Microsoft launches Surface Pro 6, Surface Laptop 2 and Surface Studio 2 and more | What a shame

[GoodBytes]

5 hours ago, AluminiumTech said:

That is factually incorrect. USB-C does not require PCI-E lanes. Only TB3 does which is totally independant of USB-C.

Well as TopHat explained, in the end, it uses PCIe. I don't have the source or time to look for, but that was what the Surface team said before, and that is why you have the big box Surface connector to USB Type-C accessories. Now, you can believe whatever you want. I don't care at this point. But I think Surface team knows a tad more than you on this subject, and knows a tad more on how the many design constraints that they faced designing the device.

 

Now, that said, Surface Pro 6 does have the PCIe lanes for it with the new U series CPU. So it can have, with TB3 even.

And as pointed out, it could be simply that they follow a 2 year design refresh, and that means they order parts for it for that length, and because it was not possible with the Pro 5, it isn't on the Pro 6. In any case, no one (beside you, apparently), needs dongles for USB Type-C or have an big need for it either. And, you are free to buy the competitor device if this one, which apparently is clearly, not fitting your needs.

 

Quote

Also, you're still not addressing the rest of what I said.

I am sorry, you are not the only user on this forum.

 

So you said:
 

Quote

Are you saying that having the majority of USB-A is an afterthought as most USB-A ports are on the back?

Your motherboard has a standard header connector for frontal USB ports that your case may provide (most does).

The USB on the back is for peripherals, such as mouse, keyboard, Webcam, USB hub from the monitor, USB printer, etc.

The front ones is for things you often plug in and out.. such as a USB powered headset, or USB flash drive.

 

You have none the mentioned items for the back of your system using USB Type-C.

So if you want to plug your phone (assuming you even have a USB Type-C to USB Type-C cable, and not USB Type-C to USB Type-A), you need to scrawl, each time, all the way on the back of the computer, possibly under your desk too, (wait, now you'll tell me, that you put your desktop on the table, and on reverse to have easy access to the port, and enjoy this great heater in winter time next to you... ok sorry, about that... assume that I am talking about ~99.99% of the world), to plug something or have an extension cord or a plug that goes from the back of the computer all the way on the front because the need of USB Type-C on the front is so low that case manufacture don't bother with it.. yet.

 

No reviews of sites goes "This case has no USB Type-C, worst... case... ever.... what a shame.... It is so shameful that it makes the Surface Pro 6 shamefulness, less shameful". No one cares.. Yes, it is a nice to have... but today, and in the short term, USB Type-A is alive and well, and you are not missing out on things. At no point in time, you come and you want to plug something and its exclusively on USB Type-C connector, beside an external GPU enclosure,  only to get and complain on how your CPU throttles, because it is not actively cooled with a dual heat pipe and fans, like the Surface Pro 1 and 2 was, and then you'll complain, once those are put in, that the device is too thick heavy, and short battery life.

 

Is there something else I missed?

[Sniperfox47]

35 minutes ago, GoodBytes said:

Well as TopHat explained, in the end, it uses PCIe. I don't have the source or time to look for, but that was what the Surface team said before, and that is why you have the big box Surface connector to USB Type-C accessories. Now, you can believe whatever you want. I don't care at this point. But I think Surface team knows a tad more than you on this subject, and knows a tad more on how the many design constraints that they faced designing the device.

Again what are you even talking about?

 

First of all, DMI is not PCIe. It's based off of PCIe, and shares most of the comm with PCIe, but it's not PCIe.

 

Second of all if you take a look at all of the datasheets for previous generations, supporting more than one USB 3.1 port is nothing new. The chip used in the previous generation Surface Pro supports the exact same number of USB lines out of it, 6. It supports the exact same number of PCIe 3.0 lanes out of it, 12.

 

Using a different connector, surface connect, doesn't magically bypass the amount of these lanes available. And considering it's capable of exposing raw PCIe over Surface Connect, it's certainly got those lanes free to share with Thunderbolt 3.

 

And if that's still not clear enough look at literally any other laptop on the market using these chips. Almost every single device on the market, detachable or not, at this size has *at least* 2 USB ports on it. Hell many premium devices *with this exact same chip* support thunderbolt 3. Hell the Spectre 13t using the step down chip from the Surface Pro 5 has two thunderbolt 3 ports *plus* a third USB 3.1 port.

 

I don't think anybody's asking for that, but why is it too much to ask for a second USB port that can be useful for anything in place of a MiniDisplayport port that's only useful if you're connecting to a wired external display?

 

35 minutes ago, GoodBytes said:

Now, that said, Surface Pro 6 does have the PCIe lanes for it with the new U series CPU. So it can have, with TB3 even.

