Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [The Inquirer] Nvidia's GTX680 gets thrashed by AMD's mid-range Radeon HD 7870 in GPU compute
New Posts  All Forums:Forum Nav:

[The Inquirer] Nvidia's GTX680 gets thrashed by AMD's mid-range Radeon HD 7870 in GPU compute - Page 13

post #121 of 186
Quote:
Originally Posted by Majin SSJ Eric View Post

Um, because that's not what Geforce cards are designed for??? I thought you were smarter than that Duckie...

GeForce, Tesla, and Quadro use the same GPU die (aside from binning). They are the same cards except for firmware locks, RAM size, and inclusion of display output.

If your workload does need Quadro/Tesla and your servers are fault-tolerant and you do not need support, why spend 3-5 times as much per card?

...because I know this and you don't.... what does that mean then?
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
post #122 of 186
I'm most likely just getting a 7970 because of this, I need my bitcoins.
post #123 of 186
Quote:
Originally Posted by DuckieHo View Post

GeForce, Tesla, and Quadro use the same GPU die (aside from binning). They are the same cards except for firmware locks, RAM size, and inclusion of display output.
If your workload does need Quadro/Tesla and your servers are fault-tolerant and you do not need support, why spend 3-5 times as much per card?
...because I know this and you don't.... what does that mean then?

Ahhhh, there lies your answer Einstein. Nvidia prices these cards for a reason...
post #124 of 186
Quote:
Originally Posted by DuckieHo View Post

GeForce, Tesla, and Quadro use the same GPU die (aside from binning). They are the same cards except for firmware locks, RAM size, and inclusion of display output.
If your workload does need Quadro/Tesla and your servers are fault-tolerant and you do not need support, why spend 3-5 times as much per card?
...because I know this and you don't.... what does that mean then?

you buy quadro because it comes with quadro drivers, without it you dont get the gpu acceleration for professional apps.
not to mention quadro come with much more vram, thats very important in the design field, try rendering architectural designs in huge resolutions and you will realize very fast how important it is to have atleast 4gb or ram for comfort. 2gb is like very average but acceptable tho
i used a gtx 460 to render before and it took forever, mostly because geforce drivers didnt support gpu acceleration for the software i was using(not to mention i was restricted in resolutions as well

nvidia takes the professional graphics business more seriously than amd in my opinion, thats why they make sure they dont slap too much vram in their gaming cards so their professional cards stay appealing and stand out, 1.5gb for 580 and now 2gb, enough for gaming, but minimal for professional design wink.gif
post #125 of 186
Why is this generating so much buzz?
post #126 of 186
Quote:
Originally Posted by sergionography View Post

you buy quadro because it comes with quadro drivers, without it you dont get the gpu acceleration for professional apps.
not to mention quadro come with much more vram, thats very important in the design field, try rendering architectural designs in huge resolutions and you will realize very fast how important it is to have atleast 4gb or ram for comfort. 2gb is like very average but acceptable tho
i used a gtx 460 to render before and it took forever, mostly because geforce drivers didnt support gpu acceleration for the software i was using(not to mention i was restricted in resolutions as well
nvidia takes the professional graphics business more seriously than amd in my opinion, thats why they make sure they dont slap too much vram in their gaming cards so their professional cards stay appealing and stand out, 1.5gb for 580 and now 2gb, enough for gaming, but minimal for professional design wink.gif

Well, that and Nvidia purposefully cripples OpenGL performance on their GeForce cards; to the point where AMD has laptop graphics cards that are better at OpenGL (see: SpecviewPerf11 benches.). Now, there are a lot of high school and college students with laptops/desktops and such that can't afford Quadro cards just to get decent performance for CAD modeling, and they're pretty much forced to go AMD if they want decent gaming performance and good OpenGL performance. Quadro provides neither.
 
Lanbox Lite
(16 items)
 
 
CPUGraphicsRAMHard Drive
i3-2310M Intel HD 3000 8GB DDR3 Samsung 830 64GB SSD 
Hard DriveOSMonitorCase
640GB Hitachi HDD Windows 7 13.3" LCD Magnesium Alloy, 3.2lbs 
CPUMotherboardGraphicsRAM
Phenom II X4 955 BE Gigabyte GA-MA78GPM-UD2H MSI Hawk R5770 3x2GB G.Skill DDR2 800 4-4-4-12 
Hard DriveHard DriveOptical DriveCooling
1TB Samsung Spinpoint F3 250 GB WD Caviar Black Samsung 20X DVD-R/RW Thermaltake MaxOrb 
CoolingOSMonitorPower
Noctua NF-B9-1600 Windows 7 Pro 64-bit BenQ E2420HD, 24" 1920x1080 TT Purepower 500W 
Case
TT Lanbox Lite 
  hide details  
Reply
 
Lanbox Lite
(16 items)
 
 
CPUGraphicsRAMHard Drive
i3-2310M Intel HD 3000 8GB DDR3 Samsung 830 64GB SSD 
Hard DriveOSMonitorCase
640GB Hitachi HDD Windows 7 13.3" LCD Magnesium Alloy, 3.2lbs 
CPUMotherboardGraphicsRAM
Phenom II X4 955 BE Gigabyte GA-MA78GPM-UD2H MSI Hawk R5770 3x2GB G.Skill DDR2 800 4-4-4-12 
Hard DriveHard DriveOptical DriveCooling
1TB Samsung Spinpoint F3 250 GB WD Caviar Black Samsung 20X DVD-R/RW Thermaltake MaxOrb 
CoolingOSMonitorPower
Noctua NF-B9-1600 Windows 7 Pro 64-bit BenQ E2420HD, 24" 1920x1080 TT Purepower 500W 
Case
TT Lanbox Lite 
  hide details  
Reply
post #127 of 186
Quote:
Originally Posted by jrbroad77 View Post

Well, that and Nvidia purposefully cripples OpenGL performance on their GeForce cards; to the point where AMD has laptop graphics cards that are better at OpenGL (see: SpecviewPerf11 benches.). Now, there are a lot of high school and college students with laptops/desktops and such that can't afford Quadro cards just to get decent performance for CAD modeling, and they're pretty much forced to go AMD if they want decent gaming performance and good OpenGL performance. Quadro provides neither.

Well cry me another sob story about these poor students. They are not the market that Nvidia is trying to sell Geforce GAMING cards to. If they lose a few hundred out of a hundred thousand customers because of that, well, they must have thoroughly studied the ramifications for that and determined that the loss was inconsequential. In other words, AMD can have them....
post #128 of 186
Quote:
Originally Posted by GingerJohn View Post

Is it so wrong of me to want a card that does both gaming and has good compute performance?
I guess I am a terrible human being. Thank you OCN for putting it in perspective for me.

The 7970 is the card for you to get. It is the best of both worlds.

You know those 1536 CUDA cores on the GTX 680? You want to know why there are so many of them? I can't give you a direct answer, but I can tell you why they can fit more of them into a smaller die: less compute performance.

For every feature on a GPU, there must be a part of the die allocated to that feature. AMD wanted to increase compute performance while simultaneously increasing the gaming performance. Thus, the full GCN die has 4 billion transistors. Nvidia's GTX 680, which has 1536 CUDA cores (3 times that of a GTX 580) is about 30% faster in games than the GTX 580, but actually slower on compute performance. The GTX 680 also has less transistors than the GTX 580 as well.

As much as you guys want Kepler to be this magical architecture, it just isn't. They allocated more resources to gaming performance than compute performance with this card. On GK110, it could be a completely different story. Who knows. GPU's are a balancing act of features and technologies. If you want to pack less transistors into a chip yet increase performance, you're going to have to ramp up the clock speed and rearrange the way the architecture works.. in the process, you may have to sacrifice something to get that performance.

Also, Quadros cost way more than their Geforce counterparts. Just keep that in mind. wink.gif
Everest - Intel
(19 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 4790k Gigabyte Z97X Gaming 7 MSI Geforce GTX 1080 Ti Gaming X 16GB (2x8) Patriot Viper 1866Mhz  
Hard DriveHard DriveOptical DriveCooling
Seagate 3TB, WD 500GB HDD, WD 640GB HD Samsung 850 EVO 512GB Samsung DVD-Burner Corsair H110 w/ Dual Aerocool DS 140mm fans 
OSMonitorMonitorKeyboard
Windows 10 Pro Dell S2716DG (1440p, 144hz Gsync) AOC U3477 PQU (3440x1440 IPS) Logitech G810 Orion Spectrum 
PowerCaseMouseMouse Pad
Evga SuperNOVA 750 G2 NZXT Phantom 530 Black Logitech G502 Proteus Core Corsair MM400 
AudioAudioAudio
Creative Sound Blaster E5 DAC/AMP Sennheiser HD 598 Headphones HyperX Cloud Headset 
  hide details  
Reply
Everest - Intel
(19 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 4790k Gigabyte Z97X Gaming 7 MSI Geforce GTX 1080 Ti Gaming X 16GB (2x8) Patriot Viper 1866Mhz  
Hard DriveHard DriveOptical DriveCooling
Seagate 3TB, WD 500GB HDD, WD 640GB HD Samsung 850 EVO 512GB Samsung DVD-Burner Corsair H110 w/ Dual Aerocool DS 140mm fans 
OSMonitorMonitorKeyboard
Windows 10 Pro Dell S2716DG (1440p, 144hz Gsync) AOC U3477 PQU (3440x1440 IPS) Logitech G810 Orion Spectrum 
PowerCaseMouseMouse Pad
Evga SuperNOVA 750 G2 NZXT Phantom 530 Black Logitech G502 Proteus Core Corsair MM400 
AudioAudioAudio
Creative Sound Blaster E5 DAC/AMP Sennheiser HD 598 Headphones HyperX Cloud Headset 
  hide details  
Reply
post #129 of 186
im sorry but people who want professional stuff, get it, that v3700 that was up in the news feed a few days ago, it pillaged all the gaming cards and it was a huge $110 dollars,
if u want to play games this is the type of card for u, we dont really care about gpu compute as gamers,
sure folding is cool and all but we do that mainly cpu ,
sigh
getting sweet
(14 items)
 
  
CPUMotherboardGraphicsRAM
i5-3570k msi Z77 Mpower sli GTX560ti cosair vengence 2x81866-CL9 
Hard DriveOptical DriveCoolingOS
1TB standard custom water  windows 7 64-bit 
MonitorPowerCase
24' cosair 860i antec 100 
  hide details  
Reply
getting sweet
(14 items)
 
  
CPUMotherboardGraphicsRAM
i5-3570k msi Z77 Mpower sli GTX560ti cosair vengence 2x81866-CL9 
Hard DriveOptical DriveCoolingOS
1TB standard custom water  windows 7 64-bit 
MonitorPowerCase
24' cosair 860i antec 100 
  hide details  
Reply
post #130 of 186
Quote:
Originally Posted by george_orm View Post

im sorry but people who want professional stuff, get it, that v3700 that was up in the news feed a few days ago, it pillaged all the gaming cards and it was a huge $110 dollars,
if u want to play games this is the type of card for u, we dont really care about gpu compute as gamers,
sure folding is cool and all but we do that mainly cpu ,
sigh

My stock 5870 gets 10 times the folding points that my overclocked quad core CPU gets each day.

GPU's due to their parallel nature are able to crunch massive amounts of data very quickly. Most CPU's simply cannot match that sort of pace.
Everest - Intel
(19 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 4790k Gigabyte Z97X Gaming 7 MSI Geforce GTX 1080 Ti Gaming X 16GB (2x8) Patriot Viper 1866Mhz  
Hard DriveHard DriveOptical DriveCooling
Seagate 3TB, WD 500GB HDD, WD 640GB HD Samsung 850 EVO 512GB Samsung DVD-Burner Corsair H110 w/ Dual Aerocool DS 140mm fans 
OSMonitorMonitorKeyboard
Windows 10 Pro Dell S2716DG (1440p, 144hz Gsync) AOC U3477 PQU (3440x1440 IPS) Logitech G810 Orion Spectrum 
PowerCaseMouseMouse Pad
Evga SuperNOVA 750 G2 NZXT Phantom 530 Black Logitech G502 Proteus Core Corsair MM400 
AudioAudioAudio
Creative Sound Blaster E5 DAC/AMP Sennheiser HD 598 Headphones HyperX Cloud Headset 
  hide details  
Reply
Everest - Intel
(19 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 4790k Gigabyte Z97X Gaming 7 MSI Geforce GTX 1080 Ti Gaming X 16GB (2x8) Patriot Viper 1866Mhz  
Hard DriveHard DriveOptical DriveCooling
Seagate 3TB, WD 500GB HDD, WD 640GB HD Samsung 850 EVO 512GB Samsung DVD-Burner Corsair H110 w/ Dual Aerocool DS 140mm fans 
OSMonitorMonitorKeyboard
Windows 10 Pro Dell S2716DG (1440p, 144hz Gsync) AOC U3477 PQU (3440x1440 IPS) Logitech G810 Orion Spectrum 
PowerCaseMouseMouse Pad
Evga SuperNOVA 750 G2 NZXT Phantom 530 Black Logitech G502 Proteus Core Corsair MM400 
AudioAudioAudio
Creative Sound Blaster E5 DAC/AMP Sennheiser HD 598 Headphones HyperX Cloud Headset 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Rumors and Unconfirmed Articles
Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [The Inquirer] Nvidia's GTX680 gets thrashed by AMD's mid-range Radeon HD 7870 in GPU compute