple CPU and archiecture generations on the same socket and chipset is nice on the consumer product line so long as it does not inhibit progress.

 

 

I know how it works

 

but if intel used 2066 for mainstream in first place and then 3647 for server/hedt they wouldnt be limiting their consumers on core count either and its not like they cant offer better clocked cpus on top

 

[leadeater]

22 minutes ago, pas008 said:

or if they did like x38 and x48 days where hedt cpus was mixed with mainstream

That was only possible because the memory controller back then was in the chipset not on die in the CPU, as things have been moved off the chipset and in to the CPU die (for performance reasons and to simplify designs) the ability to have wider set of products on the same platform has reduced. It's a side effect of progress and we wouldn't have the performance we do today if when didn't move things out of the chipset like the memory controller as that was a huge performance limiter and the chipsets ran hot as hell too.