Old cpus questions.

[Truekeeper]

Well, thanks for all the replies.

I'm going to try both of them and i'll tell what happened.

Wish me luck.

 

[Crashman]

Look up the pre-release info on Tom's Hardware Guide (good luck), there's a signalling difference or something that Intel said would be problematic. They changed two pin definitions for the express purpose of preventing Tualatin operation on pre-Tualatin motherboards, and the adapters jump a couple pins to falsify the "ok signal".

It wasn't a voltage issue because VRM 8.4 (Coppermine standard) went down to 1.30V. Tualatin specification did provide for a new VRM, but that was mostly done for support of "low voltage" Tualations (such as mobile processors). Desktop Tualatins ran at ~1.425V, the 25mv change wasn't supported in VRM 8.4 but that's not enough of a difference to call for a VRM change for the standard desktop processor versions.

 

[cwj717]

They probably require a new VRM... the Powerleap adapters have a VRM on the adapter. EDIT: I can't seem to find that page on toms...

 

[Crashman]

You're thinking of Powerleap Slot-1 adapters? That's to adapter VRM 8.2. VRM 8.2 was for early (pre-Coppermine) Pentium II's, and only went as low as 1.80V. Furthermore, Powerleap's Tualatin adapter used the same VRM 8.4 spec as its previous Coppermine adapter.

Intel even released a NEW version of the 815 chipset to support the signal change, so like I said, look up the pre-release info.

 

[cwj717]

I though it was still needed... my slot 1 boards all support 1.0GHz P3s

 

[Crashman]

No, read this:

Q. I have a Slot 1 board and have heard about using a Powerleap upgrade or Upgradeware adapter to make the newer Tualatin CPU's work, what would you recommend?

A. It depends on whether your motherboard uses the original VRM 8.2 voltage regulator or the later VRM 8.4 version. VRM 8.4 was released for Coppermine core CPU support, and is present on most Slot-1 boards that support 100MHz bus or higher. Early 100MHz bus Slot 1 boards and most 66MHz bus Slot 1 boards use the earlier VRM 8.2 standard.
While VRM 8.4 allows voltages down to 1.3v, VRM 8.2 supports voltages only as low as 1.80v.

The Upgradeware adapter is cheaper, simpler, and more reliable than the Powerleap IP3/T, because it uses the motherboard’s VRM. Since VRM 8.2 boards can’t detect a CPU at anything lower than 1.80v, the Upgradeware adapter should only be used on VRM 8.4 boards.

Conversely, the Powerleap IP3/T adapter is compatible with both VRM 8.2 and 8.4 motherboards, but it’s a waste of money and a sacrifice in reliability to put this adapter on one that already meets VRM 8.4 specification.

If your board was designed or revised for use with Coppermine CPU's, it should work with a Tualatin core processor on the Upgradeware adapter at reasonable voltages (1.50v recommended). Otherwise, you'll need a Powerleap adapter, as these have their own built-in VRM, which allows the board to detect the adapter itself at 2.05v while the CPU operates at the lower voltage provided by the Powerleap VRM.

Source

Damn, now there's an expert who sounds like he knows enough about this subject that he doesn't need to make assumptions, maybe he was actually reading this info when it was current! Hmmm....

 

[cwj717]

I didn't say I was an expert... I just had not heard that before. I figured there was a reason Intel changed the pinout. I have a few S370 boards that just would (EDIT) NOT take the Taulatin CPUs, with an adapter. My Slot-1 boards won't take ANY slocket adapter... 🙁 EDIT: I had no reason to Believe what you had said... you didn't have any sources for that info, and I couldn't find it.

 

[Crashman]

I though it was still needed... my slot 1 boards all support 1.0GHz P3s

My Slot 1 boards supported up to P4 1.4GHz (133x10.5) Tualatins using the VRM-less Upgradeware adapter. I bet they were the same boards as yours:

Abit BE6-II
Abit BH6 Rev. 2.0
Asus P2B-F Rev. 1.x(4?)
Asus P3B-F
Asus P2B-B (Tualatin on AT, w00t!)

 

[Crashman]

It's darned near impossible to look up information that old, it often takes hours, which is why I told you to do it.

 

[cwj717]

The boards I still have are my Supermicro P3DRE and an IBM Intellistation... I don't have my other boards anymore. I think one was an ASUS, its been awhile though. The Supermicro board and the IBM system both take my Slot-1 933MHz P3s just fine... but they don't even post with any slocket I have tried. EDIT: I found my Intel 840OR board... it would not take the slockets either.

 

[cwj717]

Well, thanks

 

[I]

The Celeron might not work, most Socket 370 boards don't support Tualatin cores.

I'd try it, and if it doesn't work, put in the other one.

You can get a Tualatin adaptor off eBay for a few bucks though, and they work good. The good thing about the Celeron is it won't matter if he only has PC100 RAM.

They often blow out the motherboard capacitors unless you're talking about the type with onboard regulation that takes a PSU plug for power.

 

[I]

What a great info you guys just provided me.

I'll go celeron then.

