Monday, April 8th 2013

Intel Core "Haswell" Processors Begin Shipping to OEMs

Intel reportedly began shipping its 4th generation Core processors, codenamed "Haswell," to its OEM partners (companies that sell pre-built PCs). These shipments could go into building the first batches of PCs running the chips, which will be launched alongside retail trims of the chips, in June 2013. In June, on the sidelines of Computex 2013, one of the biggest annual PC technology tradeshows, Intel is expected to launch its Core "Haswell" processors, various motherboard vendors are expected to launch compatible socket LGA1150 motherboards.

Among the Core "Haswell" chips Intel is expected to launch for the desktop platform, are the Core i7-4770K flagship quad-core part, and the Core i5-4670K, both of which feature unlocked base-clock multipliers, appealing to overclockers and PC enthusiasts. There are twelve other models on offer, with locked base-clock multipliers, a majority of which come in energy-efficient brand extensions such as -T and -S.Source: C|Net
Add your own comment

30 Comments on Intel Core "Haswell" Processors Begin Shipping to OEMs

#2
T3RM1N4L D0GM4
Good, first batches of PC running a defective chipset too...
or oems will use only third party usb3 controller?
Posted on Reply
#3
Ronnyv1
@T3RM1N4L D0GM4 didnt they resolve that?
Posted on Reply
#4
fusionblu
by: Ronnyv1
@T3RM1N4L D0GM4 didnt they resolve that?
They fixed it, but since the defective stuff has already been made or that production of the defective C1 chipset can't be stopped the defective chipset will still be released sometime in June and you should wait till mid-August/September if you want to avoid the defective chipset and get a C2 chipset motherboard.
Posted on Reply
#5
Dave65
Really isn't much of a gain from Ivy Bridge to Haswell is there?
Posted on Reply
#6
fusionblu
by: Dave65
Really isn't much of a gain from Ivy Bridge to Haswell is there?
The only gain here is if you always want to have the next generation technology from Intel, but even though I have a Sandybridge system set-up I still intend to stick with it as what it can do (as it still is now) is more than good enough for what I want to do in terms of gaming and general use.
Posted on Reply
#7
Dave65
by: fusionblu
The only gain here is if you always want to have the next generation technology from Intel, but even though I have a Sandybridge system set-up I still intend to stick with it as what it can do (as it still is now) is more than good enough for what I want to do in terms of gaming and general use.
I did upgrade from Sandy Bridge to Ivy Bridge but my Sandy was great for what I do..Just had that itch:)
I do see some improvements in benching but in gaming you don't see it..
Posted on Reply
#8
Jorge
Yeah I'll bet customers who don't know about the defect will be thrilled when they discover Intel selling known defective goods. In fact the SEC might not be too keen on this business model?
Posted on Reply
#9
jihadjoe
Every processor manufacturer is effectively selling defective goods, Intel probably less than most.

AMD never fixed the original Phenom TLB bug, and then there's the Bulldozer TLB bug, another division bug for Llano and Athlon II (errata 665), and then the stack pointer corruption bug discovered last year (errata 721). There are hundreds of bugs in any modern CPU. These are often never fixed, and it's left to the OS, the motherboard vendors and programmers to implement workarounds.

The standard response to any of these? Something like this:
727 Processor Core May Hang During CC6 Resume
Description
During a resume from core C6 (CC6) state, the processor may hang.
Potential Effect on System
Processor core hang, usually resulting in a system hang.
Suggested Workaround
BIOS should set MSRC001_1000[15] = 1b.
Fix Planned
No fix planned
The errata above affects Bulldozer and current FX processors. Yes, that's a sleep resume bug, just like Intel's and possible more troublesome because it will crash the entire system.
Posted on Reply
#10
NeoXF
by: fusionblu
They fixed it, but since the defective stuff has already been made or that production of the defective C1 chipset can't be stopped the defective chipset will still be released sometime in June and you should wait till mid-August/September if you want to avoid the defective chipset and get a C2 chipset motherboard.
I really hope board makers will tag them accordingly... I hate confusion...
Posted on Reply
#11
Melvis
by: jihadjoe
Every processor manufacturer is effectively selling defective goods, Intel probably less than most.

