Core i7-4770K: Haswell's Performance, Previewed

[EzioAs]

@jesot

We can assume that it'll cost the same as the 3770K at launched but apparently, a couple webshop list it even higher http://us.hardware.info/news/34022/intel-core-i7-4770k-haswell-appears-in-dutch-webshops.

I'm still leaning on same price as the 3770K though, or maybe at most $20 higher.

 

[ojas]

[citation][nom]mouse24[/nom]Anyone else annoyed by the naming convention as of late? Dual core i5s randomly with a T moniker... whats that about.[/citation]
It's been like this for a while, but hey, I actually think it makes sense. Even AMD has an A6 which is different from the others, but i don't remember any indication of the change from the part number.

And i think Intel/AMD's naming scheme is still miles better than Snapdragon 200, 400, 600, 800 where all of them have dual-core models and the 600 and 800 have some quad core parts. I won't even pretend to understand their naming scheme.
[citation][nom]mouse24[/nom]Ouch, they should just call it an i3 + HT... Though I suppose those would sell less compared to calling it an i5 to the masses.[/citation]
i3s already have HT. It's AES-NI.
[citation][nom]de5_Roy[/nom]according to intel, desktop core i3 shouldn't doesn't have turbo boost, so cpus with turbo boost get i5 moniker.[/citation]
Don't forget the hardware accelerated AES-NI support.

Honestly it's not that bad.
-K = unlocked multiplier, max TDP for the series (95/77/84w)
-S = 65w for a quad core, max 4 threads.
-U = 17w, 2C/4T, higher turbos on i7s
-Y = 10 to 13w TDP parts
-T = ok this is more confusing this time. Usually it would just be a 35w part, quad-core for an i7 or dual core for an i5. This time -T parts are either 45w, 4C/4T or...if it's a 35w part, then 4C/4T for an i7, or a 2C/4T i5.

So it seems the 45w i5 and i7 are differentiated on the basis of performance, while the 35w parts are differentiated on core count. Caches are of different sizes seem to fixed according to the i5/i7 designation, except the 4570T, of course...They probably should have changed the naming scheme to distinguish 45w parts.

I have a funny intuition that this time, -T series i3s will be 17w or something, -S parts will be 35w with the standard i3s being 45w. Pentiums might be 35w/17w only, Celerons should be capped at 17w.

-U parts might continue to sit at 17w, though we might see them at 13w as well...-Y series SKUs will hit an 8w TDP (with a 4w SDP, probably), mark my words.

side note: I find it strange that a i7-3632QM could turbo from 2.2GHz to 3GHz on all four cores within a 35w TDP but the i7-4765T will only go from 2 to 2.6...

 

[blazorthon]

[citation][nom]cangelini[/nom]Thanks--and yeah, VT-d is being excluded from these K-series parts, too. Funny thing is that it'll be enabled on the -4770, but not the -4770K.[/citation]
Typical corporate greed. Governments do lots, but won't step in to protect consumers from such practices. What a waste.

The K series CPUs don't have it even in the current series from Ivy and Sandy. How is this corporate greed? They decided that a virtualization feature was not necessary in an overclocking CPU. It is present in most cheaper models.

 

[ojas]

On TSX, -K models, along with the i5-4430{series ID} don't get TSX, everything else does. Why? I don't know.

Within a series i don't see too much ambiguity, for example, for unmarked processors, the i5-4670>4570?4330.
Within the -S series i5s, same logic applies, 4670S>4570S>4330S.
Within the -T series, i7-4770T>4765T and i5-4670T>4570T.
Apart from this, there's a -K series i7, a plain i7 and an -S series i7.

Looking at it another way, there are only four main processor model numbers:
i7-4770,-K,-S,-T
i5-4670,-K,-S,-T
i5-4570,-S,-T
i5-4330,-S
and the fifth is the i7-4765T

I'm not sure this is much to complain about (except bunching 45w and 35w both under -T). It's the same with AMD.

 

[de5_Roy]

Don't forget the hardware accelerated AES-NI support.

core i3 doesn't have aes ni, vt-d, pcie 3.0... there might be other differentiators i'm too lazy to look for...
http://ark.intel.com/products/65693/Intel-Core-i3-3220-Processor-3M-Cache-3_30-GHz?wapkw=core+i3+3220
http://ark.intel.com/products/65703/Intel-Core-i5-3470T-Processor-3M-Cache-up-to-3_60-GHz?wapkw=core+i5+3470t
3470t has the distinction of only desktop dual core cpu possessing pcie 3.0 support. strangely, core i3 3220's specs say it supports pcie 2.0 x8+x4+x4 modes. even core i5 2500k doesn't have that!!

