A64 101

[hookedm]

The board does not come with overclocking features like FSB, multiplier, voltage adjustments at all. It is really a pity as we are unable to determine if there is a FIX PCI/PCI-E within the chipset that most overclockers are looking for. The only adjustments we can find are options for ASync 166,133,100 operation. HTT width can be set to operate in 8 or 16 bits. HTT freq can be set to 1GHz, 800MHz, 600Mhz, 400 MHz.

http://www.ocworkbench.com/2004/msi/ms-7093/b7.htm

What can I do with my AMD FX-55 and this POS mobo?


M

 

[Gautam]

Try this version of clockgen. You probably won't have a locked PCI bus, and the HTT probably won't go further than 220, but at least its something.

To be perfectly frank, it would be well worth your money to switch boards, seeing as you've got such an expensive processor as it is. If money is an issue, you'd be better off selling the combo and opting for a San Diego 3700+ or 4000+ paired with a DFI nF4 mobo.

 

[Rattle]

does clockgen work on my board?

 

[Gautam]

It does but I wouldn't see any reason not to OC in BIOS...

 

[Rattle]

yes i've had success with just bios, I read on another forum that clockgen didnt work on this board.

 

[Porkinstine]

Great guide keep on postin great info

 

[Audioaficionado]

Time to add in X2 processors too now that they're out.

Great sticky

 

[Gautam]

Thank you again. Added the Toledo, Manchester, and Newark cores. Also made some other corrections and updates.

 

[Ryn]

Hey guys, I have a question here on cpu voltage. I'm running an A64 3200+ Newcastle on a DFI NF3 UT250GB. I'm at 2.4Ghz right now and my voltage is 1.7V. Do you think i can still push the vcore further? I'm using watercooling btw and full load while F@H is between 40-42C. Thanks in advance!

 

[Gautam]

I wouldn't get nervous until close to 1.9v, but you will probably find 1.75-1.85 to be the sweetspot.

 

[Ryn]

I wouldn't get nervous until close to 1.9v, but you will probably find 1.75-1.85 to be the sweetspot.

Thanks for the reply dude! Anyway, the reason I asked is that most people out there said that 1.75V should be the max for a Newcastle core wether air or watercooling. Anyway, thanks again.

 

[hitechjb1]

The A64 NewCastle core is based on 130 nm SOI process.

From AMD technical documents, it specifies that 1.65 V is the absolute max rating for the 130 nm CPU VDD voltage. It says

"Stresses greater than those listed in Table 14 may cause permanent damage to the device and motherboard. Systems using this device must be designed to ensure that these parameters are not violated. Violation of these ratings will void the product warranty. Exposure to absolute maximum rating conditions for extended periods may affect device reliability."

For details, refer to this post
Note on Absolute Max Voltage rating
and
AMD A64 data sheet 31410, 31411, 31412.

 

[Gautam]

Wouldn't the same disclaimer apply to running anything (including clock speeds) out of spec?

 

[hitechjb1]

Voltage, frequency and temperature are the three basic variables in running/overclocking a chip.

Voltage is the cause, frequency and temperature are the effect. High (sufficiently) voltage alone, even without clocking, can damage a chip because of leakage current, dielectric breakdown, etc. But frequency alone under low enough voltage (say stock) would not damage a chip, except the system may not be able to run correctly.

So voltage has to be applied carefully to avoid causing damage to a chip. Running voltage out of absolute max spec may damage a chip, but frequency would not. Higher frequency would dissipate more active power, hence higher temperature, but with sufficient low voltage, damage would not occur except system cannot run correctly.

AMD specifies the max absolute voltage for 130 nm AXP as 2.05 - 2.20 V, depending on the exact CPU type.
For the 130 nm A64, the max absolute voltage is specified as 1.65 V (from my reading of the tech documentations).
Max Vcore for Tbred B and Barton (page 5)
How much voltage can be applied to a CPU (page 5)
Note on Absolute Max Voltage rating

Further, from what I have seen on the net, and specifically from this thread,
A64 Overclocking Result Collection,
for 130 nm NewCastle
754 ClawHammer, NewCastle,
the majority of the results are 1.70 V or below.


Voltage, temperature and frequency: the basic variables of overclcoking

CPU voltage: from stock to max absolute, from efficient overclocking to diminishing return (page 19)

 

[Gautam]

Heh, I didn't mean to be taken literally, but that's a good point.

While I agree completely with everything you've stated, I tend to think that AMD was just being a little bit more (OK, a lot more) conservative than before. Consider that in the past they listed the absolute voltage as being about 35% higher than that of stock in the case of the AthlonXP's, and have now dropped to recommending a measly 10% increase... It also looks a little cryptic. I can't even pretend to understand a fraction of this stuff, but it looks like what they used to refer to as VCC_CORE is now VDD? And if so, why do they no longer provide maximum and minimum values due to perturbation? And the lower bound being -.3v raises an eyebrow...I know it's the only data we have to go on, but this large discrepancy makes me wary of comparing the AXP data and the A64 data, and I honestly can't say I can interpret either for sure. Seems like they had a philosophy shift of sorts. It's a tangent, but its relevant. AFAIK, they are now rating absolute case temperatures in the low 50C range...a target that many OEM's would be quite hard pressed to hit. So, in short, I'm questioning the validity of the absolute rating.

Traditionally, it was rather common to overvolt by around 25% as a rule of thumb, and I've tended to stick to that. Will running at 1.85v damage the processor? Yes. Will running at 1.65v do so as well? Absolutely. The point is that, IMHO, neither would damage the processor quickly enough to become an issue. You yourself stated that a 30% overvolt would result in a 10% decrease in failure time. With the typical life expectancy of a processor being around 10 years, I don't see such a decrease as being too significant, especially with most people switching every two years, if not yearly, or even less. That's my rationale for asserting that relatively significant overvolts will not have a noticeable effect in the period in which the processor is actually used.

PS Thanks again for documenting your sources so clearly, would've had one heck of a time sifting through all the docs for that specific data.

 

[RotKT]

thanks

 

[harryinny3]

Any word On How good the A64 socket 754 3000+ venice cores Are? Yes they are out and You Can find them at Mwave and Newegg.

Harry

 

[Gautam]

Yeah, they look interesting. Guess someone will need to find out the hard way. If they don't hit 2.8 easy, I would still say that mobile DTR claws are the best choice for 754.

 

[harryinny3]

WOW i thought this thread was dead LOL. Tah i got one of those months ago. It went to 2.5 on air easy, but i was and am still limited by my memory. I have since bought a 3700+ San Dieago and have it at 2.7 on air and still limited by my cruddy memory.

Harry

 

[magx]

i've got a Asus A8v Deluxe Mobo and when i set my ram voltage to 2.8v my system locks up in windows no matter what i do. Any ideas? Also, i've got a amd 64 3500+ winchester @ 2.2 ghz right now...all i wanna know is HOW DO I GET THIS SUCKER AT 2.4 GHZ with my memory staying at 200 mhz (400 mhz) with the hypertransport bus at 1000 mhz? is there a way i can download a modded bios to allow me to adjust the hypertransport bus by 10 mhz each or something? Mine is set like:200 mhz 400 mhz 600 mhz 800 mhz then 1000 mhz...

I'm also confused about the memory dividers too...

here's the 3 main things in my system that i'm using to overclock:

OCZ 520W Powerstream
OCZ EL Platinum Rev 1 2-3-2-5 1t (2x 512 MB)
AMD 64 3500+ Winchester @ 2.2 GHZ with 0 stepping.