Intel "Haswell" Speculation thread

Discussion of all forms of processors, from AMD to Intel to VIA.

Moderators: Flying Fox, morphine

Re: Intel "Haswell" Speculation thread

Postposted on Thu Jul 26, 2012 8:21 pm

it doesnt matter what haswel looks, performs or anything to me. looks like my socket 775 system will easily last me until then, thus successfuly and completely skipped multiple generations before even having to upgrade/rebuild. as long as it isnt too late in the year, i am a little excited for haswell, as i will get such a performance increase, it wont be funny.

kudos to the rest of those that havent splurged on new core pc stuff in years(minus ssd's and usb 3.0 hubs lol)!
td1353l
Gerbil
 
Posts: 27
Joined: Mon May 14, 2012 9:16 pm

Re: Intel "Haswell" Speculation thread

Postposted on Thu Jul 26, 2012 9:59 pm

just brew it! wrote:I find it pretty hard to get excited about what are essentially minor tweaks. IMO the last real revolutions in x86 were the introduction of 64-bit instructions and integrated DRAM controllers. (OK, hardware virtualization support was pretty huge too, but that's not something everyone uses.)


I think I'm on the same page as JBI here. Buying an Athlon 64 chip really felt like buying something that was had a lot more than my Athlon XP had. My Duron 600 (overclocked of course) also kicked the pants off of my K62-333. Intel has done a great job of improving their product time and time again, but it really is just incremental. The difference between generations is pretty small, really, and I can't imagine someone three to four years from now throwing out their system because they didn't wait for Haswell - the people who care about things like the new instruction sets are very, very few and far between. I never heard anyone say something like man, I need to upgrade to a system with SSE2 :P Ultrabooks are cool though, and I'm glad they're continuing to look that way.

If my current system could take 16GB of RAM, I'd probably stick with it, the Core 2 Duo is still chugging along just fine for me.
absurdity
Gerbil Elite
 
Posts: 871
Joined: Sat Mar 02, 2002 7:00 pm
Location: VT

Re: Intel "Haswell" Speculation thread

Postposted on Thu Jul 26, 2012 10:34 pm

Yes but the upgrade from a Core2Duo to any Intel chip out right now is pretty big and should only get bigger.

But I'm in the same boat, Core2Quad does most thing pretty ok still so I wait.
tfp
Grand Gerbil Poohbah
 
Posts: 3070
Joined: Wed Sep 24, 2003 11:09 am

Re: Intel "Haswell" Speculation thread

Postposted on Sat Jul 28, 2012 2:40 pm

They actually DO care! Intel's employees have been very cooperative in Linux forums, offering to diagnose AND fix graphics errors, performance regressions etc. All of this is possible because their graphics drivers are open source, so there's no reason to hide anything. :)
Visigoth
Gerbil In Training
Gold subscriber
 
 
Posts: 6
Joined: Thu Feb 02, 2012 9:11 pm

Re: Intel "Haswell" Speculation thread

Postposted on Sun Jul 29, 2012 10:58 am

Do you think we'll see improvement as big as Nehalem/Westmere to Sandy Bridge?
Or, do you think it will be a smaller improvement like Sandy Bridge to Ivy Bridge?

I think that consumer Haswell-E probably won't have disabled cores due to 22nm.
Intel Core i7-875K, Asus P7P55D-E Pro, Win 7 Home Premium
MSI GTX 560 Ti OC, Mushkin 2x2GB DDR3-1333, Corsair TX650
Cooler Master Hyper 212+, Logitech Z-2300, ASUS Xonar DX
Samsung Spinpoint F3 1TB, Dell Ultrasharp U2410, Antec P183
DeadOfKnight
Gerbil Elite
Gold subscriber
 
 
Posts: 635
Joined: Tue May 18, 2010 1:20 pm

Re: Intel "Haswell" Speculation thread

Postposted on Sun Jul 29, 2012 1:53 pm

Well, from the GPU side of things, expect a big improvement. From 16 EU's to 40 EU's (on the GT3 at least) is nothing to sneeze at, so you'll be getting about 2.5x the performance without any tweaks! This will make Haswell IGP graphics useful for things other than playing HD video, like OpenCL-supported programs (assuming Intel's driver team fix that issue, as it's admittedly behind the competition on that front).

