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 13

post #121 of 292
AMD Transplant
(11 items)
 
To be a NAS
(13 items)
 
Death Kühler
(14 items)
 
CPUMotherboardGraphicsRAM
965BE GA-78LMT-S2P MSI TFIII 7950 Corsair Vengence 
Hard DriveCoolingOSMonitor
Samsung 830 Noctua NH-D14 Win7 Pro FX2490HD 
PowerCaseMouse
Seasonic X660 Antec Three Hundred Saitek Rat 7 
CPUMotherboardGraphicsRAM
AMD Athlon 64 X2 4200+ Gateway GM5072 AMD HD5550 2.0 Gb 
Hard DriveMonitorKeyboardPower
Seagate Barracuda 500Gb Samsung FX2490 MS Keyboard 3000 300w generic 
Mouse
MS Intellipoint 3000 
CPUMotherboardGraphicsRAM
2500k GigaByte Z68M-D2H-B3 MSI 560 Ti TFIII Corsair Vengance 
Hard DriveCoolingOSMonitor
Samsung 830 Antec Khuler 920 Win7 HP Samsung FX2490 
KeyboardPowerCaseMouse
Microsoft Keyboard 3000 Seasonic X-660 Antec Three Hundred Microsoft Mouse 3000 
Mouse PadAudio
None - Blue Track baby Senn HD428 
  hide details  
Reply
AMD Transplant
(11 items)
 
To be a NAS
(13 items)
 
Death Kühler
(14 items)
 
CPUMotherboardGraphicsRAM
965BE GA-78LMT-S2P MSI TFIII 7950 Corsair Vengence 
Hard DriveCoolingOSMonitor
Samsung 830 Noctua NH-D14 Win7 Pro FX2490HD 
PowerCaseMouse
Seasonic X660 Antec Three Hundred Saitek Rat 7 
CPUMotherboardGraphicsRAM
AMD Athlon 64 X2 4200+ Gateway GM5072 AMD HD5550 2.0 Gb 
Hard DriveMonitorKeyboardPower
Seagate Barracuda 500Gb Samsung FX2490 MS Keyboard 3000 300w generic 
Mouse
MS Intellipoint 3000 
CPUMotherboardGraphicsRAM
2500k GigaByte Z68M-D2H-B3 MSI 560 Ti TFIII Corsair Vengance 
Hard DriveCoolingOSMonitor
Samsung 830 Antec Khuler 920 Win7 HP Samsung FX2490 
KeyboardPowerCaseMouse
Microsoft Keyboard 3000 Seasonic X-660 Antec Three Hundred Microsoft Mouse 3000 
Mouse PadAudio
None - Blue Track baby Senn HD428 
  hide details  
Reply
post #122 of 292
Quote:
Originally Posted by guyladouche View Post
And weren't you saying before that you fold, and BD isn't suitable for that either, when in fact it nets the same ppd as a 2500k?
Not to be nit picky about it, but what you forgot was the power consumption which is approaching double that of the 2600k.

Test Setup....Idle (Watts)....CPU Loaded (Watts)
i7 2600K........97 W.............158 W
FX-8150........121 W............246 W

http://www.overclockers.com/amd-fx-8...ocessor-review

That puts a rather large dent in the performance per watt category with folders, which is one of their primary concerns aside from ppd. El Gappo did a test @ 4.9 Ghz as well, where it pulled 22,025.2 ppd. However, at those current draw levels, no one would want the power bill that would beget.

Quote:
Originally Posted by El Gappo
BIG ADV project 2686, 2500NB 2300cl7 ram, 4956mhz core frequency. Given the power draw... don't think it's a folders chip.


Post #3 @ XS
Pit Stop
(35 items)
 
  
CPUMotherboardGraphicsRAM
1090T Gigabyte 990FXA-UD5 MSi ref. 6950 2GB unlocked 4GB STT WX200UB2G7 
Hard DriveHard DriveOptical DriveCooling
Samsung F3 Crucial M4 Teac slim slot load DIYINHK Toshiba pwm pump controller upgrade 
CoolingCoolingCoolingCooling
Yate Loon D12SH-12 Silverstone SST-AP181 Koolance DDC pump housing/heasink Sunon 60 mm cooling fan for pump housing 
CoolingCoolingCoolingCooling
Bitspower 7/16" Black Sparkle compression fitt... Bitspower Black Sparkle 90 degree double rotary... Bitspower 45 degree rotary fittings Primochill LRT UV blue tubing 
CoolingCoolingCoolingCooling
XSPC Rasa cpu block XSPC RX-240 radiator XSPC DDC res. top Laing DDC-1  
OSMonitorKeyboardPower
7 Professional Samsung EX-2220 Das Professional Seasonic's dead :( 
CaseMouseMouse PadAudio
Lian Li T60-B PureTrak Valor Ratpadz GS Auzentech X-plosion 7.1 
AudioAudioAudioOther
AKG K701's Lil Dot MK.III hp amp Burr-Brown OPA627SM opamp upgrade Custom built MTM style transmission line 
OtherOtherOther
Various amps. Custom built MTM style transmission line 15" Dayton Titanic MK.III 
  hide details  
Reply
Pit Stop
(35 items)
 
  
CPUMotherboardGraphicsRAM
1090T Gigabyte 990FXA-UD5 MSi ref. 6950 2GB unlocked 4GB STT WX200UB2G7 
Hard DriveHard DriveOptical DriveCooling
Samsung F3 Crucial M4 Teac slim slot load DIYINHK Toshiba pwm pump controller upgrade 
CoolingCoolingCoolingCooling
Yate Loon D12SH-12 Silverstone SST-AP181 Koolance DDC pump housing/heasink Sunon 60 mm cooling fan for pump housing 
CoolingCoolingCoolingCooling
Bitspower 7/16" Black Sparkle compression fitt... Bitspower Black Sparkle 90 degree double rotary... Bitspower 45 degree rotary fittings Primochill LRT UV blue tubing 
CoolingCoolingCoolingCooling
XSPC Rasa cpu block XSPC RX-240 radiator XSPC DDC res. top Laing DDC-1  
OSMonitorKeyboardPower
7 Professional Samsung EX-2220 Das Professional Seasonic's dead :( 
CaseMouseMouse PadAudio
Lian Li T60-B PureTrak Valor Ratpadz GS Auzentech X-plosion 7.1 
AudioAudioAudioOther
AKG K701's Lil Dot MK.III hp amp Burr-Brown OPA627SM opamp upgrade Custom built MTM style transmission line 
OtherOtherOther
Various amps. Custom built MTM style transmission line 15" Dayton Titanic MK.III 
  hide details  
Reply
post #123 of 292
Anyone else wondering why AMD hasn't released a software fix? I'm not talking BIOS update or waiting for MS to patch Win7. If Bulldozer performs better in a 4M/4C arrangement- why not restrict it to this configuration until there is a need for >4 cores. Meaning once 4M see near 100% load then goto the 4M/8C configuration.

With all the powersavings features AMD implemented on a per module basis, I gotta believe this can be done in realtime. But even if they can't implement some sort of state change on the hardware, couldnt AMD atleast release software that modifies Task Scheduler CORE affinities?

I'm sure I'm extremely oversimplifying the issue, but it seems to me something could be done here.
Spit in God's Eye
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7-5960x @ 4.26 ghz core / 3.53 ghz uncore - 1.... AsRock X99 Extreme3 EVGA GTX980 Ti 16gb (4x4gb) Crucial DDR4 2133 CL15  
Hard DriveHard DriveHard DriveOptical Drive
A-Data SP600 256gb SSD (C:) Samsung 840 EVO 1TB SSD (D:) Seagate 2TB Hybrid Drive (E:) LITE-ON 24x DVDRW 
CoolingOSMonitorKeyboard
Corsair H110 Win 10 x64 HP 2511x ( 25" 1080p ) AZIO L70 
PowerCaseMouseAudio
Corsair TX750 Antec 300 (Modded) Logitech M100 Onboard 
  hide details  
Reply
Spit in God's Eye
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7-5960x @ 4.26 ghz core / 3.53 ghz uncore - 1.... AsRock X99 Extreme3 EVGA GTX980 Ti 16gb (4x4gb) Crucial DDR4 2133 CL15  
Hard DriveHard DriveHard DriveOptical Drive
A-Data SP600 256gb SSD (C:) Samsung 840 EVO 1TB SSD (D:) Seagate 2TB Hybrid Drive (E:) LITE-ON 24x DVDRW 
CoolingOSMonitorKeyboard
Corsair H110 Win 10 x64 HP 2511x ( 25" 1080p ) AZIO L70 
PowerCaseMouseAudio
Corsair TX750 Antec 300 (Modded) Logitech M100 Onboard 
  hide details  
Reply
post #124 of 292
Quote:
Originally Posted by guyladouche View Post
The biggest failure was marketing. The second biggest failure is a lack of intelligence and appreciation of context on the side of consumers (which gets back to marketing). No one wants to sit and comprehensively read through reviews. They like to see colorful bar charts and draw unsubstantiated uneducated conclusions from them, despite no correlation between the bar chart results and actual real-world performance.

BD is a fine processor, especially for people outside of the computer hardware enthusiast community. Lots of real-world benchmarks (synthetics are meaningless) have shown it can keep up just fine with current offerings, and even surpass it (in some instances). Not high-end offerings like 2600k, but if you're not in the computer hardware enthusiast community, you wouldn't be considering that anyway.

My gf does a lot of multitasking, running virtual machines for environment testing for work, lots of photoshop work, and to be honest, the fx-8120 (or 50) is much more appealing than getting a 2600 or 2600k.
About everything (except the Marketing part) in your post is wrong.
Even in real-world computing, BD cant surpass Intel's current mid-range offerings, and yes the 2600 is midrange and NOT high-end; that would be the 980/990x or the upcoming SB-E in terms of multi-threading tasks to which you are obviously relying to.

So I dont see how this is a fine chip when it cant beat Intels mid-range offerings that have been out for almost a year after 5 years of development, having a more than double transistor count, much higher power consumption and what not. The upcoming 6-core SB-E is going to outperform the 2600k by another 30-40% in heavily multi-threaded workloads, it will literally crush BD.
I know we're talking about a different price league here, but I'm simply argueing in consideration of time and technology (especially since WE ALL KNOW that the chip would've been much more expensive if it was performing better).

Looking at performance, AMD is still back in 2009 while Intel is ahead of it's time. If AMD cant get their asses up and produce a chip that performs adequately for todays time, then they can pretty much step out of the desktop market. I see no point in buying new hardware which just came out and is already outdated by 2 years. And even if we were to do a comparison by prices, the 2500k or 1100T are still much better offerings from a price-to-performance view, Bulldozer is an obsolete chip in my eyes.

Even in highly-thread applications (not synthetics) the overall performance of the chip is not considerably better than a i7-870.
You call that a fine chip?

Quote:
Originally Posted by guyladouche View Post
Going back, your statement was that BD can't handle 480's SLI'd--which they clearly can, and yield excess frames per second. Sure, there's a performance gap between a 2600k and the 8150 in metro 2033 with the tri-fire 697-'s--which I'm not surprised at--but how does this relate back to your statement that BD couldn't deliver on sli'd 480's?

And weren't you saying before that you fold, and BD isn't suitable for that either, when in fact it nets the same ppd as a 2500k?
Not sure if serious..
Guess why theres a performance gap between the 2600k and the 8150? Correct, because the FX-8150 simply bottlenecks the heck out of the cards and cant deliver data quick enough to fully utilize them. In my book that pretty much falls under "can't handle".
If I were going by the same approach that you did, I could say an ancient Athlon 64 or P4 single-core can handle two SLI'd 480's too just because it runs .. thats pretty dull, isnt it?

If a CPU cant deliver fast enough so that it bottlenecks a graphics card, it pretty much cant handle it.
Edited by toX0rz - 10/26/11 at 12:54am
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 #125 of 292
Quote:
Anyone else wondering why AMD hasn't released a software fix? I'm not talking BIOS update or waiting for MS to patch Win7. If Bulldozer performs better in a 4M/4C arrangement- why not restrict it to this configuration until there is a need for >4 cores. Meaning once 4M see near 100% load then goto the 4M/8C configuration.
I think it is just that they are either at a loss for what to do - or they are trying to fix the design itself or some parts of it for the next stepping .
Personally i do not think BD is fixable , it has some very dubious design decisions - Which show in the various benchmarks . I think their next chip will address some of the issues - but some are the Arch itself
post #126 of 292
Quote:
Even in real-world computing, BD cant surpass Intel's current mid-range offerings, and yes the 2600 is midrange and NOT high-end; that would be the 980/990x or the upcoming SB-E in terms of multi-threading tasks to which you are obviously relying to.
Over here the 8120 costs about ~$50 less than 2600 so while the BD isn't a top performer, the pricing doesn't seem bad at all.

Quote:
Anyone else wondering why AMD hasn't released a software fix?
The poor performance is at least in part due to current OSs having schedulers which simply don't know how to distribute the workload wisely, and AMD doesn't have much control over what devs do with their kernels - we'll just have to wait it out and see what happens. Intel was in a similar situation back in the days they released HT for the first time; operating systems couldn't distinguish between logical and physical cores and it often happened that on a system with more than one HT enabled CPU or core, the OS might schedule all threads to be run on one physical CPU which of course was ineffective (especially since the HT implementation on P4 wasn't very refined) while the other physical CPU sat around doing nothing.
Edited by bomfunk - 10/26/11 at 12:49am
post #127 of 292
Quote:
Originally Posted by bomfunk View Post
Over here the 8120 costs about ~$50 less than 2600 so while the BD isn't a top performer, the pricing doesn't seem bad at all.
Thats a bit much, are you sure you didnt mix it up with the 8150 ?
The 8120 has a pretty huge clock hit compared to the 8150, I'm pretty sure it would lie behind by 30+% compared to the 2600k performance-wise, and thats simply too much to justify a merely 50$ lower price, it would need to be even lower.
And I'm pretty sure the 8120 is priced on par with the 2500k while the latter easily performs on 8150 level in multi-threaded tasks and even miles better in low-threaded tasks and games...
As I said before, from a price/performance view, bulldozer is a completely obsolete chip for now, theres absolutley nothing that would justify buying it unless you are a "fanboy".. because even if you have an AMD board, the Phenom X6 would simply be the smarter choice.
Edited by toX0rz - 10/26/11 at 1:02am
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 #128 of 292
Intel Core i5 2500K Processor LGA1155 3.3GHz CPU [24]Instock$225.00
AMD FX-8120 Eight-Core 3.1GHz 16MB Socket AM3+ 125WPreorderETA$269.00
That is Brisbane AU prices
and $199 for an 1100t
post #129 of 292
Quote:
Thats a bit much, are you sure you didnt mix it up with the 8150 ?
Yes, I'm sure - it's the 8120, $259 vs $319 at PCCG.

Quote:
The 8120 has a pretty huge clock hit compared to the 8150, I'm pretty sure it would lie behind by 30+% compared to the 2600k performance-wise
According to the benchmarks I've seen, it's only around 10%-20% slower in most scenarios even in the current state of software - mind I can't tell if they're all reliable, I'm not that "in", and the only one I know is trusthworthy out of the all I remember seeing is Anand but I can't remember if he had a 8120. But based on what I've seen so far, I would think +30% slower is only a worst case scenario.
post #130 of 292
Quote:
Intel Core i5 2500K Processor LGA1155 3.3GHz CPU [24] Instock $225.00
AMD FX-8120 Eight-Core 3.1GHz 16MB Socket AM3+ 125W PreorderETA $269.00
That is Brisbane AU prices
and $199 for an 1100t
Hence why I bought the 1100t
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