Overclock.net › Forums › Industry News › Hardware News › [Toms] Triple-GPU Scaling: AMD CrossFire Vs. Nvidia SLI
New Posts  All Forums:Forum Nav:

[Toms] Triple-GPU Scaling: AMD CrossFire Vs. Nvidia SLI - Page 8

post #71 of 93
Quote:
Originally Posted by Jump3r View Post
I think they are sorta even, my SIG rig setup uses Nvidia tri-sli i have looked at crossfire 5870's but i belive at higher resolutions nvdia has really pulled it off alot better.

I run nvidia surround at a resolution of 6040x1080 BC and i can manage good frame rates. and not run out of memory. I know for fact that ill struggle from memory with AMD/ATI solution.



Did you even read the review? Your comparing 2 year old 5870 crossfire(based on what you saw, no personal experience) to tri-sli on fatter bus/more memory cards.

The whole point is that AMD came out strong with their multi-card solutions via the 6xxx series, 69xx in particular, with a setup cheaper by over 200$ and managing better frames at 2560, much less EyeFinity / Surround.

......
Heat
(14 items)
 
Asus G73JH
(9 items)
 
 
CPUMotherboardGraphicsGraphics
i7 3770K @ 4600mhz Gigabyte GA-Z77X-UD5H eVGA GTX 980 eVGA GTX 980 
RAMHard DriveOptical DriveCooling
Team 8gb @ 2400mhz  SSDs LiteOn H70Core/ Ultra Kaze PP 
OSPowerCase
W7 Ultimate X64 PCP&C MKII 950W CM Storm Scout 
CPUMotherboardGraphicsRAM
Intel i7 720 Asus HD5870M 6gb Triple Channel 
Hard DriveCoolingOSMonitor
500gb Asus Dual Fans W7 17.3" 
Mouse
Razer Orochi 
  hide details  
Reply
Heat
(14 items)
 
Asus G73JH
(9 items)
 
 
CPUMotherboardGraphicsGraphics
i7 3770K @ 4600mhz Gigabyte GA-Z77X-UD5H eVGA GTX 980 eVGA GTX 980 
RAMHard DriveOptical DriveCooling
Team 8gb @ 2400mhz  SSDs LiteOn H70Core/ Ultra Kaze PP 
OSPowerCase
W7 Ultimate X64 PCP&C MKII 950W CM Storm Scout 
CPUMotherboardGraphicsRAM
Intel i7 720 Asus HD5870M 6gb Triple Channel 
Hard DriveCoolingOSMonitor
500gb Asus Dual Fans W7 17.3" 
Mouse
Razer Orochi 
  hide details  
Reply
post #72 of 93
HD 58XX, People where winning about CFX scaling, AMD fixed that.
GTX4XX, People where winning about Temp/Noise, Nvidia fixed that.

In general GTX5XX/HD69XX where just fixes.
Ishimura
(21 items)
 
Silent Knight
(13 items)
 
 
CPUMotherboardGraphicsRAM
Intel Core i7 3770K @ 4.6GHz ASRock Z77E-ITX eVGA GTX 1080 Ti Hybrid AMD Radeon R9 16GB DDR3-2400MHz  
Hard DriveHard DriveCoolingCooling
SanDisk Ultra II 960GB Toshiba X300 5TB Corsair H100i GTX eVGA Hybrid Water Cooler  
CoolingOSMonitorKeyboard
4x GentleTyphoon AP-15 Windows 10 Pro 64-Bit Philips Brilliance BDM4065UC 4K Razer BlackWidow Chroma  
PowerCaseMouseMouse Pad
eVGA SuperNOVA 750 G3 Define Nano S Logitech G502 Proteus Core PECHAM Gaming Mouse Pad XX-Large 
AudioAudioAudioAudio
Audioengine D1 DAC Mackie CR Series CR3 Audio-Technica ATH-M50 Sennheiser HD 598 
Audio
Sony XB950BT 
CPUMotherboardGraphicsRAM
AMD Phenom II X4 955 @ 4.2GHz ASUS M4A79XTD EVO AMD Radeon HD 7970 3GB @ 1200/1500 2x 4GB G.SKILL Ripjaws X DDR3-1600 
Hard DriveHard DriveHard DriveCooling
OCZ Agility 3 60GB WD Caviar Green 1.5TB 2 x Seagate Barracuda 2TB XSPC Raystorm 
CoolingCoolingOSPower
EK-FC7970 XSPC RS360 Windows 10 Pro 64-Bit Corsair TX750 
Case
NZXT Switch 810  
  hide details  
Reply
Ishimura
(21 items)
 
