Question about small fft vs large in prime95

Steven-1979 · 12-26-08, 07:30 PM

If I have a setting that can run small fft without errors for hours (I got up to 5 hours before turning it off) but it gets an error in about 30 minutes of large fft ... is this an indication that the problem is more likely something with the MCH settings? Or could it be the extra heat large fft is causing? I'm not sure how the tests differ, but from what I understand, large fft stresses fsb more in quads? The setting is 510 x 7.

Kris · 12-26-08, 11:07 PM

As I understand it and I may be wrong. You may be looking at ram and or NB needing some more tweaking. Blend will test more ram and NB stability than small ftt. Not knowing your current settings I don't want to give advice on what changes to make.

boorishid · 12-26-08, 11:11 PM

I never use large ftt not sure of what value it has, try using blend if that fails run memtest if memtest passes than you most likely just need to boost your nb voltage.

SkiBum1207 · 12-27-08, 01:18 AM

as far as I know, large ffts are too big for the cache so they get sent to the ram, this could mean your NB/Ram needs some more love and work lol

Dapman02 · 12-27-08, 01:26 AM

Small fft=CPU
Large fft=RAM
Blend= Ram and/or Northbridge.
try loosining your timing and see if that helps, also you should increase your ram voltage.

If blend fails increase your northbridge voltage, also make sure you actively cool and upgrade your northbridge heatsink.

Steven-1979 · 12-27-08, 11:38 AM

Good info ... I've been using a 2x memory multiplier with a 400 latch ... the memory is at 1020 though. Thats less than what it's rated at (1066). Should I still increase it's volts? Everything with it is manually set to manufacturer specs:

2.1v
5-5-5-15

Is there a way to monitor NB temps?

12-26-08, 07:30 PM	Thread Starter #1
Steven-1979 Member Join Date: Nov 2008	Question about small fft vs large in prime95 If I have a setting that can run small fft without errors for hours (I got up to 5 hours before turning it off) but it gets an error in about 30 minutes of large fft ... is this an indication that the problem is more likely something with the MCH settings? Or could it be the extra heat large fft is causing? I'm not sure how the tests differ, but from what I understand, large fft stresses fsb more in quads? The setting is 510 x 7. __________________ - How to pad-mod/VID-mod your cpu Currently waiting for lga2011 to upgrade to new system CPU: Intel E8500 Motherboard: GIGABYTE GA-EP45-UD3P RAM: 4gb Corsair Dominator DDR2 1066 w/ RAM cooling fan Cooling: ~~CPU/GPU/NB = Water Cooled~~ Graphics: EVGA GeForce GTX 260 core 216 (@ 771/1566/1298) PSU: Corsair tx 650w
	QUOTE Thanks

12-26-08, 11:07 PM	#2
Kris Member Join Date: Sep 2004 Location: Eating thermal paste!@#	As I understand it and I may be wrong. You may be looking at ram and or NB needing some more tweaking. Blend will test more ram and NB stability than small ftt. Not knowing your current settings I don't want to give advice on what changes to make. __________________ FX 8150 @ 4.5 @ 1.41 Corsair H100 Push-Pull Asus Crosshair V Formula-Z 1002 bios HD 7950 3GiG 16 GB Crucial 1866 1-750G Wd Perp-Storage 1-240 GiG Mushkin SSD-OS Corsair TX 850 Light On Sata BD Drive Sony Sata DVD Drive Fractal Arc Midi
	QUOTE Thanks

12-26-08, 11:11 PM	#3
boorishid Member Join Date: Jan 2008	I never use large ftt not sure of what value it has, try using blend if that fails run memtest if memtest passes than you most likely just need to boost your nb voltage. __________________ CPU: Qx9650 C1 @ 4ghz daily Mobo: Gigabyte EP45-UD3P Ram: 8gb gSkill ddr2 1000 5-5-5-15 GPU: 2x ATI 4850's Crossfire PSU: Tagan BZ 1100 Watt HDD: 150gb Raptor, 4xWD160 raid 0 Case: Antec 900 Cooling: Swiftech Apogee Gtz, Swiftech 320 Rad w/hi speed yates, D4 Pump. OS: Vista Ultimate x64 (_..-=HeatWare=-.._)
	QUOTE Thanks

12-27-08, 01:18 AM	#4
SkiBum1207 Member Join Date: Feb 2008	as far as I know, large ffts are too big for the cache so they get sent to the ram, this could mean your NB/Ram needs some more love and work lol __________________ Intel Q6600 3.8 Ghz @1.41v (and going higher!!) -1688mhz FSB- 4 gb Kingston HyperX PC2-9200 @ 1066 4-4-4-12-Asus P5k Vanilla - VModded-Nvidia 8600 GT OCed 670/1625/880mhz-1x Raptor 150gb & 1x Hitachi Deskstar 1tb-Thermaltake Toughpower 750W-Custom Swiftec Watercooler Blue Belt Ubercloxx0r
	QUOTE Thanks

12-27-08, 01:26 AM	#5
Dapman02 Member Join Date: Jan 2008 Location: Overland Park, KS	Small fft=CPU Large fft=RAM Blend= Ram and/or Northbridge. try loosining your timing and see if that helps, also you should increase your ram voltage. If blend fails increase your northbridge voltage, also make sure you actively cool and upgrade your northbridge heatsink. __________________ Desktop: Return of Deneb Processor: AMD Phenom II 955 @ 4.2Ghz Cooling: TRUE Push/Pull GPU: Visiontek 5850 Core 875Mhz/1175mhz memory Motherboard: MSI 880G-E45 Case: CM Storm Enforcer RAM: 2x2gb Samsung generic HDD: Samsung 500GB 5400RPM OS: Windows 7 64-bit Christian Overclocker
	QUOTE Thanks

Question about small fft vs large in prime95

Internet Services

Website Development

Tech Hardware

12-27-08, 11:38 AM	Thread Starter #6
Steven-1979 Member Join Date: Nov 2008	Good info ... I've been using a 2x memory multiplier with a 400 latch ... the memory is at 1020 though. Thats less than what it's rated at (1066). Should I still increase it's volts? Everything with it is manually set to manufacturer specs: 2.1v 5-5-5-15 Is there a way to monitor NB temps? __________________ - How to pad-mod/VID-mod your cpu Currently waiting for lga2011 to upgrade to new system CPU: Intel E8500 Motherboard: GIGABYTE GA-EP45-UD3P RAM: 4gb Corsair Dominator DDR2 1066 w/ RAM cooling fan Cooling: ~~CPU/GPU/NB = Water Cooled~~ Graphics: EVGA GeForce GTX 260 core 216 (@ 771/1566/1298) PSU: Corsair tx 650w
	QUOTE Thanks

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search