AMD Carrizo Pre-release thread

[Abwx]

Some more for you.

Can you stop to pollute this thread..?.

Thoses numbers are meaningfull without a concise analysis that you would be hardly capable of doing accurately, perhaps you want more infos..?.

Create a thread about perf/Watts and i ll promise you that i ll post all tricks used by Intel to cheat the benches by manipulating power numbers.

I already pointed that they extract power from the batteries even when plugged on the main, that cant be detected with a killawatt but has been debunked by some reviews.

If you want to hear it again 59W measured with Intel Power Gadget for a device measured at 14W at the main, this render your numbers moot and just a joke for whom has the slightest knowledge of some basic electronics enginering.

 

[Dresdenboy]

Some more for you.

http://www.notebookcheck.com/Test-Dell-Inspiron-13-7348-Convertible.140896.0.html 29W
http://www.notebookcheck.com/Test-HP-Envy-15-x360-Convertible.137987.0.html 31.8W
http://www.notebookcheck.com/Test-Asus-Zenbook-UX303LA-Subnotebook.139938.0.html 30.5W
http://www.notebookcheck.com/Test-Acer-Aspire-R13-R7-371T-Notebook.137259.0.html 29.1W
http://www.notebookcheck.com/Test-Dell-Vostro-15-3558-Notebook.144790.0.html 25.5W
http://www.notebookcheck.com/Test-Fujitsu-Lifebook-S935-Subnotebook.141410.0.html 34.3W
http://www.notebookcheck.com/Test-Medion-Akoya-E7416-Notebook.135216.0.html 31W
http://www.notebookcheck.com/Test-Acer-Aspire-S7-393-Ultrabook.142295.0.html 30.6W
http://www.notebookcheck.com/Test-HP-Spectre-13-x360-Convertible.142852.0.html 30.2W
http://www.notebookcheck.com/Test-Lenovo-ThinkPad-Yoga-12-Convertible.136698.0.html 35.7W
http://www.notebookcheck.com/Test-Dell-XPS-13-2015-Core-i7-Ultrabook.136638.0.html 33.2W
http://www.notebookcheck.com/Test-HP-EliteBook-Folio-1040-G2-Ultrabook.142338.0.html 26.1W
http://www.notebookcheck.com/Test-Asus-Zenbook-UX303LA-Subnotebook.140774.0.html 28.3W
http://www.notebookcheck.com/Test-Lenovo-ThinkPad-T550-Notebook.137270.0.html 30.4W
http://www.notebookcheck.com/Test-Lenovo-ThinkPad-Edge-E550-Notebook.137307.0.html 35.6W
http://www.notebookcheck.com/Test-Dell-Latitude-14-E7450-Ultrabook.134752.0.html 29.6W
http://www.notebookcheck.com/Test-Acer-TravelMate-P645-Notebook.140243.0.html 30.4W
http://www.notebookcheck.com/Test-Dell-Latitude-3450-Notebook.138497.0.html 33.6W
http://www.notebookcheck.com/Test-Toshiba-KIRA-10D-Kirabook-2015-Notebook.142341.0.html 30.9W
http://www.notebookcheck.com/Test-Toshiba-Satellite-Z30-B-100-Notebook.144873.0.html 28.8W

Thanks. I didn't have that much time left due to important tasks.

Now as you learned sth about small and bigger sample sizes (standard stuff in math stats, machine learning etc.), let me remind you of the Carrizo sample size and our ability to draw conclusions based on it.

 

[-sandro-]

I still don't get if Carrizo is any good... all I read here is arguments.

 

[Abwx]

You have to explain more. No idea what your problem is. I have to ask again, why do you think the score implies a steady 2.4 Ghz for this particular benchmark?

Under Cinebench R15 the APU work at 2.2GHz in MT and 2.5GHz in ST.

Under Cinebench 11.5 the APU work at 1.6Ghz in MT and 2.4GHz in ST.

All previous test done with such a chip have always yielded 2GHz in MT for a 6410 or even a 6310 in some instance, here we have the same chip that cant get higher than 2GHz in a test where it always did so within its TDP, in CB 11.5 an Athlon 5350 consume 12.5-13W for instance.

It cant be a defectuous cooling, the CB R15 MT score would had correlated the thing, as well as all other MT test, only possibility is that NBC messed this test.

 

[maarten12100]

Beware of this number, often lower numbers are due to a non conforming PSU that has no power factor correction, in such small PSUs passive correction is used and it is generaly set such that there s 2W losses that are added to the normal power losses of the adaptator.

