Overclock.net › Forums › Graphics Cards › AMD/ATI › [Official] AMD Radeon HD 7950/7970/7990 Owners Thread
New Posts  All Forums:Forum Nav:

[Official] AMD Radeon HD 7950/7970/7990 Owners Thread - Page 826

post #8251 of 38704
Quote:
Originally Posted by tsm106 View Post

Quote:
Originally Posted by RagingCain View Post

You are saying you should only focus on your weakest cards overclock, but that is the same error in logic you are claiming another guy is having.  If you are running CrossfireX 24/7, then your fastest route to stability is finding your max stable OC at a voltage you are comfortable with, with Crossfire enabled.  CrossfireX can weaken (not saying it will) your overclock even further than your weakest card.  Thats factual, and its been proven many times in the past.  Crossfire and SLI adds overhead, and one can only lose overclocking potential with CrossfireX, not gain any, so you should test accordingly.

 


I've said no such thing. You're saying that and arguing about it. And as for losing overclock potential, overhead, sli, etc, that's got nothing to do with my point. Which again is that it's foolish not to know the limit of your weakest card. It's that simple. It doesn't matter if your best card overclocks 100mhz more cuz your array will lockup when you unknowingly hit your weakest cards limit! Duh, then you're like ugh what happened scratch head. Btw, I don't know what's factual because it's not fact just because you say so. I've my own findings and have not seen my cards lose their overclocking ability inside an array BECAUSE I know the limit of my weakest card.

There's also another benefit to testing individually. Once you have found your strongest to weakest card, use your best card for the main card and the weaker ones as the slave card. The main card has to do the most work, run the displays, desktop, have all the cards synch thru it. You wouldn't want your weakest card to bare the brunt of that duty.

Btw, is it fact or not though you disclaim that you're not saying it will?


Regarding Overhead:  You will have to follow benchmarkers and more to the extreme benchmarking sections of forums and what not.  I know of the info I have seen over time, but not anything that stands out right at the second as the end all be all of explaing CFX/SLI overhead.  Instead go over to HWbot for the multiple card category submission's, you see the trending of multi-cards even on the professional bencher's who cherry pick multiple GPUs.

 

One example, while not definitive, and I am sure there are more, and maybe even a few categories where they managed to get CrossfireX higher than their stock run for various reasons.  This is just to illustrate the general case of adding cards to CFX or SLI.  What you are asking me to prove beyond doubt to something you don't even believe is possible.  Much like knowing about the bottleneck bandwidth issue in the Crossfire bridges, just something that you pick up overtime.

Rna.PNG

 

 

Again, this is not PROOF, this is just an example of what you may see, but like I said its only one case, and not guaranteed to be the reason why his clocks were limited, but this is atypical of what to expect.

 

When overclocking, you should go by your weakest link, like you were saying, but in CrossfireX, your weakest link is the fact that it is enabled.  It only makes sense to test for stability in a complete Crossfire or SLI setup, not individual cards.

 

Your "strongest" card according to you should be your main card, however, this is also a choice born out of illogical thinking, same GPUs and the same clocks should be the same, when both are clocked at the same speed (i.e. your weakest card) then they are both equal in that they are both matching.  What should be your main card, is the card that can achieve the intended clocks at the lowest possible voltage of the two or three gpus.  CrossfireX / SLI always produce more heat in the top card, therefore the card requiring the least voltage creating the least excess heat should be the top card.  If you are getting that specific.  This GPU is usually associated as your "strongest" GPU, when in fact it is better to call it the more "efficient" GPU.


Edited by RagingCain - 4/2/12 at 12:00pm
Snowdevil
(16 items)
 
ASUS G750JM
(9 items)
 
 
CPUMotherboardGraphicsGraphics
[i7 4790K @ 4.4 GHz (1.186v)] [Asus Sabertooth Z97 Mark S] [nVidia Geforce GTX 1080] [nVidia Geforce GTX 1080] 
RAMHard DriveCoolingOS
[G.Skill 32GB DDR3 2133 MHz] [Crucial MX100 256GB] [Phanteks PH-TC12DX] [Win 10.1 Pro] 
MonitorMonitorKeyboardPower
[LG 29UM65 (2560x1080)] [QNIX Evo II LED (2560x1440)] [WASD v2 Tenkeyless] [NZXT Hale90 v2 ] 
CaseMouseMouse PadAudio
[ThermalTake GT10 Snow Edition] [Razer Mamba - Chroma] [Razer Kabuto] [Razer Man O' War] 
CPUMotherboardGraphicsRAM
i7 4770HQ Intel HM87 Express Chipset Geforce GTX 860M 8GB DDR3L 1600 MHz 
Hard DriveOptical DriveCoolingOS
Samsung SSD EVO DVD-RW Stock Windows 8.1 
Monitor
1920x1080 TN 
  hide details  
Reply
Snowdevil
(16 items)
 
ASUS G750JM
(9 items)
 
 
CPUMotherboardGraphicsGraphics
[i7 4790K @ 4.4 GHz (1.186v)] [Asus Sabertooth Z97 Mark S] [nVidia Geforce GTX 1080] [nVidia Geforce GTX 1080] 
RAMHard DriveCoolingOS
[G.Skill 32GB DDR3 2133 MHz] [Crucial MX100 256GB] [Phanteks PH-TC12DX] [Win 10.1 Pro] 
MonitorMonitorKeyboardPower
[LG 29UM65 (2560x1080)] [QNIX Evo II LED (2560x1440)] [WASD v2 Tenkeyless] [NZXT Hale90 v2 ] 
CaseMouseMouse PadAudio
[ThermalTake GT10 Snow Edition] [Razer Mamba - Chroma] [Razer Kabuto] [Razer Man O' War] 
CPUMotherboardGraphicsRAM
i7 4770HQ Intel HM87 Express Chipset Geforce GTX 860M 8GB DDR3L 1600 MHz 
Hard DriveOptical DriveCoolingOS
Samsung SSD EVO DVD-RW Stock Windows 8.1 
Monitor
1920x1080 TN 
  hide details  
Reply
post #8252 of 38704
metro1375.png
metro1575.png
metro1875.png
Has
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7 4770K Asus Maximus VI Impact Sapphire 290X Tri-X OC 8GB G.Skill DDR3 2800MHz PC3-22400 TridentX 
Hard DriveCoolingOSMonitor
Samsung 830 256GB Corsair H100 Windows 8.1 x64 Dell 3xU2311H 
PowerCase
SilverStone 750W SilverStone TJ07 
  hide details  
Reply
Has
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7 4770K Asus Maximus VI Impact Sapphire 290X Tri-X OC 8GB G.Skill DDR3 2800MHz PC3-22400 TridentX 
Hard DriveCoolingOSMonitor
Samsung 830 256GB Corsair H100 Windows 8.1 x64 Dell 3xU2311H 
PowerCase
SilverStone 750W SilverStone TJ07 
  hide details  
Reply
post #8253 of 38704
My rule of thumb for memory OCing has just been to push it up 150-200mhz or so and be happy with that. Real world experience has borne out that the GPU OC is way more important to overall performance. However as the guy above me has shown there is obviously some benefit to doing it.
    
CPUMotherboardGraphicsRAM
6700k (EK Supremacy Evo) *4.7 @ 1.312 Asus Maximus VIII gene GTX 1080 FE (EK nickel/plexi)* 2075/5401 16gb Corsair 3200  
Hard DriveHard DriveHard DriveCooling
Samsung 950pro M.2 256gb Samsung 850 Evo - 512 - Steam Samsung 850 Evo - 512 - Other Apps EK Res 
CoolingCoolingCoolingOS
EK pump top XSPC EX 360 XSPC EX 240 10 
KeyboardPowerCaseMouse
Ducky Shine 3 TKL  EVGA 650 Fractal Arc Mini 2 G303 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
6700k (EK Supremacy Evo) *4.7 @ 1.312 Asus Maximus VIII gene GTX 1080 FE (EK nickel/plexi)* 2075/5401 16gb Corsair 3200  
Hard DriveHard DriveHard DriveCooling
Samsung 950pro M.2 256gb Samsung 850 Evo - 512 - Steam Samsung 850 Evo - 512 - Other Apps EK Res 
CoolingCoolingCoolingOS
EK pump top XSPC EX 360 XSPC EX 240 10 
KeyboardPowerCaseMouse
Ducky Shine 3 TKL  EVGA 650 Fractal Arc Mini 2 G303 
  hide details  
Reply
post #8254 of 38704
Quote:
Originally Posted by zpaf View Post

metro1375.png
metro1575.png
metro1875.png


Very odd that your 200 MHz OC showed virtually no gain at all, and within error of margin, yet your jump from 1575 to 1875 showed in increase just shy of 10%.

