Intel Broadwell Thread

[zir_blazer]

It was a good chip for laptops. It's a great chip for AIO's.

In what Laptop did you ever saw a Broadwell SoC? It was intended for microcloud Servers and the like. But yes, the 25W QC version could happily be repurposed for Laptops.

CPU World also says that Broadwell SoC new Stepping brings a feature called "KR functionality" but they don't know what the hell it is.

 

[Sweepr]

Broadwell-EP not only alive, but coming in Q4-2015 according to an Intel rep at IDF.

At IDF15, an Intel spokesperson reassured the press that their Broadwell-EP lineup of Xeon E5-2600 V4 series processors is heading for launch in Q4 2015. The latest Broadwell-EP line of processors will be aimed towards the enterprise and workstation market that are configured in 1S and 2S platforms while the Xeon E5-4600 V4 (4S Platform) and Broadwell-EX/EN based Xeon E7-8800/4800 V4 (8S Platform) will possibly launch next year.

http://wccftech.com/intel-assures-b...-processors-q4-15-feature-22-cores-44-threads

 

[LTC8K6]

In what Laptop did you ever saw a Broadwell SoC? It was intended for microcloud Servers and the like. But yes, the 25W QC version could happily be repurposed for Laptops.

CPU World also says that Broadwell SoC new Stepping brings a feature called "KR functionality" but they don't know what the hell it is.

Key Recovery, as in encryption key recovery?

 

[Abwx]

Some review of a NUC so people can compute perf/Watt easily...

http://www.notebookcheck.com/Test-Intel-NUC5i3RYK-Mini-PC-Broadwell-Core-i3-5010U.147828.0.html

So a CPU that hoover between 18 and 24W and a CB R15 score of 220, that s about 2.32 in Cinebench 11.5...

If the chips had no boost and was strictly restricted to 15W from start to finish it would score 198 and 2.06 in CB R15 and 11.5 respectively.

 

[piasabird]

What about Desktop IRIS?

What about Xeons with Integrate IRIS 6300 graphics.

http://www.intel.com/newsroom/kits/...tex2015-5th_Gen_Intel_Core-Xeon_FactSheet.pdf

i5-5575R 4/4 2.8
Intel® Iris™ Pro
graphics 6200
$244 4M 65W

 

[Abwx]

What about Desktop IRIS?

What about Xeons with Integrate IRIS 6300 graphics.

http://www.intel.com/newsroom/kits/...tex2015-5th_Gen_Intel_Core-Xeon_FactSheet.pdf

i5-5575R 4/4 2.8
Intel® Iris™ Pro
graphics 6200
$244 4M 65W

Intel® Xeon® workstation platform (Intel S1200RP Board) with one Intel Xeon
processor E3
-
1285 v4 (quadcore, 3.5GHz, 6M
cache) BIOS S1200RP.86B.03.01.0002.041520151123, Intel HT Technology best configuration, 32GB memory (4x8GB DDR3

-1600 ECC UDIMM), Intel Iris Pro Graphics P6300 with driver 10.18.10.3980, Intel SSDSC2BB300G4, Microsoft Windows 8.1*.
3DMark Score 2881,
Source: Intel internal testing as of June 2015

Xeon processor E3 1285 v4 (quadcore, 3.5GHz, 6M cache) 95W

 

[crashtech]

I'm curious to know if 3DMark 11 is representative of the kinds of jobs an E3 Xeon will be doing, and whether the TDP figures given represent actual power consumption in this scenario. I happen to think, in the absence of actual proof, that the answer is no to both questions.

 

[Abwx]

I'm curious to know if 3DMark 11 is representative of the kinds of jobs an E3 Xeon will be doing, and whether the TDP figures given represent actual power consumption in this scenario. I happen to think, in the absence of actual proof, that the answer is no to both questions.

Intel estimated that it was relevant otherwise they wouldnt have published this number, so you re saying that you know better than them, that is, more royalist than the king himself...

 

[crashtech]

Where does Intel say that the E3 1285 v4 will use 95W in 3D Mark?

Nowhere, that's where. Prove otherwise or retract your libel.

 

[Sweepr]

$276 65W Core i5 5675C scores 3113 @ 3DMark 11.

Bristol Ridge (2016 desktop APU) is expected to be 15% faster than Carrizo at 3DMark 11 (up to 95W TDP). In other words it will barely match 1 year old (by then) Broadwell-K. Suddenly the comparison doesn't look so favourable to AMD does it?

