AMD Piledriver rumours ... and expert conjecture

Page 120 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
Reynod

Reynod

Administrator
Mar 26, 2007
14,921
16
42,215
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 ...
 
Sort by date Sort by votes
Hmm, didn't read that far, since I don't have a 990X or plans to buy one, unless I could find an 1100T to go with it 😀...
Click to expand...

Lol best chip to go with a 990fx, Boy i was so sad when i saw the 8150 benchmarks(more then most here i'm sure) i was to excited and rushed to toms conclusion at 12.01Am i think i almost cried....well ok i did cry. So what did i do i bought a processor that was released in 2010 DEC to go with my new sabertooth board that i got in early july, that was for the 8150 i also kept 350$ saved just for the 8150 but hey, i was able to keep 150$ i guess.
 
Upvote 0 Downvote
you do realize that's what FMA4 support does? Do you really think that benchmark used FMA4?

What have we been talking about the last 3 pages?

As for your other rant ... lol is all I can say. Didn't read any of what I said did you or maybe you think since its in the past, it didn't happen, only happened the way you want to see it that way you can have something to say.

Take an intro to accounting and you will see how a 2.5B loan lowers net worth by exactly 2.5B. Its not rocket science, its 3rd grade math.
Click to expand...

They still overpaid for ATI. At the time, the 8800GTX pounded the HD2900. The HD2900 got better after driver updates but still was weak in comparison. AMD overpaid for ATI.

Even if Intel paid that much, as they could easily afford it compared to AMD, it would have been too much.

As for the whole "white knights" thing, give it a rest. People can think differently. Thats what I don't get. If we don't agree, we are fanboys or "white knights" I never defended anything just made my feelings about the issue.

As for AMD/Intel, honestly AMD did as much damage to themselves as Intel did to them. Intel didn't put the TLB bug in Barcelona, causing a massive recall in the server channels nor did AMD put the SATAII bug in the P67 chipset. Intel didn't force AMD to buy ATI and overpay for them, AMD made the choice. Intel didn't force them to go FAB-less, in fact Intel originally wanted to stop it. Intel did not force AMD to create modules for BD.

Every choice has been AMDs for many years past the now long settled issue.

So now we need to just all drop it and focus on innovation. I may not trust every AMD marketing slide like some, but thats because I don't trust any slide from anyone. I only trust benchmarks from third party sources 100%. I wont say "leaked" benchmarks are fake (mainly because we do not know for sure) but I will take them with a grain of salt.

Lets move on people, nothing to see here.
 
Upvote 0 Downvote
8800GTX - 1 of the top 10 GPU's ever made in history.
back in the days (a few years ago really) I had one paired with a C2D E8300, I thought I was da' man..
😀
Click to expand...

And I had a HD2900Pro 1GB with a Q6600 laughing at peopler with a E8400 @ 4GHz and a 8800GTX not being able to max out GTA IV.

That was it though. Most other games ran better on the 8800 than my HD2900Pro......
 
Upvote 0 Downvote
8800GTX - 1 of the top 10 GPU's ever made in history.
back in the days (a few years ago really) I had one paired with a C2D E8300, I thought I was da' man.[\b].
😀
Click to expand...


2.gif


finally i managed to post the image😀

btw new hairstyle mal😛
 
Upvote 0 Downvote
I just gave the TRUTH and then you fleshed it out some more. :sol:

BTW, With Larry Ellison now admitting he lied about Itanium's roadmap, does that mean you have sworn off SPARC, what with you being the great moralist and all? :lol:
Click to expand...

What you said was true, but only literally, this I called it a loaded statement. You left out the context and by doing that would lead someone to thinking the code compiled by the Intel Compiler would run the same on both Intel and non-Intel CPU's, which it wouldn't. He asked specifically about SSE code, the Intel Compiler would run SSE4.2 on the Intel CPU and SSE2 on the non-Intel CPU even if it the non-Intel CPU supported SSE4.2. GCC compiled to run SSE2 would run SSE2 on both CPUs. GCC compiled to run SSE4.2 would also run SSE4.2 on both CPUs. Chances are the application compiled on GCC (or MSC) running SSE4.2 would run faster on the non-Intel CPU then the same application compiled with the Intel Compiler running SSE2 on the non-Intel CPU. Unfortunately doing that would prevent the GCC Application from running on anything that doesn't support SSE4.2 and limit the market share for that application.

