Haswell Refresh to launch in May, Broadwell in Q3 - how does that make sense?

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Dec 9, 1999
10,038
0
86
#26
Both chipsets support both. Its only a motherboard implementation.

You can already buy Xeons with 4Ghz and above turbo.
How sure are you of the bolded? Are there any Z87 boards out that have announced support for Broadwell?
 

sm625

Diamond Member
May 6, 2011
8,176
1
106
#27
Does it really matter, since intel @14nm is only going to be about 10% faster than intel @22nm? (Sans the gpu, which is useless anyway.)

Meanwhile, intel's mobile competition, qualcomm, apple, etc, are each looking to increase their per core performance by 50% or more in 2014. Now that is progress. Intel can choke on their big fat margins.
 

Yuriman

Diamond Member
Jun 25, 2004
5,529
2
106
#28
Does it really matter, since intel @14nm is only going to be about 10% faster than intel @22nm? (Sans the gpu, which is useless anyway.)

Meanwhile, intel's mobile competition, qualcomm, apple, etc, are each looking to increase their per core performance by 50% or more in 2014. Now that is progress. Intel can choke on their big fat margins.
10% faster in the same power envelope?

Intel is racing to scale their big cores down for mobile, while the ARM competitors are ramping up speeds but are increasing power usage.
 

witeken

Diamond Member
Dec 25, 2013
3,866
3
106
#29
Does it really matter, since intel @14nm is only going to be about 10% faster than intel @22nm? (Sans the gpu, which is useless anyway.)
That isn't very surprising since it's just a die shrink.
 

Torn Mind

Diamond Member
Nov 25, 2012
3,244
0
91
#30
Does it really matter, since intel @14nm is only going to be about 10% faster than intel @22nm? (Sans the gpu, which is useless anyway.)

Meanwhile, intel's mobile competition, qualcomm, apple, etc, are each looking to increase their per core performance by 50% or more in 2014. Now that is progress. Intel can choke on their big fat margins.
I wonder what is the battery life on a 3GHz ARM chip is like though. Would be impressive if the power draw remains the same or lower as the current top-of-the-line chip.

With that said, Intel's die shrink should either improve battery life or allow them to clock the Cherry Trail Atoms higher than their Bay Trail predecessors.
 

ElFenix

Elite Member
Super Moderator
Mar 20, 2000
97,903
172
126
#31
I wonder what is the battery life on a 3GHz ARM chip is like though. Would be impressive if the power draw remains the same or lower as the current top-of-the-line chip.
power consumption generally increases in proportion to operating frequency.
 

Torn Mind

Diamond Member
Nov 25, 2012
3,244
0
91
#33
power consumption generally increases in proportion to operating frequency.
I am aware of that relationship; I think it is related to Joule's law. The die shrink has the opposite effect however, and which effect is greater is the big question. I tend to think the clock increase will have the bigger effect, but maybe TSMC will hit a grand slam.
 

Torn Mind

Diamond Member
Nov 25, 2012
3,244
0
91
#34

jpiniero

Diamond Member
Oct 1, 2010
5,986
123
126
#35
Broadwell-K is on the roadmap. Rumors of Iris Pro graphics being included on it. Maybe it is on both, maybe just the i7. Only Intel knows right now.
I think the i5 will have it, if only to justify the extra $50-70 I am expecting Intel to want for it given it's not any faster than Haswell Refresh.
 
Oct 14, 2003
5,751
73
126
#36
It might only be two dies - DC + GT3 for the Ultrabooks, and QC + GT3e for the high end laptops, Broadwell-K and Xeons (which the latter 2 will be 50-70 more than the Haswell Refresh GT2 version). Could be a DC + GT2 for Ultrabooks as well, but only the less than 10W models.
Haswell has 8 seperate dies(of course not including server). It's likely Broadwell has at least that much.