Meanwhile Skylake GT4e is expected to deliver 50% better performance than Broadwell GT3e @ 3DMark 11, you do the math and come up with your own conclusions of how far behind AMD will be.

 

[Abwx]

Where does Intel say that the E3 1285 v4 will use 95W in 3D Mark?

Nowhere, that's where. Prove otherwise or retract your libel.

Libel..?..

You know how much an Iris pro comsume before doing such accusation..?.

50W for the GPU alone and for 10% better 3DMark score than said Carrizo..

http://www.hardware.fr/articles/940-23/hd-graphics-530-pratique-h-265-consommation.html

That s 100% more than the Carrizo GPU wich is actualy limited to barely 24W when the SoC is set to 35W :

http://www.notebookcheck.net/fileadmin/_processed_/csm_2015_ISSCC_AMD_APU-page-007_d7a0f7ba5b.jpg

 

[Abwx]

$276 65W Core i5 5675C scores 3113 @ 3DMark 11.

Yes, as said above with twice the power used by Carrizo to score 2750..

Bristol Ridge (2016 desktop APU) is expected to be 15% faster than Carrizo at 3DMark 11 (up to 95W TDP). In other words it will barely match 1 year old (by then) Broadwell-K. Suddenly the comparison doesn't look so favourable to AMD does it?

Meanwhile Skylake GT4e is expected to deliver 50% better performance than Broadwell GT3e @ 3DMark 11, you do the math and come up with your own conclusions of how far behind AMD will be.

GT4e will perform better by the virtue of more power drain than GT3e, there s no miracle possible on this area, look at the numbers above for Iris Pro 6200...

Actualy AMD dont need more than those 15% since they will improve Carrizo s perf/watt such that Bristol Ridge GPU will stay within those 24W despite 15% higher perfs, Intel will need 50W at least to get the same scores, and by the time it s released there will be fast DDR4 that could provide an additional boost..

 

[Sweepr]

Yes, as said above with twice the power used by Carrizo to score 2750..

Wake me up when Carrizo delivers this score inside an actual OEM notebook and not AMD's reference/test platform.

GT4e will perform better by the virtue of more power drain than GT3e, there s no miracle possible on this area, look at the numbers above for Iris Pro 6200...

GT3e will come down to 15-28W TDPs and GT4e takes its spot at higher TDPs. This might come as a surprise for you but not only AMD can improve perf/watt by refining existing manufacturing processes and architectures (Gen 9 vs Gen 8).

Actualy AMD dont need more than those 15% since they will improve Carrizo s perf/watt such that Bristol Ridge GPU will stay within those 24W despite 15% higher perfs, Intel will need 50W at least to get the same scores, and by the time it s released there will be fast DDR4 that could provide an additional boost..

Up to 95W TDP for 15% better performance than a 35W TDP chip is laughable. If true this only indicates that Carrizo's 'sweet pot' is at low TDPs and they failed to improve performance on desktops. Since we're doing predictions here's mine, based on current leaks 45-65W Skylake GT4e will eat Bristol Ridge for breakfest in both CPU/iGPU tasks.

 

[Abwx]

Wake me up when Carrizo delivers this score inside an actual OEM notebook and not AMD's reference/test platform.

That will be the case in thermaly non constrained laptops like the Lenovo Y700.

GT3e will come down to 15-28W TDPs and GT4e takes its spot at higher TDPs. This might come as a surprise for you but not only AMD can improve perf/watt by refining existing manufacturing processes and architectures (Gen 9 vs Gen 8).

Carrizo s GPU improvement is almost entirely due to the process, and the same work for Intel, just that they wont have a better process than the one available for Broadwell, it will be the same intrinsical perf/Watt.

Up to 95W TDP for 15% better performance than a 35W TDP chip is laughable. If true this only indicates that Carrizo's 'sweet pot' is at low TDPs and they failed to improve performance on desktops. Since we're doing predictions here's mine, based on current leaks 45-65W Skylake GT4e will eat Bristol Ridge for breakfest in both CPU/iGPU tasks.

When i posted that the Xeon was 95W i was asked if i was assuming that the GPU was 95W, look like Intel supporters are not as cautious when it comes to AMD, we wont hear about such claims being libels...

Besides you should look better at the slides you re posting, it will be a second gen EXV core at more than 4GHz, this suggest that this will be better process than the one used for Carrizo.

As for your predictions i guess that they will be as accurate as your CPU perfs/Watt estimations based on TDPs ratings of delivered cooling apparatus.

 

[Sweepr]

That will be the case in thermaly non constrained laptops like the Lenovo Y700.

So no result from OEM systems, yikes. I guess it's fine to use benchmark results from Intel's marketing and reference platforms in chip comparisons from now on.

Carrizo s GPU improvement is almost entirely due to the process, and the same work for Intel, just that they wont have a better process than the one available for Broadwell, it will be the same intrinsical perf/Watt.

Source?

Besides you should look better at the slides you re posting, it will be a second gen EXV core at more than 4GHz, this suggest that this will be better process than the one used for Carrizo.

Which makes it even worse, how can they clock the CPU over 4GHz and improve graphics scores by a mere 15%?

As for your predictions i guess that they will be as accurate as your CPU perfs/Watt estimations based on TDPs ratings of delivered cooling apparatus.

Probably more accurate than this laughable claim:

Actualy AMD dont need more than those 15% since they will improve Carrizo s perf/watt such that Bristol Ridge GPU will stay within those 24W despite 15% higher perfs, Intel will need 50W at least to get the same scores

AMD doesn't stand a chance against GT4e until 2017 at the earliest.

 

[Abwx]

So no result from OEM systems, yikes. I guess it's fine to use benchmark results from Intel's marketing and reference platforms in chip comparisons from now on.

I used Intel s official 3DMark 11 score in the post above, why didnt you protest..?..

Source?

No need of source, SKL was fabbed long before it was launched, you thought that they had a better process last year, or in early 2015.?.

Which makes it even worse, how can they clock the CPU over 4GHz and improve graphics scores by a mere 15%?

As said it would be useless to pour more power in the GPU, perfs in games will improve more than those 15% due to faster RAM.

As for your GT4e if it is some kind of Iris Pro then we wont even have to discuss perf/Watt, it s currently a power hog and they can eventualy improve its perf/Watt by 50% if they double its size, that s still 30% less than Carrizo, and it will be relegated even further by Bristol Ridge, think about it, it will use a better process and the perf improvement wont even increase the GPU TDP.

 

[IntelUser2000]

Abwx,

according to the graph, in 3DMark11 Iris Pro 6200 is only 6% faster but in games it reaches 20%.

according to this link: http://www.hardware.fr/articles/940-23/hd-graphics-530-pratique-h-265-consommation.html

the TOTAL system power consumption is 104W in F1 2014. The difference between idle and running F1 2014 is 79W, which is not GPU alone, but ramping of memory, IOs, and most importantly the CPU.

You also know well enough that Intel chips can go above TDP, but for a short time. Average power consumption on the Broadwell U is not misleading 18-24W, but looking at the picture, 10W, half of what you claim and well under 15W.

Second point is that since Haswell U, there's indications that manufacturers are resorting to using cTDPup feature, which is at 25W.

You can also see here that average power quickly goes down to Package TDP, not way above like you claim: http://www.anandtech.com/show/5878/...ation-realtime-igpu-clocks-on-ulv-vs-quadcore

 

[Khato]

Libel..?..

You know how much an Iris pro comsume before doing such accusation..?.

50W for the GPU alone and for 10% better 3DMark score than said Carrizo..

http://www.hardware.fr/articles/940-23/hd-graphics-530-pratique-h-265-consommation.html

That s 100% more than the Carrizo GPU wich is actualy limited to barely 24W when the SoC is set to 35W :

http://www.notebookcheck.net/fileadmin/_processed_/csm_2015_ISSCC_AMD_APU-page-007_d7a0f7ba5b.jpg

Well, ignoring the comparison between platform power consumption versus a PR slide... Figured I'd chime in by saying that I'm quite looking forward to getting an i7-5775c next week to play around with. I'm especially curious to see what the graphics voltage is set to and how power/voltage/frequency scales in both directions.

 

[IntelUser2000]

-->Abwx:

Let's summarize this:

-6% in 3DMark11 advantage turns to >20% in gaming for Iris Pro over 7870K.
-Iris Pro 6200 Desktop is actually 13% better in 3DMark11 than Carrizo
-No Carrizo available, only in presentations which everyone lies
-GPU power consumption!=Load system power consumption - Idle
-You are comparing Desktop Iris Pro 6200 to a mobile Carrizo setup. mobile Iris Pro 6200 will be different and more efficient

I have a Ivy Bridge U setup myself, and the chip is pretty in-line with TDP figures. Now, I said I noticed in Haswell U, some manufacturers opt to using cTDPup. You can see from the hardware.fr link power usage figure that Intel GPU setup in overall as a system still uses less power than AMD setup and have 20% better performance, even if the iGPU is less efficient. End of story really.

