Updated 2013-08-16: MacMan's Haswell Test Build: Core i7-4770K + GA-Z87X-UD5H

[lionsaysnotoatom]

MacMan's Haswell Test Build: Core i7-4770K + GA-Z87X-UD3H

Well, it looks like Mountain Lion will have native Haswell support quite soon. The 2013 MacBook Air is shipping with Mountain Lion on it. I'm guessing a minor update will be issued within the next few days adding support for Haswell and the newer Intel HD graphics.

The only worry us Haswell adopters have now is whether or not HD 4600, HD 5100, or HD 5200 gain Mountain Lion support. If not, I'm betting we'll have to wait for 10.9 Mavericks before we get full Intel HD support.

I already know my Haswell CPU, ALC892 audio, and PCI Ethernet card will work with Mountain Lion. So now I just have to wait on the Intel HD 4600 or the GTX 770 gaining native support in ML. Then I can dive in. Can't wait for some Haswell hackintosh goodness.

EDIT: Just read that the NVIDIA web drivers add full support for the GTX 770 in 10.8.3! Now just waiting for MacMan to release bootloader...

 

[genzai]

MacMan's Haswell Test Build: Core i7-4770K + GA-Z87X-UD3H

EDIT: Just read that the NVIDIA web drivers add full support for the GTX 770 in 10.8.4! May go ahead and start the hackintoshing...

1) The GTX 770 is already fully supported in vanilla 10.8.4. I use one.

2) There is no guarantee at this point that HD4600 will work, or when it will work, even if 10.9 is released. It will depend entirely on what intel chips go into real macs and what type of broad or narrow support the intel drivers will have. In the past the intel driver support has been very narrow. For instance in IVB the HD4000 and HD2500 are from my understanding pretty much exactly the same except for the number of cores, and yet the HD2500 never seemed to get proper support (correct me if wrong). It was the same situation with SNB and the HD3000 vs HD2000. That said it seems more likely this time than ever that at least some macs will ship with a HD4xxx video, but we cant know for sure until it does.

3) There is no reason to think the NEC hubs wont work. Why is everyone assuming they wont? Have i missed something? The older pre IVB motherboard often shipped with NEC CONTROLLERS (not hubs) and these did not work when a VIA hub was attached (or any hub for that matter) because of the crappy third party drivers available at the time. However, ever since native OS X USB 3.0 drivers have been out, VIA hubs off the native controllers have worked fine and i see no reason to think the NEC ones wont work just as well. As an aside, thanks to zenith's new generic USB 3.0 driver, the old NEC controllers are also working fine with hubs now. In other words the hubs are a feature, not a deterrent.

4) The macbook air is shipping with a custom version of 10.8.4, it is unlikely there will be an update that will allow other computers to bridge the gap, that update will be 10.8.5 and wont be out for a while. IF for some reason apple issues an update specifically for the new airs, it will be a delta update meant for the specific shipping version of 10.8.4 on the airs. applying the update to a vanilla 10.8.4 would quite likely break things due to version mismatches. Early haswell adopters have 2 options. Get a copy of the 10.8.4 thats ships with the new airs (there are ways), or wait until 10.8.5 or 10.9 ships (or become a dev so you can use early versions). A third potential option would be to run a hacked kernel and run with unofficial haswell support until 10.8.5 or 10.9 ships.

g\

 

[MyNewRig]

MacMan's Haswell Test Build: Core i7-4770K + GA-Z87X-UD3H

I would like to comment on your points from my personal observations:

1) Natively or not at least we have official drivers so this point is covered.

2) Because of the point above, it does not matter if they support the HD4600 or not, i looked all over the place for a Retail Haswell CPU that has the Iris HD5000 and could not find any, so we don't have any options but to get the HD4600, if they end up getting supported fine, of not we rely on discrete cards and we will still have a perfectly working systems with the options we have, and if someone wants to use the Integrated graphics then maybe they should stick with Ivy.


