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]

Yep. Socket 7 was K6, Super Socket 7, or Super 7, was K6-2 and K6-III.

One funny not there, people say Intel changes sockets more often than AMD but if you count them, not including non-proprietary sockets like Socket 7 or prior, AMD has had 11 sockets from their first socket (Super Socket 7) and Intel has had 10 including the unreleased LGA 1150. All are desktop sockets.

 

[king smp]

Yep. Socket 7 was K6, Super Socket 7, or Super 7, was K6-2 and K6-III.

One funny not there, people say Intel changes sockets more often than AMD but if you count them, not including non-proprietary sockets like Socket 7 or prior, AMD has had 11 sockets from their first socket (Super Socket 7) and Intel has had 10 including the unreleased LGA 1150. All are desktop sockets.

true but AM series had backwards compatibility to a certain extent
between AM2 AM2+ AM3 AM3+there was alot of possible CPUs to use
there isnt that kind of compatibility with the Intel sockets so that is probably why the impression that Intel
has had more socket changes

 

[JAYDEEJOHN]

http://blogs.amd.com/fusion/2012/04/19/amd-%E2%80%9Ctrinity%E2%80%9D-and-%E2%80%9Cbrazos-2-0%E2%80%9D-heading-your-way/

 

[king smp]

I found this amusing from your link jaydeejohn

"Phil Hughes is a Senior PR Manager at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions."

so what purpose does he serve?

 

[jimmysmitty]

true but AM series had backwards compatibility to a certain extent
between AM2 AM2+ AM3 AM3+there was alot of possible CPUs to use
there isnt that kind of compatibility with the Intel sockets so that is probably why the impression that Intel
has had more socket changes

Possibly but compatibility was still dependant on the mobo just like Intels longest lasting socket, 775. You buy a cheap mobo, don't bet that it would support newer CPUs. But put a bit of money into your mobo and you could almost gurantee future same socket CPU support.

I have run into many AM2+ mobos that did not support Phenom II or Phenom, and it could have been due to the VRMs being too weak or the BIOS never being updated. I also had a P5K-E which was not the top of the line P35 chipset mobo yet it supported every LGA775 CPU. It still cost a bit, but it was worth being able to upgrade CPUs easily, much like I am sure my P8Z68-V Pro Gen3 will support every SB and IB CPU released.

But thats the thing about backwards compatibility that most people don't get, you cannot expect to go super cheap and get the best support. Its the same with everything. Gotta put some money in to get the best.

Asus had a low end mobo, the M5A78l-MX and it supported all AM3 and AM3+ CPUs up to 95W TDP. They released another M5A78l-MX Plus which had better VRNs to support the 125W TDP CPUs. But then again this is one of the cheapest mobos they make for AMD so support is not 100% due to the cheaper components or even less VRMs and such. But most of the $100+ AM3+ mobos supported all 125W TDP Phenom IIs and all BDs with a BIOS update.

 

[sarinaide]

What is the word on the IPC's on Piledriver, can it legitimately be said to be on a par or better than intel first gen? If that is the case I will regard it as a positive step in the right direction.

 

[jimmysmitty]

What is the word on the IPC's on Piledriver, can it legitimately be said to be on a par or better than intel first gen? If that is the case I will regard it as a positive step in the right direction.

There is no IPC info, just AMDs "marketing" numbers, basically slides with numbers but no real data.

My thoughts are that PD will have tweaks but much like Trinity, rely on mostly clock speed increases for performance gains more than actual IPC increases mainly because there isn't enough time between BD and PD to warrant any major changes that could increase IPC more than a bit.

I am expecting at least Phenom -> Phenom II IPC increase, but again Phenom II had more on the lines of major clock speed increases, first gen Phenom was 2.3GHz (9600), B3/second gen was 2.6GHz (9950) Phenom II was up to 3.7GHz a 1.1GHz increase.

On IPC, it was pretty decent still:

http://www.anandtech.com/show/2702/9

As well it made AMD overclockable again while Phenom I was not so OC friendly.

So on a positive level, at least 15% performance increase but probably 5-10% at most being IPC.

Again, thats my guess based on no info. Once we get more actual information we will be able to better guesstimate.

