E0 prescotts to have unlocked multipliers

[Mr. Tinker]

I'm sure that those of you who are really up to speed are already aware, but I think this is thread-worthy. Socket 775 E0 stepping x86-64 Prescotts will supposedly have unlocked downward multipliers.

What do you all think of this? I think it means that if you really wish for something, and are a good boy every day, dreams do come true.

linkage:
http://www.ocworkbench.com/ocwbcgi/newspro/viewnews.cgi?newsid1097218428,62560,

[jjcom]

welcom to XS! Man intel is finnally going to do that :D. mmm...x86-64 (AMD made) and now downward unlocked multi. All sounds very AMD. LOL

[Anemone]

Now if Intel would engage their brain and finally remove the oc locks from the 925X and 925XE chipsets, we'd finally be happy. Enthusiasts are the flag bearers of your products Intel...

But Intel would much rather just hand AMD marketshare :p

[|-jokker-|]

But all the way unlocked multipliers is sweet , it help´s a lot in oc i think .
It´s a good steep to begin regaining the consumers choice .
But for now i want a 3500+ :P

[jjcom]

yep....Intel doesn't like overclocking, but they can't stop us all. :D

[|-jokker-|]

No one can stop us from oc .
We will start ocing the optical mouses :D , and keyboards :D:D

[Mr. Tinker]

Now if Intel would engage their brain and finally remove the oc locks from the 925X and 925XE chipsets, we'd finally be happy. Enthusiasts are the flag bearers of your products Intel...

But Intel would much rather just hand AMD marketshare :p
I'm pretty sure that Asus found a way around the oc locks in the P5AD2.

Yeah, it does sound like Intel is just trying to build an A64 or an Opteron, but I think I'd still go with P4 because its high clockspeed makes it a good renderer.

[lutjens]

Posted this about 5 days ago...

http://xtremesystems.org/forums/showthread.php?t=43078

[jjcom]

yeah something seems different here. It used to be AMD would have trouble with getting money from their processors because they couldn't come up with something just as good or better. Now intel isn't having sells issues but their coping AMD. It would be intersting to see an Intel made Opteron....

[detonator]

is the unic way to stay in the market I think...

[lutjens]

If the multipliers of the E0 stepping are in fact going to come unlocked (it's looking that way), it's because the downward unlocking happens to coincide with Intel's new Speedstep.

We can't forget that they aren't giving us what we want because we want it...it's simply a fortunate happenstance. This new Speedstep required downward unlocked multipliers and they were unlocked for that reason only. They weren't unlocked to satisfy the enthusiast community, or curry favor among us (I think Intel feels that satisfying the small enthusiast community is not worth their effort or time). I also think Intel still holds us in the same contempt that they always have and see no reason why this would change.:rolleyes:

If Intel were trying to extend an olive branch to the enthusiast community, they would have unlocked the ultra-expensive, low volume EE processors. The processor's prohibitive cost would have ensured that remarking (Intel's chief excuse for the multplier lock) would have no chance of being an issue, as occurs with the Xeon. The move would have served to greatly enhance the enthusiast's opinion of Intel. But again, Intel feels it doesn't need to satisfy us and that we should happily take what we're given.

The E0 stepping doesn't change my opinion of Intel. I'll take the downward unlocked chips and overclock them, but until Intel actually gives us what we want and deserve (a completely unlocked EE), they continue to serve their own interests and not the enthusiast.

[Donnie27]

I also think Intel still holds us in the same contempt that they always have and see no reason why this would change.:rolleyes:

The E0 stepping doesn't change my opinion of Intel. I'll take the downward unlocked chips and overclock them, but until Intel actually gives us what we want and deserve (a completely unlocked EE), they continue to serve their own interests and not the enthusiast.

One of the best posts I've read in months. Yes, it is 100% Speed Step, one of our Intel Server Reps told us (at work). He said Intel would add this to help cool Prescott/Nacona or etc.. I don't know why many others didn't complain about the locks on the EE. The EE should have just been ES sold to us, heck, have write once ROM chip ID to keep away the Shady VAR's.

This is also Good and Bad news. What if the processor thinks it doesn't need full power to run something? I've seen PhotoShop run slower with Speed Step enabled.

Donnie

[jjcom]

yeah since speed step is controlled through software...there must be a way to disable it. I just thought of something...couldn't Intel screw UP overclocking with this?
14xmulti x 266 =3.73ghz
then the computer decides it wants more power and ups the muilti. I thought that is was controlled through software, but if its not then say your using a 3.6ghz chip....
18multi x 266 = 4.78ghz
It is controlled through software right?

[lutjens]

jjcom,

I looked in the Enhanced Speedstep specification sheet again and read that "the top frequency of the processor cannot be exceeded." Whether this means the processor monitors the actual clock setting and restricts automatic increases of the multiplier when this speed is exceeded remains to be seen. The processor may also simply assume a default FSB and increase to the default multiplier, and not exceeding the maximum default multiplier.

Also read this: "Voltage/Frequency selection will be software controlled by writing to processor MSR’s (Model Specific Registers) thus eliminating chipset dependency."

To me, this reads as the processor cannot act on it's own to change multplier and voltage settings. Software has to command it to. The easiest way to avoid the scernario you described would simply be not to use the software. "Software" in itself is vague...this could mean a program on the machine or programming in the BIOS. The latter is the only concern. I think most likely the BIOS will control it, with the ability to disable Speedstep in the BIOS itself.

Disabling Speedstep in this manner should have no impact on the processor's downward unlocked condition. You should still be able to specify a multiplier in the BIOS and have it remain.

Keep in mind, I've just interpreted the specifiation as I've read it. I could be very wrong on some points, so take this interpretation with a grain of salt.;)

[jjcom]

Man....that kind a sounds like the processor and the "software" have controll...Both things working together...This sounds like how on the 3ghz and up prescotts you can use the 14x multi. I would guess this to work with the BIOS manly since there are other OSes that run on a P4. So Intel would have alot of work if they wanted to make a Linux, Windows, Free BSD, etc. and they wouldn't release a source code...since some "hackers" could modify it.

[Anemone]

Yeah, Intel should have and still should unlock the EE's, as it would go a long way towards even moderately justifying their outrageous price.

The downward multiplier explains a lot as to why they stuck the 10% cpu speed lock on the 915/925 chipsets. They didn't want people taking then down a few grades then jolting the FSB through the roof, which is still what people are gonna want to do :p Still means they haven't fixed the Prescott running at full speed unless people are seeing loaded Pressies getting better temps on the new steppings...

Open the multi's on EE's - unlock the 925 at least - and let enthusiasts wave the flags again. But that would assume Intel had smarts...

[jjcom]

yeah I still wonder who came up with a multi. lock on the chipsets.....he must not like us. Yeah unlocked multi...I just don't see that happening. Intel just...wants to annoy us or something. ATi and VIA both make Intel chips sets so we could just go to them and leave Intel and their overclocking limited boards behind.

[lutjens]

VIA does make a chipset for the P4, the PT880. It doesn't perform as well as an Intel chipset though, if I remember correctly.

As for the Intel chipset lock, it could also be that the introduction of the lock on the 925/915 series of chipsets was only part of a bigger plan. Intel knew it would be introducing the E0 stepping with it's downward unlocked multipliers. I'm guessing that it was a two pronged move by Intel to introduce their new Speedstep and deal overclocking a blow at the same time.

I'm also willing to bet that the 925 chipset was introduced a little early on purpose...that way Intel could watch and see what the Taiwanese board makers did with it...and if they could get around the FSB lock. Notice how the E0 stepping chips were seemingly held back...there have been E0 samples for a long time now. My guess is that Intel is working on a way to superlock the chipset in a new stepping and is holding back the supplies of E0 chips for a while until they get the lock right (and they will get the lock right...Intel has some of the most proficient engineers on the planet working for them).

This is why I have chosen not to go down the LGA775 path...and instead stuck with the i875 chipset and the very capable Hance Rapids 6300ESB Southbridge. It's also why I'm looking forward to getting a P4 3.33 GHz 548 mobile chip with it's 133 FSB (25x downward unlocked multiplier), cheap price ($252 USD), high yield and Socket 478 package. I also know that there will be future P4 chips for the Socket 478 platform, including a mobile 3.6 GHz variant to be released in the later portion of Q1 2005 (model 558) which will represent Intel's top yield at least halfway into 2005. The only innovation I will be missing out on is the EMT64 technology, which Intel has decided is not required for the mobile crowd (even though it costs them absolutely nothing to add it:rolleyes:). Which is no great loss IMHO. When 64 bit processing hits mainstream and it's proven valuable to performance, I'll go with the company that does it best...AMD. They should have the dual cores out in full force by them.:)

I can only hope at that time that there will be a comparable chipset to the 6300ESB that's overclockable as well. Here's hoping...:)