Overclock.net banner

1 - 12 of 12 Posts

·
Registered
Joined
·
742 Posts
Discussion Starter #1
Ok....so I have been playing around with my x6800...different fsb / and multi combos with vcore variance...Something seems odd to me:<br />
<br />
If I keep the multi down to stock (11) and increase the FSB a ton (like up to 1200mHz, ram at 900Mhz), I can run at 3.3 gHz on lowest vcore with stability...if I want o keep the fsb down to stock, yet change the multi to 12, I need to increase vcore...<br />
<br />
Does it make sense that if I only use FSB to increase speed, that my voltages can stay low? If so, why the unlocked multi? I guess it would be different if temps werent an issue (running water).
 

·
Registered
Joined
·
1,831 Posts
so what your saying is that in order to overclock using the cpu multipliers that you need to overvolt, but when you overclock to the same speed with the fsb you dont need to overvolt.<br />
yea that doesnt make sense, you're right.... it shouldn't work that way.<br />
<br />
btw overclocking the fsb to 1200mhz isnt that much lol im running my fsb above 1500mhz... of course i need better northbridge cooling before i can overvolt that for a higher fsb clock.. but w/e
 

·
Registered
Joined
·
742 Posts
Discussion Starter #3
thanks..I thought I was losing my mind,
 

·
Registered
Joined
·
742 Posts
Discussion Starter #4
arent I better off with a lower multi and a higher FSB and ram speed anyway?
 

·
Registered
Joined
·
1,831 Posts
not in my opinion. some people like ram speed, but the only real issue with ram speed comes with quad core... i guess the 4 cores could process faster than it could access the information so only a certain percentage of each core could be used. i mean come on, when ram speed bottlenecks your cpu speed you know you've got a sweet processor.<br />
<br />
supposedly this is what AMD is planning, since intel's FSB is going to bottleneck very soon...<br />
AMD's hypertransport will perform about as much work as intel's fsb at 2ghz... of course, this means that ram speed or fsb won't bottleneck if AMD makes a processor with four or more cores.
 

·
Premium Member
Joined
·
19,078 Posts
Are you talking about setting the multipler to "Auto"?
 

·
Registered
Joined
·
112 Posts
holy god!! Hiba: you got a killing machine! 4x 150GB Raptor, 7900GTX and E6800! so rich! throw us some money lol.
 

·
Administrator
Joined
·
6,696 Posts
<div style="margin:20px; margin-top:5px; ">
<div class="smallfont" style="margin-bottom:2px">Quote:</div>
<table cellpadding="6" cellspacing="0" border="0" width="99%">
<tr>
<td class="alt2" style="border:1px inset">

If I keep the multi down to stock (11) and increase the FSB a ton (like up to 1200mHz, ram at 900Mhz), I can run at 3.3 gHz on lowest vcore with stability...if I want o keep the fsb down to stock, yet change the multi to 12, I need to increase vcore...<br />
<br />
Does it make sense that if I only use FSB to increase speed, that my voltages can stay low? If so, why the unlocked multi? I guess it would be different if temps werent an issue (running water).

</td>
</tr>
</table>
</div>It makes perfect sense, and that's how it is designed to work <img src="/images/smilies/smile.gif" border="0" alt="" title="Smile" class="inlineimg" /><br />
<br />
When changing the multiplier you require additional hardware components to come online in the core itself to increase the clock speed via multiplication. <br />
When you increase the multiplier a single additional multiplier mini chip must come online to increase the clock speed. <br />
<br />
The PLL will assign a new designated clock speed when it detects this chip coming online.<br />
<br />
More voltage will therefore (usually) have to be applied to keep this additional mini component working and to power the chip itself, especially as the clock speed has been dramatically increased due to this chip.<br />
With additional components coming online more current has to pass through the chip to keep all components online. V = IR. Resistance will remain near constant (temperature effects this value) so therefore following this law voltage must increase to co-inside with current.<br />
<br />
Increasing the Front-Side Bus speed will not require a significant amount (if any) of additional voltage. Using this method no additional components are required to come online therefore current will remain the same from that prospective. Only the core clocks and bus itself will increase. Therefore voltage will only need to increase when they reach their limits.<br />
As the bus controller is not inside the core itself the "Bus" will have less of a voltage requirement than the clock frequency and multiplier system.<br />
<br />
This is why you do not require as much voltage to be applied if you do not change the multiplier <img src="/images/smilies/smile.gif" border="0" alt="" title="Smile" class="inlineimg" />
 

·
Premium Member
Joined
·
3,697 Posts
And again TheManual lives upto his name <img src="/images/smilies/wink.gif" border="0" alt="" title="Wink" class="inlineimg" />
 

·
Registered
Joined
·
377 Posts
ya he went all out.. that cooler he has is $150 lol.. how does that work btw? whats the difference between that cooler and maybe a big tt with a fm121 on it?
 

·
Registered
Joined
·
742 Posts
Discussion Starter #12
hey.,..thanks for the compliments, and thanks to the manual for the dissertation on chip operations! The cooler works great man...I have had stability @ around 3.6 gHz with vcore around 1.375 and pulling around 55C full load double prime smalls.
 
1 - 12 of 12 Posts
Top