This can be checked by measuring idle power when no device is unplugged, and that s why i use systematicaly the deltas and do no generaly pay attention to this number, although if it s inflated by the laptop itself it s visible, despite said PFC.

If they include AKKULAUFZEIT numbers on their full review we can easily calculate the average idle power consumption if we assume the battery does really have the listed capacity.

 

[Abwx]

Some more for you.

http://www.notebookcheck.com/Test-Dell-Inspiron-13-7348-Convertible.140896.0.html 29W

Yes, 29W in one of those paid review to please a difficult customer since it s rare that Dell allow to publish the power numbers of their items, check at NBC, so it s no surprise that there would be some inconsitencies once we look further..


http://www.notebookcheck.com/filead...ron_13-7348/Dellinspiron7348_Stresstest39.jpg

So 18W just for the cores while the uncore generaly take 7-8W at this throughput, no wonder that they didnt select this data in the HWinfo window..

That s already 30W at the main and we have not accounted for the rest and screen that is supposed to be at maximum settings, although it s a 13.3, as a reminder that you know wich devices to select, no place for random choice when it comes to advertise you beloved brand in these threads..

Keep on posting your links, that s a pleasure to stomp those urban legends as well as aknowledge your lack of understanding of everything related to electricity, perhaps if we were using slides.....

 

[Abwx]

If they include AKKULAUFZEIT numbers on their full review we can easily calculate the average idle power consumption if we assume the battery does really have the listed capacity.

AMD has published the numbers for a 50Wh battery, 3.69W for Carrizo and 4.8W for a 5200U, screen is at 100nits apparently :

 

[maarten12100]

AMD has published the numbers for a 50Wh battery, 3.69W for Carrizo and 4.8W for a 5200U, screen is at 100nits apparently :

But that is with a decently built reference platform on which AMD and Shuttle spent money. We don't know if anybody will make a notebook with Carrizo that isn't horrible.

 

[mikk]

Under Cinebench R15 the APU work at 2.2GHz in MT and 2.5GHz in ST.

Under Cinebench 11.5 the APU work at 1.6Ghz in MT and 2.4GHz in ST.

It didn't work at this frequency. If you refer to other devices....irrelevant for this test. But here again no links and nothing for "the APU works at"

It cant be a defectuous cooling, the CB R15 MT score would had correlated the thing, as well as all other MT test, only possibility is that NBC messed this test.

There is only one small chance for a proof that NB messed something up (how could they do this btw?), provide us with other tests from this particular device. And even in this case bios version might differ. Their device was supplied by Toshiba, for this reason I believe it worked as intended.

 

[Abwx]

But that is with a decently built reference platform on which AMD and Shuttle spent money. We don't know if anybody will make a notebook with Carrizo that isn't horrible.

I guess that it s a basic configuration, these are SoCs after all, there s not a lot of circuits lefts to solder.

Display seems to be a 14" 768p Samsung LTN140AT29, at 100nits.

http://www.notebookcheck.com/fileadmin/_processed_/csm_carrizo_footnotes_p1_1_534de8e965.jpg

 

[Phynaz]

Yes, 29W in one of those paid review to please a difficult customer since it s rare that Dell allow to publish the power numbers of their items, check at NBC, so it s no surprise that there would be some inconsitencies once we look further..


http://www.notebookcheck.com/filead...ron_13-7348/Dellinspiron7348_Stresstest39.jpg

So 18W just for the cores while the uncore generaly take 7-8W at this throughput, no wonder that they didnt select this data in the HWinfo window..

That s already 30W at the main and we have not accounted for the rest and screen that is supposed to be at maximum settings, although it s a 13.3, as a reminder that you know wich devices to select, no place for random choice when it comes to advertise you beloved brand in these threads..

Keep on posting your links, that s a pleasure to stomp those urban legends as well as aknowledge your lack of understanding of everything related to electricity, perhaps if we were using slides.....

Intel paid for the review. The last grasping post for the truly desperate.

 

[Abwx]

Intel paid for the review. The last grasping post for the truly desperate.

Said difficult customer is Dell since they send directly some gear, the Carrizo laptop tested by NBC was supplied by a shop that often send them what they want to advertise, at least this is no cherry picked sample.

