Dear Paul,
Thanx you for your excellent explanation !
I'm a relative newbie on the subject and appreciate your help very much...
I have OC'ed my system now to 13 x 160.4 = 2084.7. The bios says it detecs a
2700+ chip and that the memory is running on 160 Mhz Dual mode !
I had a big increase in speed tested in Scisoft Sandra and my Cubase SX is
running all my virtual instruments without problems now !!
My temperature is still 45 'C, so that's fine..
I can still go higher but than i have to physically unlock the lower
multipler values and that's still a little too scary... I would love to get
my fsb to 200 so I get maximum results, but that's for later..
I have ran CPU-Z and outputed all the result to a textfile... Is that of any
use to publish it here ? Or would that be a waste of usenet-space ?
another thank you!!
cheers,
Jeroen
| In article <
[email protected]>, "Jeroen"
|
| > Hi there,
| >
| > Can anybody help me with synchronising my Corsair CMX256A 3500C2 memory
?
| > I'm trying to get the most out of the onboard nforce2 sound. I read on
the
| > cubase forum that i should get my memory on 200 Mhz in sync with the
CPU?
| >
| > I have the following system:
| >
| > A7N8X Deluxe rev.1.x, Bios 1003
| > AMD Athlon 2400+
| > 2 sticks of Corsair CMX256A 3500C2 memory, installed on bank 1 & 3
| >
| > When I try to lower the multiplier to 12 and set the FSB to 166 MhZ my
| > system reports 1800+ and memory at 100 Mhz and my benchmarks are way
lower
| > than normal...
| >
| > i found the memory timings to be:
| >
| > =============================
| > Testing parameters for XMS3500 CAS 2:
| > CPU Speed: Manual
| > CPU External Frequency: 145
| > CPU to Memory FSB ratio: 4:6
| > SDRAM Configuration: Manual
| > SDRAM CAS Latency: 2T
| > SDRAM RAS to CAS Delay (tRCD): 3T
| > SDRAM RAS Precharge (tRP): 3T
| > SDRAM Active to Precharge Delay (tRAS): 7T
| > SDRAM Command Control: 1T
| > SDRAM Bank Interleave: 4 Banks
| > Memory Voltage: motherboard default*
| > ================================
| >
| > in the following thread someone claims to have his pc running with a
2400
| > chip and the same two 3500 sticks as mine with no problem but he does
not
| > mention the excact settings:
| >
| >
|
http://groups.google.com/groups?hl=...c3500&hl=en&lr=&ie=UTF-8&safe=off&sa=N&tab=wg
| >
| > I really appreciate your help !
| >
| > byebye,
| >
| > jeroen
|
| I see one 2400+ listed - a Model 8 Thoroughbred CPUID 0681
| 2000 (2400+) 1.65 V 62.0 W Mult=15x CLK=133MHz Tmax=85°C
|
| It has a high multiplier. On a motherboard that doesn't drive the
| top multiplier bit, the multiplier values don't work the way they
| are supposed to.
|
|
http://fab51.com/cpu/barton/athlon-e23.html
|
| Look at the table entitled "L3 : SYSCLK Multiplier".
| It says "Close L3-FID 4 : 5X to 12.5X
| Open L3-FID 4 : Over13X" <--- This is the processor you've got
|
| That means, since your processor has a 15X multiplier, then the
| multipliers from 13X upwards are available. The L3-FID4 bridge
| is open circuit. A socket mod or bridge mod can make the lower
| multipliers available. (This is true for motherboards that only
| drive FID3..FID0 and not FID4. There is at least one brand of
| motherboard that drives all five bits, making a socket mod
| unnecessary for unlocked Athlon processors.)
|
| So, your first problem is, the multiplier you are setting, is
| not the multiplier value that the processor is getting.
|
| To get the lower multipliers on an unlocked Athlon, see this page:
|
|
http://www.ocinside.de/go_e.html?/html/workshop/pinmod/amd_pinmod.html
|
| Select
| "Socket view"
| "AMD T-Bred"
| "Vcore selection"
| "Unlock - Multi via BIOS up to 12.5x" <-- Shows how to close FID4
| "FSB selection"
|
| This will show a view of the motherboard socket, with a wire mod added
| to set FID4, so that the lower multipliers are selected. You can also
| do this modification a better way, by selecting "PCB view", and this
| shows what the back of the motherboard looks like. Soldering a wire or
| using conductive paint, to join the two pins shown, will provide
| the lower multipliers. At least, this is my understanding of how this
| stuff works (I don't own an Athlon board and haven't tested this).
|
| To set up the board, you need to sort out the multiplier, then set
| the "CPU External Frequency" to 133MHz if the FSB is rated for FSB266.
| The "Memory Frequency" is set to [Sync] or [100%], for the memory
| and the CPU to use the same clock frequency.
|
| So, for your experiments, you need to leave the CPU clock at a lower
| frequency, then do the FID4 mod, then play with the multiplier value
| and check in Windows, that the multiplier value is the one you expected.
| With the FID4 mod in place, the multipliers from 5X to 12.5X should be
| available. Once you understand the relationship between the value shown
| in the multiplier menu and the real value seen in Windows with a utility,
| you are ready for making real adjustments. Then, you can set the clock
| to 166MHz, multiplier to 12X, giving a core of 2000MHz, and with the
| "Memory Frequency" set at [Sync] or [100%], the memory clock will
| in this case be 166MHz and the actual memory rate is DDR333.
|
| If you choose not to modify L3-FID4, then I think 13X is the lowest
| multiplier you can get. That allows a CPU external clock of 154MHz,
| which allows the memory to run at DDR308. If you use more external
| clock than 154, that would be an overclock of the core, and perhaps
| more Vcore would then be needed for stability.
|
| My best guess,
| Paul
