• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Intel Broadwell Thread

Page 63 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Enigmoid said:
On a preliminary process and stepping and without any other power saving architectural changes taken into account.
Click to expand...

The process is the same 14nm they use today and they used Core-M with all its power savings. But 30% less is very very nice especially at the same performance, you will not find more vs any other process in the past.
 
The 14nm process from mid-2013 is a lot different from the 14nm process today...

1jjlgd.png


Transistor and interconnect hadn't even reached goals in November.
 
AtenRa said:
The process is the same, only think different from a year ago are yields.
Click to expand...
This is probably the case. It's unlikely that Intel lowered or raised its standards in regards to what qualified as being a "good" die -- they were simply got better over time at producing dies that met those standards.
 
III-V said:
This is probably the case. It's unlikely that Intel lowered or raised its standards in regards to what qualified as being a "good" die -- they were simply got better over time at producing dies that met those standards.
Click to expand...

Then what does the yellow box of not meeting transistor and interconnect goals mean according you?

And then there's still the case of architecture:

“So the effect was that our slow learning on yield in fact delayed the entire product development cycle.” --William Holt, Intel, IM’13
 
witeken said:
Then what does the yellow box of not meeting transistor and interconnect goals mean according you?
Click to expand...
It means that their goals were not met by most or even all of the dies they were producing.
And then there's still the case of architecture:

“So the effect was that our slow learning on yield in fact delayed the entire product development cycle.” --William Holt, Intel, IM’13
Click to expand...
That a pretty vague statement... you could draw dozens of hypotheses from that.
 
FYI, the Dell XPS 13 with the 5500u is on display at Microsoft stores (at least the one near me).

The one I saw had the lower-resolution non-touch screen. Very nice looking device and it appeared well built. I didn't have time to play much with it. Next time, I'll check about running some quick benchmarks (e.g., Octane).
 
dahorns said:
FYI, the Dell XPS 13 with the 5500u is on display at Microsoft stores (at least the one near me).

The one I saw had the lower-resolution non-touch screen. Very nice looking device and it appeared well built. I didn't have time to play much with it. Next time, I'll check about running some quick benchmarks (e.g., Octane).
Click to expand...

I couldn't help but order one the other day. Core i5, 8GB RAM, 128GB SSD, no touch. Should be a perfect traveling machine.
 
Honestly I didn't expect any bigger improvements in graphics considering that BW is still using 128bit DDR3 memory and at a measly 1866MHz at most I would like to see DDR4 at 2400/2666MHz I think those speeds should be doable in laptops. BW-U is already extremely expensive I think we could live with +60$ per laptop and 30% more performance. They could also work on their color compression, NV managed to achieve better Bandwith utilization with Maxwell. AMD also improved in that department with Tonga. Radeon 285 in iMacs looks nice.
 
dahorns said:
Also, he has posted Titanfall:

http://youtu.be/bHhzjvirzDE

Dragon Age Orgins:

http://youtu.be/GGFwX_rIJjE

Minecraft:

http://youtu.be/MnXhDTLeqJk

Sims 4:

http://youtu.be/hKytVS_hGgs

It would be nice if he'd tell us which system he is using.

Finally, a comparison between HD 5300 (HP Envy x2 13"/5y10), HD 4400 (SP2/4200 U), and HD 4200 (SP3/4020Y)

http://youtu.be/1ntnHlkgtkQ

Edit: Found 5y10 running Call of Duty Advanced Warfare:

http://youtu.be/wlNXQcWAxbo
Click to expand...

GRID Autosport in Benchmark and Attract Mode now too:
https://www.youtube.com/watch?v=OND8ioDGzI0
https://www.youtube.com/watch?v=fptGmhwjmPk

Performance actually seems a bit CPU limited in some scenarios - planning on investigating a bit and posting up some info on it.
 
Do you have driver 4080 installed? Could you make a AF flower screenshot? Maybe there are some differences to HSW.
 
Lepton87 said:
Honestly I didn't expect any bigger improvements in graphics considering that BW is still using 128bit DDR3 memory and at a measly 1866MHz at most I would like to see DDR4 at 2400/2666MHz I think those speeds should be doable in laptops. BW-U is already extremely expensive I think we could live with +60$ per laptop and 30% more performance. They could also work on their color compression, NV managed to achieve better Bandwith utilization with Maxwell. AMD also improved in that department with Tonga. Radeon 285 in iMacs looks nice.
Click to expand...
There are some cache changes in Gen 8 that will help Broadwell make improvements without using DDR4. Part of why AMD has hit such a bandwidth wall is because they do not make such extensive use of caches like Intel does, but they're expensive, of course.
 