 

[Abwx]

Abwx,

according to the graph, in 3DMark11 Iris Pro 6200 is only 6% faster but in games it reaches 20%.

I m aware that 3DMark 11 score is not representative of games scores, in this case the 20% are entirely due to eDRAM.

according to this link: http://www.hardware.fr/articles/940-23/hd-graphics-530-pratique-h-265-consommation.html

the TOTAL system power consumption is 104W in F1 2014. The difference between idle and running F1 2014 is 79W, which is not GPU alone, but ramping of memory, IOs, and most importantly the CPU.

Actualy the GPU use more power than the CPU, IrisPro is not that powerfull that the CPU must be at full througput to load it at 100%, in Hardware.fr link you can check in LUXMARK power comsumption on GPU mode with CPU unloaded, that s 50W difference when losses are accounted, on CPU + GPU loading the CPU add 21W...

You also know well enough that Intel chips can go above TDP, but for a short time. Average power consumption on the Broadwell U is not misleading 18-24W, but looking at the picture, 10W, half of what you claim and well under 15W.

Second point is that since Haswell U, there's indications that manufacturers are resorting to using cTDPup feature, which is at 25W.

You can also see here that average power quickly goes down to Package TDP, not way above like you claim: http://www.anandtech.com/show/5878/...ation-realtime-igpu-clocks-on-ulv-vs-quadcore

Stress test start shortly at 36W and the device get rapidly to 28W, deltas at the SoC level are hence 24W and 17.5W, to wich we an add 0.5-0.7W that are within the idle power comsumption..

Indeed those figures correlate perfectly with your numbers, of course that the 5010U can be set to a strict 15W or 10W but then it wont achieve the same scores.

You should realize that if 10W was feasible at 2.1GHz then the Y variants that are 1GHz would be at less than 2.5W real TDP.

Yet they are rated 4.5W, and applying a raw square law point to almost 20W at 2.1GHz, let s assume that it s 18W due to the uncore not scaling as much..

When Intel spec a same line, that is the 2C/4T, from 1.9GHz to 2.5GHz it s obvious that the latter will have a TDP that is (2.5/1.9)^2 = 1.73x higher than the former, yet they are all specced 15W, wich is physicaly impossible..

Better to aknowledge that their ratings are a mess than keeping negating laws of physics, because that s the only thing left to try "explaining" the unexplainable...

 

[Abwx]

-You are comparing Desktop Iris Pro 6200 to a mobile Carrizo setup. mobile Iris Pro 6200 will be different and more efficient

It cant be more efficient than Carrizo s GPU by the virtue of the same laws i explained above, it start from too low to close the gap if it use the same process as BDW.

You can see from the hardware.fr link power usage figure that Intel GPU setup in overall as a system still uses less power than AMD setup and have 20% better performance, even if the iGPU is less efficient. End of story really.

That s due to process, not uarch, and at high frequencies Intel has quite an advantage, you had to ressort to Kaveri to try making a point, but what differentiate this chip from Carrizo is essentialy a more efficient process at average frequencies.

 

[mikk]

It means AMD is losing in real world by a much bigger margin. As we know AMD does much better in 3dmark.

 

[Sweepr]

I used Intel s official 3DMark 11 score in the post above, why didnt you protest..?..

Only because this score was lower than some actual results for Broadwell-K, like I proved above. And we all know 3DMark 11 is a best case for AMD, in real world games I wouldn't be surprised to see Skylake-U GT2 coming close or beating the overhyped Carrizo (A10 8700P).

No need of source, SKL was fabbed long before it was launched, you thought that they had a better process last year, or in early 2015.?.

Yes we do need a source, Skylake GT4e will be launched later than the first Skylake GT2 models, probably 6 months after Broadwell GT3e. In case you missed there's also an improved Gen 9 architecture on top of that.

As for your GT4e if it is some kind of Iris Pro then we wont even have to discuss perf/Watt, it s currently a power hog and they can eventualy improve its perf/Watt by 50% if they double its size, that s still 30% less than Carrizo, and it will be relegated even further by Bristol Ridge, think about it, it will use a better process and the perf improvement wont even increase the GPU TDP.

Would you call AMD's Kaveri/Godavari chips power hog?

They use more power and deliver worse performance than Broadwell-K, just like 95W Bristol Ridge will do against Skylake GT4e a year from now (except the performance gap will be much larger).

 

[crashtech]

I see Broadwell on US Newegg now.

It seems a bit overpriced.

 

[DrMrLordX]

A bit overpriced, yes, but not horribly-so. Not $500+ from Japanese importer with an Amazon store overpriced.