Overclock.net › Forums › AMD › AMD - General › Adressing the Notion AMD CPUs Will Bottleneck a High End SLI/CF Setup
New Posts  All Forums:Forum Nav:

Adressing the Notion AMD CPUs Will Bottleneck a High End SLI/CF Setup

post #1 of 14
Thread Starter 
Hello everyone!

I posted recently in this forum looking for advice about my next upgrade and got quite a bit of responses saying that a second GTX 570 would be bottlenecked by my 955 B.E even at 4.2Ghz.

I decided to give it a chance considering the cards will outlast my current platform regardless, and the performance gains would be noticeable nonetheless. I placed my Newegg order last night with the usual 3-day UPS shipping and was surprised to see it scheduled for delivery 2/10/2011. Its sitting in the UPS facility a couple hours away as of now...Newegg rocks as usual when it comes to getting me my parts fast and cheap .

So I have decided to create this thread to post my findings when it comes to sli scaling with a midrange AMD platform (Motherboard, CPU, Ram) that can be had for $350 shipped currently on Newegg. I have no other systems to compare it to, but I do have a multitude of benchmarks I can run, as well as enough games to gather a fairly large amount of data.

Many people have the belief that they absolutely need an i7 platform that can end up costing quite a bit more (unless you live close to a Frys or M/C in which case its a bit closer in cost) than my current platform. I am going to put out as much evidence as I can one way or another and allow you all to pass judgment on it.

I am not going to be looking at performance in encoding and other cpu based tests. This is purely about gaming performance. I think we all understand that intel is generally the faster platform when it comes to most things other than gaming, I just plan to see if it is also necessary when it comes to gaming.

My rig that will be used:

AMD 955 B.E C3 Revision at 4.2 GHz (24/7 O/C)
MSI NF980a-H65 980a motherboard with NB @ 2800
4GB (2x2GB) Gskill Ripjaws DDR3 @ 1600 8-8-8-24-1t
EVGA GTX570 (Tested at a single card running 925/1850/2000 @1.1V) SLI Test will be done at whatever clocks are stable
Seasonic X750 PSU

1920x1080 Resolution
This is possibly the biggest part of this thread and the argument about cpu bottlenecks. As the resolution increases the reliance on gpu power also increases. As the resolution decreases the CPU becomes more important. I cannot test at anything above 1920x1080, and will not bother testing below this mark as no one would purchase $700 of cards to use on a lower res (even this res is pushing it, I know....you've all told me)

So I will come back after work tomorrow and post up some before and after pics as well as some initial benchmarks if I have some time. I have no days off until Sunday, so don't expect too much in the way of major updates until then.

If you have any specific requests about what benchmarks/games I should run feel free to post and I will do my best to give it a shot.

Also if this is in the wrong section mods feel free to move it, just posted it here because I thought it fit here the best.
Edited by CRosko42 - 2/9/11 at 11:19pm
The Money Pit....
(14 items)
 
  
CPUMotherboardGraphicsRAM
i5 2500k Asus P8P67 Deluxe EVGA GTX 680 @ 1290 Boost clock 8gb GSkill Sniper 1600 8-8-8-24 
Hard DriveCoolingCoolingOS
Crucial M4 64GB + 1.5TB Seagate 7200rpm EK HF Supreme +Koolance VID-NX680 RX240 + 655 + Bitspower top + ek 250 adv res Windows 7 Ultimate 64 Bit 
MonitorPowerCase
Yamakasi Catleap Q270 1440p Seasonic X750 Corsair 800D 
  hide details  
Reply
The Money Pit....
(14 items)
 
  
CPUMotherboardGraphicsRAM
i5 2500k Asus P8P67 Deluxe EVGA GTX 680 @ 1290 Boost clock 8gb GSkill Sniper 1600 8-8-8-24 
Hard DriveCoolingCoolingOS
Crucial M4 64GB + 1.5TB Seagate 7200rpm EK HF Supreme +Koolance VID-NX680 RX240 + 655 + Bitspower top + ek 250 adv res Windows 7 Ultimate 64 Bit 
MonitorPowerCase
Yamakasi Catleap Q270 1440p Seasonic X750 Corsair 800D 
  hide details  
Reply
post #2 of 14
Thread Starter 
Reserved
The Money Pit....
(14 items)
 
  
CPUMotherboardGraphicsRAM
i5 2500k Asus P8P67 Deluxe EVGA GTX 680 @ 1290 Boost clock 8gb GSkill Sniper 1600 8-8-8-24 
Hard DriveCoolingCoolingOS
Crucial M4 64GB + 1.5TB Seagate 7200rpm EK HF Supreme +Koolance VID-NX680 RX240 + 655 + Bitspower top + ek 250 adv res Windows 7 Ultimate 64 Bit 
MonitorPowerCase
Yamakasi Catleap Q270 1440p Seasonic X750 Corsair 800D 
  hide details  
Reply
The Money Pit....
(14 items)
 
  
CPUMotherboardGraphicsRAM
i5 2500k Asus P8P67 Deluxe EVGA GTX 680 @ 1290 Boost clock 8gb GSkill Sniper 1600 8-8-8-24 
Hard DriveCoolingCoolingOS
Crucial M4 64GB + 1.5TB Seagate 7200rpm EK HF Supreme +Koolance VID-NX680 RX240 + 655 + Bitspower top + ek 250 adv res Windows 7 Ultimate 64 Bit 
MonitorPowerCase
Yamakasi Catleap Q270 1440p Seasonic X750 Corsair 800D 
  hide details  
Reply
post #3 of 14
Subscribe.
post #4 of 14
It'll rock.
Enjoy.
post #5 of 14
I don't see a problem running multiple cards with a 4Ghz Phenom II as I don't think there will be a bottleneck. Though you would get more performance from an i5/i7 setup.

If you want MAX performance you have to go Intel, there is no avoiding that. If you want great/amazing performance for the money you go AMD.
     
CPUMotherboardGraphicsRAM
Intel Core i3 4150 Gigabyte  Gigabyte Geforce GTX 1050 Ti 8GB DDR3 
OSMonitorPowerCase
Windows 10 x64 Samsung BX2450 300 Watt Bitfenix 
Audio
Small Amp + Pair of Realistic Minimus 26 
CPUMotherboardGraphicsRAM
Intel Core i5 4570 H81 EVGA Geforce GTX 750 Ti 8GB 1600Mhz DDR3 
Hard DriveOSMonitor
Western Digital 1TB 7200 RPM  Windows 10 Pro x64 HP w2007 
  hide details  
Reply
     
CPUMotherboardGraphicsRAM
Intel Core i3 4150 Gigabyte  Gigabyte Geforce GTX 1050 Ti 8GB DDR3 
OSMonitorPowerCase
Windows 10 x64 Samsung BX2450 300 Watt Bitfenix 
Audio
Small Amp + Pair of Realistic Minimus 26 
CPUMotherboardGraphicsRAM
Intel Core i5 4570 H81 EVGA Geforce GTX 750 Ti 8GB 1600Mhz DDR3 
Hard DriveOSMonitor
Western Digital 1TB 7200 RPM  Windows 10 Pro x64 HP w2007 
  hide details  
Reply
post #6 of 14
Quote:
Originally Posted by Horsemama1956 View Post
I don't see a problem running multiple cards with a 4Ghz Phenom II as I don't think there will be a bottleneck. Though you would get more performance from an i5/i7 setup.

If you want MAX performance you have to go Intel, there is no avoiding that. If you want great/amazing performance for the money you go AMD.
Word to your mother.
    
CPUMotherboardGraphicsRAM
FX 8120 ASUS M4A89GTD PRO/USB3 890GX 2 x Asus 5770 CuCore 2 x 2GB Kingston HyperX 1800Mhz 
Hard DriveCoolingOSMonitor
64 GB Kingston V-Series Err, 10 Gallons of Mineral Oil and a Automotive... Windows 7 x64 SP1 24" Asus LCD 
KeyboardPowerCaseMouse
Razer BlackWidow Ultimate 600W Seasonic Aquarium with ~10 gallons of mineral oil Razer DeathAdder 
Mouse Pad
Razer Kabuto 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
FX 8120 ASUS M4A89GTD PRO/USB3 890GX 2 x Asus 5770 CuCore 2 x 2GB Kingston HyperX 1800Mhz 
Hard DriveCoolingOSMonitor
64 GB Kingston V-Series Err, 10 Gallons of Mineral Oil and a Automotive... Windows 7 x64 SP1 24" Asus LCD 
KeyboardPowerCaseMouse
Razer BlackWidow Ultimate 600W Seasonic Aquarium with ~10 gallons of mineral oil Razer DeathAdder 
Mouse Pad
Razer Kabuto 
  hide details  
Reply
post #7 of 14
Well...see, that's the thing.
If an AMD quad gets a certain score, and then an Intel gets a higher score on the same games, with the same cards, it's going to be referred to as a "bottleneck", even if it ends up being a frame faster. I don't think it will be because the AMD CPU is holding back the GPUs, it will be because the Intel is faster.
Many people may understand that 3DMark and so forth are synthetic benches, but are indicative of performance as it relates to games. Which is true to a point. Taking away certain benchmarks and resolutions to prove a point undermines the whole concept of benchmarking.
Everyone KNOWS that Intel chips are faster, just as everyone knows that AMD based setups tend have a better price/performance ratio.
I love my 955. I just got it a month or so ago. But I think this thread is going to end up being more of an advocation/justification for sub $200 CPUs than anything else.
And unless you have a comparable Intel chip and mobo...I'm not sure as to what it's going to prove.
Blue Fury
(19 items)
 
  
CPUMotherboardGraphicsRAM
AMD A8 6600k asus Sapphire 290X Tri-OCX 4GB 2x4 (8)GB G-Skill Ripjaws 1600 
Hard DriveHard DriveHard DriveHard Drive
OCZVertex3 60GB Seagate 1TB WD 500GB Samsung 
Hard DriveOptical DriveCoolingOS
Samsung Evo 850 ASUS DVDRW24x Corsair H100 W10 
MonitorKeyboardPowerCase
55" 4k Hisense Razer Blackwidow Ultimate 2013 Corsair CX750 AzzA Hurrican 
MouseMouse PadAudio
Razer Deathadder 3500 Diablo3 RoS LE onboard 
  hide details  
Reply
Blue Fury
(19 items)
 
  
CPUMotherboardGraphicsRAM
AMD A8 6600k asus Sapphire 290X Tri-OCX 4GB 2x4 (8)GB G-Skill Ripjaws 1600 
Hard DriveHard DriveHard DriveHard Drive
OCZVertex3 60GB Seagate 1TB WD 500GB Samsung 
Hard DriveOptical DriveCoolingOS
Samsung Evo 850 ASUS DVDRW24x Corsair H100 W10 
MonitorKeyboardPowerCase
55" 4k Hisense Razer Blackwidow Ultimate 2013 Corsair CX750 AzzA Hurrican 
MouseMouse PadAudio
Razer Deathadder 3500 Diablo3 RoS LE onboard 
  hide details  
Reply
post #8 of 14
I originaly had a 4ghz PhenomII with a 2600 NB in my sig rig with the sli 480's using a 780a chipset. When I ran the crysis benchmark my FPS went up a whooping 3-4fps with the addition of a second 480.

Cue i7. It went up 30fps. That is a significant difference.
Frank
(13 items)
 
Dominatrix
(10 items)
 
 
CPUMotherboardGraphicsRAM
Core i7 4770K Asus Maxiums VI Extreme Gigabyte G1 Gaming 1080 16gb G.Skill Trident X 
Hard DriveHard DriveCoolingOS
Samsung 256gb 850 Pro 2Tb WB Black Corsair H100i GTX Win 10 x64 
MonitorKeyboardPowerCase
AOC U3477PQU Leopold Cherry Browns  EVGA 850W SuperNOVA G2 Fractal Design Define R5 
Audio
Asus Xonar D2X  
CPUMotherboardGraphicsRAM
i3 4370 Maximus VII Impact Asus 960 Strix 4gb Vengeance Pro 2400 2x8Gb 
Hard DriveOptical DriveOSPower
Samsung 128gb 830 3 x WD Red 3Tb Windows 8.1 x64 Corsair RM550 
Case
Fractal Node 304 
  hide details  
Reply
Frank
(13 items)
 
Dominatrix
(10 items)
 
 
CPUMotherboardGraphicsRAM
Core i7 4770K Asus Maxiums VI Extreme Gigabyte G1 Gaming 1080 16gb G.Skill Trident X 
Hard DriveHard DriveCoolingOS
Samsung 256gb 850 Pro 2Tb WB Black Corsair H100i GTX Win 10 x64 
MonitorKeyboardPowerCase
AOC U3477PQU Leopold Cherry Browns  EVGA 850W SuperNOVA G2 Fractal Design Define R5 
Audio
Asus Xonar D2X  
CPUMotherboardGraphicsRAM
i3 4370 Maximus VII Impact Asus 960 Strix 4gb Vengeance Pro 2400 2x8Gb 
Hard DriveOptical DriveOSPower
Samsung 128gb 830 3 x WD Red 3Tb Windows 8.1 x64 Corsair RM550 
Case
Fractal Node 304 
  hide details  
Reply
post #9 of 14
because the game is CPU intensive
Core 2 Haf
(16 items)
 
  
CPUMotherboardGraphicsRAM
C2Q Q9550 ASUS P5E3 Deluxe Wifi-AP GTX 480 Kingston HyperX DDR3 KHX1600C9D3K2/8G 
RAMHard DriveCoolingOS
PNY 1333 55gb, 320Gb, 500GB, 1TB Hyper 212 EVO Win 7 Ultimate 64bit 
OSOSPowerCase
Lubuntu 13.04 x64 OS X 10.8.5 TX750 Haf 912 
  hide details  
Reply
Core 2 Haf
(16 items)
 
  
CPUMotherboardGraphicsRAM
C2Q Q9550 ASUS P5E3 Deluxe Wifi-AP GTX 480 Kingston HyperX DDR3 KHX1600C9D3K2/8G 
RAMHard DriveCoolingOS
PNY 1333 55gb, 320Gb, 500GB, 1TB Hyper 212 EVO Win 7 Ultimate 64bit 
OSOSPowerCase
Lubuntu 13.04 x64 OS X 10.8.5 TX750 Haf 912 
  hide details  
Reply
post #10 of 14
And that's the thing. Whichever game is tested will produce different results, and based on those results people will either say it's the CPU holding the GPU back, or the games is CPU intensive, or what have you.
The only way I think this can be tested would be to get a whole range of GPUs and then make the AMD and Intel setups the control factor. You could then figure out the ratio of scaling in proportion to the CPUs themselves.
Which is sort of like how most professional benches work, except they ALWAYS use the highest end rig so as to not "limit" the GPU. I understand the rationale, but I fail to see the practicality in testing say, a 5750 on an i7 980X OC'd through the roof with 12GB of RAM.
People who can afford those CPUs don't buy that level of GPU...
Anyway, get very comparable Intel/AMD setups and a big bunch of cards and then you can test for "bottlenecking".
Blue Fury
(19 items)
 
  
CPUMotherboardGraphicsRAM
AMD A8 6600k asus Sapphire 290X Tri-OCX 4GB 2x4 (8)GB G-Skill Ripjaws 1600 
Hard DriveHard DriveHard DriveHard Drive
OCZVertex3 60GB Seagate 1TB WD 500GB Samsung 
Hard DriveOptical DriveCoolingOS
Samsung Evo 850 ASUS DVDRW24x Corsair H100 W10 
MonitorKeyboardPowerCase
55" 4k Hisense Razer Blackwidow Ultimate 2013 Corsair CX750 AzzA Hurrican 
MouseMouse PadAudio
Razer Deathadder 3500 Diablo3 RoS LE onboard 
  hide details  
Reply
Blue Fury
(19 items)
 
  
CPUMotherboardGraphicsRAM
AMD A8 6600k asus Sapphire 290X Tri-OCX 4GB 2x4 (8)GB G-Skill Ripjaws 1600 
Hard DriveHard DriveHard DriveHard Drive
OCZVertex3 60GB Seagate 1TB WD 500GB Samsung 
Hard DriveOptical DriveCoolingOS
Samsung Evo 850 ASUS DVDRW24x Corsair H100 W10 
MonitorKeyboardPowerCase
55" 4k Hisense Razer Blackwidow Ultimate 2013 Corsair CX750 AzzA Hurrican 
MouseMouse PadAudio
Razer Deathadder 3500 Diablo3 RoS LE onboard 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD - General
Overclock.net › Forums › AMD › AMD - General › Adressing the Notion AMD CPUs Will Bottleneck a High End SLI/CF Setup