|
| Notices |
Welcome to the DriverHeaven.net forums. You are currently viewing our boards as a guest which gives you limited access to view most discussions and access our other features. By joining our free community you will have access to post topics, communicate privately with other members (PM), respond to polls, upload content and access many other special features. Registration is fast, simple and absolutely free so please, join our community today! If you have any problems with the registration process or your account login, please contact contact us. |
 |
Feb 28, 2003, 08:14 AM
|
#1
|
|
DriverHeaven Junior Member
Join Date: Nov 2002
Posts: 43
|
Nforce2: 133 cpu: 266 memory speed better than 333!
Just made a very interesting discovery. I have a Leadtek Nforce2 motherboard, an Athlon 2100+ cpu (133 fsb) and two new sticks of Kingston 2700 DDR (256mb) ram. I "upgraded" from two sticks of Crucial 2100 DDR when the price dropped recently.
When the bios auto set the timings, the memory was set to "Sync". This synced the memory with the processor speed and set the memory to 266. Despite the 2.5 rating, I was easily able to lower the timings at 266 to 5-2-2 with complete stability. Did a bunch of benchmarks.
I then switched the memory speed in bios to "Spd", which reads the information from the ram. It set the memory speed to the correct 333 speed with significantly higher timings. Even after tweaking the timings down, the benchmarks were much slower than when I was at "Sync" and 266.
I suppose this is a function of two things. The first is that the Nforce boards, with their dual memory setup, is most efficient syncing the cpu and memory speeds. The second is that memory timings are as important as memory speeds.
Anyone out there get the opposite results? I am glad about the purchase because I now have inexpensive memory running stabily at very low timings at stock voltage and I will most likely see the jump in performance when I upgrade to a 166/333 cpu down the road.
|
|
|
Feb 28, 2003, 09:33 AM
|
#2
|
|
DriverHeaven Extreme Member
Join Date: May 2002
Location: Ohio
Posts: 8,481
|
You are indeed correct and this has been proven. Most people that buy a 266 FSB processor just mainly buy the faster ram so they can overclock. You "will" notice an improvement with 333 FSB/333 Memory.
|
|
|
Feb 28, 2003, 11:01 AM
|
#3
|
|
Flash Banner Hater
Join Date: Jun 2002
Location: UK
Posts: 2,963
|
It's possible that 266/333 might deliver an improvement if using onboard graphics in the IGP model, as the graphics would be working at the higher RAM speed.
Apart from some oddities, every review seems to have the Nforce2 being better with synchronous memory.
It's possible that DDR400 might tip the balance the other way, but there's still the problem of taking an even worse hit on the timings!
The basic issue, is that for a CPU using a 64 bit DDR266 FSB, the best possible througput IS DDR266 with minimum timings.
VIA chipsets were always better at handling split RAM/FSB timings, and running SDRAM at 133 to a Duron at DDR200 FSB made sense.
And, good point, even if you cant unlock the multiplier and convert to 166 (DDR333) FSB for the same speed, the faster memory allows FSB overclocking without the momory becoming an issue.
|
|
|
Feb 28, 2003, 04:10 PM
|
#4
|
|
DriverHeaven Addict
Join Date: Sep 2002
Location: Elysium
Posts: 328
|
Your 2100+ should easily hit the 166FSB with a lower multiplier, maybe even without changing the multipier. Thats where you would really see a difference in performance, running sync @ 166... the nf2 boards are "supposed" to unlock the chips for you, but a buddy tells me his MSI (ilsr ver.) wont do it. If you have PC2700, you'd want to be using it, I know I would. I would try to get the FSB up to 166 and run sync from there, I know 2 folks with the MSI board who use 166/166, one has a 1700+ and PC2700, the other has a 2400+ and PC3200, but they both choose the sync setting for maximum performance. (They both use MSI boards, one IGP. one SPP)....
good luck!
|
|
|
Mar 4, 2003, 05:25 PM
|
#5
|
|
DriverHeaven Junior Member
Join Date: Nov 2002
Posts: 43
|
Interesting overclocking results
Unfortunately, the mutiplier on my AMD 2100+ is locked (I have the Palomino version). Therefore, I cannot reduce the multiplier and bump the cpu fsb to 166/333. My attempt at 166 resulted in no bootup/lock early in the process. However, I have been experimenting with fsb numbers between 133 and 166. As long as the memory speed is set to "Sync", the memory speed automatically adjusts up with corresponding increases in performance which appear to be greater than what would be achieved with just bumping the cpu. I like these Nforce2 boards more each day.
|
|
|
Mar 4, 2003, 10:47 PM
|
#6
|
|
DriverHeaven Extreme Member
Join Date: Dec 2002
Location: U.S.A.
Posts: 16,122
|
Re: Interesting overclocking results
Quote:
Originally posted by gyoung
Unfortunately, the mutiplier on my AMD 2100+ is locked (I have the Palomino version). Therefore, I cannot reduce the multiplier and bump the cpu fsb to 166/333. My attempt at 166 resulted in no bootup/lock early in the process. However, I have been experimenting with fsb numbers between 133 and 166. As long as the memory speed is set to "Sync", the memory speed automatically adjusts up with corresponding increases in performance which appear to be greater than what would be achieved with just bumping the cpu. I like these Nforce2 boards more each day.
|
well unlock the darn thing !!!!!!!!!!!!!!!!!
|
|
|
Mar 5, 2003, 02:50 AM
|
#7
|
|
DriverHeaven Addict
Join Date: Sep 2002
Location: Elysium
Posts: 328
|
Agreed, unlock it.
The "down the road" upgrade you refer to is upon you, if you can get the multiplier unlocked. The 2100+ Palomino is the one with the deep cut brigdes if memory serves, though it can still be unlocked. It has become evident that the nforce2 chipset favours "sync", as you stated, and also indication that dual channel DDR400 is only slightly favourable (and considerably cheaper even without the "compliant DDR400" hassles) than sync at 166/333.... you are sitting pretty with your current hardware if you can get it to work for you!
The 2100+ Palomino isnt the easiest to unlock, though..
http://www.pcstats.com/articleview.cfm?articleID=1009
Good luck!
|
|
|
Mar 28, 2003, 08:50 PM
|
#8
|
|
Long Live The P6!
Join Date: Nov 2002
Location: The Great White North
Posts: 119
|
The same thing was observed when VIA released the Apollo Pro chipset, which would let you run 100FSB processors with PC133. It was found that PC100 2/2/2/5/9 was the fastest setting.
- JW
|
|
|
May 1, 2003, 04:56 PM
|
#9
|
|
DriverHeaven Junior Member
Join Date: Oct 2002
Posts: 54
|
Only a moron would run their ran out of sync. Always run your ram in sync, get as much FSB mhz as you possible can, and then tighten the timings as tight as you can. in sync FSB mhz is your biggest performance gain. Timings come second. Thus a 200fsb with timings of 8-3-3-2 would run faster than 166mhz FSB with timings of 2-2-2-5
|
|
|
May 3, 2003, 05:29 AM
|
#10
|
|
DriverHeaven Senior Member
Join Date: Apr 2003
Location: Australia
Posts: 1,802
|
Re: Nforce2: 133 cpu: 266 memory speed better than 333!
Quote:
Originally posted by gyoung
Just made a very interesting discovery. I have a Leadtek Nforce2 motherboard, an Athlon 2100+ cpu (133 fsb) and two new sticks of Kingston 2700 DDR (256mb) ram. I "upgraded" from two sticks of Crucial 2100 DDR when the price dropped recently.
When the bios auto set the timings, the memory was set to "Sync". This synced the memory with the processor speed and set the memory to 266. Despite the 2.5 rating, I was easily able to lower the timings at 266 to 5-2-2 with complete stability. Did a bunch of benchmarks.
I then switched the memory speed in bios to "Spd", which reads the information from the ram. It set the memory speed to the correct 333 speed with significantly higher timings. Even after tweaking the timings down, the benchmarks were much slower than when I was at "Sync" and 266.
I suppose this is a function of two things. The first is that the Nforce boards, with their dual memory setup, is most efficient syncing the cpu and memory speeds. The second is that memory timings are as important as memory speeds.
Anyone out there get the opposite results? I am glad about the purchase because I now have inexpensive memory running stabily at very low timings at stock voltage and I will most likely see the jump in performance when I upgrade to a 166/333 cpu down the road.
|
nope , definitely not on me , see those"188"on my signature, those are actually PC2100!!! and i set them to asynchronous instead of synchronous...........It gave me a super boost, so you could say that i had the opposite results of yours.............................
|
|
|
|
|
|