Haswell mobo news/reveals??

C

c3k

2[H]4U
Joined
Sep 8, 2007
Messages
2,330
Okay, Haswell should be released in early June, about 6 weeks from today. What kind of boards are the various vendors bringing to the party? Has anyone (cough, [H], cough) gotten any samples to review?

It seems like a drought. Or a blackout.

Ken
 
Haswell will be using LGA1150 which is obviously a new socket "again" what a shame.
http://hexus.net/tech/news/cpu/53097-intel-core-i7-4770k-haswell-cpu-benchmarked-priced/

Course the new boards will also have a "bug" but will be sold anyways as its not deemed a major issue, idk seems annoying as hell :p
http://www.bit-tech.net/news/hardware/2013/03/08/haswell-z87-bug/1

but yeh, no reviews out yet for the cpu or motherboard as a whole, ENG samples is one thing, finished products another, NDA is still active best I can tell, all we seem to have is pictures of the boards and specs of the chips nothing more.
 
The most I've seen are a few pics of prerelease boards. Nothing concrete yet. I'm waiting on this, too.
 
Verass said:
Will that be fixed by January next year?
Click to expand...

Rumor was it was supposed to be fixed by whatever stepping follows the release stepping, but I don't know if we've had official word yet. I'd be very surprised if it wasn't fixed by Jan.
 
its a bug with the chipset not cpu, otherwise they could have fixed before launch, but it will require a new revision for the already out there boards. I can post many links to it, but why bother, nice they know about the bug, not so nice they will decide to ship with it anyways.

http://news.cnet.com/8301-1001_3-57578223-92/intel-confirms-usb-bug-in-haswell-chipset/
Intel confirms its a bug in the chipset, I am sure they know what they are talking about, no?

The final version sold to customers(Asus and the like) July 15th
 
It's a chipset bug, not CPU. Kind of similar to the SATA bug with P67.

However, it may not be as an annoying issue as it appears to be, as it only causes USB 3.0 devices to be dropped upon waking from S3 state. Haswell is implementing a new sleep state (not sure what it's called) that promises the speed of S1 waking with the power savings of S3 sleep, and USB 3.0 devices are not affected by this new sleep state.
 
Tsumi said:
It's a chipset bug, not CPU. Kind of similar to the SATA bug with P67.

However, it may not be as an annoying issue as it appears to be, as it only causes USB 3.0 devices to be dropped upon waking from S3 state. Haswell is implementing a new sleep state (not sure what it's called) that promises the speed of S1 waking with the power savings of S3 sleep, and USB 3.0 devices are not affected by this new sleep state.
Click to expand...

Sounds interesting. source of info?
 
How long does it usually take for Asus to release ROG motherboards after a new CPU gets released?
 
not usually that long these days, recently as an example for AMD 900 series boards when I had gotten my M5A99X EVO, they had http://comfortingbobcyx.blogspot.ca/2013/02/asus-amd-900-series-motherboard-specs.html all on launch day, so I suppose it depends, however for most folks, the one or 2 steps below ROG is more then ample enough, heck in my case this board is perfect, plays nicely with dual cards and the like and actually had a little cleaner of a design layout, though the sabertooth and Formula were pimpin nice looks.

On another note, that new S state is not quite the way you understand it or were wording it, it is meant for Mobile type usage specifying a platform as a whole to accommodate this Haswell U/ULT as well as possible display changes, not quite desktop for the moment.
S0-S6 are standard
with Haswell(clovertail) S0ix is also present from S0i1 to S0i13

sourced from article you posted
"Today, transitioning into S3 sleep is initiated by closing the lid on your notebook or telling the OS to go to sleep. In Haswell (and Clovertrail), Intel introduced a new S0ix active idle state (there are multiple active idle states, e.g. S0i1, S0i3). These states promise to deliver the same power consumption as S3 sleep, but with a quick enough wake up time to get back into full S0 should you need to do something with your device. "

quite neat though, and good someone is taking initiative to adjust the platform as a whole unit and not just the cpu or specific chipset functions, be pretty hard to say "we knocked down power by X over the past X years (but the Ethernet controller uses at least the same amount)" would it not :p
 
It's still a chipset design, one that will most likely be universal across desktop and mobile. Sure, most of the focus will be on the advantages for mobile, but desktops have sleep states as well, and there has been significant focus on improving efficiency for desktops.
 
this is true will not argue that, as it stands today, ULP being smartphones tablets and such, then laptop or mobile type platforms, desktop and servers are all quite separate parts of the market, at some point I am quite sure the market as a whole will unify many things, something AMD is also betting the farm on by having a unified socket in the near future(one socket but different features per design cannot unify everything) but as it stands now, this is not quite the case. To control as many things as they state for desktop as they will for laptop or smartphone well that is not nearly the same thing, desktop as a whole for most of us are very much anything but a standard thing, hell think even the difference of motherboard makers even using the same chipset, they use so very many different parts, laptops as a whole not nearly as complex(think like console vs pc very similar if you break it to this level)

Intel took a big PR hit when word leaked they may be trying to get rid of the upgrade factor i.e socket-less, I find it very hard to believe they would then make it so infinitely worse by basically forcing makers to only use X things, for say an Ultrabook, hell yeh, that is their branding, but as a desktop type environment as a whole not likely, at least not for now, I do not believe very many if any makers would truly stand behind that, mostly cause there is not an outright need. Sleep states is one thing, but going by the article alone, for the extended states there is and has to be a need for so much more control on many things, I do not see that happening for desktop quite yet, I know many desktop things that bugger out if things do not resume properly from the various states would hate to see them add 13 more :p Chipset for mobile, chipset for desktop are vastly different things, do they use some of the same building blocks of course they do and need to, but they are vastly different as well.

Just them lowering overall wattage for cpu, chipset, gpu, modern ram, and the lower power states(cool and quiet as an example) has made a profound difference on average and overall power use for desktop, but they still need enough power for the grunt, something mobile does not to a point. Desktop and server side very much seem to want the most performance per watt possible(overall efficiency) other designs try to get as much performance out of a given power budget, different environments require different things, will this trickle down, I am sure it will, but I bet you cookies it will be implemented vastly different, cause it would need to be, well that and Intel is not making self branded motherboard anymore, just the rights for others to make them.
 
Laptops, ultrabooks, and x86 tablets all use the same core parts as desktops. The only difference is in the packaging.

There is no fundamental difference x86 devices. The only possible difference is the new sleep states requiring use of the iGPU, in which case laptops with Haswell and discrete GPUs will most likely be utilizing Optimus to be able to use the new sleep states, while desktops with discrete GPUs can't. That's the only thing I can see limiting the new sleep states to laptops/tablets only.
 
optimus is not even Intel that's Nvidia, last I checked most platforms these days that are Intel either use Intel period or discrete (AMD or Nvidia) with Intel very much favoring just themselves Ultrabooks almost exclusive territory. Optimus and switching technology overall seems to cause much grief, far different then the chip itself controlling these power states as the article is seeming to imply, and why they want to control more then just the cpu, everything else needs to advance in the same direction to bear the best fruit sort of speak.

Laptop, Ultrabook etc very much are different in the sence that their BOM are quite controlled overall vs what standard ATX motherboard use, you mean to tell me that laptop, tablet, desktop etc only have a form factor change that is the only difference, I think there is quite a bit more then that, even the cpu design tends to be not a simple low wattage design, they are very much related to their desktop counterparts, but it has also been a few years since laptop type chips were desktop chips, they are quite custom now to fit in the power budget as a simple example, Even AMD laptop chips are related to their desktop counterparts, but they are not quite the same beast either, semi-custom(consoles etc) is a very good word to use to understand it.

There is fundamental design difference between the different market points as I have stated and tried to point out, otherwise well we could just take any chip on anything and use it anywhere, it does not work that way power, packaging, drivers, parts all tend to be modified at least a bit, and not simply by choice to fit it into a smaller box.

Either way, Desktop may see some of this, but it definitely will not see all of it. Desktop GPU are starting to get real nice features for power saving as well, Radeon 7790 is a good point to make, just imagine one or 2 generations from now what it will be like :)
 
Last edited:
I only brought up Optimus because the new sleep states might require the primary GPU to be the iGPU in Haswell, and using Optimus, the discrete GPUs are connected through the iGPU. If it doesn't require use of the iGPU, then it's a reasonable assumption the features will be the same on both desktops and laptops.

The CPUs in laptops are the exact same ones as used in desktops, just binned for low voltage/low power consumption. The i7 3630QM is the exact same die as the i7 3770k. I would venture to guess that the HM77 chipset is also pretty much identical to the H77 chipset. Probably, just once again, binned for low voltage/low power consumption. VRMs are the same. Other external controllers are the same. Discrete laptop GPUs are just low power versions of desktop GPUs.