And as pointed out, it could be simply that they follow a 2 year design refresh, and that means they order parts for it for that length, and because it was not possible with the Pro 5, it isn't on the Pro 6. In any case, no one (beside you, apparently), needs dongles for USB Type-C or have an big need for it either. And, you are free to buy the competitor device if this one, which apparently is clearly, not fitting your needs.

Again what are you talking about? I linked the 7th gen and 8th gen CPU data sheet... The new Surface Pro 6 has literally exactly the same IO capabilities as the old model >.> They both have more than enough FlexibleIO lanes to have 2 USB lanes for 2 different ports on the side, 4 PCIe lanes for the SSD, and 4 more PCIe lanes, Gigabit Ethernet, and another USB lane over Surface connect...

35 minutes ago, GoodBytes said:

Your motherboard has a standard header connector for frontal USB ports that your case may provide (most does).

The USB on the back is for peripherals, such as mouse, keyboard, Webcam, USB hub from the monitor, USB printer, etc.

The front ones is for things you often plug in and out.. such as a USB powered headset, or USB flash drive.

I mean that's kind of subjective but I'm with you so far...

35 minutes ago, GoodBytes said:

You have none the mentioned items for the back of your system using USB Type-C.

Lost me. I have an external Hard Drive, External SSD, and monitor to plug in that can all use USB-C. While many people may not currently have these things, if they plan to keep their computer for more than a few years it's pretty likely that things they pick up down the road will use it.

 

35 minutes ago, GoodBytes said:

So if you want to plug your phone (assuming you even have a USB Type-C to USB Type-C cable, and not USB Type-C to USB Type-A), you need to scrawl, each time, all the way on the back of the computer, possibly under your desk too, (wait, now you'll tell me, that you put your desktop on the table, and on reverse to have easy access to the port, and enjoy this great heater in winter time next to you... ok sorry, about that... assume that I am talking about ~99.99% of the world), to plug something or have an extension cord or a plug that goes from the back of the computer all the way on the front because the need of USB Type-C on the front is so low that case manufacture don't bother with it.. yet.

Again what are you even talking about? Many new cases come with at least one frontal USB-C port. Hell even a lot of prebuilts are coming with it now:

Notice that port at the bottom left? Thats an HP Omen.

 

But also stop to consider that you're comparing a dead market segment against a still fairly lively market segment... Why would you be surprised that nobody's investing in new designs for a literally dead market. The only part of the Desktop market that's not totally and completely dead is the gaming space, and surprise surprise pretty much every new gaming designed computer coming out has front panel USB-C...

 

35 minutes ago, GoodBytes said:

No reviews of sites goes "This case has no USB Type-C, worst... case... ever.... what a shame.... It is so shameful that it makes the Surface Pro 6 shamefulness, less shameful". No one cares.. Yes, it is a nice to have... but today, and in the short term, USB Type-A is alive and well, and you are not missing out on things. At no point in time, you come and you want to plug something and its exclusively on USB Type-C connector, beside an external GPU enclosure,  only to get and complain on how your CPU throttles, because it is not actively cooled with a dual heat pipe and fans, like the Surface Pro 1 and 2 was, and then you'll complain, once those are put in, that the device is too thick heavy, and short battery life.

Just because nobody complains about something doesn't make it not problematic. The industry *is* moving to USB-C. Whether you like it or not it is happening. If you only keep your device for a year or two it may never impact you, but if you at all care for the lifespan of your devices in 5 to 10 years you'll likely be going out of your way to find replacement peripherals that *aren't* USB-C.

 

Having USB-C doesn't prevent you from also including USB-A. By all means leave the current port, just replace the typically useless MiniDisplayPort port with a USB-C with DisplayPort port.

35 minutes ago, GoodBytes said:

Is there something else I missed?

I mean I'd love to hear an explanation of how you think USB works over the Surface Connect port without connecting to USB or PCIe on the CPU. That would be a fun read.

[TopHatProductions115]

@GoodBytes 
There is a good chance that the ports served on the new Surface devices are served over Intel's (chipset - proprietary) DMI interconnect or something of similar nature - which is disappointing. I'd prefer as many ports as possible to be passed through PCI-e and other industry standard datapaths. 

• https://en.wikipedia.org/wiki/Direct_Media_Interface

 

@Sniperfox47 

Are there any modern mobile machines that simply pass the ports through the available PCI-e lanes from the chipset and CPU? If so, that would be my preference.

[GoodBytes]

52 minutes ago, TopHatProductions115 said:

@GoodBytes 
There is a good chance that the ports served on the new Surface devices are served over Intel's (chipset - proprietary) DMI interconnect or something of similar nature - which is disappointing. I'd prefer as many ports as possible to be passed through PCI-e and other industry standard datapaths. 

• https://en.wikipedia.org/wiki/Direct_Media_Interface

Possibly. And I agree.

 

[Sniperfox47]

9 minutes ago, TopHatProductions115 said:

@GoodBytes 
There is a good chance that the ports served on the new Surface devices are served over Intel's (chipset - proprietary) DMI interconnect or something of similar nature - which is disappointing. I'd prefer as many ports as possible to be passed through PCI-e and other industry standard datapaths. 

• https://en.wikipedia.org/wiki/Direct_Media_Interface

 

@Sniperfox47 

Are there any modern mobile machines that simply pass the ports through the available PCI-e lanes from the chipset and CPU? If so, that would be my preference.

This is going to get complicated if you want to read it it's as clean of a breakdown as I can give. If not the TLDR is that you're not really losing anything by having those "PCIe" lanes between the CPU and chipset.

 

First off DMI is not lost PCIe ports. It's merely how the Memory Controller for the CPU and Chipset talk to each other. It gets called PCIe a lot but it's really not PCIe. Intel uses DMI, basically a super stripped down barebones version of PCIe, because it allows them to have less overhead in communication and smaller controllers on both ends.

 

AMD does actually use PCIe to talk between their memory controller in the APU and chipset but those are necessary for that communication and you can't "steal" them for other things because then you have no chipset and that means half your computer is missing.

 

On the Intel side you have two mobile processor families. The H/HQ/MQ line use a separate chipset and the U/Y line uses a chipset built into the CPU chip (at least on modern chips).

 

With the H series you have a 4 lane DMI between the CPU and a chipset on the laptop's board the exact same way you do on a desktop. The CPU provides PCIe lanes of it's own for a graphics card and the Chipset provides all the normal chipset stuff (more PCIe for IO like SSDs, SATA, USB, i2c, gigabit ethernet, all the typical commections). It works basically exactly the same as a desktop, just soldered instead of socketed.

 

With the U and Y series for lower power systems, the "chipset" (PCH) is integrated into the CPU package. The CPU then talks to the PCH with little bit weirder DMI setup inside the chip. Because of the extra space taken up by the PCH the graphics PCIe lanes off the memory controller get cut. This means all your IO comes off the PCH. Because of limited space for connections under the chip Intel has something called "Flexible IO". A basic explanation is that you have a certain number  of connections and you program the CPU with what protocol or mix of protocols you want it to send over that lane. For example on some lanes you may have the option to have PCIe, USB, or both. On others you may have the option of PCIe, Gigabit Ethernet, or both. If you choose both you need an extra chip (demuxer) to split up the signals.

 

AMD on the other hand is a bit more straightforward. Their older chips used the same method as the Intel H series. Chipset (fch) on the motherboard and then talk to it via PCIe (sometimes branded as UMI). The CPU provided it's own PCIe lanes for graphics and the Chipset provided it's own for IO.

 

Ryzen Mobile works a little different. Honestly I'm not too sure as I haven't dealt with it at all. I do know that it supplies 12 lanes of PCIe off the APU for Graphics and SSDs, but I don't know if there's an internal interconnect or what there is with regards to chipset.

[Doobeedoo]

Not having USB Type-C is such a crucial error that I can't believe they didn't include it across the board. Also, the Surface Laptop doesn't have it and not even dedicated GPU option, yet the Surface Book has both?! Like what?! You'd think laptop name would make more sense though...

[mr moose]

15 hours ago, LAwLz said:

 

Nope.

 

That's not how software works.

Windows doesn't have that specific optimizations or patches written for it. They don't code one patch for a specific CPU model, and as for GPU drivers that is entirely done by other companies like Nvidia and Intel.

And that's why we have issues that are specific to CPU's?  like windows 7 and 8.1 not even running on latest CPU's and AMD CPU's that wouldn't even boot?

Windows does have CPU specific issues sometimes it's neglect and sometimes it's intentional. 

Quote

 

Microsoft on January 18 released build number 16299.194, which resolved the issue stopping AMD processors from booting. For those who already manually fetched that from the Microsoft Update Catalog, the new KB4058258 update will only install other fixes. This update will download and install automatically from Windows Update.

There are no new features in this update, just a range of bug fixes, quality improvements and fixes for compatibility issues.

The key changes include a fix for a compatibility problem that distorted colors when a PC is connected to displays that support wide color gamut. It also fixes "a condition where second monitor is connected to legacy AMD display adapters flashes after waking from sleep".

 

https://www.zdnet.com/article/windows-10-update-microsofts-latest-bug-fixes-include-amd-reboot-patches/

 

https://www.extremetech.com/computing/248069-unofficial-patch-unblocks-windows-7-8-1-updates-kaby-lake-ryzen

[LAwLz]

14 hours ago, mr moose said:

And that's why we have issues that are specific to CPU's?  like windows 7 and 8.1 not even running on latest CPU's and AMD CPU's that wouldn't even boot?

Windows does have CPU specific issues sometimes it's neglect and sometimes it's intentional. 

https://www.zdnet.com/article/windows-10-update-microsofts-latest-bug-fixes-include-amd-reboot-patches/

 

https://www.extremetech.com/computing/248069-unofficial-patch-unblocks-windows-7-8-1-updates-kaby-lake-ryzen

Meltdown and Spectre have been very much edge cases, and that's mostly because of some differences in how AMD and Intel processors interprets certain instructions which hasn't been an issue until now (lfence and CPUID instructions are being used in ways they weren't meant to be used, and because of this non-standard use exact implementation is very important for spectre patches). My guess is that Microsoft assumed lfence was a serialized instruction which stopped speculative execution on both Intel and AMD processors, and used that in their patch. It is this way on all Intel processors, but it wasn't serialized on some AMD processors (it was on some AMD families, and in some it wasn't, but from now on it will be on all AMD processors).

 

As for the blocking of Windows 7 updates on Ryzen and Kaby Lake, that is 100% an artificial limitation imposed by Microsoft to force people to Windows 10. It is not some "optimization". If it weren't for contractual obligations I'm sure Microsoft would have just stopped making Windows 7 update completely, rather than this half-measure where they only do it for newer processors. Also, the patch itself isn't hardware specific. The CPU control code runs on all processors, and then just sets a flag to true or false before the update service tries to connect. The code itself is not hardware specific.

 

 

Things like CPUs has to follow fairly strict guidelines when they are designed. The reason for that is so that OS makers don't need to code specific patches for specific hardware. It would be a monumental undertaking which would never work. The entire reason why we have things like compilers, high level languages, HALs, driver models and so on is to avoid needing platform specific code and patches.

 

People really need to stop assuming companies like Apple are sitting there, writing assembly code with exact optimization for individual chips. They don't. 

[mr moose]

7 hours ago, LAwLz said:

Meltdown and Spectre have been very much edge cases, and that's mostly because of some differences in how AMD and Intel processors interprets certain instructions which hasn't been an issue until now (lfence and CPUID instructions are being used in ways they weren't meant to be used, and because of this non-standard use exact implementation is very important for spectre patches). My guess is that Microsoft assumed lfence was a serialized instruction which stopped speculative execution on both Intel and AMD processors, and used that in their patch. It is this way on all Intel processors, but it wasn't serialized on some AMD processors (it was on some AMD families, and in some it wasn't, but from now on it will be on all AMD processors).

 

As for the blocking of Windows 7 updates on Ryzen and Kaby Lake, that is 100% an artificial limitation imposed by Microsoft to force people to Windows 10. It is not some "optimization". If it weren't for contractual obligations I'm sure Microsoft would have just stopped making Windows 7 update completely, rather than this half-measure where they only do it for newer processors. Also, the patch itself isn't hardware specific. The CPU control code runs on all processors, and then just sets a flag to true or false before the update service tries to connect. The code itself is not hardware specific.

 

 

Things like CPUs has to follow fairly strict guidelines when they are designed. The reason for that is so that OS makers don't need to code specific patches for specific hardware. It would be a monumental undertaking which would never work. The entire reason why we have things like compilers, high level languages, HALs, driver models and so on is to avoid needing platform specific code and patches.

 

People really need to stop assuming companies like Apple are sitting there, writing assembly code with exact optimization for individual chips. They don't. 

You completely missed the point,  As I said:

 

21 hours ago, mr moose said:

 

Windows does have CPU specific issues sometimes it's neglect and sometimes it's intentional. 

 

 

As I have shown it happens, MS has a plethora of data for millions of crashes, it's really not that unrealistic of a concept that they looked at that data and noticed the most stable combination was last gen intel and last gen Nvidia.  

 

Also I never mentioned meltdown or spectre, even thought hey were in the article I linked, the bit I quoted to had nothing to do with them.   I realise they are the exception not hte rule.

[Mooshi]

Kinda disappointing, this thing. Guess I'm still waiting for the iPad Pro to get a rework.

[Rayhunter]

I was really hoping the Surface Laptop 2 was going to have USB C, Shame.