AMD Piledriver rumours ... and expert conjecture

Page 210 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.
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 ...
 
^^^^^ How are any of the above posts constructive expert conjecture or rumors?
Just sayin'...

AFAIK, the only thing we know from AMD slides is that AMD estimates Piledriver to be 10% better than BD based Zambezi/FX . FWIW, one test of Zambezi on Win 8 vs. Win 7 did show ~5% improvement for Zambezi/FX-8150. So it we see a 15+% improvement over the current FX CPU performance, it's a start.

http://www.pcstats.com/articleview.cfm?articleID=2622
 
"AMD to update Trinity to Trinity 2.0"
http://semiaccurate.com/2012/08/06/amd-to-update-trinity-to-trinity-2-0/

AMD is planning a Trinity refresh in 1H/2012

Thought we were already in 2H/2012? *checks calendar* Yup.. Guess Charlie meant 1H/2013 😛 Anyway Kaveri now pushed out to at least a year from now.

Also, from http://money.msn.com/business-news/article.aspx?feed=Bcom&date=20120806&id=15416380

Advanced Micro announces private offering of $300 million of senior notes Co announced that it intends to commence a private offering, subject to market and other conditions, of $300 million aggregate principal amount of senior notes due 2022. AMD intends to use the net proceeds for general corporate purposes and working capital, which may include the following: (i) the repayment or repurchase of some or all of its outstanding 5.75% Convertible Senior Notes due August 2012.

Personally I'd be a bit worried having to borrow money to repay a due loan - this is the sort of behavior people way over their budget use to get by - borrow from one credit card to pay another.. However S&P500 reaffirmed their BB- rating on AMD and said it was a sound business move. Compared to defaulting or whatever the alternative is, I guess so.
 
 
^ Well now it looks like some $500M, not $300M, and at 7.25% interest unsecured senior notes (which means they are senior to common stock, in the event of a bankruptcy, so ordinary stockholders such as one guy who used to post here would be SOL most likely).

Apparently not only is the $$ going to be used to pay off the cheaper, immediately-due senior notes (along with cash from one report), but according to one report (http://money.msn.com/business-news/article.aspx?feed=Bcom&date=20120806&id=15419397) AMD owes a lot of $$ to Global Foundries for the 28nm waiver of exclusivity.

This IMO is ridiculous. How did AMD's senior mis-management allow the company to get into such a position with GF? Seems like every quarter AMD is taking a hit on payments to GF. Thought this was supposed to be a sale of AMD's fabs to ATIC, but it seems like AMD practically gave the fabs away and now is paying $$ for ATIC to keep them..

AMD is ripe for the picking if ATIC wants to get them for the proverbial dime in the next couple years, after they've bled the company dry. I think the BoD should sue Hector Ruiz for incompetency if not outright fraud. Get some of that $16M a year pay back from the guy..

As for BD optimized for server vs. desktop, first of all I don't think it's doing that well in the server segment. I didn't see anything about AMD recouping marketshare in their Q2 report, and we all know desktop tanked bigtime. IIRC Intel said their new SB server lineup was doing well. And concerning 22nm, I have no idea when GF or TSMC will get there - probably not until 2014 given the amount of trouble they've had on 28nm with yields..
 
it was the opposite, iirc a lot of people were raving how great it was going to be. the architecture preview seemed picture perfect, etc.
btw, how will amd market piledriver anyway? during bulldozer advertising they pitched a high performance desktop cpu that beats intel'x top core i7 980x.
a while ago they stated that they aren't chasing after intel anymore (cpu performance). more over, they will have to pit their top pd against the 3960x if they do the same thing they did with bd.
so what will be the attraction(!) piledriver? "up to 15% better application performance"? rcm? "phenom ii class cpu performance"? "4+ ghz 8 core cpu"(redux)?
trinity at least has the igpu and hsa going for it.
 
first leak is out. http://wccftech.com/amd-fx8300-vishera-x86-piledriver-8-core-cpu-pictured-benchmarked/

3.3 ghz on the 8300.

the aida memory test was done at 4.8 ghz so its a bit hard to see how much improvement there is on the memory end.
 
AMD Athlon II X4 640 (Propus 4c) @3 GHz, DDR3-1333, 2 MB L2: 3.38
AMD Phenom II X4 945 (Deneb 4c) @3 GHz, DDR3-1333, 2 MB L2, 6 MB L3: 3.56

3.56 to 3.38 = 5.05 % diff in multi

AMD Athlon II X4 645 (2010, Propus): 0.83
AMD Phenom II X4 980 (2009, Deneb): 0.88

0.88 to 0.83 = 5.68% diff in single (both at 3Ghz)

That's according to Tom's own Cinebench testing and my math... Is it the universal math or my own? hahaha

Also, how representative is Cinebench regarding the larger picture? How does the higher access time for L3 in PD will compare to the K10 gen? Will the larger size affect it as well?

So many questions, so little answers!

Come on Toms, some leaks! xD

Cheers!
 
http://www.anandtech.com/show/4955/the-bulldozer-review-amd-fx8150-tested/6

L3 more than 2x the speed of main memory. I don't know where you get your data.

It will depend on a lot of factors such as which CPU, the IMC and what memory as well the mobo can affect latency.

For example, Intels IMC is typically much faster than AMDs as is their cache.

DDR3 1866 CAS9 will be much fastert than DDR3 1333 CAS9.

So multiple factors will change the results and every system will be different.

L3 overall has very little benefit to most applications but is useful when utilized, i.e. Intel method of using it to store all L1 and L2 instructions so it doesn't have to go to main memory or HDD/SSD for it.
 
If piledriver does not beat the Phenom x6 1100T all around(including single threaded tests) it will be a failure, and the more Amd holds off the release the more of a failure its going to be.
I remember just 3 years ago i recommend Amd over Intel 85% of the time, but its getting very difficult now I still recommend Amd over Intel on 650$ Laptops or cheaper and i still recommend their video cards but their desktop CPU's are failures now a days. Whats odd is it was the other way around back in 2008-2009 i told everyone to stay away from Hot Amd laptops and get Nvidia over ATI.

I think the number one problem with Advance micro devices is not caring enough about their engineer department and what they have to say, i'll never forget what some engineer said when making Bulldozer "well its not meant to be a high-end design") Then what the hell was all that money/time spent on making a mediocre design i was pretty sure the Phenom II was a medico design?

When you spend 3+ years on something and spend a lot of money this company clearly didn't have you expect to see results i personally think if they didn't buy ATI this company would be bankrupt by now. APU's and graphic cards is all this company makes that's worth buying, and in some rare occasions their Opterons.

When i go back and read about this company in the 90's it really surprises me how this company Made the K7 and really showed Intel up when they made the first 1GHZ CPU, not to mention they were always 1 to 2 generations behind Intel before the K7. Now we look at this company today and their almost in the same boat, i would love to say its not because Amd is bad its because Intel is so good but i would be lying to myself and to everyone reading this. It was Amd whole got rid of their good old boys in their engineer department and it was Amd whole got rid of hand craft designs, This New CEO is making bad and good decisions and one good decision is getting rid of most of the marketing department another is bringing back one of the best engineers in history.

I really hope Amd starts caring more about their engineers who are the most important people at Amd.

 
pd doesn't look like a failure to me.
amd 'topped out' with phenom ii's cpu performance. with bulldozer, they took a new direction which didn't start where phenom ii ended, in terms of cpu performance. imo bd's performance peak is quite a bit lower than phenom ii's - which was a mature lineup. now amd is slowly climbing up the cpu perf ladder again with pd (higher perf than bd but not necessarily phenom ii). besides, high cpu performance isn't all that important anymore..... for most pc users.
may be steamroller is where amd will get substantial cpu performance increase, disregarding other factors and only comparing among amd cpus.
however, i think it's better to keep one's expectations low and not believe hypes too much. :)
 
Bulldozer was aimed for servers at the sweet spot where intel sells like 80% of their servers. AMD could never hoped to have beaten intel in that front. Bulldozer's delays were related to glofo IIRC. Not sure about the rest. My guess is expectations weren't hit and AMD was hit too hard with Intel's dirty play and had no money to change strategy at that point. It forever for intel to even pay the money that the court demanded.
 


IIRC Intel paid AMD $1.25BN in November 2009, about 2 years before BD was released. Unfortunately it seems AMD used most of that cash to pay off their first iteration of the senior notes debt, and maybe some to ATIC for the GF spinoff, and probably some golden parachute for Hector Ruiz as well.

If AMD had been able to get BD out the door back in 2009 (which IIRC was the first targeted release date), it might have reversed AMD's fortunes at least in server since it would have been competing with Nehalem or prior Xeons.
 



Workstation APU Oxymoron.

AMD has lost their marbles. The marketing department doesn't even know what a "Workstation" is.
 
Status
Not open for further replies.