AMD never fixed the original Phenom TLB bug, and then there's the Bulldozer TLB bug, another division bug for Llano and Athlon II (errata 665), and then the stack pointer corruption bug discovered last year (errata 721). There are hundreds of bugs in any modern CPU. These are often never fixed, and it's left to the OS, the motherboard vendors and programmers to implement workarounds.

The standard response to any of these? Something like this:

The errata above affects Bulldozer and current FX processors. Yes, that's a sleep resume bug, just like Intel's and possible more troublesome because it will crash the entire system.
The difference is this isnt a bug thats is so rare that you will never see it in your life time, with this new line up from intel its called a fault that HAS to be fixed and is but still releases the faulty parts anyway, there is a big difference between the to.
Posted on Reply
#12
Aquinus
Resident Wat-man
The flexible bclk is for all Haswell chips, right? Might make it worth it to get an i3 when they come out. If the base clock is as flexible as it is on my 3820, you won't need a K edition to get a decent overclock. I do enjoy my 129.5x34 = 4.4Ghz. :cool: Who needs turbo? :p
Posted on Reply
#13
Animalpak
Im in a hurry for upgrade but the stepping process need to be more mature i think...

November for me is the month to buy haswell.
Posted on Reply
#14
happita
by: Animalpak
Im in a hurry for upgrade but the stepping process need to be more mature i think...

November for me is the month to buy haswell.
My sentiments exactly. I'm kinda in the mood to upgrade too, but not until they fix the bug that messed up the USB 3.0 and to improve the CPU a bit in the next stepping.
Posted on Reply
#15
Mathragh
It surprises me that almost no news get published about the arguably most revolutionary piece of tech in Haswell, namely the integrated VRM.

Found an article about it after I also kinda forgot about it : http://www.extremetech.com/computing/152423-intel-goes-toe-to-toe-with-analysts-denies-haswell-has-power-issues.

I can see the integrated VRM's being a blessing for overclocking, but also another way for intel to lock away any potential in non-K models.
And of course, if the iVRM's perform near their intended goal, system poweruse should be quite a bit lower for mobile devices.
Posted on Reply
#16
NeoXF
If I'll see a deal with the i7-4770K before Z87s with fixed USB3.0 errata, I might jump the gun... seeing as it will be good to go in a few weeks after that with an awesome C2 board... if shipping and labelling is done properly for once in my country that is.

Otherwise, might as well wait for a FX Steamroller or a A12 Kaveri APU (6-core, w/ L3, and all the new heterogeneous voodoo that will come with it).

But for something even more on-topic... those puppies better clock to 5GHz easy / have no massive heat-spike issues like IB and so on...


Speaking of which, relating to both my subjects... I hope the Broadwell upgrade path is straightforward... I don't care much about DDR4... it's useless outside of APU/workstation/server markets at the moment, but I DO hope there'll be a mainstream move to 6-core with Broadwell... (for Intel)... and 10 to 12 core (for AMD).
Posted on Reply
#17
DarthCyclonis
Yeah, think I will skip Haswell and see what Broadwell brings or wait for Skylake. I never like to buy the first generation because of the wonderful issues discussed already. Besides, My i7-970 is more than enough right now.
Posted on Reply
#18
NeoXF
by: DarthCyclonis
Yeah, think I will skip Haswell and see what Broadwell brings or wait for Skylake. I never like to buy the first generation because of the wonderful issues discussed already. Besides, My i7-970 is more than enough right now.
Yeah... And if I'm not mistaken that 970 is a 6-core... so in older multi-threaded software it'd still be faster for sure.


I'd like to see Intel move to 8-core on X79/X99... so as there being a place for 6-core CPUs on 1150. I had more arguments for the move to mainstreamise (Intel) 6-core CPUs... but I forgot them at the moment LOL. Either way, if Kaveri/FX Steamroller delivers, Intel will finally start to feel a real burn, and maybe x86 tech will move already...
Posted on Reply
#19
cadaveca
My name is Dave
by: Mathragh
It surprises me that almost no news get published about the arguably most revolutionary piece of tech in Haswell, namely the integrated VRM.
Having an integrated VRM would explain the higher TDP of these chips, since VRMs on the chip would need cooling too.
Posted on Reply
#20
revin
by: cadaveca
Having an integrated VRM would explain the higher TDP of these chips, since VRMs on the chip would need cooling too.
I'm confused[ when ain't I lol} I thought from the release thread the VRM was just some kind of controller, and the TDP was due to the HD4600 iGPU ?
Posted on Reply
#21
cadaveca
My name is Dave
by: revin
I'm confused[ when ain't I lol} I thought from the release thread the VRM was just some kind of controller, and the TDP was due to the HD4600 iGPU ?
This is what was given out some time last year(SAUCE):






