Overclock.net › Forums › Intel › Intel CPUs › losen timming = decrease performance
New Posts  All Forums:Forum Nav:

losen timming = decrease performance

post #1 of 11
Thread Starter 
hey, i managed to oc my system to 3.15GHZ, with timmings of 4-4-4-12, but the thing is that i tried to oc it to fsb at "460" pc wont boot,
so my question is, if i lower the timmings to, lets say 5-5-5-15 and at a fsb of "500" will i see a performance increase or decrease from the losen timmings ?
Light's Conroe
(13 items)
 
  
CPUMotherboardGraphicsRAM
Intel C2D E6300 Gigabyte P965 S3 XFX Geforce 7600GT Corsair XMS2 DDR2 2X1GB 
Hard DriveOptical DriveOSPower
Seagate 320GB DVD+CD-R Drive Windows XP OCZ 700W GXS 
Case
NZXT 
  hide details  
Reply
Light's Conroe
(13 items)
 
  
CPUMotherboardGraphicsRAM
Intel C2D E6300 Gigabyte P965 S3 XFX Geforce 7600GT Corsair XMS2 DDR2 2X1GB 
Hard DriveOptical DriveOSPower
Seagate 320GB DVD+CD-R Drive Windows XP OCZ 700W GXS 
Case
NZXT 
  hide details  
Reply
post #2 of 11
Intel CPUs don't care as much about timings as AMD CPUs do. Loosening the timings will hurt performance a little, but the planned increase in FSB you're hoping to get will more than make up for it.

Thus, I expect a significant performance increase!
    
CPUMotherboardGraphicsRAM
E8400 @ 3.8 GHz DFI LANPARTY DK X38-T2R eVGA 8400GS 2GB Mushkin DDR2 800 (single stick) 
Hard DriveOptical DriveOSMonitor
250GB WD SATA None Windows XP Pro 32-bit HP Pavilion f1703 
KeyboardPowerCaseMouse
N/A Corsair HX520 Coolermaster Cavalier N/A 
Mouse Pad
N/A 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
E8400 @ 3.8 GHz DFI LANPARTY DK X38-T2R eVGA 8400GS 2GB Mushkin DDR2 800 (single stick) 
Hard DriveOptical DriveOSMonitor
250GB WD SATA None Windows XP Pro 32-bit HP Pavilion f1703 
KeyboardPowerCaseMouse
N/A Corsair HX520 Coolermaster Cavalier N/A 
Mouse Pad
N/A 
  hide details  
Reply
post #3 of 11
hmm i never really understood the preformance trade off thing with FSB and RAM, so can someone please elaborate. im guessing that if you lower the timings then the RAM performance would decrease, but the added FSB will give the CPU added performance. so... its like a trade off, and it depends on whether or not RAM or CPU is "MORE" important to you. ... is any of this correct or am i crazy??
post #4 of 11
Plus the fact that when you increase the FSB, you ALSO increase the frequency of the RAM. In his example, assuming RAM runs 1:1 with the FSB, at an FSB = 460, the effective RAM speed will be DDR2 920. When you increase the FSB to 500, the effective RAM speed will also go up, in this case to DDR2 1000. The increase in RAM frequency alone should offset the looser timings, esp. on an Intel rig.

On top of that, the CPU frequency also increases, boosting performance even more, assuming he does not reduce the CPU multiplier

Again, the intel memory controller is not as sensitive toward timings as AMD's is. Loosening timings on an AMD rig will have more impact than doing so on an Intel rig (Intel just loves raw bandwidth).
    
CPUMotherboardGraphicsRAM
E8400 @ 3.8 GHz DFI LANPARTY DK X38-T2R eVGA 8400GS 2GB Mushkin DDR2 800 (single stick) 
Hard DriveOptical DriveOSMonitor
250GB WD SATA None Windows XP Pro 32-bit HP Pavilion f1703 
KeyboardPowerCaseMouse
N/A Corsair HX520 Coolermaster Cavalier N/A 
Mouse Pad
N/A 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
E8400 @ 3.8 GHz DFI LANPARTY DK X38-T2R eVGA 8400GS 2GB Mushkin DDR2 800 (single stick) 
Hard DriveOptical DriveOSMonitor
250GB WD SATA None Windows XP Pro 32-bit HP Pavilion f1703 
KeyboardPowerCaseMouse
N/A Corsair HX520 Coolermaster Cavalier N/A 
Mouse Pad
N/A 
  hide details  
Reply
post #5 of 11
Thread Starter 
sweet, then that means if i lower the timmings, and raise it to 500mhz fsb, i shouldnt get decrease performance instead increase performance right ?
Light's Conroe
(13 items)
 
  
CPUMotherboardGraphicsRAM
Intel C2D E6300 Gigabyte P965 S3 XFX Geforce 7600GT Corsair XMS2 DDR2 2X1GB 
Hard DriveOptical DriveOSPower
Seagate 320GB DVD+CD-R Drive Windows XP OCZ 700W GXS 
Case
NZXT 
  hide details  
Reply
Light's Conroe
(13 items)
 
  
CPUMotherboardGraphicsRAM
Intel C2D E6300 Gigabyte P965 S3 XFX Geforce 7600GT Corsair XMS2 DDR2 2X1GB 
Hard DriveOptical DriveOSPower
Seagate 320GB DVD+CD-R Drive Windows XP OCZ 700W GXS 
Case
NZXT 
  hide details  
Reply
post #6 of 11
Exactly you should see a significant performance increase.

Run SuperPi. That's generally a benchmark that's pretty sensitive to memory.
    
CPUMotherboardGraphicsRAM
E8400 @ 3.8 GHz DFI LANPARTY DK X38-T2R eVGA 8400GS 2GB Mushkin DDR2 800 (single stick) 
Hard DriveOptical DriveOSMonitor
250GB WD SATA None Windows XP Pro 32-bit HP Pavilion f1703 
KeyboardPowerCaseMouse
N/A Corsair HX520 Coolermaster Cavalier N/A 
Mouse Pad
N/A 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
E8400 @ 3.8 GHz DFI LANPARTY DK X38-T2R eVGA 8400GS 2GB Mushkin DDR2 800 (single stick) 
Hard DriveOptical DriveOSMonitor
250GB WD SATA None Windows XP Pro 32-bit HP Pavilion f1703 
KeyboardPowerCaseMouse
N/A Corsair HX520 Coolermaster Cavalier N/A 
Mouse Pad
N/A 
  hide details  
Reply
post #7 of 11
It wont perform as well as if you kept the timings the same at higher speeds, but loosening the timings and going higher on mhz will outperform slower speeds w/ tighter timings.

So loosen the timings a little and overclock more! Though try 5-4-5-12

Tight timings + high speeds is everyones dream... though its just not feasible. Where's the BH-5 of ddr2?
AMD Gamer
(13 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8320 Asus M5A99FX-Pro MSI GTX980 Gaming Crucial Ballistix 4x4gb 1934mhz @ 8-8-8-24 
Hard DriveCoolingOSMonitor
Samsung Evo 500GB Thermaltake Water 2.0 Windows 10 LG 27" 4k 
PowerCaseMouseMouse Pad
Huntkey X7 900w Thermaltake Urban S31 Logitech G500 Rocketfish Knockoff of Razer propad 
Audio
Logitech Z5500 & G430 
  hide details  
Reply
AMD Gamer
(13 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8320 Asus M5A99FX-Pro MSI GTX980 Gaming Crucial Ballistix 4x4gb 1934mhz @ 8-8-8-24 
Hard DriveCoolingOSMonitor
Samsung Evo 500GB Thermaltake Water 2.0 Windows 10 LG 27" 4k 
PowerCaseMouseMouse Pad
Huntkey X7 900w Thermaltake Urban S31 Logitech G500 Rocketfish Knockoff of Razer propad 
Audio
Logitech Z5500 & G430 
  hide details  
Reply
post #8 of 11
Thread Starter 
bh-5 ?

one more thing, if my oc didnt turn out too good, and i decide to go back to original oc, do i still need to run tests to see if its stable ? 3150oc is being tested with orthos for 20hours w/o errors.
Light's Conroe
(13 items)
 
  
CPUMotherboardGraphicsRAM
Intel C2D E6300 Gigabyte P965 S3 XFX Geforce 7600GT Corsair XMS2 DDR2 2X1GB 
Hard DriveOptical DriveOSPower
Seagate 320GB DVD+CD-R Drive Windows XP OCZ 700W GXS 
Case
NZXT 
  hide details  
Reply
Light's Conroe
(13 items)
 
  
CPUMotherboardGraphicsRAM
Intel C2D E6300 Gigabyte P965 S3 XFX Geforce 7600GT Corsair XMS2 DDR2 2X1GB 
Hard DriveOptical DriveOSPower
Seagate 320GB DVD+CD-R Drive Windows XP OCZ 700W GXS 
Case
NZXT 
  hide details  
Reply
post #9 of 11
Ok

Maybe you need a bit more vcore for a higher overclock. Take somewhat smaller steps ... inch your way to the goal!
    
CPUMotherboardGraphicsRAM
E8400 @ 3.8 GHz DFI LANPARTY DK X38-T2R eVGA 8400GS 2GB Mushkin DDR2 800 (single stick) 
Hard DriveOptical DriveOSMonitor
250GB WD SATA None Windows XP Pro 32-bit HP Pavilion f1703 
KeyboardPowerCaseMouse
N/A Corsair HX520 Coolermaster Cavalier N/A 
Mouse Pad
N/A 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
E8400 @ 3.8 GHz DFI LANPARTY DK X38-T2R eVGA 8400GS 2GB Mushkin DDR2 800 (single stick) 
Hard DriveOptical DriveOSMonitor
250GB WD SATA None Windows XP Pro 32-bit HP Pavilion f1703 
KeyboardPowerCaseMouse
N/A Corsair HX520 Coolermaster Cavalier N/A 
Mouse Pad
N/A 
  hide details  
Reply
post #10 of 11
There are two components to memory performance. Latency and Bandwidth.

Latency is affected by 2 things: clock speed and timings (RAS, CAS, etc..). Latency is basically how long it takes the memory to start sending data after a request. If your workload involves a lot of small memory reads, then latency becomes important as your memory bus will spend a lot of time waiting for the memory to respond to each request. However, there are also "burst" transfers in which you can read a series of concurrent addresses and not have to wait for the memory latency and this is where the other memory performance metric comes in...

Bandwidth is the amount of data the memory can transfer over time (per second usually) when the maximum burst transfer is utilized. The larger the chunk of memory you want to read, the better you utilize bandwidth. This is the more important performance metric as people have said.

So looser timings even at the same FSB frequency are not a big (<5%) performance hit for most types of applications, but where it can pay big performance increases are when the looser timings allow you to clock your FSB higher. Higher FSB = more memory bandwidth, more bandwidth from the CPU to the northbridge and all the functions it provides.

Also keep in mind dropping your timings from CAS4 to CAS5 doesn't mean you're going to have 25% more latency if it allows you to increase your FSB frequency.
A CAS of 4 at a FSB of 400MHz (800MHz DDR, 1600 QDR) means your memory latency is 4 clock cycles which = 10nanoseconds (nS). If you drop to CAS 5 but get your FSB up to 450MHz, now you have to wait 5 clock cycles latency, but your clock cycles are shorter, so CAS 5 * 450MHz FSB = 11.1nS. If it allows you to get to 480Mhz then your latency is only 10.4nS.

So as you can see, there almost is no trade-off. Higher FSB is almost always better even at the expense of slightly looser timings. The only time I wouldn't trade off 1 clock cycle of timing for higher FSB, is if the higher RAS CAS settings only allowed you to get <15MHz additional FSB.
     
CPUMotherboardGraphicsRAM
7700K Delid @5.1GHz @1.4V Asus Z270F Strix Asus RX480 Strix @ 1400MHz Corsair Vengance PC3200 2x8GB 
Hard DriveHard DriveHard DriveOptical Drive
Sandisk Ultra II 960GB 2 x Intel 530 SSD 2 x 3TB HDD LG BD-R 
CoolingOSMonitorPower
Noctua NH-D15 Windows 10 Home Dell 27" 1080p Seasonic Prime 750W Titanium 
CaseMouse
Fractal Define R5 Razer DeathAdder 
CPUMotherboardGraphicsRAM
i7 2600K @4.6GHz @1.4V Gigabyte GA-Z68-D3H-B3 AMD RX480 4x4GB GSkill RipjawsX 1600 CL8 
Hard DriveOptical DriveCoolingOS
2 X Intel 530 240GB + 4TB HDD LG BD-RW Cooler Master Hyper 212 Windows 10 
MonitorKeyboardPowerCase
Dell 27" 1080p Microsoft Seasonic Prime 750W Titanium Fractal Define R5 
Mouse
Razer DeathAdder 
  hide details  
Reply
     
CPUMotherboardGraphicsRAM
7700K Delid @5.1GHz @1.4V Asus Z270F Strix Asus RX480 Strix @ 1400MHz Corsair Vengance PC3200 2x8GB 
Hard DriveHard DriveHard DriveOptical Drive
Sandisk Ultra II 960GB 2 x Intel 530 SSD 2 x 3TB HDD LG BD-R 
CoolingOSMonitorPower
Noctua NH-D15 Windows 10 Home Dell 27" 1080p Seasonic Prime 750W Titanium 
CaseMouse
Fractal Define R5 Razer DeathAdder 
CPUMotherboardGraphicsRAM
i7 2600K @4.6GHz @1.4V Gigabyte GA-Z68-D3H-B3 AMD RX480 4x4GB GSkill RipjawsX 1600 CL8 
Hard DriveOptical DriveCoolingOS
2 X Intel 530 240GB + 4TB HDD LG BD-RW Cooler Master Hyper 212 Windows 10 
MonitorKeyboardPowerCase
Dell 27" 1080p Microsoft Seasonic Prime 750W Titanium Fractal Define R5 
Mouse
Razer DeathAdder 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › losen timming = decrease performance