athlon64 overclock record. all owners please post

[paul007]

Well my Prometeia is finally running again and this is at stock voltage of 1.7V

sorry stock voltage is 1.5v .2v less than that nice clocks anyways.

[andyddr]

paul007 I think he mean 1.7v is the max on the board...so `stock` as in with no Vmod (yet).

[paul007]

ahh gotcha

[Cole]

Yeh sorry ... I meant without a VCore mod yet ... my bad.

[SKYNET]

[lordrage]

I didnt write mine down

but it is an A64 3000+
stable at 10*251 1.7v
Waterchill cooled (not refrigerated)

[macci]

Here is some data on A64 RAM dividers w/ different CPU multiplier settings:

By Morkai of Muropaketti.com:

4x250Mhz 2900/2900MB/s
4.5x250Mhz 3280/3240
5x250Mhz 3645/3600
5.5x250Mhz 3566/3530
6x250Mhz 3891/3850
6.5x250Mhz 3600/3600
7x250Mhz 3880/3880
7.5x250Mhz 3620/3620
8x250Mhz 3860/3860
8.5x250Mhz 3635/3635
9x250Mhz 3860/3860
9.5x250Mhz 3685/3685
10x250Mhz 3880/3880

Interesting Sandra mem bench numbers I'd say.

As you know the RAM speed is divided from overall CPU MHz and not from FSB speed on A64 systems. So for example at 2500MHz with 10x multiplier the RAM speed is 2500/10=250MHz.
The thing here is that running w/ x.5 multiplier (9.5 for example) makes no sense at all. The RAM divider wont change when you adjust multiplier from 10x to 9.5x and so the performance will be the same if CPU speed remains the same. So 250x10=2500MHz (RAM at 2500/10 = 250MHz) is just as fast as 9.5x263=2500MHz (RAM at 2500/10 = 250MHz).
CPU-Z will report that that the RAM speed goes up (from 250 to 263) when you go from 10x250 to 9.5x263 but as you can see from the Sandra MEM scores the perfromance wont go up (9x250Mhz=3860/3860, 9.5x250Mhz=3685/3685, 10x250Mhz=3880/3880). Obviously the CPU-Z is not reporting the RAM speed properly when '1:1' mode is used w/ x.5 CPU multiplier.
So every highspeed RAM OC shot posted w/ x.5 CPU multiplier is false as the RAM is not running in true 1:1 mode w/ x.5 multiplier.

The 4x multiplier is yet another thing here too. If you check the numbers I quoted there it seems that CPU/5 is the lowest ram divider the A64 supports. So '381x4 1:1' does not mean that the RAM is running at DDR762. It seems that its actually running at 304MHz (1524MHz/5).

[A64'uSER]

Do any of you guys know how i can change my multiplier??

[[_Silence_]]

Originally posted by A64'uSER
Do any of you guys know how i can change my multiplier??

Get a program called Clock Gen (www.cpuid.com) and u change the multi in windows.

[Tom Holck]

Originally posted by macci
Here is some data on A64 RAM dividers w/ different CPU multiplier settings:

By Morkai of Muropaketti.com:

Interesting Sandra mem bench numbers I'd say.

As you know the RAM speed is divided from overall CPU MHz and not from FSB speed on A64 systems. So for example at 2500MHz with 10x multiplier the RAM speed is 2500/10=250MHz.
The thing here is that running w/ x.5 multiplier (9.5 for example) makes no sense at all. The RAM divider wont change when you adjust multiplier from 10x to 9.5x and so the performance will be the same if CPU speed remains the same. So 250x10=2500MHz (RAM at 2500/10 = 250MHz) is just as fast as 9.5x263=2500MHz (RAM at 2500/10 = 250MHz).
CPU-Z will report that that the RAM speed goes up (from 250 to 263) when you go from 10x250 to 9.5x263 but as you can see from the Sandra MEM scores the perfromance wont go up (9x250Mhz=3860/3860, 9.5x250Mhz=3685/3685, 10x250Mhz=3880/3880). Obviously the CPU-Z is not reporting the RAM speed properly when '1:1' mode is used w/ x.5 CPU multiplier.
So every highspeed RAM OC shot posted w/ x.5 CPU multiplier is false as the RAM is not running in true 1:1 mode w/ x.5 multiplier.

The 4x multiplier is yet another thing here too. If you check the numbers I quoted there it seems that CPU/5 is the lowest ram divider the A64 supports. So '381x4 1:1' does not mean that the RAM is running at DDR762. It seems that its actually running at 304MHz (1524MHz/5).

Interesting Macci,

Do you think it's the same thing on FX-51?

I now have the Corsair ER3500 they can do FSB 250-260 depending on the cas timings.
I run some SuperPi M32: 250 x 12.5 = 3125 the time was only a few sek better than 240 x 13 = 3120.
Timings was 2-6-2-2

Recording to you, then 12 x 255 = 3060 is to slow, and 12 x 260 = 3120 must have slow timings. None of them is better than 13 x 240 cas 2-6-2-2. Right?
I might go 13 x 3120 cas 2-5-2-2 that will be the best...
Still waiting on my new Cascade, have made some change it will be a killer sub -120 C on load. The "old" one then going on the GPU. Have serius condensation problem on the ATI XT some of the small resistans almost korrode..

[A64'uSER]

Originally posted by diogojcr
Get a program called Clock Gen (www.cpuid.com) and u change the multi in windows.

Well, i already have that program, but when i'm try to change the multiplier, my computer freeze... ?

[alz85]

Which stepping has the highest possiblity to clock high ?

[paul007]

CAAOC seems very good

[SKYNET]

memory 1:1 2.5-2-2-5

[sandman]

How did you keep teh PCI and agp speeds right on?

[macci]

SKYNET, your RAM is running at 265MHz
Lets see a true 279RAM speed at 9x279?

[SKYNET]

Originally posted by macci
SKYNET, your RAM is running at 265MHz
Lets see a true 279RAM speed at 9x279?

ahhh, right

279 : 10 x 9.5 = 265

thats #%&#$§%

[alz85]

Originally posted by macci
SKYNET, your RAM is running at 265MHz
Lets see a true 279RAM speed at 9x279?

So if, for example, my cpu is running at 9.5 x 260, and ram is set at 1:1 in bios, the ram is running at only 247 MHz ?

I have a 3000+ CAAMC 0349 SPAW, which does 2.45 GHz primestable at 1.7v, keep it or sell it ?

[SKYNET]

Originally posted by alz85
So if, for example, my cpu is running at 9.5 x 260, and ram is set at 1:1 in bios, the ram is running at only 247 MHz ?

jes

260 : 10 x 9.5 = 247

[Formann]

Finaly got a new one..

A64 3200+
CAAMC 0339WPMW (same as my previous )
"9" greenie

This chip is about 10c warmer than my last one.. its a weird chip. The differece between load and idle-temp is ~1.5c
First i thought it was the sensor.. but when I gave it a try in my Vapo PE I saw the same.. the evaporator-temp idle is -17c and load -15c.


Look at the cpu-temp.. I almost feel sorry for my PE.



Sometime in the near future i plan to 404mod my PE.. not sure it will work out tho.. the BD50F´s doesn´t like 404 very much.

[andyddr]

I know this is a bit off topic but anyone can with a K8NPRO (with F11 bios) tell me what you see under Cpu-Z`s `Mainboard` tab. Where it says `Bios version` do you see F7 or F11? Just not sure my system is `reading` my flashed bios (although in the boot screen it shows F11).

[s e t h]

same thing with Cole's board andy. it's not you its common.

[Pejsen]

Originally posted by macci

Very interessting find Macci, or should i say Morkai

Seems that 8.5x is the same story as 9.5x, and the lower you go the worse it gets

*edit* Except for the 6x multi.

[paul007]

know this is a bit off topic but anyone can with a K8NPRO (with F11 bios) tell me what you see under Cpu-Z`s `Mainboard` tab. Where it says `Bios version` do you see F7 or F11? Just not sure my system is `reading` my flashed bios (although in the boot screen it shows F11).

F11 M8

[Matshui]

Sorry, a bit off topic....
As I recently got my 3000+ and AN50R, I think it is time for some overclocking of this CPU...
It is a CAAOC XPMW (don't remember the 4 digits before XPMW, but can tell you tomorrow)...

If I then try and OC, doesn't matter if it is 50Mhz or 200Mhz, SuperPi gives me an "NOT_EXACT_IN_ROUND" fail..

I am currious to know what to do about it.....!


Ohh yeah, VERY usefull information you found the Macci..!