 

[g4114rd0]

.AMD Begins Volume Manufacturing of A-Series "Trinity" APUs..."
http://www.xbitlabs.com/news/cpu/display/20120419172156_AMD_Begins_Volume_Manufacturing_of_A_Series_Trinity_APUs.html

 

[sarinaide]

There is no IPC info, just AMDs "marketing" numbers, basically slides with numbers but no real data.

My thoughts are that PD will have tweaks but much like Trinity, rely on mostly clock speed increases for performance gains more than actual IPC increases mainly because there isn't enough time between BD and PD to warrant any major changes that could increase IPC more than a bit.

I am expecting at least Phenom -> Phenom II IPC increase, but again Phenom II had more on the lines of major clock speed increases, first gen Phenom was 2.3GHz (9600), B3/second gen was 2.6GHz (9950) Phenom II was up to 3.7GHz a 1.1GHz increase.

On IPC, it was pretty decent still:

http://www.anandtech.com/show/2702/9

As well it made AMD overclockable again while Phenom I was not so OC friendly.

So on a positive level, at least 15% performance increase but probably 5-10% at most being IPC.

Again, thats my guess based on no info. Once we get more actual information we will be able to better guesstimate.

It would be nice to get that kind of gain, the Zambezi's minus the IPC's are good chips, the extra muscle will come in useful. If it can somehow fit between a Phenom II and Intels first gen it will be a really good progression.

 

[jdwii]

I found this amusing from your link jaydeejohn

"Phil Hughes is a Senior PR Manager at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions."

so what purpose does he serve?

Reminds me of a different guy who was around. 😗

What is the word on the IPC's on Piledriver, can it legitimately be said to be on a par or better than intel first gen? If that is the case I will regard it as a positive step in the right direction.

??? what gen the Pentium 1 or the 386 i just want to be clear, I'm pretty sure at best its going to be on par with the Phenom II in IPC but with a high clock speed also, i wouldn't be surprised to see a 4.2Ghz 8 core piledriver with the IPC of the Phenom II and a 125 watt TDP rating.

 

[g4114rd0]

Where will AMD make 28nm APUs?..."

http://semiaccurate.com/2012/04/20/where-will-amd-make-28nm-apus/

 

[gamerk316]

Yep. Socket 7 was K6, Super Socket 7, or Super 7, was K6-2 and K6-III.

One funny not there, people say Intel changes sockets more often than AMD but if you count them, not including non-proprietary sockets like Socket 7 or prior, AMD has had 11 sockets from their first socket (Super Socket 7) and Intel has had 10 including the unreleased LGA 1150. All are desktop sockets.

Due in part to the very long life of LGA 775. Still, try using a first gen C2D on a first gen 775 chipset. Chipsets compatability matters more then the actual socket when all is said and done.

 

[gamerk316]

So on a positive level, at least 15% performance increase but probably 5-10% at most being IPC.

 

[-Fran-]

I found this amusing from your link jaydeejohn

"Phil Hughes is a Senior PR Manager at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions."

so what purpose does he serve?

Easy:

If he's right about something, then it was in AMD's best interest. If he wasn't, then it's his own fault and not AMD's.

That goes to every person in PR (or even the company).

And the socket thingy... Well, I don't like changing sockets on EVERY god damn generation, but if the change is well justified for ME (notice: different for everyone), I won't say anything about it and get it. In my own experience: AM3 -> AM3+. That gave me SLI support, enough justification for me at least; BD support was a plus.

I've always criticized Intel for "forcing" people to upgrade the whole platform when they come up with something new, but at the same time, it drives new technologies forward I guess...

Cheers!

 

[JAYDEEJOHN]

http://imgs.xkcd.com/comics/percentage_points.png

Sorta like saying system is 95, while cpu only is 77?

 

[dogman_1234]

Reminds me of a different guy who was around. 😗
.

*cough*

Well...he sure hasn't been around since we hosed him with all we have...both Baron and Chad stress tested that guy for sure. I fell bad for this poor soul!

 

[fazers_on_stun]

^ Actually I saw a recent post by JF-AMD over on AMDZone, so he probably still lurks here & there - a marketing director would want to keep his eye on what the great unwashed 😛 populace - or enthusiasts - take yer pick, are thinking..

As for Baron, seems he is now some wanna-be Internet DJ or something, according to a link that Badtrip found. Hmm, 48 yrs old and still hasn't found his calling in life 😀..

