AMD Carrizo Pre-release thread

Page 29 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Abwx

Lifer
Apr 2, 2011
11,506
4,280
136
22nm already trumps AMD's 28nm. Any improvement 14nm brings is just piling it on. And do you have any source for the enterprise segment moving to AIOs? That is news to me.

Current Intel s 14nm is inferior to their 22nm and a tweaked 28nm should be competitive enough with this early 14nm, i m curious to see the results, there is an article at PCWatch where the process used is somewhat explained, to summarize Carrizo will have a Beema like efficency.

As for AIOs, dont know how it s elsewhere but here in France generic offices PCs are more and more using such formats, towers and desk cases are so XXth century, also VESA standard cases behinds panels are generaly not used in pro environments.
 
Mar 10, 2006
11,715
2,012
126
Current Intel s 14nm is inferior to their 22nm and a tweaked 28nm should be competitive enough with this early 14nm, i m curious to see the results, there is an article at PCWatch where the process used is somewhat explained, to summarize Carrizo will have a Beema like efficency.

As for AIOs, dont know how it s elsewhere but here in France generic offices PCs are more and more using such formats, towers and desk cases are so XXth century, also VESA standard cases behinds panels are generaly not used in pro environments.

Intel's 14nm isn't inferior to its 22nm...c'mon, man.
 

NTMBK

Lifer
Nov 14, 2011
10,296
5,274
136
Current Intel s 14nm is inferior to their 22nm and a tweaked 28nm should be competitive enough with this early 14nm, i m curious to see the results, there is an article at PCWatch where the process used is somewhat explained, to summarize Carrizo will have a Beema like efficency.

As for AIOs, dont know how it s elsewhere but here in France generic offices PCs are more and more using such formats, towers and desk cases are so XXth century, also VESA standard cases behinds panels are generaly not used in pro environments.

My Haswell Venue 11 requires active cooling to reach maximum performance, but a passively cooled Broadwell Venue 11 is faster. 14nm is not worse than 22nm... certainly not worse at low power/mobile, anyway.

Edit: I don't see AIO taking off in business- LCDs have a much longer life than a PC. Better to VESA mount something like a NUC, replace the PC every 3 years and keep the panel for 10.
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
21,976
11,510
136
Agreed, I'm still rocking a Gateway FPD1730 from . . . 2003? Maybe earlier?

Let's steer this train back on-topic.

So Carrizo is going to be in the mobile space (if AMD can get design wins) and in desktop AIOs in BGA form. FM3 Bristol Ridge and the improperly-labeled "Godavari" Kaveri refresh chips (Godavari isn't their name, The Stilt corrected me on this) are what we see on socketed desktop platforms.

Barring the possibility of someone selling Carrizo BGA boards with enthusiast-tuned BIOS options, I see Carrizo as having no particular interest or value to the enthusiast community except as a preview of what Bristol Ridge might be capable of doing later on (at higher clockspeeds).

You can pretty-much put a fork in this topic, it's done. Unless someone's really pumped about Carrizo in the mobile space?
 

Abwx

Lifer
Apr 2, 2011
11,506
4,280
136
Intel's 14nm isn't inferior to its 22nm...c'mon, man.

It is, it need 10% higher voltage for almost 20% lower frequency,
1.1V at 2.7GHz where HW is at 1.0V for 3.3GHz, what about when higher clocked DT parts will be released.?.

At 3.3 they need theoricaly 1.18V, that is 18% more than 22nm and 40% more power losses at equal parasistic capacitances, to compensate those latter should be reduced by 30%, wich is not easy when shrinking a circuit, each time the distance between two parts/conductors is halved capacitances will be doubled, whether this is between conductors or transistors terminals.

Anyway expect modest frequencies for DT parts, if anything AMD s soon to be released Kaveri refreshes low frequencies are an indication that they dont expect a formidable opposition from thoses parts, for Carrizo it s litteraly a bonanza given that all its purpose is power efficency.
 

NTMBK

Lifer
Nov 14, 2011
10,296
5,274
136
Abwx, do you have full voltage/frequency curves for each of the four 14nm Intel processes, and their 22nm counterparts? Can you point out the precise crossover points in those curves?
 

AtenRa

Lifer
Feb 2, 2009
14,003
3,361
136
Im predicting Carrizo at 15W to have higher iGPU performance than 15W Broadwell HD5500. And that is 28nm planar against 14nm FF.
No matter how you see it, it would be humiliating at best if this will come true.
 
Aug 11, 2008
10,451
642
126
igpu is all that matters, I guess, except in the marketplace, where they have had a better igpu for years, and look at the market share gains it has brought them.
 

Abwx

Lifer
Apr 2, 2011
11,506
4,280
136
Abwx, do you have full voltage/frequency curves for each of the four 14nm Intel processes, and their 22nm counterparts? Can you point out the precise crossover points in those curves?

You can check by yourself, there s enough data at notebookcheck, 14nm needs higher voltage at whatever frequency, this could be of little importance at say 1GHz but the difference become substancial at 2.7, actualy at this frequency all voltage advantage against GF HP 28nm is lost, i dont know for the other parameter wich is the parasistic capacitance but in principle they are bigger with finfets, wich doesnt bode well for this 14nm process.

AMD has an opportunity assuming GF 28nm has been refined for Carrizo, according to PCWatch these are smaller transistors than what is used with Kaveri, with lower conductivity but also lower capacitance, and this is this latter parameter that matter for efficency, better transistor conduction (read high gm and Idsat) wouldnt increase efficency but only max operating frequency, this is the trade off that has been made with Carrizo.
 

DrMrLordX

Lifer
Apr 27, 2000
21,976
11,510
136
You can check by yourself, there s enough data at notebookcheck, 14nm needs higher voltage at whatever frequency, this could be of little importance at say 1GHz but the difference become substancial at 2.7, actualy at this frequency all voltage advantage against GF HP 28nm is lost, i dont know for the other parameter wich is the parasistic capacitance but in principle they are bigger with finfets, wich doesnt bode well for this 14nm process.

AMD has an opportunity assuming GF 28nm has been refined for Carrizo, according to PCWatch these are smaller transistors than what is used with Kaveri, with lower conductivity but also lower capacitance, and this is this latter parameter that matter for efficency, better transistor conduction (read high gm and Idsat) wouldnt increase efficency but only max operating frequency, this is the trade off that has been made with Carrizo.

And so this is the reason why Carrizo is to have no performance desktop presence. AMD seems to be struggling to transition to a performance node smaller than 32nm. Bleh. Oh well.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
And so this is the reason why Carrizo is to have no performance desktop presence. AMD seems to be struggling to transition to a performance node smaller than 32nm. Bleh. Oh well.

Tied themselves to the mast of the wrong ship...
 

Abwx

Lifer
Apr 2, 2011
11,506
4,280
136
And so this is the reason why Carrizo is to have no performance desktop presence. AMD seems to be struggling to transition to a performance node smaller than 32nm. Bleh. Oh well.

It should be good up to 65W although obviously it s optimised for a 12-45W range, FM2+ will be served by Kaveri refreshes, thoses latters lower frequencies than the replaced parts are quite curious, there should be some leaks in the coming months as well as more precise Carrizo numbers.
 

el etro

Golden Member
Jul 21, 2013
1,581
14
81
AMD could sacrifice some Carrizo efficiency in order to bring the uarch to the desktop...
 

DrMrLordX

Lifer
Apr 27, 2000
21,976
11,510
136
Tied themselves to the mast of the wrong ship...

Yarr, best not to be keelhaulin yerself AMD! Har har.

Seriously though, it depends on where AMD can get Carrizo in prominent designs where they can make sales/pick up market share. With the tablet market going into a state of malaise, one must wonder where the next growth opportunity in chip sales will be. Notebooks? AIO desktop?

If Carrizo winds up in the same place as mobile Kaveri, epic fail. If Carrizo winds up replacing E1 and E2 Jaguar chips during a surge in AIO purchases, epic win. I don't know what crack Wall Street is on, but they seem to be thinking "epic win", even in the server space? Unless people are doing multinode custom jobs with Kaveri/Carrizo (which would be . . . interseting), the server sector is basically dead for AMD until maybe Zen. Or are people bullish on Seattle?

