Overclock.net › Forums › Components › Memory › So I adjusted my memory timings...
New Posts  All Forums:Forum Nav:

So I adjusted my memory timings... - Page 2

post #11 of 17
Certain scenarios benefit very much from oc or high spec ram.
i5 3570k
(16 items)
 
 
Fat Belly
(9 items)
 
CPUMotherboardGraphicsRAM
i5 3570k Asus P8Z77-VPRO Gigabyte 670 oc Patriot Viper Extreme  
Hard DriveHard DriveOptical DriveCooling
Samsung 830 64 GB  Crucial M4 64GB ihas424 Scythe mugen rev.b 
OSMonitorKeyboardPower
Windows 7 64 Bit Apple Cinema 30 Optimus Maximus  XFX 650  
CaseMouseMouse PadAudio
Lian Li T60R Razer Abyssus Razer Sphex Sennheiser 558 
CPUMotherboardGraphicsRAM
athlon 64 4000+ Asus A8N-SLI Deluxe gts 250 4gb ddr-400 
Hard DriveOSPowerCase
Seagate 150GB windows xp Antec Basiq Series VP450 thermaltake armor 
CPUMotherboardGraphicsRAM
e4300 ga-965p-ds3 ati rage xl 4gb ddr2-667 
Hard DriveOSPowerCase
samsung 2tb Windows 7 Thermaltake Tr2 aerocool engine t3 
Other
Perc 6i 
  hide details  
Reply
i5 3570k
(16 items)
 
 
Fat Belly
(9 items)
 
CPUMotherboardGraphicsRAM
i5 3570k Asus P8Z77-VPRO Gigabyte 670 oc Patriot Viper Extreme  
Hard DriveHard DriveOptical DriveCooling
Samsung 830 64 GB  Crucial M4 64GB ihas424 Scythe mugen rev.b 
OSMonitorKeyboardPower
Windows 7 64 Bit Apple Cinema 30 Optimus Maximus  XFX 650  
CaseMouseMouse PadAudio
Lian Li T60R Razer Abyssus Razer Sphex Sennheiser 558 
CPUMotherboardGraphicsRAM
athlon 64 4000+ Asus A8N-SLI Deluxe gts 250 4gb ddr-400 
Hard DriveOSPowerCase
Seagate 150GB windows xp Antec Basiq Series VP450 thermaltake armor 
CPUMotherboardGraphicsRAM
e4300 ga-965p-ds3 ati rage xl 4gb ddr2-667 
Hard DriveOSPowerCase
samsung 2tb Windows 7 Thermaltake Tr2 aerocool engine t3 
Other
Perc 6i 
  hide details  
Reply
post #12 of 17
My..what nice RAMs you have there..drool.gif
 
Thuban Powah!
(14 items)
 
 
CPUMotherboardGraphicsRAM
Xeon X5650 Asus P6X58D-E Sapphire HD 7950 G.Skill Trident-X 2600 CL10 
Hard DriveCoolingPowerCase
Samsung 840 pro Bong cooler / RASA waterblock Rosewill Capstone Gold 750 None 
MouseMouse PadAudio
cheap! Newegg box panel ibeats with onboard. 
CPUMotherboardGraphicsRAM
AMD Phenom II X6 1075T w/2 cores locked @ 4.2+ M4a89GTD-Pro XFX Radeon HD 5870 8gb Samsung ddr3 1600 11-11-11-28@ 8-8-8-24 
Hard DriveHard DriveOptical DriveCooling
OCZ Agility2 40gb WD Blue 500GB Lite-On RASA waterblock 
OSMonitorKeyboardPower
Windows & Linux Samsung 1080p 2 ms Dell Thermaltake TR2-800 
CaseMouse
Cooler Master Elite 330 (was) cheap 
  hide details  
Reply
 
Thuban Powah!
(14 items)
 
 
CPUMotherboardGraphicsRAM
Xeon X5650 Asus P6X58D-E Sapphire HD 7950 G.Skill Trident-X 2600 CL10 
Hard DriveCoolingPowerCase
Samsung 840 pro Bong cooler / RASA waterblock Rosewill Capstone Gold 750 None 
MouseMouse PadAudio
cheap! Newegg box panel ibeats with onboard. 
CPUMotherboardGraphicsRAM
AMD Phenom II X6 1075T w/2 cores locked @ 4.2+ M4a89GTD-Pro XFX Radeon HD 5870 8gb Samsung ddr3 1600 11-11-11-28@ 8-8-8-24 
Hard DriveHard DriveOptical DriveCooling
OCZ Agility2 40gb WD Blue 500GB Lite-On RASA waterblock 
OSMonitorKeyboardPower
Windows & Linux Samsung 1080p 2 ms Dell Thermaltake TR2-800 
CaseMouse
Cooler Master Elite 330 (was) cheap 
  hide details  
Reply
post #13 of 17
Quote:
Originally Posted by ASSEMbler View Post

Certain scenarios benefit very much from oc or high spec ram.

If you have test data from a reputable source that shows significant gains in real applications with the RAM running above ~1333 MHz., please share. Gains from tighter latencies at or above ~1333 MHz. are even less than the minute gains in real apps from increases in frequency as one clock cycle in real time decreases as the frequency increases, thus latency changes become less and less real time change per clock cycle.

The extensive independent DDR3 RAM testing by Tom's Hardware, Anandtech and X-Bit's labs on both AMD and Intel desktop PCs did not show any significant gains once the RAM frequency was at or above ~1333 MHz. Even OCN's "damric" Far Cry 2 testing data, where he did a 50% OC from 1333 MHz. CL9 to 2000 MHz. CL8, showed no significant gains.

http://www.overclock.net/t/1237178/is-o-cing-my-ram-worth-the-trouble/10#post_16866022

"Significant" for discussion of RAM changes means when you make a change to the RAM frequency or latency when the RAM is @ ~1333 MHz. - that you can actually see or feel the change when running real applications. Synthetic benches often show theoretical gains that do not actually occur in real applications.

http://www.overclock.net/t/1264141/memory-gskill-tridentx-2400-mhz-at-crosshair-v-formula-and-fx-8150

I always recommend that people conduct their own testing using real applications so they can see firsthand that there are no significant gains once you get above ~1333 MHz. because DDR3 RAM at ~1333 MHz. is simply not a system bottleneck. More bandwidth doesn't buy you any significant performance gain if there is no bottleneck.

As mentioned in other threads Llano is an exception as it's GPU section of the APU can take advantage of the increased frequency of higher frequency RAM above 1333 MHz.
Edited by AMD4ME - 6/24/12 at 5:26pm
post #14 of 17
ok..my "real test" is IBT ...there is about a 10 GFLOP difference between running @1600 and 2133Mhz..with
the higher bandwidth providing more throughput.
 
Thuban Powah!
(14 items)
 
 
CPUMotherboardGraphicsRAM
Xeon X5650 Asus P6X58D-E Sapphire HD 7950 G.Skill Trident-X 2600 CL10 
Hard DriveCoolingPowerCase
Samsung 840 pro Bong cooler / RASA waterblock Rosewill Capstone Gold 750 None 
MouseMouse PadAudio
cheap! Newegg box panel ibeats with onboard. 
CPUMotherboardGraphicsRAM
AMD Phenom II X6 1075T w/2 cores locked @ 4.2+ M4a89GTD-Pro XFX Radeon HD 5870 8gb Samsung ddr3 1600 11-11-11-28@ 8-8-8-24 
Hard DriveHard DriveOptical DriveCooling
OCZ Agility2 40gb WD Blue 500GB Lite-On RASA waterblock 
OSMonitorKeyboardPower
Windows & Linux Samsung 1080p 2 ms Dell Thermaltake TR2-800 
CaseMouse
Cooler Master Elite 330 (was) cheap 
  hide details  
Reply
 
Thuban Powah!
(14 items)
 
 
CPUMotherboardGraphicsRAM
Xeon X5650 Asus P6X58D-E Sapphire HD 7950 G.Skill Trident-X 2600 CL10 
Hard DriveCoolingPowerCase
Samsung 840 pro Bong cooler / RASA waterblock Rosewill Capstone Gold 750 None 
MouseMouse PadAudio
cheap! Newegg box panel ibeats with onboard. 
CPUMotherboardGraphicsRAM
AMD Phenom II X6 1075T w/2 cores locked @ 4.2+ M4a89GTD-Pro XFX Radeon HD 5870 8gb Samsung ddr3 1600 11-11-11-28@ 8-8-8-24 
Hard DriveHard DriveOptical DriveCooling
OCZ Agility2 40gb WD Blue 500GB Lite-On RASA waterblock 
OSMonitorKeyboardPower
Windows & Linux Samsung 1080p 2 ms Dell Thermaltake TR2-800 
CaseMouse
Cooler Master Elite 330 (was) cheap 
  hide details  
Reply
post #15 of 17
Quote:
Originally Posted by Schmuckley View Post

ok..my "real test" is IBT ...there is about a 10 GFLOP difference between running @1600 and 2133Mhz..with
the higher bandwidth providing more throughput.

Try using real applications if you want to see real performance, not synthetic benches that show theoretical performance. FWIW more bandwidth does not equal increased performance if the current bandwidth is not a bottleneck, which DDR3 RAM @ ~1333 MHz. is NOT - which is the whole point. This is why there are only minute gains with OC'ing frequencies or tighter latencies above ~1333 Mhz.
Edited by AMD4ME - 6/25/12 at 6:17am
post #16 of 17
Quote:
Originally Posted by AMD4ME View Post

Try using real applications if you want to see real performance, not synthetic benches that show theoretical performance. FWIW more bandwidth does not equal increased performance if the current bandwidth is not a bottleneck, which DDR3 RAM @ ~1333 MHz. is NOT - which is the whole point. This is why there are only minute gains with OC'ing frequencies or tighter latencies above ~1333 Mhz.

Minute gains=still gains tongue.gif
One thing I've noticed with higher frequency is being able to use the desktop a second or two faster...all the systray stuff gets loaded faster.
 
Thuban Powah!
(14 items)
 
 
CPUMotherboardGraphicsRAM
Xeon X5650 Asus P6X58D-E Sapphire HD 7950 G.Skill Trident-X 2600 CL10 
Hard DriveCoolingPowerCase
Samsung 840 pro Bong cooler / RASA waterblock Rosewill Capstone Gold 750 None 
MouseMouse PadAudio
cheap! Newegg box panel ibeats with onboard. 
CPUMotherboardGraphicsRAM
AMD Phenom II X6 1075T w/2 cores locked @ 4.2+ M4a89GTD-Pro XFX Radeon HD 5870 8gb Samsung ddr3 1600 11-11-11-28@ 8-8-8-24 
Hard DriveHard DriveOptical DriveCooling
OCZ Agility2 40gb WD Blue 500GB Lite-On RASA waterblock 
OSMonitorKeyboardPower
Windows & Linux Samsung 1080p 2 ms Dell Thermaltake TR2-800 
CaseMouse
Cooler Master Elite 330 (was) cheap 
  hide details  
Reply
 
Thuban Powah!
(14 items)
 
 
CPUMotherboardGraphicsRAM
Xeon X5650 Asus P6X58D-E Sapphire HD 7950 G.Skill Trident-X 2600 CL10 
Hard DriveCoolingPowerCase
Samsung 840 pro Bong cooler / RASA waterblock Rosewill Capstone Gold 750 None 
MouseMouse PadAudio
cheap! Newegg box panel ibeats with onboard. 
CPUMotherboardGraphicsRAM
AMD Phenom II X6 1075T w/2 cores locked @ 4.2+ M4a89GTD-Pro XFX Radeon HD 5870 8gb Samsung ddr3 1600 11-11-11-28@ 8-8-8-24 
Hard DriveHard DriveOptical DriveCooling
OCZ Agility2 40gb WD Blue 500GB Lite-On RASA waterblock 
OSMonitorKeyboardPower
Windows & Linux Samsung 1080p 2 ms Dell Thermaltake TR2-800 
CaseMouse
Cooler Master Elite 330 (was) cheap 
  hide details  
Reply
post #17 of 17
Quote:
Originally Posted by Schmuckley View Post

Minute gains=still gains tongue.gif
One thing I've noticed with higher frequency is being able to use the desktop a second or two faster...all the systray stuff gets loaded faster.

No one is arguing that there aren't minute gains nor have I ever suggested that people NOT OC. The point is there is little value in OC'ing the RAM above 1333 MHz. if you can't see it or feel it - especially if you need to pay a premium for faster RAM.

Many people buy 1333/1600 MHz. RAM, OC it and get on with life because you simply can not see or feel RAM OC'ed beyond ~1333 MHz. on a typical AMD/Intel desktop PC, with Llano being the exception because the GPU section can use the increased frequency. Paying a premium for higher rated frequency or lower latencies is typically only of value to those who run synthetic benches all day long instead of actually using their PC.

No one is being discouraged from OC'ing. I am reporting what the results are from extensive RAM testing by numerous PC industry sources and the links I post even show OCN member results with no "significant" gains from RAM OC'ing above ~1333 MHz. People are free to do whatever makes them happy with their PC. I am simply reporting what the results are when you OC RAM on a typical AMD/Intel desktop PC at or above ~1333 MHz. Each person should determine what works best for them.

It's worth reading the links to get a better understanding of why DDR3 RAM OC'ed above ~1333 MHz. does not produce any "significant" gains in real applications.

When it comes to "increased bandwidth", maybe a simple example would be useful.

With DDR3 RAM running @ ~1333 MHz. it's like having four cars, one in each lane, on a four lane highway. Adding more lanes does not increase the speed at which the cars travel, it just provides additional capacity for more cars. In PC speak the cars are the data being moved. If your system can only produce four packets of data (i.e. "cars"), at a time, then more lanes isn't going to increase data throughput.
Edited by AMD4ME - 6/25/12 at 8:29am
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Memory
Overclock.net › Forums › Components › Memory › So I adjusted my memory timings...