Intel HEDT Skylake-X And Kaby Lake-X Detailed

Shintai

Supreme [H]ardness
Joined
Jul 1, 2016
Messages
5,691
Have you never upgraded a mainboard for future compatibility? Have you never been in a situation wherein you had an issue with your mainboard, but could not afford to purchase a new CPU as well as mainboard to replace it?

You are using a backwards compatible socket to justify non-backwards compatibility. Companies change sockets, yes, but this does not mean it is a requirement to throw out all compatibility. There is no reason that A: a socket should change every 2 years and B: that change in socket should have absolutely no backwards or forwards CPU compatability. You quickly turned this into an AMD versus Intel thing, which didn't work out for you. Intel has been pretty slimy with doing everything they can to force as many CPU and mainboard upgrades as they can, and their revolving-door, completely incompatible sockets are an example of that.

You are also striving to justify anti-consumer actions. I don't quite get it, and I don't really think you'll ever change.

First one no. And I cant imagine anyone that doesn't live on the street to have that issue.

And hold on there cowboy. This isn't about AMD vs Intel. That's something you are busy with and it shows in your post. Obviously you have seen yourself angry on one part. Its simply the defacto standard for the industry due to the every more tight electrical specs. Like it or not. This isn't about evil companies or forcing you to do something. Then you can keep on excusing it as much as you will and put any tags on the other party. It just doesn't change anything. I will leave it there.
 

KazeoHin

Supreme [H]ardness
Joined
Sep 7, 2011
Messages
8,040
First one no. And I cant imagine anyone that doesn't live on the street to have that issue.

And hold on there cowboy. This isn't about AMD vs Intel. That's something you are busy with and it shows in your post. Obviously you have seen yourself angry on one part. Its simply the defacto standard for the industry due to the every more tight electrical specs. Like it or not. This isn't about evil companies or forcing you to do something. Then you can keep on excusing it as much as you will and put any tags on the other party. It just doesn't change anything.
Once again, the defacto standard is Intel's standard. Companies aren't evil: they are more like machines that react to their environment. Competition forces companies to play nice in order to earn your purchase. Without competition, they use other methods. Nothing Evil about it. I run multiple Intel systems in my house, so this isn't a moral crusade: instead, I, as the consumer, would appreciate having more options and I know that these options could exist, but Intel chooses not to entertain the idea.

Electrical tolerances and specs are not even a factor; Even Intel themselves don't use this as an excuse. We have VRMs on-chip since Haswell, and not to mention, we have VRMs period. PCI-E specs have not changed since inception, DDR3 and DDR4 specs have not changed since introduced. Electrical specs is a pretty poor excuse to defend your point. Additional features are also not an excuse.

A drastic departure in how that power is delivered is worthy of compatibility being dropped (I.E the VRMS introduced in Haswell), but you're telling me the company that went from the Pentium 4 to the Core 2 Quad on the same socket could not have designed Sandy Bridge to work on 1156, or have designed Skylake work on 1150?

No, they chose not to, mostly because they could.
 

noko

Supreme [H]ardness
Joined
Apr 14, 2010
Messages
4,745
Man this sounds rather boring to tell the truth, wonder performance wise the real increase if any in anything over the 2011v3 cpu's and motherboards. Wait built in WiFi in the cpu :banghead:
 
D

Deleted member 88301

Guest
Requiring a new socket is BS. It's all about squeezing the money out of consumer wallets.
 

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
55,749
Enough of this new socket horseshit. Please...
You realize 2011 launched 5 years ago tomorrow right?

It had an almost 3 year run before the current revision, 2011-v3 came out almost 2 1/4 years ago and will be pushing close to 3 when this new one comes out.

Are you wanting to run your Skylake-X in a 486 motherboard or something?
I was going to say something similar. Adding to that, people think they want compatible sockets for years and years but you don't. Ask AMD how that really works out in the long run. You end up constrained by an older platform and its electrical design.
 

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
55,749
Requiring a new socket is BS. It's all about squeezing the money out of consumer wallets.
I love it when people say things like this. To some extent, I know where it comes from but you couldn't be more wrong. You wouldn't want to take a processor and design it around constraints of an 8 year old VRM design. Also, supporting CPUs on legacy hardware requires more complex UEFI/BIOS coding to support such a broad range of CPU microcode. Remember the 990FX fiasco? That was due to the motherboards not being properly tested and validated for pre-Bulldozer CPUs. Putting Bulldozer on 890FX based motherboards also prevented you from taking full advantage of all supported C-States as their VRM's couldn't support it. All the 990FX motherboards we tested without having Bulldozer CPUs were a fucking unmitigated cluster fuck. Legacy support of CPUs while optimizing for the future was exactly the reason for this bullshit.

Not to mention, supporting those older motherboards means being stuck to compatible PCIe and memory controllers or worse yet integrating more than one controller so that the CPU can be used on more than one motherboard.
 

Kongar

Gawd
Joined
Oct 25, 2004
Messages
730
my i5 3570 overclocked still beats the snot out of anything I throw at it. I just don't think there's much reason at all for Intel to push the speed limits anymore. My desire to tinker is met with "but there's nothing wrong with this PC). Now it's all about power and heat reduction and adding features. I admittedly haven't jumped on the 4k bandwagon yet though, so maybe that time is getting close.
 

drakken

[H]ard|Gawd
Joined
Aug 19, 2004
Messages
1,196
I have a socket 2011 and what is missing from basin falls is more important that what is there. It looks like some of the parts they tried jamming on the cpu are gettng moved off to the pci-e bus so that if we get usb 4 or something or 802.42 the motherboard or north bridge can be swapped out while the memory controller is still on the cpu. It looks like we have a DMI or direct memory interconnect between the cpu and what looks like a north bridge for internal external parts that connect to the mother board. What is more interesting is that means that the videocard and system memory are one step from the cpu while the slower tech like hard drives and network adapters are two steps away. Having used networked sas and open acc, I am guessing they are trying to take advantage of out of order code to toss stuff to parts that are always ready for data while simply que data pulls to happen less often so that when the data from the hard drive over the sata bus gets to the memory the operations are ready to use it.
 

KazeoHin

Supreme [H]ardness
Joined
Sep 7, 2011
Messages
8,040
I love it when people say things like this. To some extent, I know where it comes from but you couldn't be more wrong. You wouldn't want to take a processor and design it around constraints of an 8 year old VRM design. Also, supporting CPUs on legacy hardware requires more complex UEFI/BIOS coding to support such a broad range of CPU microcode. Remember the 990FX fiasco? That was due to the motherboards not being properly tested and validated for pre-Bulldozer CPUs. Putting Bulldozer on 890FX based motherboards also prevented you from taking full advantage of all supported C-States as their VRM's couldn't support it. All the 990FX motherboards we tested without having Bulldozer CPUs were a fucking unmitigated cluster fuck. Legacy support of CPUs while optimizing for the future was exactly the reason for this bullshit.

Not to mention, supporting those older motherboards means being stuck to compatible PCIe and memory controllers or worse yet integrating more than one controller so that the CPU can be used on more than one motherboard.
I get where you are coming from, and you have more knowledge in this department than I, but I don't see why Skylake could not have been designed to work with existing 1151 sockets, or why we needed the 1155 socket over the 1156 socket. I know that these CPU/chipset designs as they stand now are completely incompatible, but it seems awfully convenient to Intel's wallet that they just so happen to change sockets every two years. Are you saying that there was no way to design 1156 more compatible with Sandy-bridge's uncore by design from the get-go? You're saying there was no way for Intel to have designed sandy bridge to have been compatible with 1156?

Like I said. If this was somehow hurting Intel's wallet, I would completely believe that it was completely necessary and unavoidable, but it seems to be doing the opposite.
 
Top