Thanks very much, all of you. 8)

Another thing:
The guy also want to purchase a usb 2.0 tv receiver for portable reasons but his mobo only has usb 1, so if i put a usb 2.0 pci card do you think the
usb 2.0 tv will perform ok, i mean i heard usb 2.0 tv receivers are cpu hungry.

What you think ?

USB TV tuners are not CPU hungry, they have to have MPEG conversion hardware in them (else extremely low resolution or framerate) because USB2 bandwidth isn't high enough to transfer raw video. Thus, they're only slightly worse than MPEG2 hardware encoding type PCI cards, providing your add-on USB2 PCI card has a decent chipset like NEC, not crap like the earlier Via chipsets.

 

[I]

Hell yeah they were. I remember that they could overclock like crazy and STILL suck. They were real treasures, those Celerons.
-cm

An o'c Tualatin Celeron outperformed most (earlier models of) P4. Sometimes there were other platform specific issues like if it was put into an adapter onto a Via 693 chipset with it's poor memory bandwidth, an 810 with integrated video, or a Sis (anything really, Sis sucked badly in this era), or 440BX due to limitations of low AGP speed and ATA33.

Perhaps you are thinking of the P4 variant of a Celeron instead of the Tualatin, P3 variant. Completely different design.

 

[m25]

The P3 has a 133MHz bus... the Celeron is still on a 100MHz bus.

but the celeron is a good 40% faster and has the same cache size. Tualatins were real monsters; differently from almost all other celerons they performed almost just slightly worse than an equally clocked P3.

 

[I]

The adapters didn't work in some boards... same with the slocket adapters as well... but it wasn't just in Intels boards. I have a Supermicro board that won't take them and an old IBM system that doesn't either. I don't really think Intel did that on purpose... its just that some older boards were made before the Tualatin cores existed.

Most boards would run them with a hack or two. Notible were some Sis chipset based boards that could not use a simple pin adapter. Most other boards depended on whether the voltage regulator supported (a low enough) voltage default of the Tualatins, and if they didn't then you'd choose the lowest voltage they did support which was often 1.5V. If it was 1.8V (very old board?) that might be too high but in that case the board itself is going to have so many limitations the CPU probably won't matter much, everything else will cripple the system till so many add on cards were used that it made no sense to reuse the motherboard.

 

[cwj717]

The P3 has a 133MHz bus... the Celeron is still on a 100MHz bus.

but the celeron is a good 40% faster and has the same cache size. Tualatins were real monsters; differently from almost all other celerons they performed almost just slightly worse than an equally clocked P3. The P3 has a 33% faster system bus... look at the link to a toms comparison I posted in this thread.

 

[I]

No, Intel did do it on purpose because the company felt there would be stability issues with Tualatins on Coppermine boards.

Better would be to say they didn't want the support issues of a non-qualified configuration and they weren't going to advocate backwards tech upgrades even if it made sense economically to the system owner because it didn't to Intel.

Intel is extremely conservative, they always take the "it isn't supported" attitude when "isn't supported" only means "we don't advocate it no matter why", and doesn't necessarily mean "it won't work" nor even "we have any reason why it wouldn't".

It was rather pathetic of them though, to artificially halt their bios like that when the processor had to be working already to have decompressed and began to run it. Their artificial limitation could only be imposed on a system that WAS working in that case.

 

[cwj717]

The adapters didn't work in some boards... same with the slocket adapters as well... but it wasn't just in Intels boards. I have a Supermicro board that won't take them and an old IBM system that doesn't either. I don't really think Intel did that on purpose... its just that some older boards were made before the Tualatin cores existed.

Most boards would run them with a hack or two. Notible were some Sis chipset based boards that could not use a simple pin adapter. Most other boards depended on whether the voltage regulator supported (a low enough) voltage default of the Tualatins, and if they didn't then you'd choose the lowest voltage they did support which was often 1.5V. If it was 1.8V (very old board?) that might be too high but in that case the board itself is going to have so many limitations the CPU probably won't matter much, everything else will cripple the system till so many add on cards were used that it made no sense to reuse the motherboard. My slot-1 boards take P3 933MHz with 1.75v.

 

[I]

Look up the pre-release info on Tom's Hardware Guide (good luck), there's a signalling difference or something that Intel said would be problematic. They changed two pin definitions for the express purpose of preventing Tualatin operation on pre-Tualatin motherboards, and the adapters jump a couple pins to falsify the "ok signal".

Yes that is another shady thing they did unnecessarily.

It wasn't a voltage issue because VRM 8.4 (Coppermine standard) went down to 1.30V. Tualatin specification did provide for a new VRM, but that was mostly done for support of "low voltage" Tualations (such as mobile processors). Desktop Tualatins ran at ~1.425V, the 25mv change wasn't supported in VRM 8.4 but that's not enough of a difference to call for a VRM change for the standard desktop processor versions.

It was sometimes a voltage issue. Intel can issue any VRM standard they like, but it does't in any way obligate a board manufacturer to follow it. It also doesn't address someone using a slotket adapter on slot 1.

There was more than one default voltage for the Tualatin family as well. As core speed ramped so did default vCore.

Even so, I'd agree that it was often not the voltage preventing use, the pin changes alone would account for it not working, though we seem to be ignoring the socket 370 systems that did natively support Tualatins, it was rather silly for us to take this question without even having info on what board is being used.

 

[I]

My slot-1 boards take P3 933MHz with 1.75v.

The main thing wrong with so many of the slot 1 boards is they weren't engineered for enough current, this was the era when even Abit had picked bad caps and many others used rows of the tiny 8mm diameter caps. The best slot 1 board to run a Tualatin from a build and discrete electronics perspecitive was ironically the worst when it came to having a crippled bios that halted and no oc provisions (Intel Retail like a SE440BX).

There were a few oddball hybrids that ran Tualatins good though like a Tyan something-or-other with Via 694X chipset, supported a lot more memory and AGP 4X. That one in particular had a bios bug though where if it posted at the default Tualatin Celeron 100MHz FSB that locked the AGP divider and raising FSB to 133 in the bios menu kept the AGP out of spec. Solution was to manipulate the FSB Sel pins so it thought the CPU had default 133MHz FSB, or I think some Tyans had through-holes where FSB jumpers could be soldered on. In fact I'm certain of it as I recall soldering some onto a Trinity 400 (S1854)

 

[I]

No, read this:


Q. I have a Slot 1 board and have heard about using a Powerleap upgrade or Upgradeware adapter to make the newer Tualatin CPU's work, what would you recommend?

A. It depends on whether your motherboard uses the original VRM 8.2 voltage regulator or the later VRM 8.4 version. VRM 8.4 was released for Coppermine core CPU support, and is present on most Slot-1 boards that support 100MHz bus or higher. Early 100MHz bus Slot 1 boards and most 66MHz bus Slot 1 boards use the earlier VRM 8.2 standard.
While VRM 8.4 allows voltages down to 1.3v, VRM 8.2 supports voltages only as low as 1.80v.

The Upgradeware adapter is cheaper, simpler, and more reliable than the Powerleap IP3/T, because it uses the motherboard’s VRM. Since VRM 8.2 boards can’t detect a CPU at anything lower than 1.80v, the Upgradeware adapter should only be used on VRM 8.4 boards.

Conversely, the Powerleap IP3/T adapter is compatible with both VRM 8.2 and 8.4 motherboards, but it’s a waste of money and a sacrifice in reliability to put this adapter on one that already meets VRM 8.4 specification.

If your board was designed or revised for use with Coppermine CPU's, it should work with a Tualatin core processor on the Upgradeware adapter at reasonable voltages (1.50v recommended). Otherwise, you'll need a Powerleap adapter, as these have their own built-in VRM, which allows the board to detect the adapter itself at 2.05v while the CPU operates at the lower voltage provided by the Powerleap VRM.

Source

Damn, now there's an expert who sounds like he knows enough about this subject that he doesn't need to make assumptions, maybe he was actually reading this info when it was current! Hmmm....

We can somewhat ignore that because the VRM standard was only a suggestion, but more importantly it was only a minimum in that the chip a board designer used might've easily gone lower than 1.8V on a pre-VRM 8.4 era board. It was quite common to find boards that supported 1.5V though I don't recall the chip they used anymore... been a few years since I needed to know.

To get to the point, it should not be overgeneralized, rather the specific target board should be considered including looking up the spec sheet for the VRM controller it used. Some boards might've even used a different VRM controller on different revisions, or even on same revision if they were using some questionable component sourcing for lowest prices like PCChips seemed to do.

 

[cwj717]

The adapters didn't work in some boards... same with the slocket adapters as well... but it wasn't just in Intels boards. I have a Supermicro board that won't take them and an old IBM system that doesn't either. I don't really think Intel did that on purpose... its just that some older boards were made before the Tualatin cores existed.

Most boards would run them with a hack or two. Notible were some Sis chipset based boards that could not use a simple pin adapter. Most other boards depended on whether the voltage regulator supported (a low enough) voltage default of the Tualatins, and if they didn't then you'd choose the lowest voltage they did support which was often 1.5V. If it was 1.8V (very old board?) that might be too high but in that case the board itself is going to have so many limitations the CPU probably won't matter much, everything else will cripple the system till so many add on cards were used that it made no sense to reuse the motherboard. I did find a site where someone got a slocket adapter to work in a Supermicro P3DRE like mine... but I couldn't get it to work. None of the adapters I got worked either 🙁 I eventually gave up on all my dual P3 boards, the slot-1s and the socket 370s... It was a real pain because at that time they were really expensive boards, with Dual processor/SCSI/PCI-X and RDRAM support.

 

[Truekeeper]

I am glad a simple silly question of me turned into a so rich information debate !

Sadly the celeron dont even post, i think the mobo don't support it.

Well the guy must be settle with the p3.

Thanks people you were really helpful.

-----------------------------
Is this moss in my eyes ?

 

[m25]

And its not even worth getting an adapter to try it; most boards don't support tualatin voltages.