Piledriver - Page 32 - Overclock.net

Forum Jump: 
Reply
 
Thread Tools
post #311 of 627 Old 04-27-2012, 12:18 PM
Audiophile
 
phillyd's Avatar
 
Join Date: Oct 2011
Location: Louisville, KY
Posts: 7,358
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 194
Quote:
Originally Posted by Tweeky View Post

It's a long time till September 27
I'll be 18!

Ryzen & Vega Enthoo Evolv ATX TG Case Mod Build Log
Audio (Click to show)
Sources: The Spectre | Lenovus Maximus | Huawei Honor 8
DAC's: HiFimeDIY Sabre SPDIF | iBasso D-Zero
Amplification:ASL MG-Head OTL Mark III
Headphones: Sennheiser HD650 | Shure SE215K | Logitech UE6000
Speakers
Home Theater/Lenovus Maximus: Pioneer Elite SC-85 > Music Angel SDXE M9 > Klipsch KG5.2 w/Crites tweeters/crossovers
The Spectre: Adam Ax7
Build Logs (Click to show)Overclocking Accolades (Click to show)

Chimp Challenge Participant 

phillyd is offline  
Sponsored Links
Advertisement
 
post #312 of 627 Old 04-27-2012, 02:14 PM
Overclocker
 
computerparts's Avatar
 
Join Date: Aug 2007
Location: In the shadows
Posts: 1,798
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 153
Quote:
Originally Posted by Tslm View Post

The good news is AMD have been quiet about Piledriver. When AMD are quiet its usually a good sign, and when theyre making noise its usually bad laughingsmiley.gif
Although for their GPUs its the opposite I think

I think they were marketing based on the design and not the final product. Then the review sites took that info and exaggerated it to a great degree in where everyone expected an SB killer. From what I researched, there were 2 main things that ruined BD. First one was the absence of hand tuning where they could have made up for the loss in IPC. Second was the bad power consumption as a result of the faulty fab processing. Either way since they're being extraordinarily quiet about it, it looks like they will hopefully wait for a final product prior to marketing this time.

Quote:
Originally Posted by Masta Squidge go_quote.gif

Clearly not, you don't know how it works. The CPU has to build every individual frame before it sends the data to the GPU to be rendered.
FX-8120 vs. 2500k BenchmarksAMD FX Clubsniper.gif
computerparts is offline  
post #313 of 627 Old 04-27-2012, 04:49 PM
AMD Overclocker
 
MacLeod's Avatar
 
Join Date: Jan 2010
Location: Chattanooga, TN
Posts: 1,353
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 95
Quote:
Originally Posted by Mms60r View Post

I found this comparison of i7 2600, i7 3770, and an fx 8150.

http://www.hardwaresecrets.com/article/Core-i7-3770K-vs-AMD-FX-8150-and-Core-i7-2600K-CPU-Review/1537

Except for the multi threaded tests were the fx was equal to the i7s, the fx was give or take about 15% slower than the i7. Everything I've read about piledriver suggests it will improve on fx performance by 15%. I'm just generalizing here but it seems safe to say to expect piledriver to be on par with those i7s. I have an fx 6100 so for me the fx 8350 will make a nice upgrade sept 27.

This. Bulldozer got a worse than deserved rap mostly from Intel trolls and most people assume Intel is 50% or more ahead in performance. That just ain't the case and since Ivy Bridge is only ~5% ahead of Sandy, AMD's deficit isn't as far back as you think. So while I know Piledriver wont be faster than Intel, I don't think its a stretch at all to say that it can get to within ~5% and all they have to do is be cheaper as in no $300 price points, and AMD would be back on track.

Sent from my HTC Thunderbolt using Tapatalk 2

MacLeod is offline  
Sponsored Links
Advertisement
 
post #314 of 627 Old 04-27-2012, 05:00 PM
New to Overclock.net
 
S.M.'s Avatar
 
Join Date: Jul 2008
Location: Virtual Machine
Posts: 3,188
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 2 Post(s)
Liked: 215
Quote:
Originally Posted by computerparts View Post

Anyone see this yet? Some new roadmaps and looks like AMD will be bringing out the big guns with Steamroller.

Absolutely! AMD reinvented the wheel with a completely unique CIC archtype. The amounts of refinement to this micro-architecture is literally untapped.

An analogy is AMD brought back the 2-stroke engine.

It's a shame AMD doesn't have their own foundry or extensive R&D, because the amount of engineering and money required to refine Bulldozer is something I'm afraid AMD might not have in order to compete with Intels tournout speed. It's another shame that so called "computer enthusiasts" only correlate themselves to end products; They're really more "gamers" than "enthusiasts". If people really cared about the technology, they would realize that AMD is manning-up pretty damn hard.

Fuji X-Pro1 | XF 23mm F1.4 | XF 35mm F2 | XF 18-135mm F3.5-F5.6
Duron 1.6 - XP 2400+ - A64 3000+ DTR - E2180 - E8400 - Q6600 - 1100T - FX8150 - FX8350 - R7 1700
S.M. is offline  
post #315 of 627 Old 04-27-2012, 09:51 PM
 
Projector's Avatar
 
Join Date: Mar 2012
Posts: 985
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Manning up lachen.gif
Every single post I have ever seen come from you has been nothing but amd sunshine and sparkles.
Projector is offline  
post #316 of 627 Old 04-27-2012, 10:20 PM
WaterCooler
 
Mms60r's Avatar
 
Join Date: Apr 2012
Location: Vestal, NY
Posts: 163
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 3
Quote:
Originally Posted by MacLeod View Post

Quote:
Originally Posted by Mms60r View Post

I found this comparison of i7 2600, i7 3770, and an fx 8150.

http://www.hardwaresecrets.com/article/Core-i7-3770K-vs-AMD-FX-8150-and-Core-i7-2600K-CPU-Review/1537

Except for the multi threaded tests were the fx was equal to the i7s, the fx was give or take about 15% slower than the i7. Everything I've read about piledriver suggests it will improve on fx performance by 15%. I'm just generalizing here but it seems safe to say to expect piledriver to be on par with those i7s. I have an fx 6100 so for me the fx 8350 will make a nice upgrade sept 27.

This. Bulldozer got a worse than deserved rap mostly from Intel trolls and most people assume Intel is 50% or more ahead in performance. That just ain't the case and since Ivy Bridge is only ~5% ahead of Sandy, AMD's deficit isn't as far back as you think. So while I know Piledriver wont be faster than Intel, I don't think its a stretch at all to say that it can get to within ~5% and all they have to do is be cheaper as in no $300 price points, and AMD would be back on track.

Sent from my HTC Thunderbolt using Tapatalk 2

I agree for $250 like the 8150 is now I'd happily sign up i7 performance. Truth be told the my 6100 is probably more than I need but I'm kidding anyone I'm getting the 8350 when it comes out.
Now if I could just get my Crosshair V to work with it.
Mms60r is offline  
post #317 of 627 Old 04-27-2012, 11:00 PM
4.0ghz
 
Heavy MG's Avatar
 
Join Date: Mar 2011
Posts: 2,314
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 64
Quote:
Originally Posted by Projector View Post

Manning up lachen.gif
Every single post I have ever seen come from you has been nothing but amd sunshine and sparkles.

Hypocrite much? Nearly every post of yours has been the exact opposite of his.

Heavy MG is offline  
post #318 of 627 Old 04-28-2012, 03:11 AM
 
2advanced's Avatar
 
Join Date: Jan 2012
Location: Central Cali
Posts: 924
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 67
Any truth to this? Looks like Piledriver will carry over some of the issues of BD.

Bulldozer is performing badly mostly because of:
1) Combination of small L1 caches and slow L2 caches. This problem stays with piledriver.
2) L1 instruction cache aliasing problems and write-through L1 caches causing excessive L2 traffic. This problem stays with piledriver
3) The made couple of small mistakes somewhere and it cannot reach the clock speeds it was supposed to reach/what speeds most of it's pipeline would allow. Piledriver will fix this.
4) To get full floating point performance, you have to use AMD's own FMA4 instructions. No legacy software uses those, and not all new software is going to use them because intel is not going to implement those same instructions. Piledriver is going to support Intel Haswell-compatible FMA3, so new code optimized to intel will give full fpu performance on piledriver, no need for amd-specific optimizations.
2advanced is offline  
post #319 of 627 Old 04-28-2012, 09:10 AM
AMD Overclocker
 
MacLeod's Avatar
 
Join Date: Jan 2010
Location: Chattanooga, TN
Posts: 1,353
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 95
Quote:
Originally Posted by 2advanced View Post

Any truth to this? Looks like Piledriver will carry over some of the issues of BD.

Bulldozer is performing badly mostly because of:
1) Combination of small L1 caches and slow L2 caches. This problem stays with piledriver.
2) L1 instruction cache aliasing problems and write-through L1 caches causing excessive L2 traffic. This problem stays with piledriver
3) The made couple of small mistakes somewhere and it cannot reach the clock speeds it was supposed to reach/what speeds most of it's pipeline would allow. Piledriver will fix this.
4) To get full floating point performance, you have to use AMD's own FMA4 instructions. No legacy software uses those, and not all new software is going to use them because intel is not going to implement those same instructions. Piledriver is going to support Intel Haswell-compatible FMA3, so new code optimized to intel will give full fpu performance on piledriver, no need for amd-specific optimizations.

Hard to tell. Those issues with Bulldozer are true but if a bunch of forum nerd know this you gotta think that the AMD engineers who actually have degrees and years of experience will know this as well. wink.gif

So surely theyll fix these issues. Trinity is looking pretty good and while you cant tell exactly how that translates over to Vishera, you have to say that the 10-17% gain that Trinity is getting wouldnt be that much of a stretch to say that Vishera shouldnt enjoy as much but should even be more because they dont have to be concerned with energy efficiency as much as with Trinity.
Quote:
Originally Posted by Mms60r View Post

I agree for $250 like the 8150 is now I'd happily sign up i7 performance. Truth be told the my 6100 is probably more than I need but I'm kidding anyone I'm getting the 8350 when it comes out.
Now if I could just get my Crosshair V to work with it.

Why wouldnt your Crosshair V work with it? Its an AM3+ socket board and Vishera will work in AM3+. May need a BIOS update but that should be it. That was the whole reason I bought this Sabertooth (well aside from its sheer awesomeness) so I could drop in Piledriver when it released.

MacLeod is offline  
post #320 of 627 Old 04-28-2012, 10:07 AM
Classic Otaku
 
Join Date: Jan 2011
Posts: 5,180
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 408
Quote:
Originally Posted by 2advanced View Post

Any truth to this? Looks like Piledriver will carry over some of the issues of BD.
Bulldozer is performing badly mostly because of:
1) Combination of small L1 caches and slow L2 caches. This problem stays with piledriver.
2) L1 instruction cache aliasing problems and write-through L1 caches causing excessive L2 traffic. This problem stays with piledriver
3) The made couple of small mistakes somewhere and it cannot reach the clock speeds it was supposed to reach/what speeds most of it's pipeline would allow. Piledriver will fix this.
4) To get full floating point performance, you have to use AMD's own FMA4 instructions. No legacy software uses those, and not all new software is going to use them because intel is not going to implement those same instructions. Piledriver is going to support Intel Haswell-compatible FMA3, so new code optimized to intel will give full fpu performance on piledriver, no need for amd-specific optimizations.

I don't know how Piledriver is going to handle them. Honestly, on the last point, misconception is that Intel dropped ball on AMD with FMA3.

That is not true at all. Intel announced they will be using FMA3 at late 2008, to match AMD's FMA3. AMD announced at mid 2009 they will be using FMA4 instead. It seems that these two companies are always in disagreement.

It is correct that upcoming Trinity and Vishera will be using FMA3 in additional to FMA4. FMA3 allows easier coding, while FMA4 allows more flexibility. At this point, unless Intel adopts FMA4, it looks like FMA3 will become more mainstream.

Chimp Challenge Participant 

trumpet-205 is offline  
Reply

Quick Reply

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off