 

[-Fran-]

^ Actually I saw a recent post by JF-AMD over on AMDZone, so he probably still lurks here & there - a marketing director would want to keep his eye on what the great unwashed 😛 populace - or enthusiasts - take yer pick, are thinking..

As for Baron, seems he is now some wanna-be Internet DJ or something, according to a link that Badtrip found. Hmm, 48 yrs old and still hasn't found his calling in life 😀..

Wasn't that called the "big second middle age crisis"? The big one is at 40, right? lol

Cheers! 😛

 

[Cazalan]

Wasn't that called the "big second middle age crisis"? The big one is at 40, right? lol

Cheers! 😛

With people potentially living to 200+ by 2050 it's not really mid life anymore.
At the turn of 1900 people only lived to 46.

 

[jdwii]

Check this out!

http://www.engadget.com/2012/04/20/amd-sasa-marinkovic/

This guy is in complete denial.

 

[MU_Engineer]

Yep. Socket 7 was K6, Super Socket 7, or Super 7, was K6-2 and K6-III.

One funny not there, people say Intel changes sockets more often than AMD but if you count them, not including non-proprietary sockets like Socket 7 or prior, AMD has had 11 sockets from their first socket (Super Socket 7) and Intel has had 10 including the unreleased LGA 1150. All are desktop sockets.

AMD has fewer sockets since the Socket 7 days

Intel sockets:
- Slot 1
- Socket 370 (okay, I'll roll all 3 370 variants- PPGA 370, PGA 370, FCPGA2- into one here...)
- Socket 423
- Socket 478
- Socket 775
- Socket 771 (Skulltrail)
- Socket 1156
- Socket 1366
- Socket 1155
- Socket 2011

AMD sockets
- Slot A
- Socket A
- Socket 754
- Socket 940
- Socket 939
- Socket AM2/AM2+
- Socket L1 (QuadFX)
- Socket AM3/AM3+
- Socket FM1

Intel has used one more socket than AMD has. Intel actually had a Timna socket too that I could have counted, although it never made it into production. I count 9 AMD sockets and 10 Intel sockets. What is the 11th AMD socket? Super 7 is Socket 7 with a different VRM scheme and support for a 100 MHz FSB.

 

[Cazalan]

Hence, overpopulation.
Que World War 3

That's one effect, but overpopulation is primarily due to the sheer amount of energy we have at our disposal. More energy = more food production. The population surge started after the discovery of oil.

 

[king smp]

That's one effect, but overpopulation is primarily due to the sheer amount of energy we have at our disposal. More energy = more food production. The population surge started after the discovery of oil.

and the discovery of free pron on internet LOL

 

[notsleep]

i thought about replacing my current 1090t system with a piledriver one but maybe i just need to oc my current system, get an ssd, and get a water cooler instead? that'll probably hold me over for another year or so. 😀

 

[MU_Engineer]

^
MU_E
FM2 is ten for AMD.
11 is still unofficial I think but I hope you will know.

FM2's launch is imminent (June?) so that sounds appropriate to include it.

you could count AM2 and AM2+ as two and not one socket.
same for AM3 and AM3+.
no.?

I wouldn't count AM2/AM2+ as different sockets. They are physically identical with the same pinouts, and any AM2/AM2+ socket is electrically compatible with any AM2/AM2+ CPU. The reason you can't put AM2+ CPUs in all AM2 boards is because the board maker decided to release AM2+ compatible BIOSes. What you can't do is put AM3 CPUs in AM2 boards, but those sockets are not physically identical (AM2/AM2+ has 940 pins, AM3/AM3+ has 938 pins.)

There is a little more difference between AM3 and AM3+ than there was between AM2 and AM2+ as AM3+ has larger pin holes, but I would still consider them the same socket as the pinout is identical. You can also run any AM3 CPU in any AM3+ board, and the reverse may occasionally work depending on the board. However, I chose not to count the electrically dissimilar Socket 370 variants as individual sockets so I will not count the different AM3 variants as different sockets. Also, I am not counting the various electrically somewhat dissimilar LGA775 variants as different sockets even though there is no one board that can support all LGA775 CPUs made due to FSB and VRM changes over the life of the socket.