Can anyone tell me why anyone should bother with overclocking an Athlon T-bird using dip switches or in the BIOS? With the associated headaches, 'overheating, lockups, etc', when all you have to do is alter the L1 through L7 bridges, and have the real thing. Upto 1.2 ghz from what I've read.
I am new to this so maybe I'm missing something. I know that overclocking the FSB enhances performance but that can be done after altering the processor itself.
Your idea would probably work assuming that all chips are equal, but they're not.
I don't know that this is all accurate, but from I've read read over the years, this is how they decide which processor is which. This applies to Intel and I assume also to AMD.
They create a large "batch" of CPUs and then take a representative sample and test them (I don't know how or how many). During their testing they determine that all of their samples will run *within specifications* at, say, 800 MHz. So all of these are put out as 800 MHz. processors. Maybe the next batch will test higher and be 900 MHz., etc. You can see the possibilities...maybe some of the chips will actually run higher.
Then you take marketing/sales into account. *Perhaps* they need more 800 MHz. chips, but their yields are really good and they take a batch that tests at 900, and set them up and package them as 800's. For all I know, they may determine that they need 800 MHz. chips and not bother to test them any higher. Obviously they make more money selling faster chips, and it probably costs the same to make each chip.
By the time you get the cpu, you're not quite sure what you have, except that the manufacturer guarantees that it will run at a certain minimum speed, within its specifications. It's possible that you could just do what you suggested, close the bridges on the AMD chip and slap it in without changing anything else. (I haven't really followed that type of change, so I <i>assumme</i> that it would be enough to tell the BIOS the new multiplier, but I don't know).
But whether you change the multiplier via the jumpers or the BIOS, you are now running the physical cpu outside of its tested environment. Maybe it will work and maybe it won't. Maybe it will work if you up the voltage...further placing it outside of the tested environment. And as you pointed out, higher voltage means a hotter cpu.
Finally, as you get closer to the physical limits of the cpu making process, your chances of overclocking are slimmer.
I'd much rather make the changes in the BIOS. :^)
I'm certainly open to any corrections or other theories.
BTW, I believe they use a similar "culling" process when they make RAM chips.
December 31, 2000 7:24:30 PM
Thanks Mike, what you said makes a lot of sense. So does that mean if I have to boost the voltage past 1.85 to overclock an 800mhz T-bird to 1ghz, then I would also have to boost the voltage of that same processor if I modified its identification with the L1 - L7 bridges? If that were the case it sure would make more sense just to modify the multiplier through the dips or the BIOS.
An AMD rep came to a club, I go to, last Nov., he said that the Athlons that came out in 1999 were all capable of 1ghz but for market purposes (as you stated) they were clocked slower. So faster doesn't always make the most market sense. You went into it in more detail than he did and seem to make more sense.
<b>So does that mean if I have to boost the voltage past 1.85 to overclock an 800mhz T-bird to 1ghz, then I would also have to boost the voltage of that same processor if I modified its identification with the L1 - L7 bridges?</b>
I believe it does, but I don't know for sure. It sounds logical as we're dealing with a physical limitation of the cpu.
<b>An AMD rep came to a club, I go to, last Nov., he said that the Athlons that came out in 1999 were all capable of 1ghz but for market purposes (as you stated) they were clocked slower.</b>
I don't know as I'd believe that. I didn't pay that much attention to AMD specifics until I decided to get an Athlon a few months ago. (I didn't like any of the previous motherboards.) Regardless, I *think* the Thunderbirds came out around Jan. 2000, didn't they? Which would mean that he was talking about the original Athlon. I would think that many people using gold finger devices or whatever to overclock their Athlons would have discovered that pretty quickly, regardless of the cache speed limitations. That's a guess as I don't know that much about them and have never seen one.
Woah... your all messed up man. To overclock any processor, the bios has to know about it. Using dip switches and softmenus's allows you to overclock. First, you connect the L7 bridges to overclock the CPU then you make changes in the BIOS. You can't do one without the other man. They go together.
December 31, 2000 9:42:52 PM
I think L7 controls the voltage. I just glanced at Tom's article on changing the voltage and multipliers. It looks like you change the multiplier using a combination of L3, L4 and L6. (I'm not talking about just unlocking the processor.) Are you saying that after making the appropriate changes to the voltage and multiplier settings on the cpu, any BIOS won't recognize it, even those of say the A7V or the KT7? (I don't know how the cpu tells the BIOS what it is and what controls this process.)
Well, I meant a combination of all the bridges- not the L7. What I am saying is Athlon's are clock locked- sort of. Yes, it's easy to unlock them- but if you try to overclock using the BIOS with out unlocking the CPU- it won't work. So, what I am saying is you have to screw with both the board and the CPU. The guy who originally posted the question was saying you don't have to do crap to the BIOS to overclock a T-bird when in all actuallity you have to. I'm sure the t-birds send a signal or register to the BIOS saying "i'm 800mhz", but when you unlock a T-bird at 800mhz and overclock it to 1ghz- the BIOS has to know. That's why you have to make changes in the BIOS so it'll know it's overclocked or clocked at 1ghz. that sounds really confusing. i hope you understand it =)
December 31, 2000 10:50:51 PM
Except your talking about overclocking and I'm not. I'm talking about changing the identification of the processor so that it actually is a 1.2 ghz processor not an overclocked slower processor.
Our whole point is that you <i>can't</i> change a processor and make it something it's not.
In fact, overclocking is kind of like that... If you have a 1GHz, change the multiplier and voltage to a 1.2GHz (12x and 1.85 I think), and you have a 1.2GHz Thunderbird. Runs the exact same as a non overclocked 1.2Ghz thunderbird, except for heat of course. (And maybe stability if you got a Thunderbird from a bad batch).
December 31, 2000 11:28:19 PM
Thanks. Yeah, I realize the BIOS needs to know. I guess I'm looking for a more in depth description of things like what transpires between the cpu and the northbridge/BIOS to actually set the clock speed, how the cpu bridges control the signals through the cpu pins, etc. I went back to Tom's article and see that he references an anandtech.com article that seems to discuss this in more depth. Guess I'll shoot over there one of these days...when I get the urge.
The bridges are just switches. If they were all closed then all multiplier and voltage options would be available. If some the bridges are open then the CPU is locked at a specific multiplier and/or voltage and can't be changed from the BIOS or on the motherboard. (This is not quite true because lower mulitpliers can still be selected but not higher ones)
The reason the bridges are locked for a specific multiplier and voltage is those are the conditions at which a Thunderbird has been tested and certified to run. If it is less than the maximum 1.2 Ghz then it was either not tested at a higher speed or it was tested at a higher speed and did not perform within specification at that speed. That is, it did not function at that speed, it was unstable at that speed, it ran too hot, or some other reason kept it from running within specification. Maybe it didn't run at the nominal voltage. Whatever the reason it was certified to run only at the slower speed, the speed at which it is certified.
By closing the bridges you make it possible to select higher multipliers and voltages but it does not mean the CPU can do it. Some can't. You can up the voltages and increase CPU cooling in an attempt to make the CPU stable and maybe you will succede but now you have it running out of specification. Another CPU that IS certified to run at that same speed doesn't need the voltage adjustment and doesn't need the additional cooling to do the same job.