 

Either the voltage was too low for the 1575 MHz, meaning ECC was getting involved, and 1875 MHz had adequate voltage, or something else is going on here.

 

Very excited to get some numbers out when I get home.  Analyze the numbers and flesh out some kind of scaling.

Snowdevil
(16 items)
 
ASUS G750JM
(9 items)
 
 
CPUMotherboardGraphicsGraphics
[i7 4790K @ 4.4 GHz (1.186v)] [Asus Sabertooth Z97 Mark S] [nVidia Geforce GTX 1080] [nVidia Geforce GTX 1080] 
RAMHard DriveCoolingOS
[G.Skill 32GB DDR3 2133 MHz] [Crucial MX100 256GB] [Phanteks PH-TC12DX] [Win 10.1 Pro] 
MonitorMonitorKeyboardPower
[LG 29UM65 (2560x1080)] [QNIX Evo II LED (2560x1440)] [WASD v2 Tenkeyless] [NZXT Hale90 v2 ] 
CaseMouseMouse PadAudio
[ThermalTake GT10 Snow Edition] [Razer Mamba - Chroma] [Razer Kabuto] [Razer Man O' War] 
CPUMotherboardGraphicsRAM
i7 4770HQ Intel HM87 Express Chipset Geforce GTX 860M 8GB DDR3L 1600 MHz 
Hard DriveOptical DriveCoolingOS
Samsung SSD EVO DVD-RW Stock Windows 8.1 
Monitor
1920x1080 TN 
  hide details  
Reply
Snowdevil
(16 items)
 
ASUS G750JM
(9 items)
 
 
CPUMotherboardGraphicsGraphics
[i7 4790K @ 4.4 GHz (1.186v)] [Asus Sabertooth Z97 Mark S] [nVidia Geforce GTX 1080] [nVidia Geforce GTX 1080] 
RAMHard DriveCoolingOS
[G.Skill 32GB DDR3 2133 MHz] [Crucial MX100 256GB] [Phanteks PH-TC12DX] [Win 10.1 Pro] 
MonitorMonitorKeyboardPower
[LG 29UM65 (2560x1080)] [QNIX Evo II LED (2560x1440)] [WASD v2 Tenkeyless] [NZXT Hale90 v2 ] 
CaseMouseMouse PadAudio
[ThermalTake GT10 Snow Edition] [Razer Mamba - Chroma] [Razer Kabuto] [Razer Man O' War] 
CPUMotherboardGraphicsRAM
i7 4770HQ Intel HM87 Express Chipset Geforce GTX 860M 8GB DDR3L 1600 MHz 
Hard DriveOptical DriveCoolingOS
Samsung SSD EVO DVD-RW Stock Windows 8.1 
Monitor
1920x1080 TN 
  hide details  
Reply
post #8255 of 38704
Results of my four cards.
All done with 1.25 volts on the core and 1.75 volts on the memory

Card 1: ASIC 63.6%, 1200/1850
Card 2: ASIC 73.7%, 1250/1900
Card 3: ASIC 76.2%, 1275/1850
Card 4: ASIC 81.0%, 1225/2000

Now to fire up the R4E for some 4-way, go for 75k+ in vantage
post #8256 of 38704
Quote:
Originally Posted by RagingCain View Post



Very odd that your 200 MHz OC showed virtually no gain at all, and within error of margin, yet your jump from 1575 to 1875 showed in increase just shy of 10%.

Either the voltage was too low for the 1575 MHz, meaning ECC was getting involved, and 1875 MHz had adequate voltage, or something else is going on here.

Very excited to get some numbers out when I get home.  Analyze the numbers and flesh out some kind of scaling.

On all screenshots I am with default core/mem volts (1.175/1.6)
Has
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7 4770K Asus Maximus VI Impact Sapphire 290X Tri-X OC 8GB G.Skill DDR3 2800MHz PC3-22400 TridentX 
Hard DriveCoolingOSMonitor
Samsung 830 256GB Corsair H100 Windows 8.1 x64 Dell 3xU2311H 
PowerCase
SilverStone 750W SilverStone TJ07 
  hide details  
Reply
Has
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7 4770K Asus Maximus VI Impact Sapphire 290X Tri-X OC 8GB G.Skill DDR3 2800MHz PC3-22400 TridentX 
Hard DriveCoolingOSMonitor
Samsung 830 256GB Corsair H100 Windows 8.1 x64 Dell 3xU2311H 
PowerCase
SilverStone 750W SilverStone TJ07 
  hide details  
Reply
post #8257 of 38704
Quote:
Originally Posted by zpaf View Post

Quote:
Originally Posted by RagingCain View Post

Very odd that your 200 MHz OC showed virtually no gain at all, and within error of margin, yet your jump from 1575 to 1875 showed in increase just shy of 10%.

Either the voltage was too low for the 1575 MHz, meaning ECC was getting involved, and 1875 MHz had adequate voltage, or something else is going on here.

Very excited to get some numbers out when I get home.  Analyze the numbers and flesh out some kind of scaling.

On all screenshots I am with default core/mem volts (1.175/1.6)

 

Want to run the first two tests again?  The first test could be on the high end of the bell-curve for stock, and the second test could be on the low-end of the curve for the first OC results.  Explaining how close they are.

 

If it was stable enough for 1875 MHz @ 1.6v. I would assume 1575@ 1.6v was stable too.

Snowdevil
(16 items)
 
ASUS G750JM
(9 items)
 
 
CPUMotherboardGraphicsGraphics
[i7 4790K @ 4.4 GHz (1.186v)] [Asus Sabertooth Z97 Mark S] [nVidia Geforce GTX 1080] [nVidia Geforce GTX 1080] 
RAMHard DriveCoolingOS
[G.Skill 32GB DDR3 2133 MHz] [Crucial MX100 256GB] [Phanteks PH-TC12DX] [Win 10.1 Pro] 
MonitorMonitorKeyboardPower
[LG 29UM65 (2560x1080)] [QNIX Evo II LED (2560x1440)] [WASD v2 Tenkeyless] [NZXT Hale90 v2 ] 
CaseMouseMouse PadAudio
[ThermalTake GT10 Snow Edition] [Razer Mamba - Chroma] [Razer Kabuto] [Razer Man O' War] 
CPUMotherboardGraphicsRAM
i7 4770HQ Intel HM87 Express Chipset Geforce GTX 860M 8GB DDR3L 1600 MHz 
Hard DriveOptical DriveCoolingOS
Samsung SSD EVO DVD-RW Stock Windows 8.1 
Monitor
1920x1080 TN 
  hide details  
Reply
Snowdevil
(16 items)
 
ASUS G750JM
(9 items)
 
 
CPUMotherboardGraphicsGraphics
[i7 4790K @ 4.4 GHz (1.186v)] [Asus Sabertooth Z97 Mark S] [nVidia Geforce GTX 1080] [nVidia Geforce GTX 1080] 
RAMHard DriveCoolingOS
[G.Skill 32GB DDR3 2133 MHz] [Crucial MX100 256GB] [Phanteks PH-TC12DX] [Win 10.1 Pro] 
MonitorMonitorKeyboardPower
[LG 29UM65 (2560x1080)] [QNIX Evo II LED (2560x1440)] [WASD v2 Tenkeyless] [NZXT Hale90 v2 ] 
CaseMouseMouse PadAudio
[ThermalTake GT10 Snow Edition] [Razer Mamba - Chroma] [Razer Kabuto] [Razer Man O' War] 
CPUMotherboardGraphicsRAM
i7 4770HQ Intel HM87 Express Chipset Geforce GTX 860M 8GB DDR3L 1600 MHz 
Hard DriveOptical DriveCoolingOS
Samsung SSD EVO DVD-RW Stock Windows 8.1 
Monitor
1920x1080 TN 
  hide details  
Reply
post #8258 of 38704
Quote:
Originally Posted by RagingCain View Post


Want to run the first two tests again?  The first test could be on the high end of the bell-curve for stock, and the second test could be on the low-end of the curve for the first OC results.  Explaining how close they are.

If it was stable enough for 1875 MHz @ 1.6v. I would assume 1575@ 1.6v was stable too.

You want metro or avp ?
Tell me exactly the clocks you want for these tests.
Has
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7 4770K Asus Maximus VI Impact Sapphire 290X Tri-X OC 8GB G.Skill DDR3 2800MHz PC3-22400 TridentX 
Hard DriveCoolingOSMonitor
Samsung 830 256GB Corsair H100 Windows 8.1 x64 Dell 3xU2311H 
PowerCase
SilverStone 750W SilverStone TJ07 
  hide details  
Reply
Has
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7 4770K Asus Maximus VI Impact Sapphire 290X Tri-X OC 8GB G.Skill DDR3 2800MHz PC3-22400 TridentX 
Hard DriveCoolingOSMonitor
Samsung 830 256GB Corsair H100 Windows 8.1 x64 Dell 3xU2311H 
PowerCase
SilverStone 750W SilverStone TJ07 
  hide details  
Reply
post #8259 of 38704
Quote:
Originally Posted by zpaf View Post

Quote:
Originally Posted by RagingCain View Post


Want to run the first two tests again?  The first test could be on the high end of the bell-curve for stock, and the second test could be on the low-end of the curve for the first OC results.  Explaining how close they are.

If it was stable enough for 1875 MHz @ 1.6v. I would assume 1575@ 1.6v was stable too.

You want metro or avp ?
Tell me exactly the clocks you want for these tests.


Metro, same settings, 1125 GPU w/ 1375 MHz Mem, then again with 1125 GPU w/ 1575 Mem.

Snowdevil
(16 items)
 
ASUS G750JM
(9 items)
 
 
CPUMotherboardGraphicsGraphics
[i7 4790K @ 4.4 GHz (1.186v)] [Asus Sabertooth Z97 Mark S] [nVidia Geforce GTX 1080] [nVidia Geforce GTX 1080] 
RAMHard DriveCoolingOS
[G.Skill 32GB DDR3 2133 MHz] [Crucial MX100 256GB] [Phanteks PH-TC12DX] [Win 10.1 Pro] 
MonitorMonitorKeyboardPower
[LG 29UM65 (2560x1080)] [QNIX Evo II LED (2560x1440)] [WASD v2 Tenkeyless] [NZXT Hale90 v2 ] 
CaseMouseMouse PadAudio
[ThermalTake GT10 Snow Edition] [Razer Mamba - Chroma] [Razer Kabuto] [Razer Man O' War] 
CPUMotherboardGraphicsRAM
i7 4770HQ Intel HM87 Express Chipset Geforce GTX 860M 8GB DDR3L 1600 MHz 
Hard DriveOptical DriveCoolingOS
Samsung SSD EVO DVD-RW Stock Windows 8.1 
Monitor
1920x1080 TN 
  hide details  
Reply
Snowdevil
(16 items)
 
ASUS G750JM
(9 items)
 
 
CPUMotherboardGraphicsGraphics
[i7 4790K @ 4.4 GHz (1.186v)] [Asus Sabertooth Z97 Mark S] [nVidia Geforce GTX 1080] [nVidia Geforce GTX 1080] 
RAMHard DriveCoolingOS
[G.Skill 32GB DDR3 2133 MHz] [Crucial MX100 256GB] [Phanteks PH-TC12DX] [Win 10.1 Pro] 
MonitorMonitorKeyboardPower
[LG 29UM65 (2560x1080)] [QNIX Evo II LED (2560x1440)] [WASD v2 Tenkeyless] [NZXT Hale90 v2 ] 
CaseMouseMouse PadAudio
[ThermalTake GT10 Snow Edition] [Razer Mamba - Chroma] [Razer Kabuto] [Razer Man O' War] 
CPUMotherboardGraphicsRAM
i7 4770HQ Intel HM87 Express Chipset Geforce GTX 860M 8GB DDR3L 1600 MHz 
Hard DriveOptical DriveCoolingOS
Samsung SSD EVO DVD-RW Stock Windows 8.1 
Monitor
1920x1080 TN 
  hide details  
Reply
post #8260 of 38704
Quote:
Originally Posted by RagingCain View Post



Metro, same settings, 1125 GPU w/ 1375 MHz Mem, then again with 1125 GPU w/ 1575 Mem.

Here we are.

metro13751.png
metro15751.png
Has
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7 4770K Asus Maximus VI Impact Sapphire 290X Tri-X OC 8GB G.Skill DDR3 2800MHz PC3-22400 TridentX 
Hard DriveCoolingOSMonitor
Samsung 830 256GB Corsair H100 Windows 8.1 x64 Dell 3xU2311H 
PowerCase
SilverStone 750W SilverStone TJ07 
  hide details  
Reply
Has
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7 4770K Asus Maximus VI Impact Sapphire 290X Tri-X OC 8GB G.Skill DDR3 2800MHz PC3-22400 TridentX 
Hard DriveCoolingOSMonitor
Samsung 830 256GB Corsair H100 Windows 8.1 x64 Dell 3xU2311H 
PowerCase
SilverStone 750W SilverStone TJ07 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD/ATI
Overclock.net › Forums › Graphics Cards › AMD/ATI › [Official] AMD Radeon HD 7950/7970/7990 Owners Thread