It costs a significantly more to produce 4 different lines of CPUs/GPUs than it is to have 2 lines and just bin the parts as needed.
 
I will agree on a few things you pointed out there, and yes I suppose it does make sence to a point, but binning truly only means so much, I know binning was quite the order of the day years back, these days they do tend to go ahead and make changes at the arch/die level to keep power in check by optimizing and some changes to things, there are parts I imagine that are not quite as needed on desktop as they would be mobile and vice-versa, and there are some difference between the HM chipsets and the standard ones used for desktop http://www.pcpro.co.uk/reviews/processors/374158/intel-ivy-bridge

I suppose a lot of this does not matter at the end of the day, I suppose all I am getting at is they have much more granular control of things for mobile and such then they do for desktop parts as a whole, primarily cause they need to and secondarily cause there is not near as much incentive to control the desktop space nearly as much from say a power or temperature perspective, so yes certain things such as GPU for mobility generally are just cut down versions, but a lot of them are starting to use features you only see for mobility, so while some could be software specifically, some are and have been a hardware change/optimization as well.

Seems what this article is stating to get the most of this new S state(s) there will be much greater control on nearly every aspect of the final product, cause there would have to be to have a certain guarantee of performance, power etc, and they did state certain changes for power delivery, port changes etc, so maybe as it stands now there is not quite as much difference, but there some might be, cause they may not have the choice.

I remember back with the older Pentium for mobile, Older Turion etc they were sometimes lower wattage parts sometimes just lower voltage, but it does seem now if you compare directly the mobility parts to the desktop, voltages, wattage, multipliers, sometimes cache, and mostly shaders(gpu of course) there does tend to be a fair amount of difference, and yes sometimes a complete custom for specific(Atom as an example)

I suppose on a simple principle binning accomplishes most of this, but there is going to be and is times were binning is simply not quite enough, mobile gpu show this quite much, they have less of mostly everything so they are not simply binned and they wouldn`t just fuse off X transistors that would be very expensive to do, better to make the desktop stuff see how it pans out and shortly after that make the mobility parts based on the design of the desktop again, semi-custom.

Like this talking, makes thoughts wander lol
 
http://.com/cpus/Intel-Core-i7-3770K-vs-Intel-Core-i7-3630QM

things stand out, L3 cache, operating temperature max as examples, that is not quite just a simple binning process no more then a RAD hardened design is just a run of the mill cpu, do they similarities of course, but I do believe there is more then simple binning. Like the difference of F1 and a sports car, they both make exhaust, have tires, push X power with X fuel but a fair amount of the nuts and bolts are changed as the operating requirements make it that they need to be changed, least that is the way the industry is making it cause it needs to.

I would imagine it would be awful expensive for Intel or AMD for that matter to do a butt-load of binning compared to just making a semi-custom(chop and optimize the core design) for the task at hand considering they sell in the millions of these per month, I think it would be more financially sound to have a dedicated design based on a certain architecture(the last links kinda bring that much to mind) more then it would be to save the cherry examples for mobile(low power) they know their targets, and can build according to this, but you would take a heck of a $ loss to "hope" in getting a proper yield to bin them in different segments vs binning for X speed or X power in the same segment, you can brand them as extreme or ultra low power (low clock) but it would be much different to have to bin test to see which would make a best candidate for mobile use AND desktop use as well, that would be quite wasteful.

IvyBridge desktop
http://en.wikipedia.org/wiki/LGA_1155
Ivybridge Mobile
http://en.wikipedia.org/wiki/Socket_G2
"Socket G2, also known as rPGA 988B (there are Socket G/rPGA 989 sockets that can take Socket G2/rPGA 988B), is Intel's CPU socket for their line of mobile Core i7, the successor to the Core 2 line. It is based on Intel's Sandy Bridge architecture

This leads me to believe there are changes to the cpu in question as well, there would need to be.

Generally speaking mobile are more stripped down in regards to possible features, inclusion of ports/connectors, VRM outlay is also different as not as much wattage to deal with, seems only to make sence they would trim other things down to truly take advantage of the need, goes both way I guess, cost to not doing so, and cost in doing so(ask Nvidia with GT84/86 and G92/94)

Sorry for the long post again, I just find this interesting fact digging and the like :)
 
You must log in or register to reply here.
Back
Top