On the CPU side, they'll be streamlining more operations and make efficient use of the 22nm process, now that it has stabilized. For Ultrabooks, the PCH will be on-die (or on-package, there's a little confusion about this), so Ultrabooks will also see a big reduction in motherboard size. For desktops and "regular" notebooks, the transition of the PCH to on-die/on-package will start with Broadwell. I believe Broadwell will be a bigger change for desktop users than Ultrabook users, as Ultrabooks will have already transitioned to the on-die/on-package PCH. :)
Visigoth
Gerbil In Training
Gold subscriber
 
 
Posts: 6
Joined: Thu Feb 02, 2012 9:11 pm

Re: Intel "Haswell" Speculation thread

Postposted on Sun Jul 29, 2012 10:36 pm

Visigoth wrote:Well, from the GPU side of things, expect a big improvement. From 16 EU's to 40 EU's (on the GT3 at least) is nothing to sneeze at, so you'll be getting about 2.5x the performance without any tweaks! This will make Haswell IGP graphics useful for things other than playing HD video, like OpenCL-supported programs (assuming Intel's driver team fix that issue, as it's admittedly behind the competition on that front).

On the CPU side, they'll be streamlining more operations and make efficient use of the 22nm process, now that it has stabilized. For Ultrabooks, the PCH will be on-die (or on-package, there's a little confusion about this), so Ultrabooks will also see a big reduction in motherboard size. For desktops and "regular" notebooks, the transition of the PCH to on-die/on-package will start with Broadwell. I believe Broadwell will be a bigger change for desktop users than Ultrabook users, as Ultrabooks will have already transitioned to the on-die/on-package PCH. :)

Well I don't know if moving the PCH on die is going to improve desktops, although it could potentially mean that the motherboard socket they use might finally last us a good long time, as it clearly will not with Haswell.
Intel Core i7-875K, Asus P7P55D-E Pro, Win 7 Home Premium
MSI GTX 560 Ti OC, Mushkin 2x2GB DDR3-1333, Corsair TX650
Cooler Master Hyper 212+, Logitech Z-2300, ASUS Xonar DX
Samsung Spinpoint F3 1TB, Dell Ultrasharp U2410, Antec P183
DeadOfKnight
Gerbil Elite
Gold subscriber
 
 
Posts: 635
Joined: Tue May 18, 2010 1:20 pm

Re: Intel "Haswell" Speculation thread

Postposted on Mon Jul 30, 2012 9:57 pm

DeadOfKnight wrote:
Visigoth wrote:Well, from the GPU side of things, expect a big improvement. From 16 EU's to 40 EU's (on the GT3 at least) is nothing to sneeze at, so you'll be getting about 2.5x the performance without any tweaks! This will make Haswell IGP graphics useful for things other than playing HD video, like OpenCL-supported programs (assuming Intel's driver team fix that issue, as it's admittedly behind the competition on that front).

On the CPU side, they'll be streamlining more operations and make efficient use of the 22nm process, now that it has stabilized. For Ultrabooks, the PCH will be on-die (or on-package, there's a little confusion about this), so Ultrabooks will also see a big reduction in motherboard size. For desktops and "regular" notebooks, the transition of the PCH to on-die/on-package will start with Broadwell. I believe Broadwell will be a bigger change for desktop users than Ultrabook users, as Ultrabooks will have already transitioned to the on-die/on-package PCH. :)

Well I don't know if moving the PCH on die is going to improve desktops, although it could potentially mean that the motherboard socket they use might finally last us a good long time, as it clearly will not with Haswell.


I think that moving the PCH on die is a fairly stupid move; it's just an interface aggregator. You're still going to need external chipsets to connect various other components to like USB, SATA, Ethernet, Thunderbolt, and so on to their respective physical interfaces. Seems like it would make much more practical to continue to update the link between the CPU and the PCH to support more connections and higher bandwidths like they have been doing since LGA1366 debuted.
Canon 6D||[24-105/4L IS USM|100/2.8L Macro IS USM|70-300/4-5.6 IS USM|40/2.8 STM|50/1.4 USM|85/1.8 USM|Samyang/Bower 14/2.8 Full-Manual Rectilinear Wide-angle|
Canon EOS-M|11-22/4-5.6 IS STM|22/2 STM|EF-M 18-55/3.5-5.6 IS STM|
For sale!|24/2.8 IS USM
|
Airmantharp
Gerbil Elder
 
Posts: 5012
Joined: Fri Oct 15, 2004 10:41 pm

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 6:15 am

Airmantharp wrote:I think that moving the PCH on die is a fairly stupid move; it's just an interface aggregator. You're still going to need external chipsets to connect various other components to like USB, SATA, Ethernet, Thunderbolt, and so on to their respective physical interfaces. Seems like it would make much more practical to continue to update the link between the CPU and the PCH to support more connections and higher bandwidths like they have been doing since LGA1366 debuted.
I'm inclined to agree with you there, it's just going to take up that much more die space that could be used for extra CPU/GPU cores. They're still going to have to communicate with external chips on the motherboard, and they're already bringing the Haswell chipset down to 32nm, so I don't see the motherboard makers being able to fit a whole lot more onto ATX boards.

Maybe this is Intel drawing the line and telling us that whatever amount of cores they can fit into Haswell is good enough. I guess we've kinda seen that coming since AMD is no longer putting up a fight. The enthusiast platform may still continue to grow, but I don't know how much. Moore's law seems to be slowing to a crawl.

I think I'm not so much looking forward to Haswell as I am disappointed with what's likely to come after Haswell. I just want to grab the best performing chip they have before things really slow down. We all hate the words "future proof" used on this forum, but I don't think there's ever been a chip more "future proof" than Haswell will be.

Furthermore, I suspect GPUs will be slowing as well, as there will be little demand for more performance if the console rumors are true and game developers are still going to be targeting fairly low-end hardware. The only thing that might keep pushing it forward is if somehow 4k becomes affordable enough to go mainstream in the future.

All the chipmakers seem to be excited about lately are power efficiency and "Heterogeneous Computing" which still has a ways to go before it can prove itself.
Intel Core i7-875K, Asus P7P55D-E Pro, Win 7 Home Premium
MSI GTX 560 Ti OC, Mushkin 2x2GB DDR3-1333, Corsair TX650
Cooler Master Hyper 212+, Logitech Z-2300, ASUS Xonar DX
Samsung Spinpoint F3 1TB, Dell Ultrasharp U2410, Antec P183
DeadOfKnight
Gerbil Elite
Gold subscriber
 
 
Posts: 635
Joined: Tue May 18, 2010 1:20 pm

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 7:34 am

DeadOfKnight wrote:Fully integrated voltage regulator, thereby moving another component from the motherboard onto the CPU.

I wonder how this will influence the Haswell motherboard industry, since with Ivy Bridge and all motherboards before, a lot of attention is given to designing the VRM. In many ways it's what defines the difference between mid-end and high-end motherboards. Without a seperate VRM, I imagine motherboards will be much more similar, feature-wise and physically.
Esben_DK
Gerbil In Training
 
Posts: 4
Joined: Wed Aug 15, 2007 3:33 am
Location: Copenhagen. Denmark

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 9:49 am

Why wait for Haswell? You've got a great base system that is worth upgrading if you need more performance:

• you are running a K. Are you OC'ing?
• SSD
• are you a gamer in need of more performance? If so that GPU is holding you back.
• your motherboard supports 16GB of memory - memory is cheap!
i7-3770K@4.7 | H100 | P8Z77-V PREMIUM | 16GB | 2 GTX 770 4GB SLI | M500 960GB | EVO 840 250GB | AX850 | Obsidian 550D | R.A.T. 9 | U2713H | U2711
End User
Gerbil Elite
Gold subscriber
 
 
Posts: 951
Joined: Fri Apr 16, 2004 6:47 pm
Location: Toronto, Canada

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 10:35 am

Esben_DK wrote:
DeadOfKnight wrote:Fully integrated voltage regulator, thereby moving another component from the motherboard onto the CPU.

I wonder how this will influence the Haswell motherboard industry, since with Ivy Bridge and all motherboards before, a lot of attention is given to designing the VRM. In many ways it's what defines the difference between mid-end and high-end motherboards. Without a seperate VRM, I imagine motherboards will be much more similar, feature-wise and physically.

Yes and no.

I can't imagine them moving the entire VRM onto the CPU; it would have to be just the VRM controller chip (which is a very small fraction of the total board real estate occupied by the VRM). The VRM uses hefty power MOSFETs, inductors, and filter capacitors, none of which can be integrated onto the CPU die. So the mobos will still look essentially the same physically.

What it WILL potentially do is limit the flexibility mobo designers will have to tweak CPU supply voltages, which would in turn limit the options for overclockers and/or undervolters. Whether this happens or not will depend on how Intel implements the integrated VRM controller. If it is locked down, then there won't be any way to run the CPU at other than stock voltage; OTOH if it is sufficiently programmable, and/or if they expose the VCore feedback loop on an external pin, clever mobo designers will still be able to provide the sort of tweaking options we're accustomed to seeing on enthusiast boards.
(this space intentionally left blank)
just brew it!
Administrator
Gold subscriber
 
 
Posts: 37705
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 11:07 am

End User wrote:Why wait for Haswell? You've got a great base system that is worth upgrading if you need more performance:

• you are running a K. Are you OC'ing?
• SSD
• are you a gamer in need of more performance? If so that GPU is holding you back.
• your motherboard supports 16GB of memory - memory is cheap!

Oh my system runs fine, my brother is turning 18 next year and I plan to give this system to him.
Intel Core i7-875K, Asus P7P55D-E Pro, Win 7 Home Premium
MSI GTX 560 Ti OC, Mushkin 2x2GB DDR3-1333, Corsair TX650
Cooler Master Hyper 212+, Logitech Z-2300, ASUS Xonar DX
Samsung Spinpoint F3 1TB, Dell Ultrasharp U2410, Antec P183
DeadOfKnight
Gerbil Elite
Gold subscriber
 
 
Posts: 635
Joined: Tue May 18, 2010 1:20 pm

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 11:16 am

Esben_DK wrote:
DeadOfKnight wrote:Fully integrated voltage regulator, thereby moving another component from the motherboard onto the CPU.

I wonder how this will influence the Haswell motherboard industry, since with Ivy Bridge and all motherboards before, a lot of attention is given to designing the VRM. In many ways it's what defines the difference between mid-end and high-end motherboards. Without a seperate VRM, I imagine motherboards will be much more similar, feature-wise and physically.

I think this is an expected feature for ultrabooks. I don't think they plan on moving anything else to the CPU on desktops til Broadwell. I could be wrong though as that's what many people are saying is the cause of moving to the new 1150 socket.

If it is, I suspect JBI is right, and that it's not truly a "fully integrated VRM", but just the control portion that talks with the CPU to know how much power to deliver.

The first slide I posted claims "Overclocking Improvements", possibly achieved by better communication with the VRM. This might have been necessary due to the configurable TDP introduced in Ivy Bridge. In fact, it could be part of the reason Ivy gets so hot when overclocked, maybe the VRM thinks it has to supply more voltage than is required.

I expect that if these rumors are true then motherboards have the potential to get cheaper with Haswell. Whether they do or not is a different story. We'll start seeing motherboard pictures leak out long before Haswell arrives, then we will see how much has changed.
Intel Core i7-875K, Asus P7P55D-E Pro, Win 7 Home Premium
MSI GTX 560 Ti OC, Mushkin 2x2GB DDR3-1333, Corsair TX650
Cooler Master Hyper 212+, Logitech Z-2300, ASUS Xonar DX
Samsung Spinpoint F3 1TB, Dell Ultrasharp U2410, Antec P183
DeadOfKnight
Gerbil Elite
Gold subscriber
 
 
Posts: 635
Joined: Tue May 18, 2010 1:20 pm

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 1:55 pm

DeadOfKnight wrote:If it is, I suspect JBI is right, and that it's not truly a "fully integrated VRM", but just the control portion that talks with the CPU to know how much power to deliver.


JBI, as always, *is* right. It's not physically possible for them to integrate the VRM. Intel is almost certainly talking about the VID logic, e.g. the PWM controller, not all the chokes and the like.

It also bares remembering that Intel is up to like at least 3 voltages on the 1156 package, Vcore (obvious), Vtt (memory controller) and VAXG(graphics core). Having different phase setups and regulators for all of them is less than optimal. On-die voltage regulation allows for a more harmonious IC but also much better control. For instance, different Vcores for each core. That's difficult/wasteful to do on a motherboard because a mobo can't guarantee how many cores will be on a package. Intel, obviously, can. The benefits from that one change alone should be obvious.
Glorious
Darth Gerbil
Gold subscriber
 
 
Posts: 7854
Joined: Tue Aug 27, 2002 6:35 pm

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 4:49 pm

just brew it! wrote:I can't imagine them moving the entire VRM onto the CPU; it would have to be just the VRM controller chip (which is a very small fraction of the total board real estate occupied by the VRM). The VRM uses hefty power MOSFETs, inductors, and filter capacitors, none of which can be integrated onto the CPU die. So the mobos will still look essentially the same physically.

I also have a hard time imagining how they should be able to put all the usual VRM circuitry onboard the CPU-package, but at the IDF 2005 they showed a working on-die CMOS VRM.

Also Foxconn did in 2006 make a digital PWM VRM, without big chokes, caps and power MOSFETs. I could imagine Intel would be smart enough to figure out a way to make the VRM circuitry small enough to reside on the CPU-package. They do have a lot of experience in fast switching transistors, and if they use a high PWM frequency they could get away with much smaller chokes and caps. If Intel wants their CPUs to operate at near-threshold voltages, it might be necessary for them to make their own VRM circuit.
Esben_DK
Gerbil In Training
 
Posts: 4
Joined: Wed Aug 15, 2007 3:33 am
Location: Copenhagen. Denmark

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 5:12 pm

Pulling most of the VRM circuitry onto the CPU die may be doable for lower power devices (e.g. mobile). I can't imagine it being practical for higher-powered desktop and server CPUs though. Not only do you have the cap/choke issue, you also have additional heat since the VRM can never be 100% efficient. Any space/cost savings from integrating the VRM would likely be offset by the need for a heftier cooling solution on the CPU.
(this space intentionally left blank)
just brew it!
Administrator
Gold subscriber
 
 
Posts: 37705
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 5:38 pm

If the CPU TDP is already 77W, the extra heat generated from the VRM will account for very little. I would estimate a normal VRM to waste at most 5W-10W, since they are often used without any heatsinks. Thus putting that extra heat load on the CPU will not make it much more difficult to cool. Actually the VRMs would be better cooled as part of the HSF cooling system.
Esben_DK
Gerbil In Training
 
Posts: 4
Joined: Wed Aug 15, 2007 3:33 am
Location: Copenhagen. Denmark

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 6:47 pm

You drop the VRM into the CPU, and suddenly the motherboard just got a lot smaller and the system got easier to cool at higher power draws. Didn't Intel show a new sub-ITX form factor that might take advantage of such a design?

Heck, it'd be cool to see the VRM subsystem moved on-die on discrete GPUs as well, as that's usually around 1/3rd the length of the card, and they can use any power reduction they can get; hell, the advantage for mobile GPUs is undeniable. Might be possible to get 50% more performance per generation out of those with the ability to fine-tune power delivery!
Canon 6D||[24-105/4L IS USM|100/2.8L Macro IS USM|70-300/4-5.6 IS USM|40/2.8 STM|50/1.4 USM|85/1.8 USM|Samyang/Bower 14/2.8 Full-Manual Rectilinear Wide-angle|
Canon EOS-M|11-22/4-5.6 IS STM|22/2 STM|EF-M 18-55/3.5-5.6 IS STM|
For sale!|24/2.8 IS USM
|
Airmantharp
Gerbil Elder
 