14 nm Broadwell vs 22 nm Haswell mobile @ 15 W TDP:

Am I correct in that on average this will result in ~20% higher iGPU performance, and ~4% better CPU performance, at the same TDP?

Isn't that far lower than expected? I remember some saying that the frequency wall was not hit. I.e. FinFet @ 22 nm was a one time penalty, so that 14 nm and beyond would provide increasing frequencies again. Did that not happen after all?
 
@Fjodor2001 - 20% higher GPU perf at same TDP but TDP isnt actual power drawn. arent broadwell laptops reporting higher battery life compared to haswell?
i read dell xps13 may have 15hr battery life. thats a possible 5hr increase
 
5200u Benches from the XPS 13:

CB 11.5 Multi = 2.85
Google Octane = ~12,000
Ice Storm Unlimited Physics = ~36,000

The Ice Storm physics score matches the 4650u in my SP3.

Compared to the 4200u, there is a boost, although the Octane score is unchanged. In CB it's 15%, in Ice Storm Physics its about 25%.
 
Fjodor2001 said:
14 nm Broadwell vs 22 nm Haswell mobile @ 15 W TDP:

Am I correct in that on average this will result in ~20% higher iGPU performance, and ~4% better CPU performance, at the same TDP?

Isn't that far lower than expected? I remember some saying that the frequency wall was not hit. I.e. FinFet @ 22 nm was a one time penalty, so that 14 nm and beyond would provide increasing frequencies again. Did that not happen after all?
Click to expand...

As the chips are for a 15W TDP spec then clocks will only be raised enough to keep it within that specification. We'll have to wait to see what BDW-K can do.

Power savings seem to be in the overall package such that idle powers are quite impressive and this is what will help extend battery life. Running a HSW or BDW at full load, ie hitting the manufacturers 15W power limit, is going to use 15W on either processor with no savings at all.

Possibly peoples expectations of performance were due to the hype from the comparison between BDW-Y (Core-m) and a crippled HSW made to run with a 4.5W limit. With package power improvement in BDW this would also mean more power would be available to the IA cores than would be available to HSW IA cores.

Cinebench 11.5 on the i7-5500U hits ~3.3, also a ~15% increase over the i7-4500U however the 4500U has since been refreshed to i7-4510U with increase in clocks.

What would be interesting to see between 4500U and 5500U during say Cinebench 11.5 or Linpack would be the difference between package power and IA core power.
 
Last edited:
Fjodor2001 said:
I remember some saying that the frequency wall was not hit. I.e. FinFet @ 22 nm was a one time penalty, so that 14 nm and beyond would provide increasing frequencies again. Did that not happen after all?
Click to expand...
That was in regards to overclocking. Regardless, frequency is higher for Broadwell in most cases. GPU clocks are a bit lower, but that's to be expected, given the thermal constraints and the substantial expansion of the GPU.
 
Last edited:
dahorns said:
5200u Benches from the XPS 13:

CB 11.5 Multi = 2.85
Google Octane = ~12,000
Ice Storm Unlimited Physics = ~36,000

The Ice Storm physics score matches the 4650u in my SP3.

Compared to the 4200u, there is a boost, although the Octane score is unchanged. In CB it's 15%, in Ice Storm Physics its about 25%.
Click to expand...

also, sunspider on ie = 108ms
 
bullzz said:
arent broadwell laptops reporting higher battery life compared to haswell?
i read dell xps13 may have 15hr battery life. thats a possible 5hr increase
Click to expand...

Yes, but that could be due to other things than the CPU, like the display or SSD consuming less power than previous laptop models. Hard to tell without more data.
 
III-V said:
That was in regards to overclocking. Regardless, frequency is higher for Broadwell in most cases. GPU clocks are a bit lower, but that's to be expected, given the thermal constraints and the substantial expansion of the GPU.
Click to expand...

Anyway, I'm not seeing much frequency increase at the same TDP yet for Broadwell 14 nm compared to Haswell 22 nm so far. So far everything points to a frequency wall having been hit after all, so that newer nodes no longer provide much higher frequencies.
 
Fjodor2001 said:
Anyway, I'm not seeing much frequency increase at the same TDP yet for Broadwell 14 nm compared to Haswell 22 nm so far. So far everything points to a frequency wall having been hit after all, so that newer nodes no longer provide much higher frequencies.
Click to expand...

The best 15W U base went from 1.7 Ghz to 2.2 Ghz. The best 28W U base went from 3 Ghz to 3.1 Ghz. Both have lower turbos, which could be just simply because of the problems Intel's had with 14 nm.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top