 

[ojas]

[citation][nom]de5_Roy[/nom]core i3 doesn't have aes ni, vt-d, pcie 3.0... there might be other differentiators i'm too lazy to look for... http://ark.intel.com/products/6569 [...] re+i3+3220http://ark.intel.com/products/6570 [...] e+i5+3470t3470t has the distinction of only desktop dual core cpu possessing pcie 3.0 support. strangely, core i3 3220's specs say it supports pcie 2.0 x8+x4+x4 modes. even core i5 2500k doesn't have that!![/citation]
What!? lol i never noticed, i thought it was only with the Pentiums (without PCIe 3.0)!

I guess it'll be the same with the 4570T, only dual core to support AES-NI, VT-d, and PCIe 3.0, faster iGPU, vPro, Trusted Execution tech and Turbo Boost...

BTW: the 3240T is actually more like the 3470T (same TDP, base clock).
http://ark.intel.com/products/66168/Intel-Core-i3-3240T-Processor-3M-Cache-2_90-GHz

 

[ojas]

[citation][nom]de5_Roy[/nom]strangely, core i3 3220's specs say it supports pcie 2.0 x8+x4+x4 modes. even core i5 2500k doesn't have that!![/citation]
[can't edit anymore it seems, so i'll have to post again.]

See? Intel being competitive :lol:

 

[de5_Roy]

when one ventures into the murky swamps of intel sku-ing, one finds weird, strange and amusing things doing the sloppy swish....

[citation][nom]de5_Roy[/nom]strangely, core i3 3220's specs say it supports pcie 2.0 x8+x4+x4 modes. even core i5 2500k doesn't have that!![/citation]
[can't edit anymore it seems, so i'll have to post again.]
See? Intel being competitive :lol:

exactly. this is why, if one wishes to tri-fire or tri-sli 3x gtx titan or 7970s on teh cheap, acquire a core i3 3220 cpu along with a gigabyte ga z77x-ud5h or that msi motherboard i do not care to remember.
.....
i kid.
on a side note, i am pretty sure a regular person reading this comment is going 'wtf!'

my apologies.

 

[jonjonjon]

so all Broadwell cpu's will be BGA-only even desktop cpu's? i thought that was just a rumor. has intel confirmed that?

 

[ceh4702]

Who uses an i-7?

 

[blazorthon]

Who uses an i-7?

Most people who can afford them or can get them paid for by their employers I'd guess. People who actually make good use out of them would mostly be people who need moderately high end performance for some heavy professional software or such workloads and to an extent, also otherwise high-performance users such as many high-end gamers.

 

[slomo4sho]

7-13% faster than the 3770K and consumes 9% more power? This is definitely a disappointment.

 

[blazorthon]

7-13% faster than the 3770K and consumes 9% more power? This is definitely a disappointment.

They do not consume more power. They use less power. The VRM is now integrated into the CPU instead of on the motherboard, so their TDPs are higher (which I noticed you based your power consumption number on, so I'll state yet again that TDP is NOT power consumption and shouldn't be treated as if it is) and sure, the chips themselves may use more power, but system power consumption went down because Haswell's motherboards use much less power.

 

[slomo4sho]

[citation][nom]blazorthon[/nom]They do not consume more power. They use less power. The VRM is now integrated into the CPU instead of on the motherboard, so their TDPs are higher (which I noticed you based your power consumption number on, so I'll state yet again that TDP is NOT power consumption and shouldn't be treated as if it is) and sure, the chips themselves may use more power, but system power consumption went down because Haswell's motherboards use much less power.[/citation]

TDP is usually pretty accurate in providing load power consumption. I'll wait for the power consumption benchmarks.

 

[blazorthon]

[citation][nom]blazorthon[/nom]They do not consume more power. They use less power. The VRM is now integrated into the CPU instead of on the motherboard, so their TDPs are higher (which I noticed you based your power consumption number on, so I'll state yet again that TDP is NOT power consumption and shouldn't be treated as if it is) and sure, the chips themselves may use more power, but system power consumption went down because Haswell's motherboards use much less power.[/citation]

TDP is usually pretty accurate in providing load power consumption. I'll wait for the power consumption benchmarks.