Silent Knight
(13 items)
 
 
CPUMotherboardGraphicsRAM
Intel Core i7 3770K @ 4.6GHz ASRock Z77E-ITX eVGA GTX 1080 Ti Hybrid AMD Radeon R9 16GB DDR3-2400MHz  
Hard DriveHard DriveCoolingCooling
SanDisk Ultra II 960GB Toshiba X300 5TB Corsair H100i GTX eVGA Hybrid Water Cooler  
CoolingOSMonitorKeyboard
4x GentleTyphoon AP-15 Windows 10 Pro 64-Bit Philips Brilliance BDM4065UC 4K Razer BlackWidow Chroma  
PowerCaseMouseMouse Pad
eVGA SuperNOVA 750 G3 Define Nano S Logitech G502 Proteus Core PECHAM Gaming Mouse Pad XX-Large 
AudioAudioAudioAudio
Audioengine D1 DAC Mackie CR Series CR3 Audio-Technica ATH-M50 Sennheiser HD 598 
Audio
Sony XB950BT 
CPUMotherboardGraphicsRAM
AMD Phenom II X4 955 @ 4.2GHz ASUS M4A79XTD EVO AMD Radeon HD 7970 3GB @ 1200/1500 2x 4GB G.SKILL Ripjaws X DDR3-1600 
Hard DriveHard DriveHard DriveCooling
OCZ Agility 3 60GB WD Caviar Green 1.5TB 2 x Seagate Barracuda 2TB XSPC Raystorm 
CoolingCoolingOSPower
EK-FC7970 XSPC RS360 Windows 10 Pro 64-Bit Corsair TX750 
Case
NZXT Switch 810  
  hide details  
Reply
post #73 of 93
Quote:
Originally Posted by ZealotKi11er View Post
HD 58XX, People where winning about CFX scaling, AMD fixed that.
GTX4XX, People where winning about Temp/Noise, Nvidia fixed that.

In general GTX5XX/HD69XX where just fixes.
In a way I disagree.

AMD: Drivers have improved a lot, especially with 11.4 - still needs more fine tuning imo.

nVidia: GTX470/480 just basically ran super hot and "was designed to run hot" GTX570/580 can reach +90°c in eg. Furmark when the Power-Throttling is disabled. GTX560 isn't a problem and runs cool, but so was GTX460..
post #74 of 93
3-way and only going up to 2560x1600?

They had an opportunity to show why the current Xfire bridges are a bottleneck for 3/4way Xfire but they failed to show it.


The real test was done here: http://www.xtremesystems.org/forums/...d.php?t=266367
Anyone who uses triple monitors will need 3/4-way and can't use Xfire since one bridge is bottlenecking at those resolutions. Two 6990s with both bridges connected is the only way around it until AMD designs new bridges. But then they'll just be memory limited since even 2GB per GPU isn't enough for triple 30" monitors.

Quote:
The crossfire bridge in which all AMD cards use today is only 900mbits/sec as on the chart (top). There is no such thing as a side-port on the 69xx series.

Of course it is duplex, but the cards are only connected in serial so card #1 has to go through card #2 to talk to card #3.

In SLI you can see cards can talk directly to each other and not eat up bandwidth. Card #1 can talk directly to card #3 without interfering with the bandwidth of card #1 talking to card #2.
Quote:
"Generating that many pixels at the right quality levels would tax any single graphics chip. Making CrossFire work on this scale presents some challenges, however, as AMD readily admits. The core issue is the fact that the dedicated CrossFire interconnect used for passing completed frames between cards has "only" enough bandwidth to sustain a 2560x1600 display resolution. Even three 1080p displays will exceed its capacity. The alternative is to transfer frame buffer data via PCI Express, which is what AMD does when necessary. Using PCIe works, but it can limit performance scaling somewhat—don't expect to see the near-linear scaling one might get out of a dual-card setup in the right game with a single display. That's not to say mixing CrossFire with Eyefinity won't be worth doing. Based on AMD's performance estimates, frame rates could improve between about 25% and 75% when adding a second GPU with a 12-megapixel, six-monitor array."

http://techreport.com/articles.x/18521/2

There's also a Hardforum review of 5870s crossfire at super high resolutions in which AMD told them to put both crossfire bridges on to eliminate bandwidth problems. In three and four way crossfire you can obviously only use one interconnect per card.

There's more to be found Googling.

If I ran single 30", Tri-Quad crossfire would work fine. I also tried all versions of drivers since the 69xx series was released. On top of that, If I simply lowered the Eyefinity resolution to a certain point, the Tri-Quad crossfire would magically start working again. I found tests that the crossfire bridge was one of the main issues with 2x 5970's not working properly quad-fire. Also found references by AMD that said the crossfire bridge wasn't even designed to handle multi-display bandwidth, let alone resolutions this extreme.

I think all of this clearly points to a crossfire bridge limitation.

Edited by PoopaScoopa - 3/15/11 at 4:22pm
post #75 of 93
Quote:
Originally Posted by un-nefer View Post
Tomshardware used to provide very unbiased reviews, but not much any more.

Just looking at the test setup you see this "review" has fail all over it:

AMD cards: 3x XFX HD-695A-CNFC = $289.99ea or $869.97
Nvidia cards: 3x Asus ENGTX570/2DI/1280MD5 = $399.99ea ($339.99 with $60 mail in rebate) or $1199.97 ($1019.97 with $180 mail in rebate)

If the review was based on cards of the same cost, then they should be comparing those Asus 570's to any of the 6970 cards, as they are all under $399. Or if they wanted to get technical and include the mail in rebate as part of the 570's cost, then they should compare to the MSI or Powercolor 6970's, which cost exactly $339.99 too.

That's the first fail.


Moving on though, lets look at the date of the review and compare that to the drivers used in the test setup.
Date of review: March 14, 2011
AMD driver used: 11.1 (released January 2011)
Current AMD driver available as of review date: 11.4ep (released March 10, 2011)

Nvidia driver used: 266.58 (released January 18, 2011)
Current Nvidia driver: 266.58

So they used the most current Nvidia drivers, yet they used older AMD drivers. And to make it worse, they didn't even use the 11.1a hotfix drivers that were introduced for the newer 6900 series cards.

That's the second fail.


Now I read the article and gone right over the test setup page, and nowhere did they mention or list the catalyst application profiles used for the AMD setup - which are at 11.2 CAP3 and were released on Feb 2, 2011 and so were not included with the 11.1 cat drivers they used in the test setup.

That's the third fail.


So all up a three pretty big fails, leading to another fairly biased review from Tomshardware.

My 2c.
Best post in thread.
Main Rig
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II X6 1090T @ 4.1 GHz Asus Crosshair IV Formula Radeon HD 6990 / 6950 Trifire 16 GB G.Skill 1600 (4GB x 4) 
Hard DriveOSMonitorPower
2x WD1600 7200 RPM Sata 3.0 | Raid 0 Windows 7 Ultimate 64 Samsung 24" Thermaltake TR2 1000W 
CaseMouse
Corsair 600T Logitech G500 
  hide details  
Reply
Main Rig
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II X6 1090T @ 4.1 GHz Asus Crosshair IV Formula Radeon HD 6990 / 6950 Trifire 16 GB G.Skill 1600 (4GB x 4) 
Hard DriveOSMonitorPower
2x WD1600 7200 RPM Sata 3.0 | Raid 0 Windows 7 Ultimate 64 Samsung 24" Thermaltake TR2 1000W 
CaseMouse
Corsair 600T Logitech G500 
  hide details  
Reply
post #76 of 93
Quote:
Originally Posted by un-nefer View Post
Tomshardware used to provide very unbiased reviews, but not much any more.

Just looking at the test setup you see this "review" has fail all over it:

AMD cards: 3x XFX HD-695A-CNFC = $289.99ea or $869.97
Nvidia cards: 3x Asus ENGTX570/2DI/1280MD5 = $399.99ea ($339.99 with $60 mail in rebate) or $1199.97 ($1019.97 with $180 mail in rebate)

If the review was based on cards of the same cost, then they should be comparing those Asus 570's to any of the 6970 cards, as they are all under $399. Or if they wanted to get technical and include the mail in rebate as part of the 570's cost, then they should compare to the MSI or Powercolor 6970's, which cost exactly $339.99 too.

That's the first fail.


Moving on though, lets look at the date of the review and compare that to the drivers used in the test setup.
Date of review: March 14, 2011
AMD driver used: 11.1 (released January 2011)
Current AMD driver available as of review date: 11.4ep (released March 10, 2011)

Nvidia driver used: 266.58 (released January 18, 2011)
Current Nvidia driver: 266.58

So they used the most current Nvidia drivers, yet they used older AMD drivers. And to make it worse, they didn't even use the 11.1a hotfix drivers that were introduced for the newer 6900 series cards.

That's the second fail.


Now I read the article and gone right over the test setup page, and nowhere did they mention or list the catalyst application profiles used for the AMD setup - which are at 11.2 CAP3 and were released on Feb 2, 2011 and so were not included with the 11.1 cat drivers they used in the test setup.

That's the third fail.


So all up a three pretty big fails, leading to another fairly biased review from Tomshardware.

My 2c.
I don't know how I managed to miss that but yeah, thats really, really lame of TH.

Good post.
post #77 of 93
Quote:
Originally Posted by CULLEN View Post
In a way I disagree.

AMD: Drivers have improved a lot, especially with 11.4 - still needs more fine tuning imo.

nVidia: GTX470/480 just basically ran super hot and "was designed to run hot" GTX570/580 can reach +90°c in eg. Furmark when the Power-Throttling is disabled. GTX560 isn't a problem and runs cool, but so was GTX460..
You are wrong there. I have GTX580 and had GTX470. Playing Crysis for 1 hour in both system.

GTX580 @ 1.125v, 900Mhz 76C @ 76% fan speed and not very loud at all.
GTX 470 @ 1.087v, 800Mhz 95C @ 95% fan speed, Supper hot and extreme noise. They improved a lot trust me.
Ishimura
(21 items)
 
Silent Knight
(13 items)
 
 
CPUMotherboardGraphicsRAM
Intel Core i7 3770K @ 4.6GHz ASRock Z77E-ITX eVGA GTX 1080 Ti Hybrid AMD Radeon R9 16GB DDR3-2400MHz  
Hard DriveHard DriveCoolingCooling
SanDisk Ultra II 960GB Toshiba X300 5TB Corsair H100i GTX eVGA Hybrid Water Cooler  
CoolingOSMonitorKeyboard
4x GentleTyphoon AP-15 Windows 10 Pro 64-Bit Philips Brilliance BDM4065UC 4K Razer BlackWidow Chroma  
PowerCaseMouseMouse Pad
eVGA SuperNOVA 750 G3 Define Nano S Logitech G502 Proteus Core PECHAM Gaming Mouse Pad XX-Large 
AudioAudioAudioAudio
Audioengine D1 DAC Mackie CR Series CR3 Audio-Technica ATH-M50 Sennheiser HD 598 
Audio
Sony XB950BT 
CPUMotherboardGraphicsRAM
AMD Phenom II X4 955 @ 4.2GHz ASUS M4A79XTD EVO AMD Radeon HD 7970 3GB @ 1200/1500 2x 4GB G.SKILL Ripjaws X DDR3-1600 
Hard DriveHard DriveHard DriveCooling
OCZ Agility 3 60GB WD Caviar Green 1.5TB 2 x Seagate Barracuda 2TB XSPC Raystorm 
CoolingCoolingOSPower
EK-FC7970 XSPC RS360 Windows 10 Pro 64-Bit Corsair TX750 
Case
NZXT Switch 810  
  hide details  
Reply
Ishimura
(21 items)
 
Silent Knight
(13 items)
 
 
CPUMotherboardGraphicsRAM
Intel Core i7 3770K @ 4.6GHz ASRock Z77E-ITX eVGA GTX 1080 Ti Hybrid AMD Radeon R9 16GB DDR3-2400MHz  
Hard DriveHard DriveCoolingCooling
SanDisk Ultra II 960GB Toshiba X300 5TB Corsair H100i GTX eVGA Hybrid Water Cooler  
CoolingOSMonitorKeyboard
4x GentleTyphoon AP-15 Windows 10 Pro 64-Bit Philips Brilliance BDM4065UC 4K Razer BlackWidow Chroma  
PowerCaseMouseMouse Pad
eVGA SuperNOVA 750 G3 Define Nano S Logitech G502 Proteus Core PECHAM Gaming Mouse Pad XX-Large 
AudioAudioAudioAudio
Audioengine D1 DAC Mackie CR Series CR3 Audio-Technica ATH-M50 Sennheiser HD 598 
Audio
Sony XB950BT 
CPUMotherboardGraphicsRAM
AMD Phenom II X4 955 @ 4.2GHz ASUS M4A79XTD EVO AMD Radeon HD 7970 3GB @ 1200/1500 2x 4GB G.SKILL Ripjaws X DDR3-1600 
Hard DriveHard DriveHard DriveCooling
OCZ Agility 3 60GB WD Caviar Green 1.5TB 2 x Seagate Barracuda 2TB XSPC Raystorm 
CoolingCoolingOSPower
EK-FC7970 XSPC RS360 Windows 10 Pro 64-Bit Corsair TX750 
Case
NZXT Switch 810  
  hide details  
Reply
post #78 of 93
Quote:
Originally Posted by ZealotKi11er View Post
You are wrong there. I have GTX580 and had GTX470. Playing Crysis for 1 hour in both system.

GTX580 @ 1.125v, 900Mhz 76C @ 76% fan speed and not very loud at all.
GTX 470 @ 1.087v, 800Mhz 95C @ 95% fan speed, Supper hot and extreme noise. They improved a lot trust me.
I though that when the "Power-Throttling" was disabled it would burn just like 470/480 but if you have both I wont doubt your words. I also had GTX 470 for few days, the store I bought it from allowed me to return it and take GTX 460 instead and get refunded the difference since I was aiming to build extremely silent rig and the 470 ruined the whole idea.

On the other hand a friend of mine has GTX580, when I asked him if he was going to get the 2nd card so he could use 2 monitors + tv he wasn't sure since one card was making enough noise him. Hes got Haf X so airflow shouldn't be the problem.

What type is your card?
post #79 of 93
Quote:
Originally Posted by ZealotKi11er View Post
HD 58XX, People where winning about CFX scaling, AMD fixed that.
GTX4XX, People where winning about Temp/Noise, Nvidia fixed that.

In general GTX5XX/HD69XX where just fixes.
Charlie is that u?
Rice Burner
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920@4.2 1.35 volts EVGA Classified 759 nf200 2xZotac 580 gtx 3gb G Skill 3x4gb PC12800 
Hard DriveOptical DriveOSMonitor
X25M G2 + 10TB of "data pr0nz" Lite On Win 7 64bit 3007 wfp-hc x 3 "Surround" 5156x2560 oh yea! 
KeyboardPowerCaseMouse
Logitech G19 Corsair AX1200 Corsair 600T Razer Mamba 
Mouse Pad
Steel Series 5L 
  hide details  
Reply
Rice Burner
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920@4.2 1.35 volts EVGA Classified 759 nf200 2xZotac 580 gtx 3gb G Skill 3x4gb PC12800 
Hard DriveOptical DriveOSMonitor
X25M G2 + 10TB of "data pr0nz" Lite On Win 7 64bit 3007 wfp-hc x 3 "Surround" 5156x2560 oh yea! 
KeyboardPowerCaseMouse
Logitech G19 Corsair AX1200 Corsair 600T Razer Mamba 
Mouse Pad
Steel Series 5L 
  hide details  
Reply
post #80 of 93
in terms of price/perf. in 2 card scaling, amd stole that one.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [Toms] Triple-GPU Scaling: AMD CrossFire Vs. Nvidia SLI