AOA Forums AOA Forums AOA Forums Folding For Team 45 AOA Files Home Front Page Become an AOA Subscriber! UserCP Calendar Memberlist FAQ Search Forum Home


Go Back   AOA Forums > Hardware > EPoX MotherBoards

EPoX MotherBoards EPoX Intel and AMD Motherboards.


Reply
 
LinkBack Thread Tools Rate Thread
  #1 (permalink)  
Old 18th March, 2003, 08:56 AM
Lord Vetinari's Avatar
Member
 
Join Date: March 2003
Location: Worthing Sussex England
Posts: 325

FSB and running dual memory on 8RDA

Most people (on these forums anyway) are always on the search for higher FSB with their Nforce2 boards.
I'm one of the people as well , but needed to mod the VDD to move towards 200FSB from 180 to 185 FSB limit.

Now I've seen two "facts" mentioned on various boards.
1. Running dual memory reduces the max FSB achievable.
2. The extra memory bandwidth afforded by dual memory is a total waste of time unless you have an onboard graphics card.

Questions-
anyone know what the difference is in max FSB achievable between running dual memory and not?
and is "fact 2 " true?
ie would we be better off running at higher FSB without dual mem enabled?
__________________
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #2 (permalink)  
Old 18th March, 2003, 09:22 AM
Banned
 
Join Date: January 2003
Posts: 189

1:right as with any mobo one stick will overclock higher at lower timings with less voltage than two.. The nf2 mobo is very picky when it comes to DC mode though as the chipset itself needs more VDD and 2x the heat IMO is brought to the table.

2C:stands for not 200mhz fsb 3200mb/s=6400mb/s MAX as the AMD chipset limits this to 3200mb/s period..
but: 200mhz fsb running DC yeilds a much better "EFFICENTCY" of the total bandwith passing through per cycle.. so a 200mhz fsb single might yeild 3050/2900 where as DC yeilds 3125/2975.. its the reduced latency and efficentcy that gives the boost over single channel..

3:I see around a5-10mhz max avantage in single over DC, but DC at the same timings, but 5-10 less than single will almost equal the results so you gain nothing..

4:the best AMD mobo with current design would yeild 100% efficentcy as with intel amd has no use for the bandwith that ecseeds the limit the fsb sets.. such as ASYNC on any mobo kt333-nf2 you see nothing in gain and 9 out of 10 a loss..
__________________
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #3 (permalink)  
Old 18th March, 2003, 10:23 AM
Lord Vetinari's Avatar
Member
 
Join Date: March 2003
Location: Worthing Sussex England
Posts: 325

qwerty57

Thanks for the reply,
Of course couldn't resist trying it.
Confirms your explanation.
Keeping the CPU around 2250MHz I normally run at 195*11.5 in DC
Can just about mamange 200*11 but is slower (no 11.25 multi!)
In Non DC mode can manage 205*11 but results are only marginally better.
viz Sandra membandwidth 2945,2774 vs 3010,2763
PCmark CPU 6867 vs 6920
PCmark Mem 5650 vs 5775
3dmark 2001 15220 vs 15266
tried 213*10.5 (next sensible settings) and 2dmark and 3dmark would not run . Sandramem gave 3168,2874
__________________
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #4 (permalink)  
Old 18th March, 2003, 10:28 AM
Chief Systems Administrator
 
Join Date: September 2001
Location: Europe
Posts: 13,075

To be totally honest, I'm not sure sandra is the best test. I think we should look at using a range other "tests" that might better reflect real world, and see if the DC might make more of a difference when more random reads come into play.

Anyone got any suggestions?

Áedán
__________________
Any views, thoughts and opinions are entirely my own. They don't necessarily represent those of my employer (BlackBerry).
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #5 (permalink)  
Old 18th March, 2003, 10:57 AM
Member
 
Join Date: November 2002
Posts: 64

Science Mark
__________________
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #6 (permalink)  
Old 18th March, 2003, 11:27 AM
Lord Vetinari's Avatar
Member
 
Join Date: March 2003
Location: Worthing Sussex England
Posts: 325

Aedan,
I assumed 3dmark 2001 was a reasonably suitable test.
Not true?
__________________
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #7 (permalink)  
Old 18th March, 2003, 11:28 AM
Holst's Avatar
Member
 
Join Date: September 2001
Location: Leics UK
Posts: 4,528

if your running the 8RGA+ with onboard graphics then dual channel gives a huge performance increase.