We already know there's:
-Dual + GT2
-Dual + GT3
-Quad + GT2
-Quad + GT3
 

jpiniero

Diamond Member
Oct 1, 2010
5,986
123
126
#37
It really makes sense (2 dies only) if you think about it. Those two represent the high margin parts Intel craves while letting "new" Haswell Refresh fill in the rest of the lineup. They probably aren't comfortable putting <$275 Core on 14 nm yet given how expensive the wafers must be and how lousy the yield has been.
 

tolis626

Senior member
Aug 25, 2013
399
0
76
#38
So with all that said,does it make any sense to build a Haswell high-ish end rig now?I mean,if rumors are true,I can't be excited about Z97.Sata Express is cool,but I'm quite sure the SSDs that support it and take advantage of it will be prohibitively expensive initially.I suppose I'll be okay with a 4770k and a Asus Maximus VI Formula,but it's too expensive of a setup to just want to throw it away for something newer in a few months time.
 

Homeles

Platinum Member
Dec 9, 2011
2,585
0
0
#39
So with all that said,does it make any sense to build a Haswell high-ish end rig now?I mean,if rumors are true,I can't be excited about Z97.Sata Express is cool,but I'm quite sure the SSDs that support it and take advantage of it will be prohibitively expensive initially.I suppose I'll be okay with a 4770k and a Asus Maximus VI Formula,but it's too expensive of a setup to just want to throw it away for something newer in a few months time.
I could have sworn that SATA Express was dead. It doesn't matter anyway -- this year is the year of the affordable PCI-E SSD.
 

hhhd1

Senior member
Apr 8, 2012
667
0
71
#40
isnt't Haswell refresh = broadwell ?

just like ivy bridge is sandy bridge with smaller nm
 

Denithor

Diamond Member
Apr 11, 2004
6,302
0
81
#42
I could have sworn that SATA Express was dead. It doesn't matter anyway -- this year is the year of the affordable PCI-E SSD.
Can you share more details? I haven't been watching component markets as much recently, curious about this. Thanks!
 
Apr 22, 2012
20,395
0
106
#43
SATA Express is removed from the 9 series chipsets. And thank god for that. Worst connector in history perhaps.
 

Unoid

Senior member
Dec 20, 2012
462
0
41
#44
SATA Express is removed from the 9 series chipsets. And thank god for that. Worst connector in history perhaps.
Long live being stuck at SATA-3 for another 2 years then!
 

witeken

Diamond Member
Dec 25, 2013
3,866
3
106
#45
It really makes sense (2 dies only) if you think about it. Those two represent the high margin parts Intel craves while letting "new" Haswell Refresh fill in the rest of the lineup. They probably aren't comfortable putting <$275 Core on 14 nm yet given how expensive the wafers must be and how lousy the yield has been.
The cheaper, smaller transistors more than offset the increased wafer cost.
 

Lepton87

Platinum Member
Jul 28, 2009
2,546
0
81
#46
Everybody forgets that there already was an 4.4GHz stock Xeon Westmere CPU. Dual core and OEM only, very rare.
 

Abwx

Diamond Member
Apr 2, 2011
8,772
100
126
#48
power consumption generally increases in proportion to operating frequency.
Only if supply voltage is kept constant otherwise power will increase as the product of the frequency delta by the square of the supply voltage delta , increasing frequency by 30% and voltage by 20% will result to power being
increased by 1.3 x 1.2 x 1.2 = 1.872 ratio , that is 87.2%.
 

Lepton87

Platinum Member
Jul 28, 2009
2,546
0
81
#49
Only if supply voltage is kept constant otherwise power will increase as the product of the frequency delta by the square of the supply voltage delta , increasing frequency by 30% and voltage by 20% will result to power being
increased by 1.3 x 1.2 x 1.2 = 1.872 ratio , that is 87.2%.
Only valid if you kept temperature constant, otherwise static leakage increases so does the power consumption.
 

lagokc

Senior member
Mar 27, 2013
808
0
41
#50
Just use PCIe SSDs. Its the same thing. Just without this:
There is nothing wrong with that connector. It's certainly a lot better than PATA + MOLEX. PCIe SSDs aren't really an option for those of us off in ITX land.
 


ASK THE COMMUNITY