TDP is almost always highly inaccurate. For example even comparing i3s and i5s of Ivy Bridge shows i5s being farther from their TDP in most of the same workloads and comparing across generations to Sandy shows significant differences in even that much. Bringing other generations and AMD into it makes it even worse. TDP is not even close to accurate at predicting load power consumption since it's not even directly related to power consumption, but only loosely!

Heck, even going by different models (and even units of each model due to not all chips being equal even of the same model when it comes to power consumption) of the same family and generation will get somewhat different power consumption at the same TDP. TDP is not accurate at all for power consumption prediction.

 

[Novuake]

7-13% faster than the 3770K and consumes 9% more power? This is definitely a disappointment.

They do not consume more power. They use less power. The VRM is now integrated into the CPU instead of on the motherboard, so their TDPs are higher (which I noticed you based your power consumption number on, so I'll state yet again that TDP is NOT power consumption and shouldn't be treated as if it is) and sure, the chips themselves may use more power, but system power consumption went down because Haswell's motherboards use much less power.

LOL people will understand this once the initial REVIEWS are out. I was also like, "wait what??" when I saw higher TDPs, but that is obviously chip power itself. Everything is moving to the CPU now, seems like the motherboard is becoming an extended connector for everything instead of a controlling device.

 

[ojas]

[citation][nom]jonjonjon[/nom]so all Broadwell cpu's will be BGA-only even desktop cpu's? i thought that was just a rumor. has intel confirmed that?[/citation]
Yup they pretty much have.

 

[ojas]

[citation][nom]blazorthon[/nom]They do not consume more power. They use less power. The VRM is now integrated into the CPU instead of on the motherboard, so their TDPs are higher (which I noticed you based your power consumption number on, so I'll state yet again that TDP is NOT power consumption and shouldn't be treated as if it is) and sure, the chips themselves may use more power, but system power consumption went down because Haswell's motherboards use much less power.[/citation]
To be honest, i don't think the VRM has anything to do with it. All models have:

1) An integrated VRM.
2) Turbo frequencies in the same range (especially the 4770/K and S models).
3) Identical graphics cores with almost identical clocks.

Yet the TDPs are so different. Something else is going on here. No, i'm not suggesting they'll consume more power. I think it's to do with overclocking and/or Turbo Boost, and a different TIM b/w the IHS and the silicon.

Think about it:
the i7-4770T and 4765T are identical except the clock speeds. So 500 MHz seems to warrant a TDP increment of 10w.
By that logic, the difference b/w the 4770S and 4770/K should also be 10w or less, not 19w.

BUT! Look at the 4770S and 4330. The 4770S has more cache, a higher clock speed, higher iGPU clock, same number of cores, yet still has a 19w lower TDP! Even the 4570 is too close to the 4770S to warrant a 19w increase in TDP.

My conclusion:
They could have fit the 4330 and 4570 within 65w if they wanted to, and the 4670/K and 4770/K could have easily fit within 75w (or at the most 77w like Ivy).

They've chosen not to. And i haven't been able to figure out why this is the case ever since the first Haswell slide leaked.

 

[Intel_shills_uncovered]

Finally, AMD will kick Intel's ass in 2014, Why? Well Steamroller includes the New HSA technology created by AMD that will offload CPU work on the GPU when needed, reducing power and greatly increasing performance, not only that but Steamroller will have a 30% performance increase from Piledriver.

 

[ubercake]

Finally, AMD will kick Intel's ass in 2014, Why? Well Steamroller includes the New HSA technology created by AMD that will offload CPU work on the GPU when needed, reducing power and greatly increasing performance, not only that but Steamroller will have a 30% performance increase from Piledriver.

OMG ... How many times have we heard this???

For the sake of competition, I really hope you're right.

 

[morpheas768]

Intel does this every time and its so annoying:
Every year, recycle the same platforms, by slightly upgrading the CPU architecture, and slightly upgrading the new chipsets for mobos, which they could easily have the new additions to the chipsets (6 SATA 6gbps ports instead of 4, really?) a few years ago, but no, they need something to be able to claim an improvement over the previous gen chipset (7,8 series), and as for the CPU's the improvements are minor, and in fact so minor that there's practically no difference in gaming for enthusiasts.

The thing is, that Intel has gained an unquestionable advantage over AMD in the last years, so all they are doing now is merely maintain that advantage, and sell as much as possible by recycling the same products with minor improvements each year.
What a "beautiful" scheme to make more money. Everything in their policy is designed for that. All Intel wants is to make you buy a new mobo and CPU every year, they dont care about anything else.

Of course I do have an Intel platform, and will continue to buy as long as they offer more performance in games than AMD, but still, I do not like to be played like that as a consumer, even if I dont waste my money on that Haswell nonsense.

Flame me all you want, but the fact of the matter is that Intel's policy has been very tedious and greedy the last few years. Just count all the different sockets they have introduced, and you'll see my point (1156, 1155, 2011, 1150, and wait for 1151 etc etc).

 

[InvalidError]

[citation][nom]jonjonjon[/nom]so all Broadwell cpu's will be BGA-only even desktop cpu's? i thought that was just a rumor. has intel confirmed that?[/citation]
All the mainstream models, which I presume means pretty much everything except K/X CPUs and maybe the top i5/i7 non-K models. Most non-enthusiast people never upgrade a CPU so losing the socket should enable cheaper and more compact form factors.

If you look at Haswell, models with GT3+eDRAM IGP which should be a strong mainstream option for non-gamers will only be available in BGA too. Depending on what sort of model lineup they have planned, they may be starting to push BGA CPUs in the mainstream in just about everything except tower PCs long before Broadwell - that would be consistent with Intel's intention to migrate from standard motherboards to OEM/proprietary AiO/tablet/embedded/etc. form factors.

 

[devBunny]

[citation][nom]morpheas768[/nom]Intel does this every time and its so annoying: Every year, recycle the same platforms, by slightly upgrading the CPU architecture, and slightly upgrading the new chipsets for mobos, which they could easily have the new additions to the chipsets (6 SATA 6gbps ports instead of 4, really?) a few years ago, but no, they need something to be able to claim an improvement over the previous gen chipset (7,8 series), and as for the CPU's the improvements are minor, and in fact so minor that there's practically no difference in gaming for enthusiasts.[/citation]

Smart enthusiast would therefore refrain from buying.

[citation]All Intel wants is to make you buy a new mobo and CPU every year, they dont care about anything else.[/citation]

They must be relying on people being dumb then because smart people don't upgrade when it costs good money for poor improvement. If I were Intel I'd want as much improvement as possible every single time so that I could get the world to throw last year's snail system on the trash heap because this year's is so much better.

[citation]Of course I do have an Intel platform, and will continue to buy as long as they offer more performance in games than AMD, but still, I do not like to be played like that as a consumer, [/citation]

So you're one of the dumb ones that Intel is apparently geared towards conning? ;-)

[citation]even if I dont waste my money on that Haswell nonsense.[/citation]

Nope, you're one of the smart ones. You'll sit it out until there's something worth paying for.

[citation]Flame me all you want, but the fact of the matter is that Intel's policy has been very tedious and greedy the last few years. Just count all the different sockets they have introduced, and you'll see my point (1156, 1155, 2011, 1150, and wait for 1151 etc etc).[/citation]

Intel's "greedy" policy seems rather counter-intuitive. It makes smart people, yourself included, *not* buy their latest product!

[citation]Flame me all you want, [/citation]

You don't need flames from others, you need more inside yourself - a passion to read and pay attention to what's being said, and has been said for months. Intel is focused on power consumption and graphics performance, with a long term eye on the market that matters more and more - mobile computing. More oomph in the CPU will do no good if the chips are unsuitable for that market. (Although, as has also been mentioned time and time again, technological limitations are now preventing leaps and bounds in CPU performance anyway)

 

[Novuake]

Flame me all you want, but the fact of the matter is that Intel's policy has been very tedious and greedy the last few years. Just count all the different sockets they have introduced, and you'll see my point (1156, 1155, 2011, 1150, and wait for 1151 etc etc).

For the record, many of those socket changes have a legitimate reason. While it may not be NECESSARY persè, there are still reasons. I could tediously explain some now, but I would rather not.

 

[sanilmahambre]

Can there be anything like UNIVERSAL PROCESSOR SOCKET (UPS)?

Again change of sockets! what am i gonna do with my expensive mobo (OLD)