The risk is if "big" desktop sales (midsize tower, etc) are the next growth sector (for whatever bizarre reason, I don't see it happening though), since that's where AMD can't position Carrizo. That space is owned by Kaveri/Kaveri refresh which is, at any given TDP, probably going to be the less-competitive product. Unless 65W Carrizo BGA goes into that space and is able to knock off stuff like the A8-7600 with relative ease.

AMD could sacrifice some Carrizo efficiency in order to bring the uarch to the desktop...

I think that's what Bristol Ridge is meant to be in 2016. Maybe they're banking on node refinements and uarch refinements to roll out a product that is decisively superior to Kaveri, maybe with more clockspeed headroom. That combined with the IPC improvements of Excavator and some shader improvements should result in an improved product that can also achieve higher clockspeeds, on a platform that is perhaps less "funky" than FM2+ (personally I find my FM2+ board to be quite capable as an overclocking platform, but some other boards do some pretty weird stuff).

For AMD's sake, Carrizo better be pretty awesome at 12W TDP. They sure seem to have made a lot of design tweaks to get it down to that power envelope in quad form.
 
Last edited:

cbn

Lifer
Mar 27, 2009
12,968
221
106
Barring the possibility of someone selling Carrizo BGA boards with enthusiast-tuned BIOS options, I see Carrizo as having no particular interest or value to the enthusiast community except as a preview of what Bristol Ridge might be capable of doing later on (at higher clockspeeds).

With FM2+ having relatively expensive Mini-ITX boards, I'm thinking the possibility exists we may see BGA Carrizo Mini-ITX boards. This purely for the form factor. I guess it just depends on what Carrizo costs vs. the existing price structure on Kaveri and FM2+ Mini-ITX. If Carrizo is cheap enough then those BGA Mini-ITX might be coming.
 
Aug 11, 2008
10,451
642
126
I dont want to turn this into a stock market forum, but lets get real. AMDs stock is looking hot right now (down 8+ % on Friday BTW) because it was down from 4.00 plus 9 months ago to around 2.50 before it started to bounce back. And before that, it traded 4.00 or well above for a period of several years. So it is still well below the trading range before it tanked a few months ago. Therefore, I would not read into this that some great new product is in the works.
 

Fjodor2001

Diamond Member
Feb 6, 2010
3,989
440
126
Therefore, I would not read into this that some great new product is in the works.

That assumes the market is trading on insider info, which is illegal. If it was known to the public how Carrizo, Zen, K12 etc will perform, it would be safe to say it is reflected in the stock price. But since that info is not known, the stock price is only a guess about the future.
 

AtenRa

Lifer
Feb 2, 2009
14,003
3,361
136
I dont want to turn this into a stock market forum, but lets get real. AMDs stock is looking hot right now (down 8+ % on Friday BTW) because it was down from 4.00 plus 9 months ago to around 2.50 before it started to bounce back. And before that, it traded 4.00 or well above for a period of several years. So it is still well below the trading range before it tanked a few months ago. Therefore, I would not read into this that some great new product is in the works.

What this has to do with Carrizo ??
 

DrMrLordX

Lifer
Apr 27, 2000
21,976
11,510
136
I guess it just depends on what Carrizo costs vs. the existing price structure on Kaveri and FM2+ Mini-ITX. If Carrizo is cheap enough then those BGA Mini-ITX might be coming.

It also depends on the anticipated market for such a product. If AMD and/or their partners (whoever is making the board) think they'll sell well enough, they'll put it on the market for DIY use.

Otherwise, BGA Carrizo will be stuck in AIOs and maybe NUCs.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
Yarr, best not to be keelhaulin yerself AMD! Har har.



That assumes the market is trading on insider info, which is illegal. If it was known to the public how Carrizo, Zen, K12 etc will perform, it would be safe to say it is reflected in the stock price. But since that info is not known, the stock price is only a guess about the future.

Ironically that is exactly what has happened in the past with AMD.

http://www.cbsnews.com/news/galleon-scandal-snags-top-ibm-and-amd-execs/

Illegal, yes. Improbable, no.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
The risk is if "big" desktop sales (midsize tower, etc) are the next growth sector (for whatever bizarre reason, I don't see it happening though)

I think in order for big desktops (and this includes SFF boxes capable of taking a video card) to be the next growth sector we would need to have more competition in the metric consumers want the most and that is cpu performance value.

In contrast, AMD has chosen not to go after this metric and instead of making their quad "big" core (stars, construction cores) processors more affordable they have decided to tack on big iGPUs.

Now AMD does have cat core desktop chips with small 128sp iGPUs, but cpu performance is just not in the same league as Intel's bargain chips (ie, Pentium dual core, etc.). Further worsening this problem is the number of downclocked cat core quads I see making into these "big" desktops.
 
Last edited:

cbn

Lifer
Mar 27, 2009
12,968
221
106
Regarding the downclocked quad and dual cat core chips (including AM1) , I am not sure why this happens.

One theory I have is that AMD must stratify and bin its line-up so there is reduced competition amongst its own chips.

However, with the top bin Kaveri A10-7850K not being too competitive there is increased price and bin pressure on the rest of the line-up to reduce overlap and competition. The result is a trickle down effect on the whole desktop line-up (including cat cores) that makes all the pricing and binning less than ideal IMO.
 

DrMrLordX

Lifer
Apr 27, 2000
21,976
11,510
136
I think in order for big desktops (and this includes SFF boxes capable of taking a video card) to be the next growth sector we would need to have more competition in the metric consumers want the most and that is cpu performance value.

Actually, I'm not sure SFF boxes belong in the same segment as your typical Dude You're Getting A Dell desktop. Different form factors involved, different market. SFF is basically following in the footsteps of the iMac Mini (and actually, Shuttle probably deserves credit here too).

That aside, competition in the desktop sector really isn't what has driven desktop sales into doldroms. It's more like this:

1). People started shifting to notebooks/netbooks/tablets/smartphones
2). People started being able to keep their desktop for longer periods of time without needing to upgrade, especially in 2006/2007 and later once Core 2 chips became common.

There are still lots of people out there with Core 2-era desktops running something like Vista or 7 that see no reason to upgrade. There are fewer people flocking to tablets now, but that does not mean they're going to turn around and buy an i3 box with the money they save avoiding consumer tech purchases. AMD can't automagically stimulate sales in this sector by leapfrogging Intel in IPC or raw performance if people just don't want a new beige (or black) box to sit under their desk.

Now AMD does have cat core desktop chips with small 128sp iGPUs, but cpu performance is just not in the same league as Intel's bargain chips (ie, Pentium dual core, etc.). Further worsening this problem is the number of downclocked cat core quads I see making into these "big" desktops.

Regarding the downclocked quad and dual cat core chips (including AM1) , I am not sure why this happens.

OEMs seem to like them. Seriously, they do. They let the OEM check off all kinds of functionality checkboxes, they're cheap, and . . . they're cheap. And, for the most part, they are faster than Grandma's Netburst Celeron that she is replacing so why not? Sadly, there's a fairly long list of older CPUs that these chips do not exceed in performance.

And let's face it, those desktops with E1 and E2 chips inside really aren't classic desktops anyway, they are just AiOs in desktop clothing. It would be better if, say, 5350s were going into these machines, but noooo can't have that. Not that the 5350 is stupendous or anything. It'd just be better than the dual-core Jaguars.

One theory I have is that AMD must stratify and bin its line-up so there is reduced competition amongst its own chips.

However, with the top bin Kaveri A10-7850K not being too competitive there is increased price and bin pressure on the rest of the line-up to reduce overlap and competition. The result is a trickle down effect on the whole desktop line-up (including cat cores) that makes all the pricing and binning less than ideal IMO.

It's probably not anything quite that complicated. AMD has wafers to burn and has opportunities to push cheap CPUs into OEM systems, so this is what they cough up. If AMD stepped up and started trying to stick 5350s or FX-7500s into those machines, the OEMs would balk at the lower margins involved in selling those machines for $400 at Staples.
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |