AMD Piledriver rumours ... and expert conjecture

[Reynod]

We have had several requests for a sticky on AMD's yet to be released Piledriver architecture ... so here it is.

I want to make a few things clear though.

Post a question relevant to the topic, or information about the topic, or it will be deleted.

Post any negative personal comments about another user ... and they will be deleted.

Post flame baiting comments about the blue, red and green team and they will be deleted.

Enjoy ...

 

[jimmysmitty]

Mr. Akrou AMD- “Since you asked twice we are going to get you an answer.”

Which 2 models are used for main purposes?

left side: Llano
right side:Trinity the High Performance model + A better RAM

http://semiaccurate.com/assets/uploads/2012/02/Trinity-Improvements.png

That is a very good question, one always left out.

As I said, I can only see Trinity getting a major boost if its clocked higher. I don't expect PD to do much more above BD and Trinity is based on PD.

I also want to know how they got their 17w going since the current 32nm is not that great.

Lets also consider this is basically sayin that at 35w it would be about 2x faster than a 35w Llano.

Love the marketing slides.

 

[esrever]

Trinity supposedly 25% better cpu wise than Llano. A little bit of math using Tom's Adobe Premiere Pro benchmark:

Llano=69.2 minutes. -25%=51.9 minutes, or still under preforming a i3-2100(47:11).

That still doesn't look that good for a trinity A8, the A10 will be even better, but it would still not even come close to a 2500k(24:47).

On the graphics side, 50% is a huge jump, and would completely smash anything Intel will offer with ivy, even if Intel get's their 50% better than SB.

I highly doubt either side will get 50% better on graphics, but I think that 20-25% on trinity cpu will be possible.

TBH, This A8 vs i3 was pretty sad for AMD, considering they are marketing trinity at the high end of the mainstream market.

the A6 easily beats the i3 in laptops, the A8 competes with the i5s.

The desktop APUs are not marketed towards high end mainstream, they are marketed towards cheap All in one systems for OEMs.

The desktop Llano also beats out the i3 in any multi threaded benchmarks.

http://www.anandtech.com/show/4448/amd-llano-desktop-performance-preview/2

 

[r0ck3tm@n]

the A6 easily beats the i3 in laptops, the A8 competes with the i5s.

The desktop APUs are not marketed towards high end mainstream, they are marketed towards cheap All in one systems for OEMs.

The desktop Llano also beats out the i3 in any multi threaded benchmarks.

http://www.anandtech.com/show/4448/amd-llano-desktop-performance-preview/2

Do NOT get me interested in AMD's APUs.

 

[g4114rd0]

...many rumours can cause mistrust"

AMD Denies Building Chips at IBM's Factories.

http://www.xbitlabs.com/news/other/display/20120207170536_AMD_Denies_Building_Chips_at_IBM_s_Factories.html

 

[g4114rd0]

the A6 easily beats the i3 in laptops, the A8 competes with the i5s.

The desktop APUs are not marketed towards high end mainstream, they are marketed towards cheap All in one systems for OEMs.

The desktop Llano also beats out the i3 in any multi threaded benchmarks.

http://www.anandtech.com/show/4448/amd-llano-desktop-performance-preview/2

... APU fortune cookies!

http://www.bit-tech.net/hardware/cpus/2012/02/06/amd-a8-3870-review/5

no bench, no Glory!

 

[esrever]

... APU fortune cookies!

http://www.bit-tech.net/hardware/cpus/2012/02/06/amd-a8-3870-review/5

no bench, no Glory!

that 7zip bench doesn't make any sense... a video shouldn't change cpu performance considering its all done by the gpu unless theres some ram or power restrains that forces llano to not be able to fully make use of the cpu.

 

[g4114rd0]

that 7zip bench doesn't make any sense... a video shouldn't change cpu performance considering its all done by the gpu unless theres some ram or power restrains that forces llano to not be able to fully make use of the cpu.

-...last cookie, please take you time."


AMD says PD will not use used Transactional memory (new mantra).

http://software.intel.com/en-us/blogs/2012/02/07/transactional-synchronization-in-haswell/

 

[jdwii]

so your prediction 30% graphics and 10% cpu

what does 4110BD have to with PiledriverTrinity

BD is the same design PD is, Except it's supposed to be improved(Why didn't Amd explain how their going to do this?) and probably a new stepping. Amd will have to make Trinity have a pretty high clock rate to beat Llano. Unless the PD core is around 20% more Powerful then BD per clock or if they can Increase the clock rates by 20% which would mean a 4 core Trinity clock at 2.0-2.2Ghz(over a Llano at 1.5-1.7Ghz). Until i see the numbers i will think PD is nothing but a few changes and a newer stepping. And right now the Phenom is about 9-18% more Faster per clock then BD. Not to mention a module is pretty much 160% of 2 cores.
Go here (http://www.legitreviews.com/article/1741/11/)
See look at cinebench Under the 8150fx it gets a score of 0.97 under the single core Now if we take 0.97 and times it by 8(cores) we would get 7.76 which would mean 100% perfect scaling which is never possible now if we take 90% scaling which is what some one told all of us BD would be per core(equal to a module being 180% of a dual core which is false) we would get 6.984 But instead we get 6.01 which is around 80% scaling maybe a little less like 78%. What this means is a 4 core BD even at a clock rate of around 15% higher then the Phenom would still lose when all cores are needed such as encoding Which is why the 4110fx loses on encoding test's even compared to the Llano with no L3 (http://www.legitreviews.com/article/1766/7/).


Now look at the Phenom scaling we get 1.11 for the 1100T which is already around 14% faster then the BD even with a 9% lower clock rate, Take 1.11 and times it by 6(cores) and we get 6.66 which is not the score its a 5.86 which is around 90%(88%) scaling.

What i'm saying is the Phenom has about 10% better scaling per core and around 10-15% better IPC then BD. These are the results today and you also have to remember the Phenom has no turbo that works unlike BD where its always going into turbo at 3.9-4.2Ghz which even makes these results look worse!

 

[triny]

BD is the same design PD is, Except it's supposed to be improved(Why didn't Amd explain how their going to do this?) and probably a new stepping. Amd will have to make Trinity have a pretty high clock rate to beat Llano. Unless the PD core is around 20% more Powerful then BD per clock or if they can Increase the clock rates by 20% which would mean a 4 core Trinity clock at 2.0-2.2Ghz(over a Llano at 1.5-1.7Ghz). Until i see the numbers i will think PD is nothing but a few changes and a newer stepping. And right now the Phenom is about 9-18% more Faster per clock then BD. Not to mention a module is pretty much 160% of 2 cores.
Go here (http://www.legitreviews.com/article/1741/11/)
See look at cinebench Under the 8150fx it gets a score of 0.97 under the single core Now if we take 0.97 and times it by 8(cores) we would get 7.76 which would mean 100% perfect scaling which is never possible now if we take 90% scaling which is what some one told all of us BD would be per core(equal to a module being 180% of a dual core which is false) we would get 6.984 But instead we get 6.01 which is around 80% scaling maybe a little less like 78%. What this means is a 4 core BD even at a clock rate of around 15% higher then the Phenom would still lose when all cores are needed such as encoding Which is why the 4110fx loses on encoding test's even compared to the Llano with no L3 (http://www.legitreviews.com/article/1766/7/).


Now look at the Phenom scaling we get 1.11 for the 1100T which is already around 14% faster then the BD even with a 9% lower clock rate, Take 1.11 and times it by 6(cores) and we get 6.66 which is not the score its a 5.86 which is around 90%(88%) scaling.

What i'm saying is the Phenom has about 10% better scaling per core and around 10-15% better IPC then BD. These are the results today and you also have to remember the Phenom has no turbo that works unlike BD where its always going into turbo at 3.9-4.2Ghz which even makes these results look worse!

Any 4 core has ample power especially for the mobile market Imho

 

[viridiancrystal]

the A6 easily beats the i3 in laptops, the A8 competes with the i5s.

The desktop APUs are not marketed towards high end mainstream, they are marketed towards cheap All in one systems for OEMs.

The desktop Llano also beats out the i3 in any multi threaded benchmarks.

http://www.anandtech.com/show/4448/amd-llano-desktop-performance-preview/2

Yeah i figured the benchmark results i used were a bit odd for most cases. The big thing I should realize is that Apu's are aimed a mobile, and mobile is where they win.

 

[g4114rd0]

Athlon APUs 2!
Apu AD641XWNGXBOX AMD Athlon II X4 641K FM1 4M 2800MHZ 100W Retail!
AMD KunFusion is the marketing name generated by  most unclear  fusion processes,
at least the reverse engineering you must be 4 unlocking the GPU-less.
http://www.cpu-world.com//news_2012/2012020801_AMD_quietly_introduces_two_Athlon_II_X4_CPUs.html

edit: more info

 

[triny]

-...last cookie, please take you time."


AMD says PD will not use used Transactional memory (new mantra).

http://software.intel.com/en-us/blogs/2012/02/07/transactional-synchronization-in-haswell/

Intel is ahead in the search for the personal super computer
but Amd won't concede the fight Lano was the preliminary effort it's best to compare it to i3 first generation
Trinity should be compared to i3 second generation
not until kaveri (sp) can we compare it to Haswell

It should be interesting to see what happens in 2015
Piledriver will be like an unfinished painting still behind the 8ball
but conidering even the weakest 4 core Athhelon is more than amply equipped
to satisfy most needs. Only the Machinations of extremists persuade the masses
that AMD has little to offer.
BD is a prime example of this

 

[g4114rd0]

no GPU, not considered APU

you're Absolutely Right, but it's clear what they're trying to do.

http://www.provantage.com/amd-ad641xwngxbox~7AAMD2PP.htm

 

[triny]

no it doesn't.
and do not speak of gaming either..
http://i1130.photobucket.com/albums/m529/malmental/Averages.png

What Vcard was used? the behemoth?
mobile I3 has no chance nor i5 or i7, even with discrete

When you skew the results by testing beyond the scope of the parts it isn't going to give the proper results
http://www.youtube.com/watch?v=LxlQLzOCxEc
here we can see the the 3850 + the card it's designed to work with it plays games at roughly 60 fps
since the chip has benefits using the proper card
If you go outside the scope of it's design then your capping it.(the 6550D won't get used)

http://www.youtube.com/watch?v=RLzcat9Colc&feature=related
i7 compared to 3850

 

[triny]

not talking about mobile, I'm not anyways.

Oh ,well those videos are worth a look in any case.

 

[triny]

Trinity supposedly 25% better cpu wise than Llano. A little bit of math using Tom's Adobe Premiere Pro benchmark:

Llano=69.2 minutes. -25%=51.9 minutes, or still under preforming a i3-2100(47:11).

That still doesn't look that good for a trinity A8, the A10 will be even better, but it would still not even come close to a 2500k(24:47).

On the graphics side, 50% is a huge jump, and would completely smash anything Intel will offer with ivy, even if Intel get's their 50% better than SB.

I highly doubt either side will get 50% better on graphics, but I think that 20-25% on trinity cpu will
TBH, This A8 vs i3 was pretty sad for AMD, considering they are marketing trinity at the high end of the mainstream market.

Heck yeah the 2500k is the best but trinity should still play games above 60 fps with discrete very cost effectively
If you have the money yes i5 2500k is best. For those want highest fps sure but personally I can't tell the difference above 60 fps

 

[-Fran-]

... APU fortune cookies!

http://www.bit-tech.net/hardware/cpus/2012/02/06/amd-a8-3870-review/5

no bench, no Glory!

I kinda agree with esrever there... That 7zip + mplayer makes no sense at all, lol. They don't even mention the qualitative part of the "playback". That "bench" cries for a Video to show how mplayer actually played the video. Besides... The 3.6Ghz has less points than the same chip at 3Ghz? No, there's something wrong in that test, lol.

Anyway, those Athlon IIs are such a bad stunt... Not even OEMs want them =/

Cheers!

 

[triny]

http://www.kitguru.net/site-news/interviews/jules/brice-from-arctic-reveals-all-in-exclusive-interview/

Trinity A10 APU (replacement for Llano) which uses Piledriver cores and gives Radeon HD 7760 graphics. When asked about specific launch dates and pricing, Brice became shy.
If this is true Amd may own the mobile market especially price wise

error edit

 

[esrever]

no it doesn't.
and do not speak of gaming either..
http://i1130.photobucket.com/albums/m529/malmental/Averages.png

what part of that chart was multithreaded benchmarks I couldn't find it?

 

[g4114rd0]

I kinda agree with esrever there... That 7zip + mplayer makes no sense at all, lol. They don't even mention the qualitative part of the "playback". That "bench" cries for a Video to show how mplayer actually played the video. Besides... The 3.6Ghz has less points than the same chip at 3Ghz? No, there's something wrong in that test, lol.

Anyway, those Athlon IIs are such a bad stunt... Not even OEMs want them =/

Cheers!

...The exact provenance of APUs fortune cookies is unclear.”

-The desktop Llano also beats out the i3 in any multi threaded benchmarks.

So "just for fun" I had found two, even if some things don't work out.

.Handbrake tests multi-threaded CPU and memory subsystem performance.
.a massive file backup (with encryption) using 7-Zip, while simultaneously playing back an HD movie file using mplayer.

@ suggestions you can try at home, a counterargument can be

PC a 2.66GHz Intel Core 2 Duo E6750 at stock speed, 2GB of Corsair 1,066MHz DDR2 score of 1,000 ,

and yet, I don't entirely agree with @3.6Ghz.

 

[palladin9479]

APU's really don't make sense on the desktop platform as cheap discrete cards are available. It does make sense on mobile platforms and extremely low profile kiosk / mini-ITX platforms. They also make sense in the low budget PC's, as in sub $500 USD OEM PCs.

Now those market segments happen to be very large and lead the industry in sales. Their boring market segments, no big performance numbers, no performance crowns, no headlines, just lots of sales to people who really only care about how it looks.

 

[r0ck3tm@n]

APU's really don't make sense on the desktop platform as cheap discrete cards are available. It does make sense on mobile platforms and extremely low profile kiosk / mini-ITX platforms. They also make sense in the low budget PC's, as in sub $500 USD OEM PCs.

Now those market segments happen to be very large and lead the industry in sales. Their boring market segments, no big performance numbers, no performance crowns, no headlines, just lots of sales to people who really only care about how it looks.

I agree with this. I think for a high performance desktop the only APU that makes sense is something like an I7 2600K which supports desktop graphics only and cpu performance is not compromised.

However, if I build a computer for my brother's kids to play with, an APU is perfect. The same reasoning goes for that HTPC that I have always wanted to build.

Intel has done such a good job with the Sandy Bridge architecture that it's confusing things, you don't really loose any CPU power for there being a gpu in there. Thing is, I find myself entirely in agreement with this and wishing now that my new motherboard had an on-board graphics connector.

I am nevertheless hoping that AMD goes somewhere interesting with their APUs regarding computing power and graphics. I really, really hope they don't give up and resign themselves to the mid and low performance markets.

Finally, I still have my AMD T-shirt. Good day.

 

[triny]

I agree with this. I think for a high performance desktop the only APU that makes sense is something like an I7 2600K which supports desktop graphics only and cpu performance is not compromised.

However, if I build a computer for my brother's kids to play with, an APU is perfect. The same reasoning goes for that HTPC that I have always wanted to build.

Intel has done such a good job with the Sandy Bridge architecture that it's confusing things, you don't really loose any CPU power for there being a gpu in there. Thing is, I find myself entirely in agreement with this and wishing now that my new motherboard had an on-board graphics connector.

I am nevertheless hoping that AMD goes somewhere interesting with their APUs regarding computing power and graphics. I really, really hope they don't give up and resign themselves to the mid and low performance markets.

Finally, I still have my AMD T-shirt. Good day.

HTPC is where they shine as well as casual gaming and mobile of course.If it's true trinity's igp is 7760x the game will change.

 

[Chad Boga]

HTPC is where they shine as well as casual gaming and mobile of course.If it's true trinity's igp is 7760x the game will change.

What current discrete card would give the same performance as this 7760x?

 

[truegenius]

using a fast gpu (even if using 1k spu if tdp is still under control by some magic) will not make the trinity a good choice over dedicated card (even 6770)
because of slow ram, and using high speed ram is not cheap which is making apu a flop product for medium gaming.

Amd needs to overcome this memory problem.


(you can try this at home
use 300mhz (to simulate slow vram for igpu)for video memory of your card and first stock and then overclocked gpu to test improvements in performance/benchmark.