Overclock.net › Forums › Graphics Cards › AMD/ATI › Tri Crossfire Bottleneck?
New Posts  All Forums:Forum Nav:

Tri Crossfire Bottleneck?

post #1 of 7
Thread Starter 
I read this review.

http://www.guru3d.com/article/radeon...x-review-test/

From their conclusion an i7 OC at 3.75 ghz is a bottleneck for THREE 5770s in crossfire.

I want to know is the reason for the bottleneck is because its 3 5770 "separate" cards opposed to a single card solution?

Would lets say SINGLE card with 3x the performance of a 5770 still be considered a bottleneck?

Ignoring scaling, does having multiple cards means the CPU becomes more bottlenecked?

Main question.

Lets say 3 4870's in crossfire vs 4870x2 + 4870 in crossfire.

3 4870s use 3 pci slots.

4870x2+4870 only use 2 pci slots.

is the bottleneck the same? or is the 3 4870s have a bigger cpu bottleneck because its managing three separate cards on three pci lanes opposed to 2?

thanks.

EDIT
seperate question.

i hear people talking about installing CAPS? in crossfire threads?

what is CAPS?
Edited by Prideless - 6/20/11 at 12:40am
My Tank
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 955 BE @ 3.7ghz 1.35v ASUS M4A79XTD EVO 790X AM3 4870x2 + 4870 in CF G. Skill Ripjaws DDR3 1600 4GB 7-7-8-7-24 
Hard DriveOSMonitorKeyboard
WD Black Cavier 750GB Win7 Pro x64,WinXP64,WinServ2003 Dell 23" SP2309W Logitech Illuminated Keyboard 
PowerCaseMouseMouse Pad
Corsair 850w CM HAF 922 Logitech G9 Hatsune Miku/ Jugs! 
  hide details  
Reply
My Tank
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 955 BE @ 3.7ghz 1.35v ASUS M4A79XTD EVO 790X AM3 4870x2 + 4870 in CF G. Skill Ripjaws DDR3 1600 4GB 7-7-8-7-24 
Hard DriveOSMonitorKeyboard
WD Black Cavier 750GB Win7 Pro x64,WinXP64,WinServ2003 Dell 23" SP2309W Logitech Illuminated Keyboard 
PowerCaseMouseMouse Pad
Corsair 850w CM HAF 922 Logitech G9 Hatsune Miku/ Jugs! 
  hide details  
Reply
post #2 of 7
Can't answer the bottleneck but CAP = CrossFireApplicationProfiles
that increase scaling and can be found here: http://sites.amd.com/us/game/downloa...n_win7-64.aspx
justmypc
(15 items)
 
  
CPUMotherboardGraphicsRAM
8700K @ 50/49 Ring 1.472V HT OFF, NO AVX (delid) Gigabyte Gaming 7 1080Ti FE w/ G10+X61 @ 2076/5962Mhz 1.2V G.skill Trident Z B-die @ 3800 15-15-15-35 
Hard DriveOptical DriveCoolingOS
Samsung SM951 256GB, Samsung 850 EVO 500GB Samsung Spinpoint F3 1TB EVGA CLC280 Win 10 x64 Pro 
MonitorKeyboardPowerCase
Dell S2417DG G610 Brown Corsair AX850 Phanteks Enthoo Pro 
MouseMouse PadAudio
Mionix Naos QG SteelSeries Qck+ Xonar DG 
  hide details  
Reply
justmypc
(15 items)
 
  
CPUMotherboardGraphicsRAM
8700K @ 50/49 Ring 1.472V HT OFF, NO AVX (delid) Gigabyte Gaming 7 1080Ti FE w/ G10+X61 @ 2076/5962Mhz 1.2V G.skill Trident Z B-die @ 3800 15-15-15-35 
Hard DriveOptical DriveCoolingOS
Samsung SM951 256GB, Samsung 850 EVO 500GB Samsung Spinpoint F3 1TB EVGA CLC280 Win 10 x64 Pro 
MonitorKeyboardPowerCase
Dell S2417DG G610 Brown Corsair AX850 Phanteks Enthoo Pro 
MouseMouse PadAudio
Mionix Naos QG SteelSeries Qck+ Xonar DG 
  hide details  
Reply
post #3 of 7
It's the number of GPUs, not the pci-e slots. The GPU to GPU communication link on a single board, though, is far better & helps clear up PCI-E bandwidth linking the CPU to the graphic cards.

That aside, it's feeding the command buffer of 3 gpus & syncing frames that chews up CPU capacity.
Quadfire Turbo
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 - 2600K / 5.4 GHz Gigabyte P67A-UD7-B3 #2 Sapphie Radeon 6990 @ 1030 / 1520 Ripjaws 4GB x 4 (16GB) @ 2200 MHz 
Hard DriveOptical DriveOSMonitor
OCZ Vertex3,Corsair Force GT,Intel X25-M, +4TB HDD LG Bluray Burner Windows 7 64 5x1 P(NEC ips), S750 Samsung 120hz HD3D, HP zr30w 
KeyboardPowerCaseMouse
Logitech G13 #2 Kingwin Mach 1 / 1220 Watts XSPC H2 R.A.T. 7 
Mouse Pad
Razer Goliath 
  hide details  
Reply
Quadfire Turbo
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 - 2600K / 5.4 GHz Gigabyte P67A-UD7-B3 #2 Sapphie Radeon 6990 @ 1030 / 1520 Ripjaws 4GB x 4 (16GB) @ 2200 MHz 
Hard DriveOptical DriveOSMonitor
OCZ Vertex3,Corsair Force GT,Intel X25-M, +4TB HDD LG Bluray Burner Windows 7 64 5x1 P(NEC ips), S750 Samsung 120hz HD3D, HP zr30w 
KeyboardPowerCaseMouse
Logitech G13 #2 Kingwin Mach 1 / 1220 Watts XSPC H2 R.A.T. 7 
Mouse Pad
Razer Goliath 
  hide details  
Reply
post #4 of 7
Thread Starter 
ah ok so basically the bottleneck would be the same? because theres 3 gpus either way?
My Tank
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 955 BE @ 3.7ghz 1.35v ASUS M4A79XTD EVO 790X AM3 4870x2 + 4870 in CF G. Skill Ripjaws DDR3 1600 4GB 7-7-8-7-24 
Hard DriveOSMonitorKeyboard
WD Black Cavier 750GB Win7 Pro x64,WinXP64,WinServ2003 Dell 23" SP2309W Logitech Illuminated Keyboard 
PowerCaseMouseMouse Pad
Corsair 850w CM HAF 922 Logitech G9 Hatsune Miku/ Jugs! 
  hide details  
Reply
My Tank
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 955 BE @ 3.7ghz 1.35v ASUS M4A79XTD EVO 790X AM3 4870x2 + 4870 in CF G. Skill Ripjaws DDR3 1600 4GB 7-7-8-7-24 
Hard DriveOSMonitorKeyboard
WD Black Cavier 750GB Win7 Pro x64,WinXP64,WinServ2003 Dell 23" SP2309W Logitech Illuminated Keyboard 
PowerCaseMouseMouse Pad
Corsair 850w CM HAF 922 Logitech G9 Hatsune Miku/ Jugs! 
  hide details  
Reply
post #5 of 7
Quote:
That is however a luxury problem though but as an example, Fallout 3 at 8x AA at 1920x1200 performed as well as a 2-way 5770 setup as the processor (Core i7 @ 3.75 GHz) was literally holding back the GPUs from reaching the stratosphere.
Their charts showed no gain with the 3rd card in Fallout and worse it dropped going from 1080P to 1600P. The drop is obvious vram choke and the no scaling is probably a game/driver issue. I'd have a hard time believing the cards are cpu bound at 1080P in just Fallout. Their chart shows an issue with the 3rd card scaling from puny 1024P all the way up.
The 3930
(26 items)
 
Junior's 3930
(22 items)
 
DATA/HTPC
(20 items)
 
  hide details  
Reply
The 3930
(26 items)
 
Junior's 3930
(22 items)
 
DATA/HTPC
(20 items)
 
  hide details  
Reply
post #6 of 7
Thread Starter 
Quote:
Originally Posted by tsm106 View Post
Their charts showed no gain with the 3rd card in Fallout and worse it dropped going from 1080P to 1600P. The drop is obvious vram choke and the no scaling is probably a game/driver issue. I'd have a hard time believing the cards are cpu bound at 1080P in just Fallout. Their chart shows an issue with the 3rd card scaling from puny 1024P all the way up.
whats your experience with games and the benefits to the third card? or even a fourth?

says in your sig you have 4 in crossfire
My Tank
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 955 BE @ 3.7ghz 1.35v ASUS M4A79XTD EVO 790X AM3 4870x2 + 4870 in CF G. Skill Ripjaws DDR3 1600 4GB 7-7-8-7-24 
Hard DriveOSMonitorKeyboard
WD Black Cavier 750GB Win7 Pro x64,WinXP64,WinServ2003 Dell 23" SP2309W Logitech Illuminated Keyboard 
PowerCaseMouseMouse Pad
Corsair 850w CM HAF 922 Logitech G9 Hatsune Miku/ Jugs! 
  hide details  
Reply
My Tank
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 955 BE @ 3.7ghz 1.35v ASUS M4A79XTD EVO 790X AM3 4870x2 + 4870 in CF G. Skill Ripjaws DDR3 1600 4GB 7-7-8-7-24 
Hard DriveOSMonitorKeyboard
WD Black Cavier 750GB Win7 Pro x64,WinXP64,WinServ2003 Dell 23" SP2309W Logitech Illuminated Keyboard 
PowerCaseMouseMouse Pad
Corsair 850w CM HAF 922 Logitech G9 Hatsune Miku/ Jugs! 
  hide details  
Reply
post #7 of 7
Quote:
Originally Posted by tsm106 View Post
Their charts showed no gain with the 3rd card in Fallout and worse it dropped going from 1080P to 1600P. The drop is obvious vram choke and the no scaling is probably a game/driver issue. I'd have a hard time believing the cards are cpu bound at 1080P in just Fallout. Their chart shows an issue with the 3rd card scaling from puny 1024P all the way up.
There is obviously a cpu bottleneck in fallout when the 5870 CFX performs the same as 5770 CFX and 4870 CFX. With that said, trifire performance was not the best with the 5xxx series especially 5770s. Overall, most of the time your limited due to the bad scaling but in some instances in this review there is some obvious bottlenecking. Not that the FPS in which the bottleneck initiated are terrible at all. Actually quite nice. Not to mention a 975 has plenty of more breathing room than the clocks they used.
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
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD/ATI
Overclock.net › Forums › Graphics Cards › AMD/ATI › Tri Crossfire Bottleneck?