The Dual 1 GHz Pentium III Myth

[ArticleRobot]

The 1 GHz Intel Pentium III seems to be the subject of much controversy, as many claims have been made about its inability to run in a dual CPU configuration. HardwareCentral has been following the discussion closely and decided to put an end to all the rumors and get a couple of GigaHertz Pentium IIIs and a dual CPU motherboard and find out what exactly is the truth of the matter.

http://www.hardwarecentral.com/hardwarecentral/reviews/1673/

 

[rowhome]

You can also use pin N33 ( M33 and N33 are VccCmos voltage pins) for enabling SMP coppermines on the MSI Master Slocket which is a mod that has been on the web for a few weeks now.(connect a wire from N33 to pin 3 of jumper 3 on the MSI Master slocket).The mod came from a message board in Japan. I did dual FCPGA coppermine 500s @666 on a P2B-DS a few weeks ago with this mod using the old stepping, cA2.Also, the dual enabling pin is not M33 or N33 but X2(BR1). The reason I think the mods work is that the slockets have the coppermine circuits and they are enabled by jumpers, just no voltage to X2. The mods provide the necessary voltage so B75 on the cardedge is detected by the chipset.



[This message has been edited by rowhome (edited 04-03-2000).]

 

[prefab0]

This is the best article I have read so far on the subject. Congratulations HardwareCentral. How about testing the celemines now?

rowhome,

I agree with you, according to the specs X2 has the BR1# signal. I wasn't aware that you could provide Vcc voltage to B75 and fool the chipset (by the way, how do you figure N33 to be Vcc, the doc says reserved. Also pin M33 seems inexistant). That is very clever. Do you think this can also work on the celemines (maybe disabling AN15 also)? please post your comments.

Cheers


[This message has been edited by prefab0 (edited 04-03-2000).]

 

[Sagan]

I can't believe you think this is the best article there is on this subject. If it was called "Dual .866GHz FC-PGA Engineering Samples Myth" maybe there could be SOME truth to it. First, the 1GHz CPU's that are out now, and mentioned as being not SMP capable, are Slot-1 NOT FC-PGA, making 80% of the article meaningless to the discussion. Second, using engineering samples of ANY Intel CPU and claiming to dispell myths about shipping products is completely asinine. Last, and somewhat minor, the 820 board they tested from Iwill DOES NOT have Dual Rambus Channels, it's an 820 Chipset, not 840!

 

[XkALiBRe]

rowhome::

If you get a chance, would you mind dropping me an email? I'd like to talk with you breifly about the mods you made. Drop me a line at [email protected], if you get a minute. Thanks.

------------------
"In the computer industry, there are three kinds of lies;
lies, damn lies, and benchmarks."

 

[prefab0]

Hi Sagan,

I think this is a very good article because:
1) This is the first test of PIIIs with cB0 stepping and shows that they can run dual. Several other tests exist with cA2 stepping but were not conclusive or did not work.
2) The rewiring work on the sloket is very interesting. It raises some pin assignement questions. I am interested in running dual celemines, so every bit of info is helpfull.
3) The board used supports dual rimms (funny that the photo shows 3 memory slots) you can check it at:
http://www.iwillusa.com/products/Spec.asp?ModelName=DS133-R&SupportID=

Cheers.

 

[redpriest]

I'm going to have to agree with Sagan on this one. Just because they're the same stepping doesn't mean the real PIII 1 GHz supports dual operation!

Intel has admitted that dual operation in stepping cA2 is broken. cB0 is quite a different story -- it's been fixed, but that doesn't mean that the "real" 1 GHz will work.

Once I've seen actual 1 GHz Intel PIII processors working in dual configuration, I'll believe it.

 

[prefab0]

sagan, redpriest,

I see your point. I think the article just shows that it can be done and there is no reason for those 1Ghz not be dual capable, except for Intel design.

 

[Sagan]

Actually nobody is arguing that C0B steppings aren't Dual. And AGAIN the 1GHz out there now are ONLY SECC2, there are NO FC-PGA 1GHz. When the finnaly release "Real" 1GHz, then they will be just like all the rest of the family.

The Slotket reworking was good information, but it has NOTHING to do with title of the article. Also by the time anyone can really get there hands on the cpus (not egineering samples) then new slotkets will around too.

The board has 2 RIMM slots, but NOT dual Channel RDRAM, which is what they said. Dual Channel is a feature of the 840 Chipset, not the 820.

And your about the last statement, Intel most likely just "turned off" the SMP pin, because they didn't have the time to go through all the validation necessary. And, as mentioned, the Intel Approved cooling systems dont' allow much room on the board for 2 CPU's. When they really ship 1GHz to the general public, then we can end this

hehe.

Also, it looks like they pulled the article. Guess I was not the only one complaining. Hmm..

 

[BrianH]

"Also, it looks like they pulled the article. Guess I was not the only one complaining. Hmm.."

No, there was a major overwrite of the whole MDT site by a March 22 backup. No cause or resolution available at this time. Stay tuned.

`````````
Current backup located and in place. All ficksed.

[This message has been edited by BrianH (edited 04-04-2000).]

 

[WTammen]

Please run same test with the new:
Tyan Tiger 133 it is a dual processor Apollo chipset board. Thank you

 

[Simon Dean]

Could someone please tell me (by e-mail please) the name of the heatsinks and fans used in the article?

regards,

Simon Dean

 

[rowhome]

Hmmm, seems the article has changed a little since it was taken down. Oh well, we all make little mistakes once in a while.
To the authors: It would have been much more edifying if you had explained why you did the modifications you did (What the circuits did that were cut and what the new circuits do).
The latest thinking on the web is that even the present stepping of the FCPGA coppermine is SMP capable, there just haven't been any SMP enabled slockets ( IWill says they will have one soon) or dual FCPGA 370 motherboards to test them on. This is basicly what the mod does.
I have to agree this article is a bit of a stretch as far as really testing dual 1Gz CPU's but it may still be valid assuming the real 1Gz chips are exactly the same.

 

[H1ghlander]

I can attest that the new Abit Slotket !!! (3) work with my 600E stepping cA2 in an Epox KP6-BS board. Not rewiring or jumper settings, it just works. I'm waiting for a Tyan Tiger 133 so I can test high bus speed
as my Adaptec U2W doesn't like anything over 103Mhz.

 

[rol]

Well i would love to see them try this mod on a set of dual celmines and see if they acctually work. My opinion on the cAo stepping with smp disabled is that maybe thoughs were the test bed for disableing the celmines. Possibly they werent meant to be relieased intel has pulled this kind of bonner before. Remember the mobile pentium2's with the pentium3 id number in them that got put into pentium2 notebooks by mistake and no one found out about them till the whole stink went up about the pentium3 chip id and intel ended up doing a dance over that one. Finally releasing a proggy to disable the id at bootup? This could have been the same or similar situation. I wouldnt be really surprised if this turned out to be the case and the celmines did do smp with this mod.

 

[rol]

Well i would love to see them try this mod on a set of dual celmines and see if they acctually work. My opinion on the cAo stepping with smp disabled is that maybe thoughs were the test bed for disableing the celmines. Possibly they werent meant to be relieased intel has pulled this kind of bonner before. Remember the mobile pentium2's with the pentium3 id number in them that got put into pentium2 notebooks by mistake and no one found out about them till the whole stink went up about the pentium3 chip id and intel ended up doing a dance over that one. Finally releasing a proggy to disable the id at bootup? This could have been the same or similar situation. I wouldnt be really surprised if this turned out to be the case and the celmines did do smp with this mod.

 

[PeterLunde]

To: rowhome and others

Perhaps you could check out my Abit BP6 Dual Pentium III SMP modification at
www.interlog.com/~telsci/bp6-p3.html

and let me know if you think it will work.

Thanks in advance.

 

[Technofix]

Originally posted by PeterLunde:
To: rowhome and others

>Perhaps you could check out my Abit BP6 >Dual Pentium III SMP modification at

And the IWILL DCA200 is cheese dip ?
It appears to be working so far.
Running 'very' fast.
(SYSMARK2000 couldn't get an accurate clock)
Whatever.

 

[PeterLunde]

If you want to talk about Xeon boards, why not the Supermicro S2QR6?

Originally posted by Technofix:
And the IWILL DCA200 is cheese dip ?
It appears to be working so far.
Running 'very' fast.

 

[satow]

I haven't seen anywhere else on the web regarding the modification of the MS-6905 Master CPU converter card to run dual Pentium III coppermine FC-PGA cpus on dual cpu motherboards. Where else did you see this information posted? It's difficult to believe that your modification would work. I would like to see more proof.

Originally posted by rowhome:
You can also use pin N33 ( M33 and N33 are VccCmos voltage pins) for enabling SMP coppermines on the MSI Master Slocket which is a mod that has been on the web for a few weeks now.(connect a wire from N33 to pin 3 of jumper 3 on the MSI Master slocket).The mod came from a message board in Japan. I did dual FCPGA coppermine 500s @666 on a P2B-DS a few weeks ago with this mod using the old stepping, cA2.Also, the dual enabling pin is not M33 or N33 but X2(BR1). The reason I think the mods work is that the slockets have the coppermine circuits and they are enabled by jumpers, just no voltage to X2. The mods provide the necessary voltage so B75 on the cardedge is detected by the chipset.

[This message has been edited by rowhome (edited 04-03-2000).]

[This message has been edited by satow (edited 04-19-2000).]