AMD Carrizo Pre-release thread

[bullzz]

core-m vs carrizo

http://browser.primatelabs.com/geekbench3/compare/1319713?baseline=2208190

 

[frozentundra123456]

Wow, wonder how our resident power consumption expert is going to spin that to show AMD is more efficient.

 

[Shehriazad]

Wow, wonder how our resident power consumption expert is going to spin that to show AMD is more efficient.

Mh? At least that Dell system in that last link on the Intel side approaches 19-20W under load. (system load)

Now if AMD has that 15W TDP on their chip with the "chipset" being inside the chip....then wouldn't that possibly mean that AMD versions would suck less/comparable power while having the better GPU performance?


I can't really speak for real world CPU performance as I have absolutely no real data on that (realworld apps, games, etc)...but it doesn't look all that terrible to me.

Seems like they'll at least be comparable.


Btw I'm not trying to pull random facts or anything here...just looking up the bit of limited info that I CAN find and try to put it together.

 

[Abwx]

Wow, wonder how our resident power consumption expert is going to spin that to show AMD is more efficient.

Lol....

BDW run the geekbench test at full frequency even on MT, when limited to a power that is still well above its fake 4.5W official spec it score like a Beema on Cinebech 11.5....

http://www.notebookcheck.net/Dell-Venue-11-Pro-7140-Convertible-Tablet-Review.133634.0.html

19W to be dissipated by a 10"8 device...

And you thought that the CPU was a genuine 4.5W, that is, you compare AMD s real TDPs to Intel s fake TDPs, no wonder that you believe the urban legend that the latter has much better perf/watt..

How long will thoses fictious perfs persists once you use the item.??.

Edit : Looking at the battery life under full load it appear that on battery the item consume less than half what it drain when plugged to the main, perfs will be nowhere near what is displayed at notebookcheck, that is, the displayed perfs are completely irrelevant in respect of usage when on battery...

 

[Abwx]

core-m vs carrizo

http://browser.primatelabs.com/geekbench3/compare/1319713?baseline=2208190

Is this your own submission.??.

http://browser.primatelabs.com/geekbench3/search?q=Intel+Core+M-5Y10

 

[bullzz]

nope. picked this up from another forum
how does it matter though

 

[DrMrLordX]

In all fairness, that Dell Venue score is abnormally high for a 5Y10 of that clockspeed. The 8700P scores within spitting distance of the vast majority of 5Y10 speeds reported by primatelabs as Abwx pointed out. Take that for what it's worth, but at least according to Geekbench, Carrizo looks like it can hold its own.

Battery life is another matter, but we won't know about that until we see the battery life of actual production units.

 

[Abwx]

Actualy this 8700P base frequency is quite weird, according to AMD s slide a 15W chip should work at roughly 2.2-2.3GHz base frequency, 1.6 is quite low even for a 12W chip and would put it at Beema s level perf/Watt wise in this TDP range.

Also a Kaveri already manage 1.7-1.8 at 13-15W and the FX8800P tested at Geekbench display lower perfs than this 8700P, look like more submissions than thoses two samples are needed to have an accurate idea of how the chip perform.

 

[DrMrLordX]

My guess is the base clock is being misreported. That's probably the idle clockspeed, or a throttle point. No idea why the 8800P is reporting inferior performance.

 

[geoxile]

So when is Carrizo going to launch? I need a new laptop.

 

[Dresdenboy]

My guess is the base clock is being misreported. That's probably the idle clockspeed, or a throttle point. No idea why the 8800P is reporting inferior performance.

That's also my assumption.

 

[Abwx]

So when is Carrizo going to launch? I need a new laptop.

Within Q2 according to AT Carrizo article.

That's also my assumption.

The 8700P look to work at 1.8-1.9 at most, the 7300s you used as comparison all have a single channel RAM wich will blur the multithread score.



Edit : Look like it s even at base frequency on MT with a 2.7 turbo on ST, also FP did seemingly get a greater boost than Integer.

Comparison with an apparently non throttling Kaveri FX 7600P at 2.7 base/3.6 turbo :

http://browser.primatelabs.com/geekbench3/compare/651668?baseline=2208190


And with a Richland A8 5550M :

http://browser.primatelabs.com/geekbench3/compare/2223817?baseline=2208190

 

[cbn]

Do we know how many M.2 slots Carrizo will support?

 

[ShintaiDK]

Do we know how many M.2 slots Carrizo will support?

 

[AtenRa]

DDR-4 confirmed and die picture

 

[burninatortech4]

Damn that x86 portion is small. High density libraries!

When will OEM's ship Carrizo laptops!? :\

 

[Idontcare]

DDR-4 confirmed and die picture

Are my eyes seeing this correctly? The freaking NB has comparable die-area as the x86 processor space? Seriously?

When your SKU's ASP is predominately determined by the performance of something that occupies <20% of your silicon outlay, you should feel reasonably confident in proclaiming that somewhere in your journey to today the wheels very much came off your wagon :\

 

[AtenRa]

Carrizo is an SoC, it also includes the FCH. The NB alone is smaller than you think by looking at the pic above.

 

[NTMBK]

Are my eyes seeing this correctly? The freaking NB has comparable die-area as the x86 processor space? Seriously?

When your SKU's ASP is predominately determined by the performance of something that occupies <20% of your silicon outlay, you should feel reasonably confident in proclaiming that somewhere in your journey to today the wheels very much came off your wagon :\

"Northbridge" also includes stuff like the hardware video encode/decode blocks, if I remember correctly.

 

[pTmdfx]

A kind reminder: 14.48 mm2 does not include the L2 arrays and the cache interface unit.

 

[pTmdfx]

"Northbridge" also includes stuff like the hardware video encode/decode blocks, if I remember correctly.

Yeah. The "real" northbridge is between the core pairs, with the memory controller and the GMC block, IIRC. The rest on the left is the I/O complex and all the GPU and multimedia related stuff, which is called Graphics Northbridge.

 

[svenge]

That picture really does sum up the current state of AMD's x86 product...

 

[ShintaiDK]

Target cost reduction, yet still a 250mm2 die. That WSA still haunts them.

 

[AtenRa]

Target cost reduction, yet still a 250mm2 die. That WSA still haunts them.

You are forgetting the Carrizo die also includes the FCH, so it does bring cost down among energy consumption reduction and communication performance increases.
Also this SoC will have higher ASP than mobile Kaveri and volumes may also be higher.

A 15W Carrizo with 2133MHz DDR-4 So-Dimms will be a killer combination for the Mobile market. The perf/watt and perf/$ will be among the best in its segment. As an OEM/ODM you will be sort-sighted not to use this product.

 

[NTMBK]

Target cost reduction, yet still a 250mm2 die. That WSA still haunts them.

They completely eliminated the Southbridge die, reducing total die area and also PCB costs.