Posts: 5012
Joined: Fri Oct 15, 2004 10:41 pm

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 10:13 pm

I still don't think that's what's happening here, not for desktops anyway. Maybe for ultrabooks.
Intel Core i7-875K, Asus P7P55D-E Pro, Win 7 Home Premium
MSI GTX 560 Ti OC, Mushkin 2x2GB DDR3-1333, Corsair TX650
Cooler Master Hyper 212+, Logitech Z-2300, ASUS Xonar DX
Samsung Spinpoint F3 1TB, Dell Ultrasharp U2410, Antec P183
DeadOfKnight
Gerbil Elite
Gold subscriber
 
 
Posts: 635
Joined: Tue May 18, 2010 1:20 pm

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 11:13 pm

Airmantharp wrote:You drop the VRM into the CPU, and suddenly the motherboard just got a lot smaller and the system got easier to cool at higher power draws.

Concentrating more of the heat in a single spot doesn't necessarily make the system easier to cool, especially if it pushes the CPU closer to the top end of of its thermal envelope.
(this space intentionally left blank)
just brew it!
Administrator
Gold subscriber
 
 
Posts: 37705
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 11:21 pm

just brew it! wrote:
Airmantharp wrote:You drop the VRM into the CPU, and suddenly the motherboard just got a lot smaller and the system got easier to cool at higher power draws.

Concentrating more of the heat in a single spot doesn't necessarily make the system easier to cool, especially if it pushes the CPU closer to the top end of of its thermal envelope.


You're absolutely right; my point is based on the assumption that the VRM circuitry in-package or on-die would statically increase the total socket power usage by 5-10w as mentioned above, while bringing with it power reductions due to more accurate voltage switching. That being the case, adding even 15w to the socket isn't that big of a deal in the hottest parts, yet we might expect an overall socket increase of 1-5w while removing motherboard VRM power usage and heat output for a net positive system power usage gain.

Essentially, I believe that the system will be easier to cool because the net gain in heat should be minimal and more concentrated, which should also result in a practical form factor shrink as well.
Canon 6D||[24-105/4L IS USM|100/2.8L Macro IS USM|70-300/4-5.6 IS USM|40/2.8 STM|50/1.4 USM|85/1.8 USM|Samyang/Bower 14/2.8 Full-Manual Rectilinear Wide-angle|
Canon EOS-M|11-22/4-5.6 IS STM|22/2 STM|EF-M 18-55/3.5-5.6 IS STM|
For sale!|24/2.8 IS USM
|
Airmantharp
Gerbil Elder
 
Posts: 5012
Joined: Fri Oct 15, 2004 10:41 pm

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 11:44 pm

I'm not convinced that moving it onto the CPU would make it that much more efficient. Yes, you'll gain *some* since you'll have lower losses in the motherboard power planes and CPU socket; the magnitude of the current flowing into the socket (and hence resistive losses) drops by roughly an order of magnitude since the voltage is about 10x higher. But this is only a few percent effect at best, since resistive losses only account for a small percentage of total power used.

High power switching transistors also occupy a large amount of die space, and don't require the sophisticated manufacturing techniques of a CPU on a nanometer scale process. Integrating them onto the CPU would have a rather large negative impact on the number of good CPUs Intel can get per wafer. Wasteful. You'd probably be much better off allocating that extra die area to additional L3 cache or a beefier integrated GPU...
(this space intentionally left blank)
just brew it!
Administrator
Gold subscriber
 
 
Posts: 37705
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 11:47 pm

What if they just moved it on-package?
Canon 6D||[24-105/4L IS USM|100/2.8L Macro IS USM|70-300/4-5.6 IS USM|40/2.8 STM|50/1.4 USM|85/1.8 USM|Samyang/Bower 14/2.8 Full-Manual Rectilinear Wide-angle|
Canon EOS-M|11-22/4-5.6 IS STM|22/2 STM|EF-M 18-55/3.5-5.6 IS STM|
For sale!|24/2.8 IS USM
|
Airmantharp
Gerbil Elder
 
Posts: 5012
Joined: Fri Oct 15, 2004 10:41 pm

Re: Intel "Haswell" Speculation thread

Postposted on Tue Jul 31, 2012 11:49 pm

Airmantharp wrote:What if they just moved it on-package?

That's better. But then you lose some of the advantage too (you'll have resistive losses between the regulator chip and the CPU chip).

I remain totally unconvinced that onboard VRM makes sense for anything other than low-power mobile parts.
(this space intentionally left blank)
just brew it!
Administrator
Gold subscriber
 
 
Posts: 37705
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Intel "Haswell" Speculation thread

Postposted on Wed Aug 01, 2012 11:50 am

I found a few images from IDF 2005 that might help clarify what's happening here exactly:

Image
Image
Image

Apparently Intel does intend to integrate the VR on-die, so transitions might be even faster:

Image
Intel Core i7-875K, Asus P7P55D-E Pro, Win 7 Home Premium
MSI GTX 560 Ti OC, Mushkin 2x2GB DDR3-1333, Corsair TX650
Cooler Master Hyper 212+, Logitech Z-2300, ASUS Xonar DX
Samsung Spinpoint F3 1TB, Dell Ultrasharp U2410, Antec P183
DeadOfKnight
Gerbil Elite
Gold subscriber
 
 
Posts: 635
Joined: Tue May 18, 2010 1:20 pm

Re: Intel "Haswell" Speculation thread

Postposted on Wed Aug 01, 2012 1:49 pm

I also found a good thread on AnandTech Forums regarding Haswell's power management.
Intel Core i7-875K, Asus P7P55D-E Pro, Win 7 Home Premium
MSI GTX 560 Ti OC, Mushkin 2x2GB DDR3-1333, Corsair TX650
Cooler Master Hyper 212+, Logitech Z-2300, ASUS Xonar DX
Samsung Spinpoint F3 1TB, Dell Ultrasharp U2410, Antec P183
DeadOfKnight
Gerbil Elite
Gold subscriber
 
 
Posts: 635
Joined: Tue May 18, 2010 1:20 pm

Re: Intel "Haswell" Speculation thread

Postposted on Wed Aug 01, 2012 4:26 pm

That's hawt.
Canon 6D||[24-105/4L IS USM|100/2.8L Macro IS USM|70-300/4-5.6 IS USM|40/2.8 STM|50/1.4 USM|85/1.8 USM|Samyang/Bower 14/2.8 Full-Manual Rectilinear Wide-angle|
Canon EOS-M|11-22/4-5.6 IS STM|22/2 STM|EF-M 18-55/3.5-5.6 IS STM|
For sale!|24/2.8 IS USM
|
Airmantharp
Gerbil Elder
 
Posts: 5012
Joined: Fri Oct 15, 2004 10:41 pm

Re: Intel "Haswell" Speculation thread

Postposted on Wed Aug 01, 2012 5:24 pm

I still wonder what this means for enthusiast motherboards and overclocking.

As far as heat dissipation, would the tighter package make a big difference?

Also, perhaps the integrated VR will enable Intel to create unlockable upgrade features in the future?
Intel Core i7-875K, Asus P7P55D-E Pro, Win 7 Home Premium
MSI GTX 560 Ti OC, Mushkin 2x2GB DDR3-1333, Corsair TX650
Cooler Master Hyper 212+, Logitech Z-2300, ASUS Xonar DX
Samsung Spinpoint F3 1TB, Dell Ultrasharp U2410, Antec P183
DeadOfKnight
Gerbil Elite
Gold subscriber
 
 
Posts: 635
Joined: Tue May 18, 2010 1:20 pm

Re: Intel "Haswell" Speculation thread

Postposted on Mon Aug 06, 2012 4:11 pm

Well this is interesting:

http://pcper.com/news/General-Tech/Ivy- ... cross-eyed

I thought for sure they'd be skipping IB-E.
Intel Core i7-875K, Asus P7P55D-E Pro, Win 7 Home Premium
MSI GTX 560 Ti OC, Mushkin 2x2GB DDR3-1333, Corsair TX650
Cooler Master Hyper 212+, Logitech Z-2300, ASUS Xonar DX
Samsung Spinpoint F3 1TB, Dell Ultrasharp U2410, Antec P183
DeadOfKnight
Gerbil Elite
Gold subscriber
 
 
Posts: 635
Joined: Tue May 18, 2010 1:20 pm

PreviousNext

Return to Processors

Who is online

Users browsing this forum: Bing [Bot] and 1 guest

cron