You may be right, it could be just the iGPU, but to me, with a smaller process, the CPU and System Agent should draw less power, allowing the iGPU to be faster and larger, while remaining within the same power envelope. Mind you, it's only another 10 watts...

So I dunno whether it's the VRMs or the iGPU, but I was personally expecting higher tDP due to the iVRM.
Posted on Reply
#22
Mathragh
by: cadaveca
This is what was given out some time last year(SAUCE):

http://www.techpowerup.com/forums/attachment.php?attachmentid=50780&stc=1&d=1365696237

http://www.techpowerup.com/forums/attachment.php?attachmentid=50781&stc=1&d=1365696237


You may be right, it could be just the iGPU, but to me, with a smaller process, the CPU and System Agent should draw less power, allowing the iGPU to be faster and larger, while remaining within the same power envelope. Mind you, it's only another 10 watts...

So I dunno whether it's the VRMs or the iGPU, but I was personally expecting higher tDP due to the iVRM.
Exactly my thoughts, system poweruse will probably be lower, but the package poweruse wil be up, just not up as much as the system power will be down because of no VRM's on the mobo.
I suppose this will also allow mobo makers to lower the cost of the motherboard, and furthermore, it will probably mean that the problem of blown up VRM's on cheap motherboards will be history because there will be nothing left to blow up!

Edit: another pic i found of Haswell with supposedly external ram on chip, found here, Enjoy!


I wonder where the iVRM is on this chip btw lol
Posted on Reply
#23
MxPhenom 216
Corsair Fanboy
All I know is ill be getting Haswell. I want that new socket. It might be an OCD thing with me or something, but I like being on current sockets. LGA1155 is done, it still has the power to do everything, but it wont have anymore chips being released for it, no upgrade path for it now while the socket got LGA1150(Haswell) will be new and have about 2-3 generations on it.

Ill wait for the C2 boards to come out, don't like the idea of buying something that I know is faulty. Just peace of mind I guess.
Posted on Reply
#24
cadaveca
My name is Dave
by: Mathragh
I wonder where the iVRM is on this chip btw lol
Yeah, I dunno what's real about Haswell, and what isn't. I was expecting higher TDP due to this iVRM, and higher TDP it seems to be. But that huge eDRAM die...would explain it too.


That's one large chip, IMHO.
Posted on Reply
#25
revin
by: Mathragh
Exactly my thoughts, system poweruse will probably be lower, but the package poweruse wil be up, just not up as much as the system power will be down because of no VRM's on the mobo.
I suppose this will also allow mobo makers to lower the cost of the motherboard, and furthermore, it will probably mean that the problem of blown up VRM's on cheap motherboards will be history because there will be nothing left to blow up!
Well alrighty then, that answer's my next thought where is the VRM gonna be.
So you mean that my sweet 8-phase active-swiching/load-balancing digital PWM power circuitry is on chip,

then will that lend to "limited" overclocking ?

I mean, Intel did a great job on my M/B, especially on the killer 1866 Ram divider, but I'd hate to move on only to find that there's no more hope of hitting high 4.8-5.x+ CPU speed's.
FWIW, it was by no mean's easy for me to get there, sure I didn't have to soldier, but it took a LOT of trial and error with this "engineering" BIOS to get a good O/C.

Hell when I left[THAT's Why} the Gigabyte GA-UD3R, I couldn't get sht for O/Cing, and it was with my same Q6600 that did really well on the conroe865PE, so yea it's always a crap shoot, but that's what the $30 Warranty was for on this Z68/i7, Fry it/ they Buy it Lol.
Hope it stay's that way, Fast AND Stable
Posted on Reply
Add your own comment