Die nachfolgenden Benchmarks haben wir auf einem HP Pavilion 17-g054ng durchgeführt, welches uns freundlicherweise von Notebooksbilliger.de

But once at NBC strange things can happen :

Yet :

And of course we are warned that GPU perfs could suffer :

Anmerkung: Obwohl laut Speicher-SPD und Datenblatt des Herstellers DDR3L-1600-Module von Kingston verbaut wurden, vermeldeten unsere Tools (CPU-Z, HWInfo) einen realen Speichertakt von lediglich 666 MHz respektive DDR3-1333. Sollte es sich dabei nicht um einen Auslesefehler handeln, könnten die nachfolgenden GPU-Benchmarks um etwa 5 bis 10 Prozent beeinträchtigt sein.

 

[Dresdenboy]

Yep, HWInfo listed the same. And CPU-Z has more detection problems as it reports a Fury level mem bandwidth for the dGPU...

 

[maarten12100]

Yep, HWInfo listed the same. And CPU-Z has more detection problems as it reports a Fury level mem bandwidth for the dGPU...

relativistic time dialated delta colour compression () :sneaky:

I kid I kid

 

[Abwx]

Insightfull review, indeed, scroll down to the CPU bench scores, click CB 11.5 and then CB R15 and see how the Intel scores get enhanced in respect of the AMD scores..

http://www.notebookcheck.com/Carrizo-im-Test-Was-leistet-AMDs-A10-8700P.147092.0.html

No need of improving the uarchs, it s enough to "improve" the benches apparently.

I was told that the rendering engine was updated and hence that a revision of the bench was necessary, granted, but then why has the rendered picture been modded with the spheres sizes being noticeably reduced..??.

Less computation dependent and more bandwith dependent in the parts that are not complexe, and thus with less iteration to compute the points reflective properties..?.

Dont have an answer at this point but the scores say that something has been changed noticeably, and not randomly...

 

[jpiniero]

Seems pretty obvious now that OEMs don't care about the Windows 10 launch. Consumers definitely don't care. So if AMD was hoping for some magical improvement in the market with Windows 10 they are sadly mistaken. Corps might pick up Windows 10, but that won't be until 2016 and they don't buy AMD anyway.

 

[dark zero]

AMD ended to have a BIG flop... also DX12 wasn't the big change nVIDIA expected... so at last only Intel is the big winner this time.

 

[Abwx]

The more i look at it the less i understand NBC half review of the 7410.

The screenshot is actualy CB R15 with the CPU alleged as being at 1.6GHz in MT according to their comment below the pic.

R15 Multi @1,6 GHz fast konstant

Yet the 175 CB R15 score is obviously performed at 2.12GHz if we compare with the Beema 6410 score at 165.

http://www.notebookcheck.com/filead...0P_Carrizo-L/R15_multi_fast100proz_1_6ghz.PNG


http://www.notebookcheck.com/Analys...MD-Carrizo-L-Notebook-Plattform.147072.0.html

Edit : Same mess with CB R15 ST, stated frequency and score do not match at all :

R15 Single @1,6 GHz, sporadisch 2,4 GHz

http://www.notebookcheck.com/filead...arrizo-L/R15_single_5sek_1_6_1_sek_2_4ghz.PNG

One more time, if ever someone need a job, cookers and butchers are not specialy handicapped and can postulate, although a grade in masonry would be a plus, dont pay attention to the apparent requirements :

Jobs

Wir erweitern unser Team und suchen engagierte Redakteure mit besonderen Kenntnissen im Bereich mobiler Technologien.

• CPU/GPU-Spezialist(in)

 

[erunion]

Seems pretty obvious now that OEMs don't care about the Windows 10 launch. Consumers definitely don't care. So if AMD was hoping for some magical improvement in the market with Windows 10 they are sadly mistaken. Corps might pick up Windows 10, but that won't be until 2016 and they don't buy AMD anyway.

From what I saw AMD was really the only one expecting win10 to boost PC sales.

 

[frozentundra123456]

Even AMD marketing could not be that deluded could they? I think it was just marketing speak/wishful thinking. With it being a free upgrade, no way it was going to drive hardware sales.

 

[Ma_Deuce]

Within the circle of people I know, win8 had a very negative effect on PC sales. I don't think it's out of the realm of possibility that win10 could have a positive effect on new sales.

I can quickly think of at least a few people that are nursing along old hardware because the new stuff all had win8 on it. If 10 isn't the same disaster as 8 was, they will be purchasing new stuff before too long.

 

[MiddleOfTheRoad]

From what I saw AMD was really the only one expecting win10 to boost PC sales.

