Guess what product AMD is revealing!

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Status
Not open for further replies.

cbn

Lifer
Mar 27, 2009
12,968
221
106
I am actually hoping one of the SKUs is an AM1 or BGA dual core Kabini with 64 GCN stream processors enabled.

Yes, I really did wite that. (basically give me Zacate E350 performance on 28nm for an even lower price than we see with Sempron 2650)
 

Vesku

Diamond Member
Aug 25, 2005
3,743
28
86
Yes, it's the same die as kabini so definitely a Cat family announcement. If this is the actual die shot of beema/mullins then they didn't use high density libraries to make their performance per watt gains, must be a combination of foundry processes and binning.

lQTyhwS.jpg
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
Die shown

Toyota.jpg

Comparing the portion of the Kabini die above under the magnifiying glass to the Kabini floorplan below:



(Based on my very rough understanding of dies and floorplans) It looks like the part under the "magnifying glass" involves the iGPU and some of the I/O (including SATA, MIsc I/O, usb, FCH)

I wonder if there is any significance?
 

PPB

Golden Member
Jul 5, 2013
1,118
168
106
If no HDL, then can we say they finally applied the resonant mesh improvements in the cat line, AFAIK amd only talked about using on the big core line
 

AtenRa

Lifer
Feb 2, 2009
14,003
3,362
136
From a first look it seams the new die is smaller than the first Kabini die.
It seams to me that certain parts have been shrink, like the CPU cores, the L2 Cache, the two CU clusters (iGPU). But other parts seams to retain the same area as before.
I believe they used the HD Libraries.
 

Arkadrel

Diamond Member
Oct 19, 2010
3,681
2
0
Yes, it's the same die as kabini so definitely a Cat family announcement. If this is the actual die shot of beema/mullins then they didn't use high density libraries to make their performance per watt gains, must be a combination of foundry processes and binning.

lQTyhwS.jpg


Thanks for posting this side by side picture.

You can tell the new die shot is of a smaller chip.
Because some parts that are the same from the other die shot, look bigger here.

Thus its atleast a shrunk version of a Kabini.

What looks the same (size wise):
The memory interface
The shared 2mb L2
The usb / pci expresse stuff

What looks "new" and differnt:
The GPU portion looks ALOT smaller, like tiny.
The CPU cores look a little bit smaller too.


Im guessing this is a faster more energy effecient Kabini chip.
 
Last edited:

NTMBK

Lifer
Nov 14, 2011
10,411
5,677
136
Thanks for posting this side by side picture.

You can tell the new die shot is of a smaller chip.
Because some parts that are the same from the other die shot, look bigger here.

Thus its atleast a shrunk version of a Kabini.

What looks the same (size wise):
The memory interface
The shared 2mb L2
The usb / pci expresse stuff

What looks "new" and differnt:
The GPU portion looks ALOT smaller, like tiny.
The CPU cores look a little bit smaller too.


Im guessing this is a faster more energy effecient Kabini chip.

Um, don't read too much into the scaling of marketing images. To me it looks like they took a Kabini die shot, and slightly distorted the aspect ratio. Look at the L2 cache- structures which were square on the old shot are now rectangular, despite very clearly being the same structure. I think someone in marketing just stretched the image a bit to fit their web layout.
 

Rakehellion

Lifer
Jan 15, 2013
12,181
35
91
Thanks for posting this side by side picture.

You can tell the new die shot is of a smaller chip.
Because some parts that are the same from the other die shot, look bigger here.

Thus its atleast a shrunk version of a Kabini.

It's impossible to tell by looking at the pictures because you don't have a frame of reference.

I am hoping it's something for mobile.
 

Vesku

Diamond Member
Aug 25, 2005
3,743
28
86
Um, don't read too much into the scaling of marketing images. To me it looks like they took a Kabini die shot, and slightly distorted the aspect ratio. Look at the L2 cache- structures which were square on the old shot are now rectangular, despite very clearly being the same structure. I think someone in marketing just stretched the image a bit to fit their web layout.

Exactly, don't try to judge scale from marketing die shots. I posted side by side more so it was easier to match the structures and to me they match up 1:1. Given AMD's marketing it's possible they just used a kabini die shot they already had in their approved materials, though, rather than actual Beema/Mullins die.
 

Phynaz

Lifer
Mar 13, 2006
10,140
819
126
Exactly, don't try to judge scale from marketing die shots. I posted side by side more so it was easier to match the structures and to me they match up 1:1. Given AMD's marketing it's possible they just used a kabini die shot they already had in their approved materials, though, rather than actual Beema/Mullins die.

This
 

2is

Diamond Member
Apr 8, 2012
4,281
131
106
Another uncompetitive product?

There are half a dozen ways you could have phrased that post and made a legitimate discussion out of it. Instead you chose thread crapping, which is something we cannot tolerate.
-ViRGE
 
Last edited by a moderator:

piesquared

Golden Member
Oct 16, 2006
1,651
473
136
Those two die pics are not the same. The black areas are different.

15rdwg3.jpg

Yeah, there's some differences for sure. Beema, Mullins, Steppe Eagle is probably what this is referencing. There should be some changes to the Compute Units and at the SoC level.
 
Last edited:

~CS~

Senior member
Apr 14, 2010
706
2
81
My guess it's the new cat core, but only for the mobile space. Will probably arrive to desktop a year later to replace Kabini, and will most likely be compatible with existing AM1 motherboards. I know it's been speculated that it might need new motherboards, because of the eminent arrival of ddr4, but ddr4 will be too expensive at release for entry level platform.
 

blackened23

Diamond Member
Jul 26, 2011
8,548
2
0
I continually get a laugh out of the AMD "victim" mindset. Instead of asking AMD to deliver a product with better performance per watt, a metric that AMD objectively (for the lack of a better word) completely sucks at, they'll blame everything else. Conspiracies, bribes, compilers, the list goes on.. Let's just blame everyone except where the blame lies: AMD for delivering endless products with poor performance or poor performance per watt. Performance per watt is everything for mobile, and AMD either ends up delivering an under performing product or a product with terrible PPW.

Maybe that can change in the future, but in the meantime, perhaps some AMD fans can actually ask AMD to deliver on the metrics that matter for mobile. Apparently, intel has delivered competitive performance per watt in every computing segment. But we can't ask that of AMD. We just have to pretend their inferior performance per watt is not their fault.

And like intel stated, if beema / mullins are good products which reverse the trend of past products, then we will see products using them. Period. Bay Trail is being used in products because it's a good SOC with competetive performance per watt; is it perfect? Absolutely not. But it's a great product with good CPU performance and good performance per watt. If beema and mullins are good products, you will see them in devices. Period. So let's see if AMD can put their money where their mouth is this time. Will that happen or will it be a repeat of past products?
 
Last edited:

cytg111

Lifer
Mar 17, 2008
25,661
15,161
136
So .. AMD is revealing either an Intel chip or a fanboy on a stick ... judging from the direction this thread has taken.
 
Status
Not open for further replies.