Larry Ellison is Oracle not Sun (although Oracle now owns Sun). Oracle is worse then Microsoft, Intel and Apple when it comes to Corporatism and screwing it's customers. They don't deal with consumers so most people here wouldn't know how bad they can get. Ex: Prior to the merger all Solaris security patch's, SUN firmware / obp updates and additional companion software was freely available from Sun. After the merger, they are still freely available but only if you have a My Oracle Support (MOS) account, only way to get a MOS account is to have a support contract with Oracle. Those are $10K USD a year, for a single non-Oracle desktop PC. For Oracle server HW their ridiculously expensive, your looking at 10~20K per year per socket for the minimum HW replacement warranty. For on-site server it ratchet's up into the stratosphere. Their offering the OS for free, but making a killing on support agreements.

Also, Sun =/= SPARC. SPARC was originally developed by SUN but has since been made into a freely available architecture.

http://en.wikipedia.org/wiki/SPARC

Anyone can design a SPARC CPU if they desired, no licensing fee's required. SPARC International is the organization responsible for publishing the SPARC ISA. SUN even released the design for the T1 / T2 so that anyone could make one. It's an open CPU architecture with SUN and Fujitsu being the major manufacturers.

[
In March 2006, the complete design of Sun Microsystems' UltraSPARC T1 microprocessor was released-in open-source form at OpenSPARC.net and named the OpenSPARC T1. In 2007, the design of Sun's UltraSPARC T2 microprocessor was also released in open-source form, as OpenSPARC T2.
Click to expand...

Finally, I'm not a moralist at all. If JS and others would of just admitted that they were fine with Intel participating in unethical business practices then the debate would of ended right there. I don't object to it, business's will be business's. What I do object to is historical revisionism and people pretending it didn't happen and isn't happening right now. AMD just like SUN has been successful at IT but that success was mismanaged. They both failed to notice how critical software developers are to the success of any product. You can make the most awesome piece of HW in existence but it's merely a pile of tin if developers aren't on board making software for it.
 
Upvote 0 Downvote
Damn... BD arch really depends on re compiles and some re coding, lol

Cheers!
Click to expand...

It was a giant cluster FCK between the two companies.

At first Intel was going with three component fused instructions (FMA3). This is where you do a single operand (FM) but with three data components being worked on. Then they announced they were instead going to do a four way FM instruction (FMA4). AMD then switched their BD design's (it wasn't near finished yet) to use Intel's FMA4 (Intel produced both of those extensions btw). Later Intel announced for SB they were going with FMA3 instead as they found that applications rarely used the fourth data component. AMD was too far along in the BD design phase to alter it's CPU to use FMA3.

The kicker is the two instructions are not compatible, if you compile it for three way fused (FMA3) then it won't run on a four way CPU, if it's made for a four way (FMA4) then it won't run on a three way CPU. So for now, FMA4 might as well not even exist, same with XOP. Within a few years they will of become merged into a single extension, most likely FMA3 will be added to AMD's CPU line.

AVX is also a problem, Intel's implementation of AVX isn't exactly as published. This caused AMD's implementation to be non-functional. Remember Intel created the AVX extension, AMD can only implement what their given. Just like FMA3/4, AMD's implementation of AVX is basically dead right now. In future processor's AMD will most likely modify it to match whatever Intel is implementing.
 
Upvote 0 Downvote
AMD to Formally Announce "Trinity" Fusion APUs for Notebooks on May 15..."

http://www.xbitlabs.com/news/cpu/display/20120405175642_AMD_to_Formally_Announce_Trinity_Fusion_APUs_for_Notebooks_on_May_15.html
Click to expand...

According to performance benchmarks conducted by AMD, the Trinity 35W APU with Piledriver-class x86 cores will provide 25% better x86 performance compared to Llano 35W (with K10.5+ "Husky" x86 cores) based on results obtained in PC Mark Vantage Productivity benchmark. AMD also claims that Trinity 35W will offer up to 50% better result in 3D Mark Vantage performance benchmark compared to Llano 35W
Click to expand...

Interesting. 25% CPU, 50% GPU at 35W. From AMD so I would be careful, most likely 15% / 35% but really depends on how you measure it. I believe the 15~25% because RCM can reduce power usage by 10~25% and thus you can upclock it. If BD was right on par / slightly under K10 per core then upping the clock by 25% would resulting in that kind of gain.
 
Upvote 0 Downvote
It's loaded because you were lying by omission.

Nice strawman, I never argued who owned Sun, only that Sun =/= SPARC and that Sun now is not Sun prior to being bought.
Click to expand...

*chuckle*

Palladin is rather a sharp thinker. Don't hate him for it, remember he was born that way.

I think people use strawman arguments without realizing what they are doing, Palladin. Most people don't think very precisely, not at all. This doesn't mean they aren't intelligent, it means their thinking works differently. I say this without accusing you of being unaware but rather so that others might realize that when they argue with you, they are arguing with someone they aren't used to arguing with.

Notwithstanding that a few harsh words have been written above, I would like to say that I feel this discussion of certain aspects of *software engineering* has been very important. Unfortunately it seems that some do not feel this way however I believe that when taken in proper consideration it shall seem quite unreasonable for one to not consider how code is generated for a program, especially when such a program is used to measure performance of a cpu architecture.

Designing and writing code has been probably the most fun thing I have ever done. Please remember that all any cpu ever does is execute instructions generated by an assembler or a compiler, one after the other.
 
Upvote 0 Downvote
in this wiki article it makes it seem like a year after the apu version is released the cpu version will be released (for steamroller and excavator)is that true?




3rd Generation

As of 2011 AMD mentioned (by name) a 3rd generation Bulldozer-based line for 2013,[44] with working title Next Generation Bulldozer, on the 28nm manufacturing process.[45]

On 21 September 2011, leaked AMD slides indicated this 3rd generation of Bulldozer core was codenamed Steamroller[46][47]. The focus of Steamroller is for greater parallelism.[31] It would initially be incorporated into specific desktop and notebook markets for 2013[48]:

3rd Generation A-series APU - Desktop Budget and Mainstream market (??? platform): The Trinity Fusion APU line will be replaced by Kaveri Fusion APU line as the 3rd generation A10-, A8-, A6-, and A4-series for the desktop market.
3rd Generation A-series APU - Notebook Mainstream and Performance market (Indus platform): Will be the same as mentioned in Desktop Budget/Mainstream market. The FCH chipset will be codenamed Bolton.

According to AMD's 2012 Financial Analyst Day[49], the FX-series and Opteron lines will receive an improved version of Steamroller in 2014, a year after Kaveri Fusion APU. Both processor lines will continue to use the 2nd Generation Piledriver-based cores, thereby retaining use of Socket AM3+, Socket C32, and Socket G34 throughout 2013.
[edit] 4th Generation

On 12 October 2011, AMD revealed Excavator to be the codename for the 4th generation Bulldozer core, scheduled for 2014 release.[50] Excavator will initially be implemented in the 4th Generation A-series Fusion APU line in 2014, while a revised version will be adopted in 2015 for the FX-series and Opteron lines.[51]
 
Upvote 0 Downvote
According to performance benchmarks conducted by AMD, the Trinity 35W APU with Piledriver-class x86 cores will provide 25% better x86 performance compared to Llano 35W (with K10.5+ "Husky" x86 cores) based on results obtained in PC Mark Vantage Productivity benchmark. AMD also claims that Trinity 35W will offer up to 50% better result in 3D Mark Vantage performance benchmark compared to Llano 35W
Click to expand...

So, in a BEST CASE situation where performance scales, you get 25%. Which, for a 4 module chip, makes sense. On non-scaling SW though, you can assume much smaller gains, closer to 10% or so, which is what I've been expecting.

Remember, looking only at best cases, BD > SB.
 
Upvote 0 Downvote
According to performance benchmarks conducted by AMD, the Trinity 35W APU with Piledriver-class x86 cores will provide 25% better x86 performance compared to Llano 35W (with K10.5+ "Husky" x86 cores) based on results obtained in PC Mark Vantage Productivity benchmark. AMD also claims that Trinity 35W will offer up to 50% better result in 3D Mark Vantage performance benchmark compared to Llano 35W
Click to expand...

So, in a BEST CASE situation where performance scales, you get 25%. Which, for a 4 module chip, makes sense. On non-scaling SW though, you can assume much smaller gains, closer to 10% or so, which is what I've been expecting.

Remember, looking only at best cases, BD > SB.
Click to expand...



do you mean piledriver or bulldozer? because you put bulldozer would be better than sandy bridge.
 
Upvote 0 Downvote
...

So now we need to just all drop it and focus on innovation. I may not trust every AMD marketing slide like some, but thats because I don't trust any slide from anyone. I only trust benchmarks from third party sources 100%. I wont say "leaked" benchmarks are fake (mainly because we do not know for sure) but I will take them with a grain of salt.

Lets move on people, nothing to see here.
Click to expand...
ya, lets all forget about the past and pretend it didn't happen ...

here is something similar to that line of thinking.

http://www.cbsnews.com/2100-202_162-1124255.html
 
Upvote 0 Downvote
you do realize that's what FMA4 support does? Do you really think that benchmark used FMA4?
Click to expand...

And what does that have to do with JF and the likes of Baron, etc who spent pages in the BD thread a year ago, touting the advantages of the FPU ability to process two 128-bit instructions on legacy code? Hint: "legacy" doesn't mean FMA4..

Don't feel like dredging up old posts outta that BD thread (assuming it hasn't been deleted anyway), but the point they were making a year ago, is that a 4-module/8-core BD would have 8 128-bit FPUs for processing legacy code, whereas SB had just four 256-bit units and if it needed to process 128-bit instructions it had to steal resources elsewhere and thus would not be as efficient as the BD design.

The benchies tell a different story however.

What have we been talking about the last 3 pages?
Click to expand...

I dunno - I usually just skim your posts as they mostly appear to be whining over evil ogre Intel and its assorted past sins. Gets rather tiresome frankly - this thread is supposedly about PD 😀..

As for your other rant ... lol is all I can say. Didn't read any of what I said did you or maybe you think since its in the past, it didn't happen, only happened the way you want to see it that way you can have something to say.

Take an intro to accounting and you will see how a 2.5B loan lowers net worth by exactly 2.5B. Its not rocket science, its 3rd grade math.
Click to expand...

Whether AMD forked over cash or had to indebt itself with the senior notes is immaterial to their paying too much $$ for ATI. Money spent or funds obligated via a loan is still subtracted from the bottom line net worth.

You apparently don't understand basic shopping principles - a 2-day old fish selling for $1 a pound will be selling for half that a day later 😛..
 
Upvote 0 Downvote
ya, lets all forget about the past and pretend it didn't happen ...

here is something similar to that line of thinking.

http://www.cbsnews.com/2100-202_162-1124255.html
Click to expand...
:lol: Impressive that you could bring about the invoking of Godwin's Law.

Don't let anyone ever tell you that your perspective is in the toilet.
 
Upvote 0 Downvote
:lol: Impressive that you could bring about the invoking of Godwin's Law.

Don't let anyone ever tell you that your perspective is in the toilet.
Click to expand...

Actually, that link does have a sense of being here. Even if it invokes Godwin's Law.

"With enough BS, you can convince people of something that didn't happen, but actually did happen and should always be remembered".

On all the pages of the discussion, I don't remember any of the Intel kids said "yeah, they totally screwed with my hard earned money, but I don't care cause I love Intel", hahaha.

Cheers!
 
Upvote 0 Downvote
Whether AMD forked over cash or had to indebt itself with the senior notes is immaterial to their paying too much $$ for ATI. Money spent or funds obligated via a loan is still subtracted from the bottom line net worth.

You apparently don't understand basic shopping principles - a 2-day old fish selling for $1 a pound will be selling for half that a day later 😛..
Click to expand...
you apparently really believe that ATI would have hemoraged 2.5 billion dollars in 6 months when just a quarter before the buyout they posted a profit of 31.9 M?

http://www.reghardware.com/2006/06/30/ati_results_q3_06/

and before that, 34.1M?

http://hothardware.com/News/ATI-Reports-Results-for-Second-Quarter-of-Fiscal-2006/

People still fail to see that the 2.5B loss in ATI value was the loan that was taken out by AMD. ATI didn't take the loan prior to being bought, AMD took the loan. When the deal was finished, they move the loan AGAINST ATI assets instead of keeping it against AMD assets.

Had AMD not bought out ATI, their value would not have been cut in half in 3-6 months since they were still posting profits every quarter.
 
Upvote 0 Downvote
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom