Operton 165 - 2.6 Ghz getting lower scores than a 2.25??

Chronictank

Chronictank

Chronictank

Soldato
Joined
12 Jan 2006
Posts
2,547
I was reading :
http://www.dfi-street.com/forum/showthread.php?t=40178
and it seems accodring to those results raw cpu power is better than optimising your ram and dropping a few mhz as a result. So i increased my divider to 166.
(setup in sig)

I am a noob overclocker, but i think i have finally got my head around the basics. My main sources of info are:
Overclockers UK
Definitive OC Guide
Somewhat complete a64 oc guide

I have followed these guides to the letter, then played around but i seem to have problems getting past the 6k hurdle on 3d mark 06 using 1t timings(which i achieved on 2t timings at 2.3 Ghz).

Some stats ^^:
Max CPU FSB: 290Mhz, i didnt want to take my vcore above 1.475 so stopped here
Max RAM: 250 Mhz

Stock timings on ram: 2.5-3-3-6
Stock FSB: 200Mhz
Stock Multi: x9

Setup 1 - 5/06 divider (200Mhz)
FSB: 290
LDT/FSB: x3
LDT Bus: 16 16
CPU/FSB: x9
PCI Xpress Freq: 101Mhz
Cool n Quiet: Disabled
CPU VID StartUp: StartUp

CPU VID: 1.35 (effective 1.485 when vid special kicks in)
CPU VID Special: 110%
LDT Voltage: 1.20
Chip Set Voltage: 1.50
DRAM Voltage Control: 2.80

Memory Settings
DRAM Freq: 166 (5/06)
CPC: Enable
CAS Latency Control (Tcl): 3.0
RAS to CAS delay (Trcd): 3
Min RAS active time (Tras): 7
Row Precharge (Trp): 03
Row cycle time (Trc): 07
Row Refresh cyc time (Trfc): 11
Row to Row Delay (trrd): 1
Write Recovery time (Twr): 2
Wrote to Read Delay (Twtr): 1
Read to Write delay (trwt): 3
Refresh Period (tref): 2560 (166Mhz 3.9us)
Write CAS Latency (twcl): 1
DRAM Bank Interleave: Enabled

DQS Skew Control: Auto
DQS Skew: 0
DRAM Drive Strength: 7
DRAM Data Drive Strength: 2
Max Async Latency: Auto
DRAM Responce Time: Fast
Read Preamble Time: 6
Idle Cycle limit: 32
Dynamic Counter: Enable
R/W Queue bypass: 16
Bypass max: 4
32 Byte Granularity: Disabled

3d mark 06 score: 5328
SM 2.0: 2581
SM 3.0: 2388
CPU score: 1181

Setup 1 - 1/01 divider (200Mhz)
Using a 1/01 divider i found that the max FSB on my ram i could achieve was 250 Mhz without losing stability, even at 251 i started getting memory errors on test 8 on memtest,
at 253 i was failing both tests 5 and 8

FSB: 250
LDT/FSB: x3
LDT Bus: 16 16
CPU/FSB: x9
PCI Xpress Freq: 101Mhz
Cool n Quiet: Disabled
CPU VID StartUp: StartUp

CPU VID: 1.400
CPU VID Special: Auto
LDT Voltage: 1.20
Chip Set Voltage: 1.50
DRAM Voltage Control: 2.80

Memory Settings
DRAM Freq: 200 (1/01)
CPC: Enable
CAS Latency Control (Tcl): 3.0
RAS to CAS delay (Trcd): 4
Min RAS active time (Tras): 8
Row Precharge (Trp): 04
Row cycle time (Trc): 07
Row Refresh cyc time (Trfc): 11
Row to Row Delay (trrd): 2
Write Recovery time (Twr): 2
Wrote to Read Delay (Twtr): 1
Read to Write delay (trwt): 3
Refresh Period (tref): 3120 (200Mhz 3.9us)
Write CAS Latency (twcl): 1
DRAM Bank Interleave: Enabled

DQS Skew Control: Increase Skew
DQS Skew: 255
DRAM Drive Strength: 7
DRAM Data Drive Strength: 2
Max Async Latency: 7.0
DRAM Responce Time: Fast
Read Preamble Time: 5
Idle Cycle limit: 32
Dynamic Counter: Enable
R/W Queue bypass: 16
Bypass max: 7
32 Byte Granularity: Disabled

3d mark 06 score: 5876
SM 2.0: 2574
SM 3.0: 2472
CPU score: 1695

So increasing the clock speed but chaning the setup to asynchrous actually made the performance worse. And taking the setup onto 1t timings (2.3Ghz, ram settings on auto i got 6005 on the 3dmark) also reduced my score further.