Dual channel is of less benefit when running an AGP card, 8rda or 8rga.

The difference between dual and single channel will vary between system to system.
I dont have two identical memory sticks, so I dont know on my own system, it will run the same dual or single as the memory is the limit.

From what ive read by Oppainter and other 3dmarkers dual channel is the only way to go in 3dmark even if there is a significant difference in max FSB.

The only way will be to test yourselves, id recomend using 3dmark, pcmark and superPI to try to work it out.
__________________
No longer Epox Tech.

Best of luck in the future all my friends.
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #8 (permalink)  
Old 18th March, 2003, 11:43 AM
Lord Vetinari's Avatar
Member
 
Join Date: March 2003
Location: Worthing Sussex England
Posts: 325

Here,s the full set of results:-
195*11.5 DC
sandra CPU 8404,3402
sandra multimedia 12443,13268
mem bandwidth 2945, 2774
PCMark CPU 6867
PCmark Mem 5650
PCMark HD 1204
3dmark 2001 15220

195*11.5 Non DC
Sandra CPU 8417,3383
sandra multimedia 12468, 13294
PCMark CPU 6853
PCMark Mem 5527
PCMArkHD 1200
3DMArk 2001 15078

205*11 Non DC
Sandra CPU 8445,3420
Sandra Multimedia 12530,13355
Sandra MemBandwidth 3010, 2763
PCMark CPU 6920
PCMark mem 5775
PCMark HD 1218
3DMark 2001 15266

Total memory= 512MB memsettings 5,3,2,2 in all tests
Didn't try with just one stick of memory.
__________________

Last edited by Lord Vetinari; 18th March, 2003 at 11:55 AM.
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #9 (permalink)  
Old 18th March, 2003, 05:39 PM
Banned
 
Join Date: January 2003
Posts: 189

Quote:
Total memory= 512MB memsettings 5,3,2,2 in all tests
Didn't try with just one stick of memory.
some disagree, but the abit nf7 for example if you use 1+2 dimm and not "3" you'll see a 200-500 point increase in 3dmark switch you didn't see.. My a7n8x asus and this epox yeild results almost identical as long as I use "ANY" two slots.. thus DC is 1+2 1+3 or 2+3.. as long as to slots are used DC is on. The downfall of not using the "3" slot (1+2 only) is maybe 50 points in 3dmark.. To be honest some of my fastest 3dmark scores came out of using slots 1+2 only no slot "3" witch is supposly the DC slot.
__________________
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #10 (permalink)  
Old 18th March, 2003, 06:22 PM
Lord Vetinari's Avatar
Member
 
Join Date: March 2003
Location: Worthing Sussex England
Posts: 325

According to the MB Manual, I am using Dimm 1 and dimm2 for normal DC operation (nearest and furthest from CPU). For the non DC operation I was using Dimm 2 and Dimm3.
Haven't tried testing DC using dimm1 and Dimm3
__________________
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #11 (permalink)  
Old 19th March, 2003, 03:49 AM
timbob2469's Avatar
Member
 
Join Date: December 2002
Location: Pa. USA
Posts: 805

"To be totally honest, I'm not sure sandra is the best test. I think we should look at using a range other "tests" that might better reflect real world, and see if the DC might make more of a difference when more random reads come into play."

"Anyone got any suggestions?"

Áedán

http://www.specbench.org/gpc/downloadindex.html
This is what they use to test high end CAD-CAM workstations.
http://www.cadonline.com/reviews/hardware/0203gc/
and this is what they tested.^^^^
According to the test my pc is about 2 times faster than a Celeron 1.7GHz.
bob
edit: Although you won't find it at that site YET , the march issue of that magazine does a comparison of 6 new workstations , 3 with Nforce2 motherboards and the other 3 Intel. The Nforce2 boards rank 2 , 3 , 5 . But are the suggested buy because of the price and because the differences between the number1 and no.6 are not that much . Check out some of those videocard prices $1500-$2000 , wow!
Attached Thumbnails
FSB and running dual memory on 8RDA-specvp-6.466.jpg  
__________________
EP-8RDA+ v1.0 3827 bios
XP1700+ JIUHB 175 x 12.5=2187MHz. 1.75v
2 x 512MB Crucial pc3200 10, 2, 2, 2.0
THERMALRIGHT SLK900u 92mm Zalman fan, Enlight 420 watt, AIW 9000 Pro 64mb DDR agp4x, voltage mod 1.85v
Home made NB cooler
AOA Team fah

Last edited by timbob2469; 19th March, 2003 at 04:00 AM.
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #12 (permalink)  
Old 19th March, 2003, 12:10 PM
Member
 
Join Date: February 2003
Posts: 179

If you really want to test the effect of the boost that dual channel memmory will give you in bandwidth terms, run something like specviewperf in dual and single modes
__________________
Thoroughbred 2400+ @ 2300Mhz, Abit NF7 2.0, 1536Mb Samsung PC2700 (DDR433)
LSI 20160 Ultra SCSI 3 HBA, 1xMaxtor Atlas 10K IV 36GB w/ 8mb cache
3ware Escalade 6410 Storage Switch, Unit 0: [RAID 10] 4x120Gb IBM 7K250 w/ 8mb cache (64K Stripe)
3Com 3CR990SVR97 Server NIC, DMX 6fire 24/96 LT, Hercules Radeon 9800Pro
Windows 2000 SP4


Angels fall farthest...
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #13 (permalink)  
Old 19th March, 2003, 01:21 PM
Lord Vetinari's Avatar
Member
 
Join Date: March 2003
Location: Worthing Sussex England
Posts: 325

OK specview7.1 results 3dmax test only.
Dc=Dual channel, NDc= Non Dual Channel
SMT= standard memory timings of 5,3,2,2
LMT= Lax memory timings of 7,3,3,2.5

195*11.5 Dc SMT weighted average 11.13 retest at end 11.17
195*11.5 Dc LMT weighted average 11.15

195*11.5 NDc SMT weighted Average 10.49
195*11.5 NDc LMT weighted Average 10.49

204*11 NDc SMT weighted Average 10.77
204*11 NDc LMT weighted Average 10.71

So DC at same FSB vs NonDC gives a 6.2% advantage to DC
going from 195 DC to 204 NonDC shows a 3.6% disadvantage in the higher FSB.
Therefore it would not be worth going non DC to achieve a higher FSB according to this.

On a side issue I was suprised the memory timings showed no significant difference between 5,3,2,2 and 7,3,3,2.5 in all the tests.
195*11.5=2253 Mhz, 204*11=2254 Mhz (on my system anyway)
Any comments?

Note the previous tests appear to have shown an equivalence or slight gain of 195*11.5 DC vs 205*11 NonDC, there was in that case a slightly bigger diff in the cpu frequency 2253 Vs 2265
__________________
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #14 (permalink)  
Old 19th March, 2003, 09:01 PM
Chief Systems Administrator
 
Join Date: September 2001
Location: Europe
Posts: 13,075

Quote:
Originally posted by Lord Vetinari
OK specview7.1 results 3dmax test only.
Dc=Dual channel, NDc= Non Dual Channel
SMT= standard memory timings of 5,3,2,2
LMT= Lax memory timings of 7,3,3,2.5

195*11.5 Dc SMT weighted average 11.13 retest at end 11.17
195*11.5 Dc LMT weighted average 11.15

195*11.5 NDc SMT weighted Average 10.49
195*11.5 NDc LMT weighted Average 10.49

204*11 NDc SMT weighted Average 10.77
204*11 NDc LMT weighted Average 10.71

So DC at same FSB vs NonDC gives a 6.2% advantage to DC
going from 195 DC to 204 NonDC shows a 3.6% disadvantage in the higher FSB.
Therefore it would not be worth going non DC to achieve a higher FSB according to this.

On a side issue I was suprised the memory timings showed no significant difference between 5,3,2,2 and 7,3,3,2.5 in all the tests.
195*11.5=2253 Mhz, 204*11=2254 Mhz (on my system anyway)
Any comments?

Note the previous tests appear to have shown an equivalence or slight gain of 195*11.5 DC vs 205*11 NonDC, there was in that case a slightly bigger diff in the cpu frequency 2253 Vs 2265
Nice work there. DC obviously gives that little bit of edge that Sandra doesn't know how to test for. I suspect this is probably more due to reduced latency in more random reads/writes. Doing sequential reads/writes obviously doesn't show DC's strong points.

Moving the FSB from 195MHz to 205Mhz is approximately a 5% increase in FSB bandwidth.

Áedán
__________________
Any views, thoughts and opinions are entirely my own. They don't necessarily represent those of my employer (BlackBerry).
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #15 (permalink)  
Old 19th March, 2003, 09:21 PM
Member
 
Join Date: November 2002
Location: Miami, FL.
Posts: 53

Quote:
Originally posted by Lord Vetinari
On a side issue I was suprised the memory timings showed no significant difference between 5,3,2,2 and 7,3,3,2.5 in all the tests.
195*11.5=2253 Mhz, 204*11=2254 Mhz (on my system anyway)
Any comments?
I'm not sure why, but I had noticed the same. I actually get better results running 7-3-3 cas2 than 6-3-3 or 5-2-2 (running DC).
__________________
RuKuS

XP2400+ @ 2485MHz (11@225) 2.15v
2x256MB Corsair XMS3500C2 3.05v (5-2-2 cas2 1T)
Epox 8RDA+ Vdd 2.15v
Radeon 9500NP SoftMod (All vmods)
Onboard Sound and LAN
SCSI 160 18Gb Seagate 15k
Adaptec 19160 Controller
Adaptec 2930 Controller
Plex 40x SCSI Reader
LiteOn 40x USB 2.0 Burner
Klipsch 5.1 ProMedia
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #16 (permalink)  
Old 21st March, 2003, 02:07 AM
timbob2469's Avatar
Member
 
Join Date: December 2002
Location: Pa. USA
Posts: 805

Wow! I am really glad to see someone else trying that SPECviewperf7.0 benchmark program . It is a huge download , and would probably take days to get it with a dial-up connection. I re-read that March article and my numbers were just a little bit off , they tested 8 systems , 3 with NForce2 motherboards and the other 5 with Intel based Pentium4 2.53 GHz cpu's and up. I still know that the AMD based systems -on the whole- got better reviews because they did as well at lower proceessor speeds. In the next couple of months I fully expect to see the NForce2 mobo's the clean the Intel clock. The lone CADCAM engineer at my place of employ (30 employees) seems really interested in my "Home made" pc , as well he should , at this point he just got an upgrade cpu to a P4 2.0 GHz. I think he has a case of computer envy , I must have gave that to him when I showed him my results from SVperf7 benches. Wait till he sees the numbers from Lord Vetinari , I might even be asked to bring MY whitebox to work to save the design dept. some rendering time , or at least convince them that the 8RDA+ is worth a job there.
bob
__________________
EP-8RDA+ v1.0 3827 bios
XP1700+ JIUHB 175 x 12.5=2187MHz. 1.75v
2 x 512MB Crucial pc3200 10, 2, 2, 2.0
THERMALRIGHT SLK900u 92mm Zalman fan, Enlight 420 watt, AIW 9000 Pro 64mb DDR agp4x, voltage mod 1.85v
Home made NB cooler
AOA Team fah
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #17 (permalink)  
Old 21st March, 2003, 03:30 AM
Banned
 
Join Date: January 2003
Posts: 189

Quote:
I'm not sure why, but I had noticed the same. I actually get better results running 7-3-3 cas2 than 6-3-3 or 5-2-2 (running DC).
ok whatever you say!!!

some app's and benchmarks are not cas or timings sensative.. however run 3dmark 2k1 which is very, very sensitive to timings and you'll see what effect it has.. well here.:

I score 17.5k at 4,2,2,2 DC at 225mhz fsb
I score 16.1k at 7,3,3,2.5 DC at 225mhz fsb

1400 marks ran the test over 15 times.. to all that don't see improvements then your PC is fast enough for what you do, but you ask any person that benches or goes for all out speed and they'll laugh at the quote I posted above..
__________________
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #18 (permalink)  
Old 21st March, 2003, 04:51 AM
Member
 
Join Date: November 2002
Location: Miami, FL.
Posts: 53

Quote:
Originally posted by qwerty57
ok whatever you say!!!

some app's and benchmarks are not cas or timings sensative.. however run 3dmark 2k1 which is very, very sensitive to timings and you'll see what effect it has.. well here.:

I score 17.5k at 4,2,2,2 DC at 225mhz fsb
I score 16.1k at 7,3,3,2.5 DC at 225mhz fsb

1400 marks ran the test over 15 times.. to all that don't see improvements then your PC is fast enough for what you do, but you ask any person that benches or goes for all out speed and they'll laugh at the quote I posted above..
Relax there qwerty57!!

I was talking about Sandra, not 3DMark. In 3DMark, I actually get better results at 5-2-2-2 than at 3-2-2-2. Everyone's results vary, so no need to over-react and get all excited...
__________________
RuKuS

XP2400+ @ 2485MHz (11@225) 2.15v
2x256MB Corsair XMS3500C2 3.05v (5-2-2 cas2 1T)
Epox 8RDA+ Vdd 2.15v
Radeon 9500NP SoftMod (All vmods)
Onboard Sound and LAN
SCSI 160 18Gb Seagate 15k
Adaptec 19160 Controller
Adaptec 2930 Controller
Plex 40x SCSI Reader
LiteOn 40x USB 2.0 Burner
Klipsch 5.1 ProMedia
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #19 (permalink)  
Old 21st March, 2003, 05:27 AM
refermadness's Avatar
Banned
 
Join Date: February 2003
Posts: 120

He just going through XS withdraws. lol.

It will all be well again in a day or so qwerty57 :beer:
__________________
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
  #20 (permalink)  
Old 21st March, 2003, 06:35 AM
Member
 
Join Date: November 2002
Location: Miami, FL.
Posts: 53

Quote:
Originally posted by refermadness
He just going through XS withdraws. lol.

It will all be well again in a day or so qwerty57 :beer:
Must be!! LOL

I miss them too, but JBell sent me the new IP for now..

Linky :beer:
__________________
RuKuS

XP2400+ @ 2485MHz (11@225) 2.15v
2x256MB Corsair XMS3500C2 3.05v (5-2-2 cas2 1T)
Epox 8RDA+ Vdd 2.15v
Radeon 9500NP SoftMod (All vmods)
Onboard Sound and LAN
SCSI 160 18Gb Seagate 15k
Adaptec 19160 Controller
Adaptec 2930 Controller
Plex 40x SCSI Reader
LiteOn 40x USB 2.0 Burner
Klipsch 5.1 ProMedia
Digg this Post!Add Post to del.icio.usBookmark Post in TechnoratiFurl this Post!
Reply With Quote
Reply



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
(8RDA) Different memory timings using dual-channel. Frazas EPoX MotherBoards 8 20th April, 2004 09:08 AM
8RDA+ and new dual channel memory ProphetX EPoX MotherBoards 8 22nd February, 2004 02:51 PM
Dual Channel memory on a 8rda+ CooledByWater EPoX MotherBoards 5 25th October, 2003 04:04 PM
8rda+ could you please post your highest fsb, running in dual mode. romper77 EPoX MotherBoards 1 7th February, 2003 05:52 AM
8rda+ expert memory setting... memory running sync or not?? macro EPoX MotherBoards 1 8th January, 2003 03:21 PM


All times are GMT +1. The time now is 02:03 AM.


Copyright ©2001 - 2010, AOA Forums
Don't Click Here Don't Click Here Either

Search Engine Friendly URLs by vBSEO 3.3.0