AMD Carrizo Pre-release thread

[Vesku]

Naming implies this will be a modest bump up from the 6000 series, perhaps a health reduction in power consumption.

 

[DrMrLordX]

http://www.planet3dnow.de/cms/14527-modellbezeichnungen-von-carrizo-l-bestaetigt/

carrizo-l skus

Interesting, though definitely not the main attraction. Carrizo-L is just warmed-over Beema/Mullins, but hey if they're giving higher clockspeeds like the article says then that's an improvement of sorts.

 

[csbin]

https://www.sata-io.org/building-blocks

 

[AtenRa]

Just realized what has been written in the slide.
Hint, have a look in the purple quadrant. Is this a typo or not ??

 

[Sweepr]

Hitting desktops doesn't mean it's LGA and not BGA.

http://cdn.wccftech.com/wp-content/...ell-E-Haswell-E-Broadwell-Skylake-Haswell.jpg

http://cdn.liliputing.com/wp-content/uploads/2014/08/nuc-roadmap_00.jpg

 

[DrMrLordX]

If they are really moving Carrizo to the desktop in full force (by AMD standards, anyway) at the same time that they launch Carrizo in laptops, then maybe we really will see it in AiO desktops, which would not be bad at all. The skeptic in me expects Carrizo-L in that space instead due to it being stupid-cheap. But again, even Carrizo-L would be a step up from what you see in $400-$500 AMD AiOs right now.

edit: the other bit that's interesting is that they are launching "A-series" Opterons, which are going to be ARM-based SoCs (such as the Opteron A1100). Is this Seattle finally hitting the market?

 

[ShintaiDK]

Just realized what has been written in the slide.
Hint, have a look in the purple quadrant. Is this a typo or not ??

The Broadwell-U chips are also desktop, right?

Or is it something else when you place it in AIOs and NUCs?

There is a reason why AMD does a Kaveri refresh on the desktop.

 

[NTMBK]

"New ARM and x86 based embedded products", "Ultra low power devices", looks like Nolan and Amur aren't cancelled after all.

 

[Abwx]

"New ARM and x86 based embedded products", "Ultra low power devices", looks like Nolan and Amur aren't cancelled after all.

You know that they displayed thoses products as being for 2015 in a japanese conference last month, right.?.

 

[mrmt]

"New ARM and x86 based embedded products", "Ultra low power devices", looks like Nolan and Amur aren't cancelled after all.

"embedded products". It could be new semi-custom designs.

 

[AtenRa]

"embedded products". It could be new semi-custom designs.

Embedded has nothing to do with semi custom.

 

[NTMBK]

You know that they displayed thoses products as being for 2015 in a japanese conference last month, right.?.

No I missed that, thanks for the tip!

 

[ARMD Supremacy]

http://browser.primatelabs.com/geekbench3/compare/2080848?baseline=2208190

Currently, Broadwell i3 in higher TDP allowance NUCs and production laptop comparison sample is small but Dell XPS 13 priced at $800+ with i3-5010u and others priced in mid range.

http://browser.primatelabs.com/geekbench3/compare/2128686?baseline=2208190

 

[Abwx]

Currently, Broadwell i3 in higher TDP allowance NUCs and comparison sample is small but Dell XPS 13 priced at $800+ with i3-5010u and others price in mid range.

http://browser.primatelabs.com/geekbench3/compare/2128686?baseline=2208190

Nice find, numbers are significantly better than Beema, i suspect that it run at a little more than the displayed 1.6, neverless with such a base frequency it s very likely a 15W SKU.

Perfs/watt will be much better than the competition, Beema already trounce anything Intel in this register and Carrizo will hammer the nail even further...

 

[bullzz]

does this make you sleep better at night

 

[Abwx]

does this make you sleep better at night

And could you post something else than an ad hominem and obvious troll post.?.

Are you that annoyed that Carrizo will be much better than Intel s over hyped mobile CPUs to the point of litteraly losing control ?.

 

[Enigmoid]

Are you that annoyed that Carrizo will be much better than Intel s over hyped mobile CPUs to the point of litteraly losing control ?.

Personally, I'm very annoyed that you keep passing off the characteristics of a product that we have seen nothing of as far as definitive proof as fact rather than as speculation.

 

[Abwx]

Personally, I'm very annoyed that you keep passing off the characteristics of a product that we have seen nothing of as far as definitive proof as fact rather than as speculation.

Even if Carrizo would have been just on par with Beema/Mullins perf/watt wise it still would have been better than Haswell or Broadwell.

Indeed the numbers at Geekbench point that it s better than its low power AMD counterpart, so it should largely best Intel s symmetrical offerings.

The only speculation actualy left is by how much, in this respect i think that it will have something like 20-30% perf/Watt advantage over Beema within this latter 14W TDP :

http://browser.primatelabs.com/geekbench3/compare/2208190?baseline=2212381

And Beema has already better efficency than BDW at this TDP even on CPU mode only.

 

[Enigmoid]

Even if Carrizo would have been just on par with Beema/Mullins perf/watt wise it still would have been better than Haswell or Broadwell.

Indeed the numbers at Geekbench point that it s better than its low power AMD counterpart, so it should largely best Intel s symmetrical offerings.

The only speculation actualy left is by how much, in this respect i think that it will have something like 20-30% perf/Watt advantage over Beema within this latter 14W TDP :

http://browser.primatelabs.com/geekbench3/compare/2208190?baseline=2212381

And Beema has already better efficency than BDW at this TDP even on CPU mode only.

Looks like it performs a little better but there is no indication of power usage, other than TDP, which with boost is rather useless other than putting products in a class.

How much power do you estimate Beema uses for the CPU only while running at those scores (~1200 single, 3200 multi)?

 

[Abwx]

Looks like it performs a little better but there is no indication of power usage, other than TDP, which with boost is rather useless other than putting products in a class.

How much power do you estimate Beema uses for the CPU only while running at those scores (~1200 single, 3200 multi)?

For Beema it s about 12W on FP tasks like Cinebench and 10W on Integer, i m assuming that contrary to this latter Carrizo can reach the 15W limit, but not more since AMD use limitations irrespective of eventual thermal headrooms.

 

[positivedoppler]

Didn't AMD say it's launching in the 2nd quarter? The 2nd quarter has started.

 

[jpiniero]

Probably another week or two.

 

[geoxile]

Even if Carrizo would have been just on par with Beema/Mullins perf/watt wise it still would have been better than Haswell or Broadwell.

Indeed the numbers at Geekbench point that it s better than its low power AMD counterpart, so it should largely best Intel s symmetrical offerings.

The only speculation actualy left is by how much, in this respect i think that it will have something like 20-30% perf/Watt advantage over Beema within this latter 14W TDP :

http://browser.primatelabs.com/geekbench3/compare/2208190?baseline=2212381

And Beema has already better efficency than BDW at this TDP even on CPU mode only.

That performance looks kinda disappointing...

 

[Dresdenboy]

I did a short analysis of the two new Geekbench results:
http://citavia.blog.de/2015/04/08/some-carrizo-xv-geekbench-results-20217851/

 

[Abwx]

I did a short analysis of the two new Geekbench results:
http://citavia.blog.de/2015/04/08/some-carrizo-xv-geekbench-results-20217851/

The 8800P has an older bios, in integer it performs generaly, but not always, accordingly compared to the 8700P, in FP there s obviously something that limit its performance, hence Geekbench s odd total scores.