Jump to content


Photo
- - - - -

overclock my toledo x2 3800


  • Please log in to reply
16 replies to this topic

#1 colelt1

colelt1

    Member

  • Members
  • PipPip
  • 159 posts

Posted 27 December 2005 - 08:49 PM

Here is my 1:2, 2.5 4 4 8, 10x, results for the past 2 weeks.

1.350v = 2279mhz (baseline)(34c)
1.375v = 2321mhz (+.025v + 42mhz)(36c)
1.400v = 2360mhz (+.025v + 39mhz)(36c)
1.425v = 2388mhz (+.025v + 28mhz)(37c)
1.450v = 2401mhz (+.025v + 13mhz)(38c)
1.475v = 2430mhz (+.025v + 29mhz)(38c)
1.500v = 2460mhz (+.025v + 20mhz)(38c)
1.525v = 2470mhz (+.025v + 10mhz)(39c)
1.550v = 2480mhz (+.025v + 10mhz)(41c)
1.582v = 2500mhz (+.025v + 20mhz)(42c)

Looks like my 3800 doesnt respond to voltage worth a crap, and I have spend the past few weeks getting no better results. PWMIC and Chipset say below 43c so thats not the problem.

Ram does not like >2.5v, but is stable at DDR596 @ 2.5v
I have tried 2T, 3 5 5 9 with no help.
If I cant get anything better out of it, then I will stick for low 2.4ghz with lower heat, but I can do:
1) 10x240 with tight timings
2) 10x240 with memory on the memory 233 7:6 mult (DDR560)
3) 9x267 with g.skill suggested DDR550 timings (2.5 3 3 7 or lower)
4) 8x300 with g.skill suggested DDR600 timings (2.5 4 4 8 or lower)

FSB Bus Frequency.............................  -  See above

LDT/FSB Frequency Ratio.......................  -  Auto (also tried 3x and 4x)

CPU/FSB Frequency Ratio.......................  -  10x

PCI eXpress Frequency.........................  -  100Mhz



CPU VID StartUp Value.........................  -  STARTUP



CPU VID Control...............................  -  see above

CPU VID Special Control.......................  -  see above

LDT Voltage Control...........................  -  1.20v 

Chip Set Voltage Control......................  -  1.50v

DRAM Voltage Control..........................  -  2.50v



DRAM Configuration Settings:



DRAM Frequency Set............................  -  100=RAM/FSB:01/02

Command Per Clock (CPC).......................  -  Auto

CAS Latency Control (Tcl).....................  -  2.5

RAS# to CAS# delay (Trcd).....................  -  04 Bus Clocks

Min RAS# active time (Tras)...................  -  08 Bus Clocks

Row precharge time (Trp)......................  -  04 Bus Clocks

Row Cycle time (Trc)..........................  -  12 Bus Clocks

Row refresh cyc time (Trfc)...................  -  24 Bus Clocks

Row to Row delay (Trrd).......................  -  03 Bus Clocks

Write recovery time (Twr).....................  -  03 Bus Clocks

Write to Read delay (Twtr)....................  -  02 Bus Clocks

Read to Write delay (Trwt)....................  -  03 Bus Clocks

Refresh Period (Tref).........................  -  3120 Cycles

DRAM Bank Interleave..........................  -  Enabled



DQS Skew Control..............................  -  Auto

DQS Skew Value................................  -  0

DRAM Drive Strength...........................  -  Auto

DRAM Data Drive Strength......................  -  Auto

Max Async Latency.............................  -  Auto

DRAM Response Time............................  -  Normal

Read Preamble Time............................  -  Auto

IdleCycle Limit...............................  -  256 Cycles

Dynamic Counter...............................  -  Disable

R/W Queue Bypass..............................  -  16 x

Bypass Max....................................  -  04 x

32 Byte Granularity...........................  -  Disable(4 Bursts)


#2 lowboy

lowboy

    Total Nerd

  • Members
  • PipPipPipPipPip
  • 4113 posts

Posted 27 December 2005 - 09:08 PM

Looks like you worked it out pretty good, some cpus clock better than others. Buying cpus is a crap shoot.

#3 colelt1

colelt1

    Member

  • Members
  • PipPip
  • 159 posts

Posted 28 December 2005 - 04:16 AM

Anyone want to tweak me to see if there is something else holding me back??

This computer runs 24/7, so 1.582v is over what I want. So I guess I will drop it down to 2.4ghz (20% overclock isnt just awful) and try to tighten up my volts. At 1:1, 1.450v failed overnight test and I want 24hr stable, so I switched to 1.469v and am going to recheck everything with my DMM

All things considered, what one of these 4 options would you suggest:
1) 10x240 with tight timings (seems a waste of my ddr600)
2) 10x240 with memory on the memory 233 7:6 mult (DDR560) (have heard nothing of the 7:6 mult)
3) 9x267 with g.skill suggested DDR550 timings (2.5 3 3 7 or lower) (what I am planning on running)
4) 8x300 wih g.skill suggested DDR600 timings (2.5 4 4 8 or lower) (going to benchmark this setting too and see what give me better results)

#4 J054

J054

    Member

  • Members
  • PipPip
  • 72 posts

Posted 28 December 2005 - 04:28 AM

...
This computer runs 24/7, so 1.582v is over what I want.
...


As long as it has good temps it should be fine, I run my setup with 1.59volts to the cpu, somtimes for days and temps always stay under 35C load (i kno its a diff cpu etc.. but for me if temps are good then it will work).

#5 JVG81

JVG81

    Member

  • Members
  • PipPip
  • 231 posts

Posted 28 December 2005 - 04:47 AM

1.350v = 2279mhz (baseline)(34c)
1.375v = 2321mhz (+.025v + 42mhz)(36c)
1.400v = 2360mhz (+.025v + 39mhz)(36c)
1.425v = 2388mhz (+.025v + 28mhz)(37c)
1.450v = 2401mhz (+.025v + 13mhz)(38c)
1.475v = 2430mhz (+.025v + 29mhz)(38c)
1.500v = 2460mhz (+.025v + 20mhz)(38c)
1.525v = 2470mhz (+.025v + 10mhz)(39c)
1.550v = 2480mhz (+.025v + 10mhz)(41c)
1.582v = 2500mhz (+.025v + 20mhz)(42c)


Damn I like it, pretty methodical and everyone should do this. :)

Looks like my 3800 doesnt respond to voltage worth a crap, and I have spend the past few weeks getting no better results. PWMIC and Chipset say below 43c so thats not the problem.


It happens, you could still get that IHS off or consider buying some killerstepping off ebay or something. IMO 2.5 @ 1.58v is a nice OC anyway and personally I wouldnt be too afraid to use this 24/7.

4) 8x300 with g.skill suggested DDR600 timings (2.5 4 4 8 or lower)


Do this or 8*312 @1.58v, for some very nice overall speed. Dont be putting DDR600 on dividers and such, thats just silly.

p.s. you have a 3800+ manchester core not toledo I presume

#6 colelt1

colelt1

    Member

  • Members
  • PipPip
  • 159 posts

Posted 28 December 2005 - 05:33 AM

Yeah, I have a 5+ page overclock.txt file on my desktop with EVERY single step I took, what you are seeing is just a summary of what didnt crash.

It happens, you could still get that IHS off or consider buying some killerstepping off ebay or something. IMO 2.5 @ 1.58v is a nice OC anyway and personally I wouldnt be too afraid to use this 24/7.


The IHS is off, it didnt help my overclock any, and I have tried Artic silver cermaque and just now swiched to Artic silver 5 with no advantage. I put a tiny drop on the core, and spread it over the core with a credit card, and had it nicely covered... but maybe I should try putting a PAPER thin layer on there, and see if it helps. (the 9500 has a mirror finish polish job on the contact copper). Then lay the computer on the side to make sure its not gravity hurting me. Its always cpu0 that fails, I have yet to see cpu1 fail (and have had it do 2.5ghz on stock voltage :( )

I leave it on 24/7 normally with Counterstrike minimized when im sleeping, which still uses 50+% of my cpu, so temps would sit around 41cpu 42pwmic 44 chipset if I did 1.582v = 2500mhz, but would all be below 40 with their fans not running 100% if I did 2.4ghz @ 1.469v.

Do this or 8*312 @1.58v, for some very nice overall speed. Dont be putting DDR600 on dividers and such, thats just silly.

p.s. you have a 3800+ manchester core not toledo I presume


I wasnt saying put the ram on a divider, I was saying on the 233 multiplier to raise it even higher (2.33 * bus speed)

And its not a manchester, I got a weird retail x2 3800 toledo see stepping below:
AMD Athlon 64
ADA3800DAA5CD
CCBWE 0535SPMW
1319559I50571

It prob started life as a 4400 or 4800, had a defect, and failed a test.

#7 JVG81

JVG81

    Member

  • Members
  • PipPip
  • 231 posts

Posted 28 December 2005 - 09:41 AM

So does that cpu have half the cache disabled?

#8 Guest_Ryn_*

Guest_Ryn_*
  • Guests

Posted 28 December 2005 - 09:58 AM

And its not a manchester, I got a weird retail x2 3800 toledo see stepping below:
AMD Athlon 64
ADA3800DAA5CD
CCBWE 0535SPMW
1319559I50571

It prob started life as a 4400 or 4800, had a defect, and failed a test.


Dang! I have almost the exact steppings that you got except mine is 0535TPMW. I'll start building my rig on my sig tomorrow and will OC after a week. Hopefully I'll reach 2.5GHz with 1.5Vcore.

#9 colelt1

colelt1

    Member

  • Members
  • PipPip
  • 159 posts

Posted 28 December 2005 - 11:13 AM

So does that cpu have half the cache disabled?


Correct

Tonight I am giong to remove the 9500, put the most thin layer of AS5 I can put on there, lay my computer on its side, let it set for a few days, then see what happens. I suspect no difference, I just think I have 1 core that is a bastard.

I really wish I would of known about the 939 opterons before I built my x2

Anyone have any values they want me to tweak in 2 days to see if it helps. I am going to pull a ram chip and see what happens when I use a single 512 chip.

#10 SkeptiKal12

SkeptiKal12

    New Member

  • Members
  • 15 posts

Posted 28 December 2005 - 02:54 PM

Yeah, I have a 5+ page overclock.txt file on my desktop with EVERY single step I took, what you are seeing is just a summary of what didnt crash.



The IHS is off, it didnt help my overclock any, and I have tried Artic silver cermaque and just now swiched to Artic silver 5 with no advantage. I put a tiny drop on the core, and spread it over the core with a credit card, and had it nicely covered... but maybe I should try putting a PAPER thin layer on there, and see if it helps. (the 9500 has a mirror finish polish job on the contact copper). Then lay the computer on the side to make sure its not gravity hurting me. Its always cpu0 that fails, I have yet to see cpu1 fail (and have had it do 2.5ghz on stock voltage :( )

I leave it on 24/7 normally with Counterstrike minimized when im sleeping, which still uses 50+% of my cpu, so temps would sit around 41cpu 42pwmic 44 chipset if I did 1.582v = 2500mhz, but would all be below 40 with their fans not running 100% if I did 2.4ghz @ 1.469v.



I wasnt saying put the ram on a divider, I was saying on the 233 multiplier to raise it even higher (2.33 * bus speed)

And its not a manchester, I got a weird retail x2 3800 toledo see stepping below:
AMD Athlon 64
ADA3800DAA5CD
CCBWE 0535SPMW
1319559I50571

It prob started life as a 4400 or 4800, had a defect, and failed a test.



Hey I think I have the exact same stepping (I slapped an extra 5 in my sig so dont worry). I have gotten to 2.4 at roughly 1.425v and am running it now at 2.6 with 1.65v (in bios, in cpu-z its lower). Cooling with a Big Typhoon and at load I see 52-53 C. I've found that if you let the temps get passed ~55C it will fail even at lower frequency. I can pass 8+ hours of prime at this speed and am comfortable with the temperature.

Like yourself, I have removed the IHS and found minimal gains (but I am going to retighten the heatsink to make sure its REALLY on there).


Its not a great overclocker but I can't say Im not happy with 2.6GHz (although the voltage could be lower).

#11 colelt1

colelt1

    Member

  • Members
  • PipPip
  • 159 posts

Posted 28 December 2005 - 03:06 PM

Why are your temps so high?? Small case? I just put mine at 1.65 and ran prime for 10 minutes and it only got to 43-44c, but I do have 6 case fans + 2 psu fans + 1 pwmic fan + vid card + zalman 9500, thats alot of fans.

Thats cool, but I worry about 1.6v or higher, I dont want a sudden cpu death.... although my temps at 1.6v are only 42 running dual prime, but it is winter and when spring/summer comes around my temps may rise by a large amount.

I just applied more artic silver and I should of taken a picture, it almost looks like the center of the core has the AS5 pressed thin while it is thicker twards the outside.... so I put the thinnest layer of AS5 that was possible, like took me 30 minutes to get it that thin with no core showing through, and I have my case laying down. I will wait a few days and see what happens.

#12 mongoled

mongoled

    Member

  • Members
  • PipPip
  • 199 posts

Posted 28 December 2005 - 03:09 PM

Not tht i think its going to make any difference to your ovrclk, but you shouldnt be spreading the arctic silver 5 according to the arctic silver website. Wht they say u should do is place a very small dab on the die and use the heatsink pressur to spread the paste. The reason thy suggest not to spread the paste with CC is as not to get air bubbles in the paste.

I see tht u are also conscious of the amount of voltage you are using. Where are the values for vCore you are quoting coming from? If they are not from a DMM I would suggest you get your readings with such a device. These boards overvolt quite a bit, my board in particular is quite perculiar.

Example

vCore reading bios = 1.52v
vCore reading DMM idle = 1.55v
vCore reading DMM load = 1.58v

Dont know if other boards act this way (i started a thread at XS abt this but only one person responded)