3) I been thinking about this 24 hours a day for many days, and i have put many observations together to end up with the conclusion that Hubs could be a problem, 1st without Hubs the platform is the most native, using the Z87 chip to control USB, PCIE and the likes and using the CPU controllers to control the memory, graphics and the voltage is the most native motherboard design and is sure to prove less problematic, while you can add external hubs if needed, and 2nd all Apple machines do not use many USBs i think most use 2, 4 or 6 in the MacPro, so it seems that apple has opted to support the Z87 platform's native ports and nothing more, and when this platform can support up to 6 USB3 and 4 USB natively, why would i need more? and why would i add Hubs to complicate the design, introduce problems and gain only two extra USB3 ports in return? and 3rd it seems that all Hubs have caused a problem in the past especially NEC ones and thus they have the most probability in causing problems in the future. Because of all of the above i feel that an implementation that uses all the Z87 platform features and nothing more and keep it simple is the safest to go with.

4) Time will tell how we will Hack the Haswell platform, but it seems very possible especially when taking the safe route discussed above.

Just my two cents and would love to hear further comments from others.

 

[genzai]

MacMan's Haswell Test Build: Core i7-4770K + GA-Z87X-UD3H

I would like to comment on your points from my personal observations:

1) Well except that the web drivers don't run in 10.8.4, sure...

2) Sure, except that many people are writing as though they are under the assumption that surely 4600 will work at some point, like when 10.9 ships, and that may or may not be the case. In reality, it MAY never work. And there are other important questions related to this, like will quicksync ever work on these machines? And the reason you can't find a desktop socketed part with 5xxx is because they don't exist.

3) That was quite a diatribe... yes, why oh why would you EVER need more? *sigh* ...again please direct me to evidence that NEC hubs have ever been an issue. In fact NEC hubs are a pretty new thing, and as far as i can tell should work at least as well as the VIA hubs that work just fine now. Either you are confusing NEC hubs with the old NEC controllers, which are totally different things, or you are referencing issues that i am not aware of (which i would like to be made aware of).

4) I encourage early adopters, but it will be rough for a while... Obviously using a discrete GPU and getting a hold of a fully supported OS X build will make things easier in the near term.

Best of luck and keep the reports coming in.
g\

 

[repez]

MacMan's Haswell Test Build: Core i7-4770K + GA-Z87X-UD3H

Thank you very much, you are a life saver! i am going to place an order for the GA-Z87MX-D3H immediately, i can't see how that can go wrong, it is a fully native board, using the Z87 Chip for everything except for the Audio which you have just confirmed, no Hubs, no nothing ...

Why you don't wait until the June 15th? In that date tonymacx86.com will release the list of Haswell compatible hardware like every month...

 

[sung57]

MacMan's Haswell Test Build: Core i7-4770K + GA-Z87X-UD3H

From the view of a beginner like me, fewer non-native components means fewer troubles and easier setup.

 

[MyNewRig]

MacMan's Haswell Test Build: Core i7-4770K + GA-Z87X-UD3H

UPDATE1: I am pretty positive now that the Renesas' uPD720210 host controller interface (HCI), which is a chip that needs a DRIVER to work under Windows/Linux will also need to a Kext to be supported under OSX which has been proven problematic in the past, below i quote from a few referenced and list the reference links after each comment for you to check and provide feedback, Please notice the BOLD TEXT in the quotes:

"What is a hub controller?
A host controller interface (HCI) is a register-level interface that enables a host controller for USB or FireWire hardware to communicate with a host controller driver in software. USB Hubs expand the USB host's connection capability to allow more than one device to be connected to each host port by repeating and branching the bus. The driver software is typically provided with an operating system of a personal computer, but may also be implemented by application-specific devices such as a microcontroller. On the expansion card or motherboard controller, this involves custom logic, with digital logic engines in FPGAs plus analog circuitry managing the high-speed differential signals. On the software side, it requires a device driver (called a Host Controller Driver, or HCD)."

SOURCE: http://electronicsforu.com/electron...ith+USB-IF&id=11616&article_type=8&b_type=new


Also from another source i quote:

"These key features make it easy and inexpensive to add ports allowing SuperSpeed data transfer via a USB 3.0 interface to products such as monitors, docking stations (function extension units that connect to notebook PCs), and TVs. In addition, the μPD720210 hub controller offers substantially reduced power consumption when devices are in standby mode or not in use, which is an important consideration in view of environmental regulations in many countries. Finally, the Renesas Electronics host controller driver stack for Microsoft Windows and device drivers supported by Linux may be used free of charge with the μPD720210."


SOURCE: http://www.businesswire.com/news/ho...Electronics-Introduces-USB-3.0-Hub-Controller


Now the above surely indicates that these "honest looking" little Hubs adding practically only 2 Extra USB3 ports are not seen as an opportunity as some indicated in this thread but could indeed prove problematic and would never be natively supported under OSX, and also i see a potential sleep issues and power management problems as, according to the highlighted text above they try to "save power" for USB connected devices at idle, practically overriding the native Z87 platform native power management.

In the past the uPD720200 proved extremely problematic with 10s of kext posted to solve the problem, while most doesn't or does but with other issues like sleep and such ...


Please correct me if i am wrong, but still up to this point i see that the MX-D3H without the Hubs or bridges or all that stuff is still the safest bet unless proven otherwise

 

[disinterested_hacker]

MacMan's Haswell Test Build: Core i7-4770K + GA-Z87X-UD3H

UPDATE1: I am pretty positive now that the Renesas' uPD720210 host controller interface (HCI), which is a chip that needs a DRIVER to work under Windows/Linux will also need to a Kext to be supported under OSX which has been proven problematic in the past, below i quote from a few referenced and list the reference links after each comment for you to check and provide feedback, Please notice the BOLD TEXT in the quotes:

Every host controller in existence from UHCI/OHCI and EHCI to xHCI has needed a driver. You can't have USB without a host controller and you can't talk to the host controller without a driver unless the capability is compiled into the kernel. Type this in terminal "kextstat | grep -i hci" you're running one right now.

The 210 is actually a hub controller, not a host controller, although neither of them would be any real cause for concern, but hubs don't need drivers, they just need to follow the specification and the controller and hc driver take care of the rest. Past HCI specs are straightforward and writing generic drivers isn't really a problem (lots of work sure, but no reversing required). I'm literally working on a generic EHCI os independent driver right now. xHCI is supposed to consolidate O/U/EHCI and simplify things... I would be very surprised if this becomes a problem.

 

[MyNewRig]

MacMan's Haswell Test Build: Core i7-4770K + GA-Z87X-UD3H

Drivers being hard or easy to implements still does not provide a seamless experience and forward compatibility with every release, using the Z87 native intel controllers is sure to be supported by Apple natively without much hassle and this is the whole point of the discussion, but you saying that Hubs don't need drivers means that they will be able to interface with the Intel driver without a problem?

Also what about the Marvell 88SE9172 eSATA controller? i see in the manual that in windows you have to install special drivers and configurations for it?

The thing is that we know that anything could eventually work, with native support, custom drivers, workaround or whatever, but which Motherboard should provide the most seamless experience and straight forward installation? from my very limited knowledge i see that a board using the Z87 chipset native features to the fullest is good enough, why would i get into these complications to have a few extra ports?

To be honest for me i would really love to have the full ATX UDxH features and expendability, but with so much uncertainty and no testing results or feedback from anyone at this point, and really badly needing to buy a computer now, i am left with no option but to get the board that utilizes the Z87 fully and natively without any spices or addons since it has all the features i need, small form-factor and a good price.

If anyone knows better please provide more input

 

[klumikaze]

MacMan's Haswell Test Build: Core i7-4770K + GA-Z87X-UD3H

Stupid question perhaps... but have you tried the Developer Preview that was released the other day?

Thanks for taking one for the team and doing this by the way. I'm another anxiously awaiting Z87/1150 builder.