Some heat problem for Broadwell-D ... ?
http://www.chw.net/2015/01/rumor-broadwell-k-tendra-un-alto-consumo-al-ser-overclockeado/
http://www.chw.net/2015/01/rumor-broadwell-k-tendra-un-alto-consumo-al-ser-overclockeado/
Some heat problem for Broadwell-D ... ?
http://www.chw.net/2015/01/rumor-broadwell-k-tendra-un-alto-consumo-al-ser-overclockeado/
Core fifth generation "Broadwell-K" might not be as good overclocker as "Haswell-K".
Launched in June last year, Intel Core microprocessors K Series V2 fourth generation "Haswell-K" (also known as "Devil's Canyon") are the greatest exponents of the "Shark Bay" (LGA 1150) platform; and is expected to be replaced by fifth-generation Core microprocessors K Series "Broadwell-K".
Broadwell-K will only be compatible with socket LGA 1150 motherboards based on the Z97 and H97 "Wildcat Point" chipsets (Intel 8 Series chipsets "Lynx Point" will not support Broadwell-K) and offers the latest in architecture CPU / GPU and lithographic processes (14nm FinFET); besides having unlocked multiplier, making it a worthy successor Haswell-K.
Preliminary data suggest that the first copies of Broadwell-K for Intel, suffering from a very high level of consumption at high operating frequencies (overclock), an issue that seems to be caused by higher voltage requirements node to 14nm manufacturing Tri- Gate of Intel.
Similar rumors circulated before the launch of the third generation microprocessors Core "Ivy Bridge" (which later were confirmed by Intel).
According to the report, increasing the operating frequency above the factory (overclock), clock for clock, drinking Broadwell-K is proportionately greater than Haswell-K (unfortunately did not provide exact figures), factor which could affect your ability to overclock.
Finally, since there are still a few months before the release of fifth-generation Core "Broadwell-K", Intel still has time to refine and optimize your new micro-processor, before mass produce.
Written by Gian Maria Forni - Wednesday, January 21, 2015
According to reports from our own internal source, it seems that Intel is encountering new problems with the node to 14nm 3D-Gate. Expected by mid-2014, Broadwell was then postponded to the end of that year only in the version with unlocked multiplier (Broadwell-K) and especially this for CPU were marketed Z97 and H97 chipsets. To date, Broadwell-K, has not yet seen anything.
Google translated to English:
They're attributing the delay to this, which is rather silly. It's been delayed for a very long time, and that was because of yield issues pushing back the entire product stack.
They're different dies, though. It could certainly be true, as we're seeing in reduced turbo speeds in chips already on the market, but you can't bin a what could have been a Broadwell-K to be a Core M.Parametric yields. Don't conflate them with functional yields. This is a binning issue, always has been. That's why they released only low-clock SKUs on 14nm thus far. High clock (high bin) stuff is way to hot and consumes too much power. That is bad yield (parametric yield) at the bins you want to produce and sell. Intel never lies, but don't convince yourself they are doing everything they can to make sure you fully understand what it is they are telling you
They're different dies, though. It could certainly be true, as we're seeing in reduced turbo speeds in chips already on the market, but you can't bin a what could have been a Broadwell-K to be a Core M.
There is a good reason we ended up with a "haswell refresh" rather than broadwell-K last summer.
And no, Intel does not have the option of "refine and optimize" any farther then they already have over the past year.
It's too early to suggest we are looking at a 90nm repeat here, but in 2002 Intel was mighty proud of their upcoming 0.09um node which didn't turn out to be all that exciting when 90nm Prescott arrived 2 years later.
14nm is going to be awesome. Intel has done everything in its power to tell us as much. But don't be surprised if the operating temps are a skosh higher than 22nm and the clockspeeds pretty much the same as well (whenever they get around to launching the Haswell desktop replacements). Shades of 90nm all over again, just don't blame the messenger.
Parametric yields. Don't conflate them with functional yields. This is a binning issue, always has been. That's why they released only low-clock SKUs on 14nm thus far. High clock (high bin) stuff is way to hot and consumes too much power. That is bad yield (parametric yield) at the bins you want to produce and sell. Intel never lies, but don't convince yourself they are doing everything they can to make sure you fully understand what it is they are telling you
50% over what?50% gain in igp is pretty decent.
