Overclock.net › Forums › Industry News › Hardware News › [ExtremeTech] Analyzing Bulldozer: Why AMD’s chip is so disappointing
New Posts  All Forums:Forum Nav:

[ExtremeTech] Analyzing Bulldozer: Why AMD’s chip is so disappointing - Page 23

post #221 of 292
Quote:
Originally Posted by jivenjune View Post
While I agree that Bulldozer has potential in multi-threaded applications, so does the Thuban at a much more agreeable price point.
BD is a new CPU-uarch design philosophy for the massively parallel future.
think radeon cores (smaller) vs. nV cores (bigger).
very much of the BD wafer is on cache, not execution units/pipes.

this is the first step forward, leaving the big/fat cores era behind,
while utilizing better instruction sets w/ narrower cores; (xop/avx/fma).

fadd/fmul/fmisc are all deprecated (although fma can still do all of those).
just compare a k10 fpu vs. a BD fpu; k10 fpu is fully featured for legacy software.
while BD relies on fma to attain better performance w/ a narrower fpu unit.
all in all, BD is already there w/ the right core,
something intel is still working on and will only become available w/ haswell.

expect for these legacy bentmarks to be re-compiled for fma when haswell launches.
just to avoid the "hasfailed" moniker when its fma3 fpu unit runs legacy code.


DSC04119 by ᵿ, on Flickr
post #222 of 292
Quote:
2. X86 is a standardized architecture, while you can cater to a chip people are only going to optimize so far and the fact of the matter is that hyperthreading adapts better to current optimization standards than modules
Ehh...

x86 is not standardised. I have no idea where you got that idea from. And hyper-threading really doesn't "adapt" better to current optimisation standards... really, it does a lot to reduce performance penalty from pipeline stalls, so it's probably more useful for running ill-optimised software than adapting to any "current" standards, though HT is usually not something you generally can or would optimise for (unless you consider multi-threading optimisation).

Gentoo users rejoice.
post #223 of 292
one more thing re: power draw of bulldozer;

whats the difference in the test platforms between AM3+ and Z68 in this setup?

w/c one should draw more power just by looking at it?
amazing scientific review work, aint it.

AM3+


Z68
post #224 of 292
Quote:
Originally Posted by DigiHound View Post
I'm the author of the piece in question. 2600K scores are lower because I turned Turbo Mode off. Turbo Mode increases the clock speed of the chip by 400MHz, or by 11%. The point was not to penalize any processor, but to gain a clearer picture of Bulldozer's IPC.
No it does not when all threads are utilized. It bumps 400 MHz when only 1 core is actively being used and the supposed 5,91 Cinebench score was obviously done at Multi-Thread setting which would only apply a 100 MHz bump on each core.

Also, heres proof:
Quote:
Originally Posted by sintricate View Post
I'm a bit late but I had nothing else to do just now so I ran Cinebench to compare results with that chart with I found to be incorrect.

I turned turbo off and set my 2600k to stock speeds and got the following results:

Multi-core: 6.64 pts
Single-core: 1.39 pts

Most of you won't care but that chart rubbed me the wrong way.
See, I knew it.
The 2600k x-CPU Cinebench result in that article is totally off.

Quote:
Originally Posted by alexmaia_br View Post
You know, I get people have different opinions. That's one awesome aspect of forums.

I know people tend to stick to their beliefs very strongly, that's only human.

I do NOT, however, have a lot of patience for preachers who want to decide how others should feel about their choices.

I got a bulldozer too, and guess what? I am happy. I couldn't care less about your numbers.

Want some reasons?
I was bored with intel, I wanted to try something new, I found the idea of a new architecture intriguing.

Does it deliver mind blowing performance? No, nor do I require it.
Could I get a 2600k for the same price? Perhaps, but I didn't feel like it.

I got a new processor that is, to me, intriguing. I'm having fun setting up my own config, I know I'll play my games easily. Really, I get that you don't like the bulldozer, but can you get that other people like it? Even better: with uncalled sarcasm towards other people?

As for $$ loss, pfft. I don't care. One day having fun with a plataform I'm curious about is more than worth the difference TO ME. Again, TO ME.

you´re unhappy with bulldozer, we get it. You offend people, we .... we'll, we get it too, that's your limitation.

On my side, I respect the intel plataform a LOT, I like to read about it, even to try it on my friends houses. Hell, I'm looking forward to ivy-E in the future. But I was curious about BD, and here I am.

conclusion: chill man. Pax. you're happy, we're happy. $$, performace are not the sole reasons of choice. There's fun factor as well, among other things (curiosity etc).
I'm glad you're having fun with your bulldozer, each to his own.
Yet theres a part in your post that somehow contradicts yourself.

You said you dont need the performance of bulldozer, yet I see you having a HD 6990 in your sig rig which indicates that you actually ARE after big performance in games.

Now the funny thing is that despite saying you wouldnt need the performance that Bulldozer delivers, you actually need even MORE than that, because your FX-8150 will bottleneck the heck out of your HD 6990 in games.

In all honesty, I don't see the point of buying a CPU that will not be able to fully utilize your GPU, thats wasted resources and more importantly, wasted money since you will essentially end up with much lower performance that could most probably be achieved by an HD 6970. (I'm not exaggerating here, my system performs about as good as a much weaker GTX 580 coupled with a strong Sandy Bridge)

Don't get me wrong, its fine if you're having 'fun' and are happy with your CPU and thats pretty much the most important part, but I don't see the point in wasting money on something that's essentially throttling your system.

I realize I'm currently in the same boat, BUT the last thing that I upgraded was actually the GPU and not the CPU since I knew I will be going SB / SB-E soon anyway. So thats a bit of a different story here.
Edited by toX0rz - 10/27/11 at 12:18am
Rebel
(13 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i5-2500k 4,6 GHz @ 1,32 V AsRock Z68 Extreme3 Gen3 PNY GeForce GTX 660 Ti SLI @ 1228 / 3312 MHz 8GB G.Skill Ripjaws DDR3-1600 
Hard DriveOSMonitorPower
Samsung 1TB + 500GB Windows 7 Ultimate 64 Asus VH22H Xilence XQ 850W with a Scythe Kama Flex 
CaseMouse
Aerocool VS-9 Xmas Edition :P Logitech G500 
  hide details  
Reply
Rebel
(13 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i5-2500k 4,6 GHz @ 1,32 V AsRock Z68 Extreme3 Gen3 PNY GeForce GTX 660 Ti SLI @ 1228 / 3312 MHz 8GB G.Skill Ripjaws DDR3-1600 
Hard DriveOSMonitorPower
Samsung 1TB + 500GB Windows 7 Ultimate 64 Asus VH22H Xilence XQ 850W with a Scythe Kama Flex 
CaseMouse
Aerocool VS-9 Xmas Edition :P Logitech G500 
  hide details  
Reply
post #225 of 292
Quote:
In all honesty, I don't see the point of buying a CPU that will not be able to fully utilize your GPU, thats wasted resources and more importantly, wasted money since you will essentially end up with much lower performance that could most probably be achieved by an HD 6970. (I'm not exaggerating here, my system performs about as good as a much weaker GTX 580 coupled with a strong Sandy Bridge)
That's a rather silly statement. Do you know what games he plays? There's no way to achieve a "true" balance, one component will always "bottleneck" your system, and the worst thing is under different applications it's going to be a different component. There's just too much variance in the CPU vs GPU requirements in different games, you can't build a "bottleneck free" system unless you tailor it specifically for one and only one game (and assuming that the game doesn't have sections where the GPU vs CPU usage ratio varies, which just doesn't happen). While that system probably leans towards being CPU bound in most games, to say that it's CPU bound to the point of "wasting resources", especially without knowing what games plays is a sweeping assumption.
post #226 of 292
Quote:
Originally Posted by toX0rz View Post
No it does not when all threads are utilized. It bumps 400 MHz when only 1 core is actively being used and the supposed 5,91 Cinebench score was obviously done at Multi-Thread setting which would only apply a 100 MHz bump on each core.

Also, heres proof:

See, I knew it.
The 2600k x-CPU Cinebench result in that article is totally off.



I'm glad you're having fun with your bulldozer, each to his own.
Yet theres a part in your post that somehow contradicts yourself.

You said you dont need the performance of bulldozer, yet I see you having a HD 6990 in your sig rig which indicates that you actually ARE after big performance in games.

Now the funny thing is that despite saying you wouldnt need the performance that Bulldozer delivers, you actually need even MORE than that, because your FX-8150 will bottleneck the heck out of your HD 6990 in games.

In all honesty, I don't see the point of buying a CPU that will not be able to fully utilize your GPU, thats wasted resources and more importantly, wasted money since you will essentially end up with much lower performance that could most probably be achieved by an HD 6970. (I'm not exaggerating here, my system performs about as good as a much weaker GTX 580 coupled with a strong Sandy Bridge)

Don't get me wrong, its fine if you're having 'fun' and are happy with your CPU and thats pretty much the most important part, but I don't see the point in wasting money on something that's essentially throttling your system.

I realize I'm currently in the same boat, BUT the last thing that I upgraded was actually the GPU and not the CPU since I knew I will be going SB / SB-E soon anyway. So thats a bit of a different story here.

I don't see a contradiction. I was curious about the amd plataform. That is, correct me if im wrong, the scorpion plataform. It is the best amd has to offer (or at least the top of their line).
The point is to experiment with a completly different plataform.
Also, there's nothing written in stone for me. No, I don't care about 10fps, as long as my games are playable. As long as I can enjoy my system, I'm happy.
If it ever comes to a point where I cannot because, lets say, of the bulldozer, off it goes.

There is a limit, then, to my "I don't care" attitude. I honestly belive it will not happen, and I'll be very happy until ivy-e comes around at the end of the next year.

One last thing about me. If I'm wrong, I'll be the first to come here and say it out loud =)

I'm not here to say Bulldozer is teh power, get it now. Nowhere I am defending a rational performance argument. What I am defending is that there are other possible reasons to justify my decision. Also, the right to meet with disagreement, but no disrespect and sarcasm

Last but not least, I'm in a position of my life where, thanks God, I can afford to experiment with little things I like (no, I'm not rich) without heavy impact on my finances. If everything goes wrong, I'll just say **** it, come here to admit, and go out and get myself a better plataform. And you know what? Still worth it, for me, for the fun factor.

Pax and best wishes,
Alex.
    
CPUMotherboardGraphicsRAM
i7 3930k @ 4.4ghz (undervolted) ASUS Sabertooth x79 HIS HD7970 4x4g G.Skill Sniper 1866mhz 
Hard DriveHard DriveHard DriveOptical Drive
x2 SSD Corsair F3 (2x 120g) crucial m4 ssd (256g - steam) toshiba external usb 3.0 generic dvd r/w 
CoolingOSMonitorKeyboard
Corsair H100 windows 8 pro + start button + no metro DELL UltraSharp U2212HM leadership gamer 
PowerCaseMouseAudio
corsair ax1200 aerocool bx-500 evil black ed. C3 ares x-fi titanium HD 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i7 3930k @ 4.4ghz (undervolted) ASUS Sabertooth x79 HIS HD7970 4x4g G.Skill Sniper 1866mhz 
Hard DriveHard DriveHard DriveOptical Drive
x2 SSD Corsair F3 (2x 120g) crucial m4 ssd (256g - steam) toshiba external usb 3.0 generic dvd r/w 
CoolingOSMonitorKeyboard
Corsair H100 windows 8 pro + start button + no metro DELL UltraSharp U2212HM leadership gamer 
PowerCaseMouseAudio
corsair ax1200 aerocool bx-500 evil black ed. C3 ares x-fi titanium HD 
  hide details  
Reply
post #227 of 292
The finger pointing at the OS, or when to turn off cores or not, its a bit silly to say that its killing the BD performance.
SB has the same problem. If you could also turn off HT and get the full benefit of a core as well, this would increase performance on SB as well.

If AMD will say "its the OS fault", well, its a cheap excuse for bad design.
Main system
(16 items)
 
Editing PC
(8 items)
 
 
CPUGraphicsGraphicsRAM
E5-1680v2 AMD FirePro D700 AMD FirePro D700 64GB 1866mhz 
Hard DriveOSMonitorCase
1TB PCIE SSD OSX 10.10.x Dell U2713H Mac Pro 
  hide details  
Reply
Main system
(16 items)
 
Editing PC
(8 items)
 
 
CPUGraphicsGraphicsRAM
E5-1680v2 AMD FirePro D700 AMD FirePro D700 64GB 1866mhz 
Hard DriveOSMonitorCase
1TB PCIE SSD OSX 10.10.x Dell U2713H Mac Pro 
  hide details  
Reply
post #228 of 292
Quote:
Originally Posted by Defoler View Post
The finger pointing at the OS, or when to turn off cores or not, its a bit silly to say that its killing the BD performance.
SB has the same problem. If you could also turn off HT and get the full benefit of a core as well, this would increase performance on SB as well.

If AMD will say "its the OS fault", well, its a cheap excuse for bad design.
AMD is saying that it's the OS's fault? Where? Well, I would say that you couldn't fault MS (assuming we're talking about windows) for the situation. However, with a different architecture comes different ways to work with it to get optimal performance and windows may not be doing that. One way would be some sort of advanced task scheduler (or whatever, for lack of a better term) that uses cores/modules in BD more efficiently as they're intended, as opposed to the traditional way.

But hasn't that been addressed already? I feel like I've read something about that, and the gains when using an optimized scheduler are still, max 10% in some situations, with no effect in others. I could be wrong, and I don't have the time at the moment to look into it again.

There was some nutcase saying that with some sort of hacked/modified component in windows, he was getting like 100% increase in performance with BD in things like media encoding--don't exactly recall what was going on. It could have just been an optimized encoder that utilizes the new instruction sets that AMD dumped into BD. Not sure. Don't even know if anything came of it, or if the guy wasn't just spouting fake results.
    
CPUMotherboardGraphicsRAM
i5-2500K Biostar TP67B+ XFX HD5750 1GB 2x4GB DDR3 Corsair 1600 
Hard DriveOSMonitorPower
60GB OCZ SSD, 2x160GB HDD RAID0, 500GB+500GB+1.5TB Windows 7 Ultimate 64-bit Samsung SyncMaster 930B Antec SmartPower 450w 
Case
Antec 900 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i5-2500K Biostar TP67B+ XFX HD5750 1GB 2x4GB DDR3 Corsair 1600 
Hard DriveOSMonitorPower
60GB OCZ SSD, 2x160GB HDD RAID0, 500GB+500GB+1.5TB Windows 7 Ultimate 64-bit Samsung SyncMaster 930B Antec SmartPower 450w 
Case
Antec 900 
  hide details  
Reply
post #229 of 292
Quote:
Originally Posted by jivenjune View Post
To bench and do absolutely nothing else meaningful aside from that? Also willing to wager that most extreme OCers are working multiple platforms for varying purposes.
Most haven't touched 1155 after initial results because it 1) hits an architectural limit around 5.7-5.8 (Even though nearly any board and chip should theoretically be able to hit 5.88Ghz) and 2) isn't as fun to OC compared to say, Phenom II. You can't tweak as much for every drip of performance.
    
CPUMotherboardGraphicsRAM
Intel Core i5 3570k @ 4.5Ghz ASRock Z77 Pro3 Powercolor Radeon HD7950 3GB @ 1150/1350 4x4GB G.Skill Ares 2000Mhz CL9 
Hard DriveHard DriveHard DriveHard Drive
Samsung 840 250GB Western Digital Black 1TB WD1002FAEX Seagate Barracuda 3TB ST3000DM001 Samsung Spinpoint EcoGreen 2TB 
Optical DriveCoolingCoolingCooling
Pioneer DVR-220LBKS Noctua NH-D14 Scythe Gentle Typhoon 1850rpm Corsair AF140 Quiet Edition 
CoolingOSMonitorMonitor
Arcitc Cooling Acclero Twin Turbo II Arch Linux x86-64, amdgpu BenQ G2220HD BenQ G2020HD 
KeyboardPowerCaseMouse
Ducky Shine III Year of the Snake, Cherry Blue Silverstone Strider Plus 600w CoolerMaster CM690 II Black and White SteelSeries Sensei Professional 
Mouse PadAudioOther
Artisan Hien Mid Japan Black Large ASUS Xonar DX NZXT Sentry Mesh 30w Fan Controller 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Intel Core i5 3570k @ 4.5Ghz ASRock Z77 Pro3 Powercolor Radeon HD7950 3GB @ 1150/1350 4x4GB G.Skill Ares 2000Mhz CL9 
Hard DriveHard DriveHard DriveHard Drive
Samsung 840 250GB Western Digital Black 1TB WD1002FAEX Seagate Barracuda 3TB ST3000DM001 Samsung Spinpoint EcoGreen 2TB 
Optical DriveCoolingCoolingCooling
Pioneer DVR-220LBKS Noctua NH-D14 Scythe Gentle Typhoon 1850rpm Corsair AF140 Quiet Edition 
CoolingOSMonitorMonitor
Arcitc Cooling Acclero Twin Turbo II Arch Linux x86-64, amdgpu BenQ G2220HD BenQ G2020HD 
KeyboardPowerCaseMouse
Ducky Shine III Year of the Snake, Cherry Blue Silverstone Strider Plus 600w CoolerMaster CM690 II Black and White SteelSeries Sensei Professional 
Mouse PadAudioOther
Artisan Hien Mid Japan Black Large ASUS Xonar DX NZXT Sentry Mesh 30w Fan Controller 
  hide details  
Reply
post #230 of 292
Quote:
Originally Posted by Brutuz View Post
4. It's got 8 cores, it uses about as much power as a 32nm Intel six core? I'd say that's a fair deal.
So when Bulldozer performs worse per core than pretty much every modern Intel chip, everyone yells, "It's not about performance per core, it's about actual overall performance!"

But when it consumes power comparable (stock mind you) to a 6 core, all of sudden it isn't about overall power consumption, it's about power consumption per core. Not to mention the 6 core blows the BD out of the water in absolutely every way imaginable. It's a stupid comparison because of the price difference, but hey, you brought it up.

So basically you can look at it like this:
Does it paint a pretty picture for AMD? If no, multiple or divide by the number of cores accordingly. If yes, brag about 8 cores. AWWWW YEAHHHH.

    
CPUMotherboardGraphicsRAM
i7-2600k 4.5GHz @ 1.32V Asus P8P67 Pro EVGA GTX 580 Mushkin 2133 9-10-9-24 
Hard DriveHard DriveCoolingOS
Samsung 840  WD Black Silver Arrow Windows 7 
MonitorKeyboardPowerCase
Dell U2211H Rosewill RK-9000BR Seasonic X750 HAF X 
MouseAudio
Razer Lachesis Grado HF2 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i7-2600k 4.5GHz @ 1.32V Asus P8P67 Pro EVGA GTX 580 Mushkin 2133 9-10-9-24 
Hard DriveHard DriveCoolingOS
Samsung 840  WD Black Silver Arrow Windows 7 
MonitorKeyboardPowerCase
Dell U2211H Rosewill RK-9000BR Seasonic X750 HAF X 
MouseAudio
Razer Lachesis Grado HF2 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [ExtremeTech] Analyzing Bulldozer: Why AMD’s chip is so disappointing