Memory Bandwidth vs. Latency Timings

jmke · 12th November 2003, 12:41

Quote:

While bandwidth is still very important to the Intel Pentium 4, it's not as important as it once was in the i845PE days of single channel memory controllers. Thanks to the i865PE/i875P's dual channel memory controller things are much brighter. On average, the system with the memory running at 400 MHz (5:4 memory divider enabled) with aggressive memory timings performed 2-3% faster than the system using high speed memory with loose timings.

so get yourself some BH5-chipset memory for christmas this year! (I head OCZ is producing them a-new)

The Senile Doctor · 12th November 2003, 20:50

if I get this right...
if you run your mem async lower with faster timings, then your pc will be faster then running it sync with lower timings even when the difference in memspeed is 50 mhz??

unbelievable!

RichBa5tard · 12th November 2003, 21:32

Yup. Has been a known fact for a long time for ppl who visit xtreme.

The Senile Doctor · 13th November 2003, 11:12

so if I put my 3.0 that runs 240 fsb on 5:4 and increase the timings (now it's at 3 4-4-8) I'll get the same power?

and if you find ram that can run that 240 @ 2, 2-5-2, will the increase at this speed be worthwhile?

RichBa5tard · 13th November 2003, 11:19

Quote:

Originally posted by calantak so if I put my 3.0 that runs 240 fsb on 5:4 and increase the timings (now it's at 3 4-4-8) I'll get the same power?

yes.

Quote:

and if you find ram that can run that 240 @ 2, 2-5-2, will the increase at this speed be worthwhile?

Good question.

You could try my mushkin level 2 if you didn't live at the other side of the country.

The Senile Doctor · 13th November 2003, 11:23

coz that is what's important... running 1:1 at cas 3 4-8-4 at that speed could mean 1% with cas 2 2-5-2 or it could mean 5% which is a big difference

RichBa5tard · 13th November 2003, 11:26

On my IS7:
240 1/1 6-3-3 cl2: 92xx mem marks
265 4/5 5-2-2 cl2: 10xxx mem marks

The Senile Doctor · 13th November 2003, 11:31

and 230 5-2-2 cas 2 versus 230 6-3-3 cas 2.5 and even 7-4-4 cas 3

please compare

RichBa5tard · 13th November 2003, 11:47

i'll test it at 250 with a single stick twinmoss, and at 238 with 2 sticks mushkin tonight if i find some time

The Senile Doctor · 13th November 2003, 13:25

you should have the highest possible stable 2-5-2-cas2 overclock
dual channel and test it with 1:1, with 5:4 and with 1:1 cas 3 4-8-3, that would give us the most info on what to think of this whole situation...in 3dmark01 runs that'd take only 10 minutes tops

12th November 2003, 20:50	#2
The Senile Doctor Member Join Date: May 2002 Location: wherever the doom is Posts: 3,171	if I get this right... if you run your mem async lower with faster timings, then your pc will be faster then running it sync with lower timings even when the difference in memspeed is 50 mhz?? unbelievable! __________________ OC-2-the-death Where the Reverend is doing his Magick, all mortals be silent Doom over the world

12th November 2003, 21:32	#3
RichBa5tard [M] Reviewer/HWBot ***** Join Date: May 2002 Posts: 3,344	Yup. Has been a known fact for a long time for ppl who visit xtreme. __________________ HTPC (mac osx): Mac Mini \| Core Duo 1.6Ghz \| 2GB DDR2 \| 26\" TFT Development (mac osx): Macbook \| Core 2 2.0Ghz \| 4GB DDR2 \| 250GB HD Games (win xp): E2160 @ 2.4Ghz \| HD3850 OC \| Asrock 4coredual-vsta \| 2GB DDR2

13th November 2003, 11:12	#4
The Senile Doctor Member Join Date: May 2002 Location: wherever the doom is Posts: 3,171	so if I put my 3.0 that runs 240 fsb on 5:4 and increase the timings (now it's at 3 4-4-8) I'll get the same power? and if you find ram that can run that 240 @ 2, 2-5-2, will the increase at this speed be worthwhile? __________________ OC-2-the-death Where the Reverend is doing his Magick, all mortals be silent Doom over the world

13th November 2003, 11:23	#6
The Senile Doctor Member Join Date: May 2002 Location: wherever the doom is Posts: 3,171	coz that is what's important... running 1:1 at cas 3 4-8-4 at that speed could mean 1% with cas 2 2-5-2 or it could mean 5% which is a big difference __________________ OC-2-the-death Where the Reverend is doing his Magick, all mortals be silent Doom over the world

13th November 2003, 11:26	#7
RichBa5tard [M] Reviewer/HWBot ***** Join Date: May 2002 Posts: 3,344	On my IS7: 240 1/1 6-3-3 cl2: 92xx mem marks 265 4/5 5-2-2 cl2: 10xxx mem marks __________________ HTPC (mac osx): Mac Mini \| Core Duo 1.6Ghz \| 2GB DDR2 \| 26\" TFT Development (mac osx): Macbook \| Core 2 2.0Ghz \| 4GB DDR2 \| 250GB HD Games (win xp): E2160 @ 2.4Ghz \| HD3850 OC \| Asrock 4coredual-vsta \| 2GB DDR2

13th November 2003, 11:31	#8
The Senile Doctor Member Join Date: May 2002 Location: wherever the doom is Posts: 3,171	and 230 5-2-2 cas 2 versus 230 6-3-3 cas 2.5 and even 7-4-4 cas 3 please compare __________________ OC-2-the-death Where the Reverend is doing his Magick, all mortals be silent Doom over the world

13th November 2003, 11:47	#9
RichBa5tard [M] Reviewer/HWBot ***** Join Date: May 2002 Posts: 3,344	i'll test it at 250 with a single stick twinmoss, and at 238 with 2 sticks mushkin tonight if i find some time __________________ HTPC (mac osx): Mac Mini \| Core Duo 1.6Ghz \| 2GB DDR2 \| 26\" TFT Development (mac osx): Macbook \| Core 2 2.0Ghz \| 4GB DDR2 \| 250GB HD Games (win xp): E2160 @ 2.4Ghz \| HD3850 OC \| Asrock 4coredual-vsta \| 2GB DDR2

13th November 2003, 13:25	#10
The Senile Doctor Member Join Date: May 2002 Location: wherever the doom is Posts: 3,171	you should have the highest possible stable 2-5-2-cas2 overclock dual channel and test it with 1:1, with 5:4 and with 1:1 cas 3 4-8-3, that would give us the most info on what to think of this whole situation...in 3dmark01 runs that'd take only 10 minutes tops __________________ OC-2-the-death Where the Reverend is doing his Magick, all mortals be silent Doom over the world

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Memory Frequency vs. Timings on Intel Core i7 (DE)	jmke	WebNews	0	6th September 2009 13:33
Rambus Aims for Terabyte-Per-Second Memory Bandwidth	jmke	WebNews	0	26th November 2007 14:20
Memory Bandwidth vs. Latency Timings	Sidney	WebNews	1	13th September 2005 15:13
GEIL to Present Memory Modules with CL1.5 Timings	jmke	WebNews	1	20th May 2005 10:46
Latency Settings and their Impact on Memory Performance	jmke	WebNews	1	29th March 2005 19:29
Memory Bandwidth vs. Latency Timings	Sidney	WebNews	0	13th December 2004 19:02
Corsair TwinX 1024 PC3200XL Low Latency Memory	Sidney	WebNews	0	21st October 2004 02:40
OCZ Announces PC-3700 Enhanced Bandwidth DDR Memory	jmke	WebNews	0	21st April 2004 13:09
AMD nForce2 memory timings explored	jmke	Articles & Howto's	74	29th March 2004 13:39
INFO: Memory Timings	jmke	FAQ / INFO / HOW-TO	6	18th May 2003 11:38