The slightly higher base clocks in the Core i7-5500U (compared to the Core i7-4500U) are probably the reason for the Haswell-based unit appearing more power efficient than the Broadwell counterpart - but, make no mistake here - the Broadwell unit wins the performance per watt test quite easily.
Where did I state otherwise?@coercitiv - that 14W is a difference of max load consumption between the devices.
I believe the claim of 50% gain in igp was related to gaming performance.if you look at power consumption during each decoding stream, broadwell falls in line with haswell nuc
Which is exactly why I asked for a reference point for the 50% gain in igp, since the needed data for HD 4400 isn't there.also hd5500 replaces hd4400. hd5000 should be ideally compared to hd6000
Once we get to the power charts the Asrock Vision is no more displayed, how could people make relevant comparisons if such numbers are missing, how one does compute perf/watt.?.
Or does the reviewer prefer people to use some references rather than others..?.
Besides the article state 15W CPU, yet real TDP is 28W in this review when deducting losses, indeed Intel mobile CPUs have an habit to consume much more than the official specs, look at all thoses laptops at Notebookcheck, with 11.5W Haswell pumping as much as 18-19W at the soc level, as much as the 15W parts that also reach this number...
Notebook check doesn't measure power from the SOC.
The CPU is the successor of the Core i5-4300U. While the Core i5-5300U uses a higher base clock (1.9 GHz vs. 2.3 GHz), the Turbo clocks are almost the same (Core i5-4300U: 2.6 GHz/2.9 GHz; Core i5-5300U: 2.7 GHz/2.9 GHz). Improvements in the Broadwell architecture ensure that the Core i5-5300 achieves slightly better results than the Core i5-4300U processor. The benchmarks run by us show a performance gain of about 5 to 10 percent.
PCMark 7
Core i5 4300U: 3693 points
Core i5 5300U: 4638 points
Intel's HD Graphics 5500 GPU is responsible for graphics output. The core supports DirectX 11.2 and clocks at up to 900 MHz. It is the successor of the HD Graphics 4400. The HD Graphics 5500 has a slightly lower clock rate (900 MHz vs. 1100 MHz), but brings additional execution units and a bigger L1 cache. In the 3D Mark benchmarks the it performs slightly better than the HD Graphics 4400. The graphics performance of the Latitude can be further improved. If the computer is equipped with a second RAM module, RAM works in dual channel mode. As a result, the GPU is used more efficiently and achieves better results. This way we were able to increase the 3D Mark 2011 result by 16 percent.
3DMark 2013
Core i5 5300U: 48863 points
Core i5 4310U: 36471 points
Tomb Raider
1024x768 Low
Core i5 5300U (dual-channel): 53.5 FPS
Core i5 4310U: 42.2 FPS
1366x768 Normal
Core i5 5300U (dual-channel): 26.9 FPS
Core i5 4310U: 20.6 FPS
1366x768 High
Core i5 5300U (dual-channel): 16.5 FPS
Core i5 4310U: 12 FPS
The Latitude completed the stress test (Prime95 and Furmark running simultaneously for at least an hour) in the same way in both AC power mode and battery mode. Processor and graphics core ran at maximum speed (CPU: 2.7 GHz; GPU: 900 MHz). The notebook did not grow particularly warm. We measured a temperature of over 40 °C (104 °F) in two spot during the stress test.
...Overall, the Latitude delivers good battery runtimes and performs slightly better than its Haswell sister model despite better performance and same battery capacity.
Are they talking about performance per clock?The benchmarks run by us show a performance gain of about 5 to 10 percent.
Are they talking about performance per clock?
No, overall performance.Are they talking about performance per clock?
I agree with III-V, this is overall performance, which likely means the higher base clock increase has a non-negligible impact.Mostly, Turbo clocks are almost equal.
The really sad part is the Core i5 4310U in the Dell E5450 review was single channel as well.Broadwell-U's HD5500 is BW limited with single-channel memory.