Well, not sure if other markets were similar -- but Windows 10 laptop sales were incredibly brisk down here in Florida at the two electronics stores I was at today. Best Buy completely sold out of a few Acer / HP and Dell models. Sales at Walmart seemed more subdued, but still solid. Definitely outside the norm for a Wednesday. Could bode well for both Intel and AMD.

I think AMD expected a modest rebound from their currently historically low levels -- I don't believe Intel nor AMD expect PC sales ever to return to peak levels.

 

[majord]

Some more for you.

http://www.notebookcheck.com/Test-Dell-Inspiron-13-7348-Convertible.140896.0.html 29W
http://www.notebookcheck.com/Test-HP-Envy-15-x360-Convertible.137987.0.html 31.8W
http://www.notebookcheck.com/Test-Asus-Zenbook-UX303LA-Subnotebook.139938.0.html 30.5W
http://www.notebookcheck.com/Test-Acer-Aspire-R13-R7-371T-Notebook.137259.0.html 29.1W
http://www.notebookcheck.com/Test-Dell-Vostro-15-3558-Notebook.144790.0.html 25.5W
http://www.notebookcheck.com/Test-Fujitsu-Lifebook-S935-Subnotebook.141410.0.html 34.3W
http://www.notebookcheck.com/Test-Medion-Akoya-E7416-Notebook.135216.0.html 31W
http://www.notebookcheck.com/Test-Acer-Aspire-S7-393-Ultrabook.142295.0.html 30.6W
http://www.notebookcheck.com/Test-HP-Spectre-13-x360-Convertible.142852.0.html 30.2W
http://www.notebookcheck.com/Test-Lenovo-ThinkPad-Yoga-12-Convertible.136698.0.html 35.7W
http://www.notebookcheck.com/Test-Dell-XPS-13-2015-Core-i7-Ultrabook.136638.0.html 33.2W
http://www.notebookcheck.com/Test-HP-EliteBook-Folio-1040-G2-Ultrabook.142338.0.html 26.1W
http://www.notebookcheck.com/Test-Asus-Zenbook-UX303LA-Subnotebook.140774.0.html 28.3W
http://www.notebookcheck.com/Test-Lenovo-ThinkPad-T550-Notebook.137270.0.html 30.4W
http://www.notebookcheck.com/Test-Lenovo-ThinkPad-Edge-E550-Notebook.137307.0.html 35.6W
http://www.notebookcheck.com/Test-Dell-Latitude-14-E7450-Ultrabook.134752.0.html 29.6W
http://www.notebookcheck.com/Test-Acer-TravelMate-P645-Notebook.140243.0.html 30.4W
http://www.notebookcheck.com/Test-Dell-Latitude-3450-Notebook.138497.0.html 33.6W
http://www.notebookcheck.com/Test-Toshiba-KIRA-10D-Kirabook-2015-Notebook.142341.0.html 30.9W
http://www.notebookcheck.com/Test-Toshiba-Satellite-Z30-B-100-Notebook.144873.0.html 28.8W

You seem to be missing this part:

or consumption under load we have already said a few words during the test; it moves at about the same level of competition based Broadwell with 15 watt CPUs. In numbers, this means that immediately after the start of 3Dmark06 (38 watts) or our stress test (39.6 watts) announces the meter just under 40 watts, the fall with the onset TDP-speed regulation for about 2 minutes to just over 30 watts.

This is the same as Haswell, but, NBC have no consistentcy in what numbers they put in their graph.. Sometimes it's the power measured immediatly, other times it's after TDP limits set in.

You have to read the text!

e.g:
http://www.notebookcheck.com/Test-Dell-Latitude-14-E7450-Ultrabook.134752.0.html

Recorded power consumption: 31.6w

However, in the text: The entire unit stabilized at about 30 watts a. For short-term peak loads once 40 watts or more can flow through the outlet


and the opposite:

http://www.notebookcheck.com/Test-Lenovo-Thinkpad-L450-Notebook.140004.0.html

Recorded power: 41.2w

However: Under load the power consumption rises to 42.4 watts. This value, however, we have only measured during the first minute of the stress test. Then, the value drops to 30.8 watts,

 

[Dresdenboy]

Majord, that's very interesting. This mentioning in the text happened in one of my nine listed i5-4300U reviews, too.

 

[Willyfisch]

Does anybody know why there are no reviews of Carrizo from any major review-site like Anandtech? Carrizo notebooks have been out for a while now and all I could find was a review from the German Notebookcheck.

http://www.notebookcheck.com/Carrizo-im-Test-Was-leistet-AMDs-A10-8700P.147092.0.html