I was under the impression that putting the settings onto 1t would give the system speeds a boost, not reduce them :s

Is there any way i can boost the system performance on 1t timings, or would it be better to go back to 2.3 Ghz on 2t timings as i didnt even really scratch the service in terms of oc'ing.

But surely with higher clock speeds i should be getting higher 3d mark scores?
 
Last edited:
What processor are you using? Have you tried using the 9.5 or 10 multiplier?

If you use the .5 multipliers you can't use the memory divider though.
 
Jimi said:
If you use the .5 multipliers you can't use the memory divider though.
Click to expand...


Why cant you? the memory dividers still work. Just they use the rounded up multiplyer.
Know your stuff before posting in future


Tom
 
welshtom said:
Why cant you? the memory dividers still work. Just they use the rounded up multiplyer.
Know your stuff before posting in future


Tom
Click to expand...

Exactly, they round up. What would be the point in using a .5 multiplier if it rounds up? :rolleyes:
 
the multi doesnt change. The cpu speed will be the same

the memory divider calculates the memory speed from the nearest (up) multi.
 
sorry thought the spec would come up with my sig if i added it in the CP

-DFI LanParty UT NF4 Ultra-D (Socket 939) PCI-Express Motherboard
-Bios 06/23/2005 v6.00 PG
-AMD Opteron UP 165 Dual Core San Diego 1.8GHz 2MB Cache (Socket 939), CCBWE 0601 UPMW stepping
-Sythe Ninja Plus Heatpipe with AKSA Ultra Quiet 120mm fan
-G.Skill 2GB DDR HS PC3200 (2x1GB) CAS2.5 Dual Channel Kit (F1-3200USU2-2GBHS)
-MSI NX7900 GTX 512mb (GeForce 7900 GTX)
-Seasonic S12 600W Silent ATX2.0 Power Supply
-Windows XP Pro SP 2

Edit: lol there it is ^^
 
mmm it is in your sig now. Although it isn't allowed to have a sig that big, so before a don jumps on you, I'd read the FAQ and get it changed :)
 
hehe its more why go to 2.6 ghz on 1t timing when i reached 6k on 2t timings with cpu clocked at 2.3Ghz,
i.e whats the advantage of the higher clock speed/faster memory timings
 
Last edited:
Have you got a compare url for your 6k+ score @ 2.3ghz then would be interested to see what you scored and what settings exactly did you use, as i said i have practically the same setup and if i can get better performance from it at lower speed so much the better.
 
5900,everything on stock, ram switched to 2t timings
Last 100 is relatively easy to do with tweaking, back to getting 2.6Ghz to work ><
http://service.futuremark.com/compare?3dm06=260654
Installed the AMD dual core driver
Was curious what would happen on 2.6Ghz 2T timings,
for some reason (checking using cpu-z), it defaults to dual channel.
But heres the score:
6042
Memory untouched, interestingly enough exactly the same score as you, even tho i should have a higher score because of my gfx card and a couple of extra mhz
To confirm this i ran the same test at 1T timings;
6062
Ran it again on 2.25 with the new drivers
And the result is a lower score.
5858
Seems like the AMD Dual Core driver has a significant impact on system performance

Updated the gfx card drivers seemed to have made a impact, look more like the scores i was getting before i formatted :)
3DMark Score 6160 3DMarks

SM 2.0 Score 2631 Marks

SM 3.0 Score 2533 Marks

CPU Score 1958 Marks
 
Last edited:
bit more like it matey, one thing i noticed though, you beat me on shader model2 yet i beat you on sm3, weird huh :confused: nvidia thing maybe?? dunno, decent overall score anyway, nice one :D


Oh your linkies don't seem to work have you made them public / published them?
 
rxmac said:
bit more like it matey, one thing i noticed though, you beat me on shader model2 yet i beat you on sm3, weird huh :confused: nvidia thing maybe?? dunno, decent overall score anyway, nice one :D


Oh your linkies don't seem to work have you made them public / published them?
Click to expand...
realised i can only publish one link at a time with a free account, so left the 2.3 ghz one you wanted to see up :)
try this one
http://service.futuremark.com/compare?3dm06=260654
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom