Apparently there was the Mobile Pentium 4, a s478 CPU meant for laptops. As it supports speedstep I'd say these should have a way of lowering the multi atleast with software

http://www.cpu-world.com/CPUs/Pentium_4/TYPE- … entium%204.html

Apparently there was the Mobile Pentium 4, a s478 CPU meant for laptops. As it supports speedstep I'd say these should have a way of lowering the multi atleast with software

But very few S478 desktop boards support Speedstep (I don't know of any that do, actually), so usually those chips will only run at the lowest default multiplier.

Tetrium wrote:

Apparently there was the Mobile Pentium 4, a s478 CPU meant for laptops. As it supports speedstep I'd say these should have a way of lowering the multi atleast with software

http://www.cpu-world.com/CPUs/Pentium_4/TYPE- … entium%204.html

Most of those were socket 479, and I can assure you the socket is completely different, or at least different enough you're not going to get one to fit in a desktop board without an adapter. I've never succeeded in finding an adapter because they use a different voltage too. It looks like those are true s478 chips though. Most likely for those monstrous "desktop replacement" laptops with 17 inch screens and 10 pounds of junk in the trunk. If you can get past their lower voltage, that could work, but it might only work if the BIOS supports changing the voltage.

It would probably be a better option though as I hear rumors that a lot of engineering samples with unlocked multipliers had minimum multipliers that kept them from going to insanely low clock speeds. But then again, those are just rumors. I've never been lucky enough to get my hands on an engineering sample of any kind. 😁

Asus made a S479 adapter, but it only worked in their boards, and only a few specific models, at that. But there were two kinds of mobile P4s... the "Mobile Penium 4" which was just a S478 desktop chip with Speedstep, that's the one Tetrium is talking about, and the "P4-M" which was the S479 deal.

i think you are right. the "p4-m" and "mobile-p4" are different things, i looked into this years ago but now have forgotten all about it.

SavantStrike wrote:

Most of those were socket 479, and I can assure you the socket is completely different

You are 100% correct here mate 😉

I know about the different sockets (Intel purposefully made them different, kinda like what they did with Tualatin, which sparked the pinmod thing), but I thought these ones are pin-compatible with desktop s478. I did do some research, but I'll go research some more.
There were overclockers using mobile P4's in desktop boards so it "should" definitely be possible.

Edit:

SavantStrike wrote:

It would probably be a better option though as I hear rumors that a lot of engineering samples with unlocked multipliers had minimum multipliers that kept them from going to insanely low clock speeds. But then again, those are just rumors. I've never been lucky enough to get my hands on an engineering sample of any kind. 😁

Unfortunately I don't have an ES P4 (got very few P4's to start with, me no likey netburst very much 😁), otherwise I'd do a couple quick tests for ya 😉

www.cpu-world has people selling ES's on a regular basis, it's where I got my s370 ES which turned out to indeed have a free multiplier! 😁 I posted about it about a week ago or so?

Btw, I can't do any testing as I'm REALLY moving stuff around in the attic atm, but the results should be quite pleasing. Attic 2.0 is starting to near it's completion 😁

Edit2:

Old Thrashbarg wrote:

Asus made a S479 adapter, but it only worked in their boards, and only a few specific models, at that. But there were two kinds of mobile P4s... the "Mobile Penium 4" which was just a S478 desktop chip with Speedstep...

True, but frankly it's maybe only slightly less craptastic as the standard desktop Netburst chips, but at least they are slightly better...right??
And perhaps you could still change the default low multi of the mobiles with CrystalCPU. CrystalCPU worked perfectly fine with my Athlon 64. Might work with P4 mobiles's multi's al well.

Edit3:Btw, I don't really know a lot about Netburst, other then their stock heatsinks always made a terrible noise and that after having washed my hands I could try them more quicker by holding them to the back of my mothers P4 rig 😜

I know about the different sockets (Intel purposefully made them different, kinda like what they did with Tualatin, which sparked the pinmod thing), but I thought these ones are pin-compatible with desktop s478. I did do some research, but I'll go research some more.
There were overclockers using mobile P4's in desktop boards so it "should" definitely be possible.

Come to think of it, you're right... both the Mobile P4 and the P4-M did use S478... it was the Pentium-M that went to S479. (Fuck Intel and their stupid-assed naming scheme. 😒 )

The P4-M was technically pin-compatible with the desktop socket, and it's possible that they did work in some desktop boards, though you'd still have to contend with the significantly lower voltages and lack of a heatspreader.

I don't recall any overclockers actually using them, though... The trouble is, most people seem to use "Mobile P4" and "P4-M" interchangeably, even though they're two entirely different things. While some people may have had actual P4-Ms in their systems, usually they're talking about Mobile P4s when they say "P4-M". The Mobile P4s were relatively popular with overclockers for awhile, since the low multiplier allowed really high FSB speeds.

And perhaps you could still change the default low multi of the mobiles with CrystalCPU. CrystalCPU worked perfectly fine with my Athlon 64. Might work with P4 mobiles's multis as well.

The Speedstep on those things allows only two fixed settings... the low mode, which is 12x (or 9x on some of the later ones, I think), and the high mode, which is whatever the full speed of the chip. Nothing in between. Furthermore, the mode is switched by a signal pin on the CPU, a pin which isn't hooked up on desktop motherboards, and that signaling is controlled by the EIST support in the BIOS, which isn't present on desktop boards. Without that hardware and BIOS support, you're stuck at the lowest multiplier. Quite a few people tried to find ways around that, and AFAIK nobody ever had any success.

Old Thrashbarg wrote:

The Speedstep on those things allows only two fixed settings... the low mode, which is 12x (or 9x on some of the later ones, I think), and the high mode, which is whatever the full speed of the chip. Nothing in between. Furthermore, the mode is switched by a signal pin on the CPU, a pin which isn't hooked up on desktop motherboards, and that signaling is controlled by the EIST support in the BIOS, which isn't present on desktop boards. Without that hardware and BIOS support, you're stuck at the lowest multiplier. Quite a few people tried to find ways around that, and AFAIK nobody ever had any success.

Ouch...well then the only route for a free multi seems to be the ES.

Tetrium wrote:

Old Thrashbarg wrote:

The Speedstep on those things allows only two fixed settings... the low mode, which is 12x (or 9x on some of the later ones, I think), and the high mode, which is whatever the full speed of the chip. Nothing in between. Furthermore, the mode is switched by a signal pin on the CPU, a pin which isn't hooked up on desktop motherboards, and that signaling is controlled by the EIST support in the BIOS, which isn't present on desktop boards. Without that hardware and BIOS support, you're stuck at the lowest multiplier. Quite a few people tried to find ways around that, and AFAIK nobody ever had any success.

Ouch...well then the only route for a free multi seems to be the ES.

Or jack the FSB up to something crazy. Though trying to run a chip with a 400mhz FSB at say... 8-900mhz might not work so well either. I can see someone trying this for overclocking purposes, but industrial mobos don't usually have that kind of head room.

I hated this socket 479 crap. I was happy with the desktop S478. Call me crazy, but despite it's lower per clock performance, the P4 had a pretty significant clock lead over the Athlon up until the A64 came out. Add in significantly higher memory bandwidth and they were very attractive for gaming (and crap for everything else in retrospect). Of course then when the A64 came out they were crap in memory bandwidth by comparison, and were outclassed in every way imaginable.

My mom's P4-M in her laptop was a train wreck though. It came with a 1.8ghz Celeron and I upgraded it to a 2.4ghz P4-M. I swear it probably only used like 10 watts less energy than the 2.4ghz part I had in my desktop a year or two before. A laptop CPU with a 50-60W TDP is just crazy. For the record, it was socket 479, or I'd be putting it in a desktop board and torturing the poor thing. 60W is a good 20W away from the desktop counterpart.

SavantStrike wrote:

My mom's P4-M in her laptop was a train wreck though. It came with a 1.8ghz Celeron and I upgraded it to a 2.4ghz P4-M. I swear it probably only used like 10 watts less energy than the 2.4ghz part I had in my desktop a year or two before. A laptop CPU with a 50-60W TDP is just crazy. For the record, it was socket 479, or I'd be putting it in a desktop board and torturing the poor thing. 60W is a good 20W away from the desktop counterpart.

Agree completely. I still remember when Intel finally released a mobile netburst CPU to replace the Pentium 3-based CPU's they had been using as their mobile line (which in itself was a sign on the wall that netburst was a bad thing), it mentioned something similar to what is mentioned in the wiki:

The Mobile Intel Pentium 4 Processor[14] was released to address the problem of putting a full desktop Pentium 4 processor into a laptop, which some manufacturers were doing. The Mobile Pentium 4 used a 533 MHz FSB, following the desktop Pentium 4's evolution. Oddly, increasing the bus speed by 133 MHz (33 MHz core) caused a massive increase in TDPs, as mobile Pentium 4 processors gave off 59.8 W - 70 W of heat, with the Hyper-Threading variants giving off 66.1 W - 88 W. This allowed the mobile Pentium 4 to bridge the gap between the desktop Pentium 4 (giving off 115 W maximum), and the Pentium 4-M (giving off 35 W maximum).

Pentium 4 desktop 115W, Pentium 4-M (which was really based on the P6 core) 35W and bridging the gap with a CPU that's 88W?? Now THATS a joke 🤣!
Even if I read it today, it still makes me laugh, it's not bridging the gap, it's just a 1 step improvement with 15 more steps to take before I would call it a successful "bridging of the gap".

It's a miracle that Intel had to take so long before realizing that netburst was just a dumb idea, basically doing the opposite of what they should have been doing:Making the chip dumber so they can clock it higher. Then having to add more cache to compensate for P4's increased dumbness causing the chip to run hotter and scale worse. Intel's solution? Make the chip even more retarded so it would still scale (and adding even more cache to compensate again, making it run even more hot...really, how could Intel have been so stupid 😵 ).

My laptop uses S478 processors. It's a 14-inch. The nice thing is that there is a jumper beside the processor socket which selects the installed processor type - "Mobile CPU" or "Desktop CPU". (It shipped with a P4 Desktop CPU though, because when I looked at the Spec number on Intel Ark, it doesn't support SpeedStep).

On a room with ambient temperature around 28 degrees Celcius (a typical temperature in tropical Asian countries 😁), it reaches around 45-50 degree Celcius while used for light tasks (web browsing & Microsoft Office, plus listening to some MP3s using WinAmp).

As a comparison, a Core 2 Duo Wolfdale at the same room runs around 39 degree Celcius in idle mode, and Intel Atom 230 runs around 36 degree Celcius.

You can't really compare temps in different laptops as their cooling solutions will be completely different from one another. It's the thermal output that really matters and netburst is the worst solution in this regard.
Sure, Northwood was kinda nice but the P6 laptop CPU's were way better

Tetrium wrote:

You can't really compare temps in different laptops as their cooling solutions will be completely different from one another.

The two comparison systems are desktops, actually. 😀

MatthewBrian wrote:

My laptop uses.....

Tetrium wrote:

www.cpu-world has people selling ES's on a regular basis, it's where I got my s370 ES which turned out to indeed have a free multiplier! 😁 I posted about it about a week ago or so?

I've looked at the p4 s-codes on that site. Ones starting with Q seem to be qualification and engineering samples. What's the difference?

Is it more or less a lucky dip what an ES is capable of, ie. I won't know until I try it?

I didn't find any P4 ES's for sale on there but there's an ebay seller lzf70000 who has a few. Tempting to get one to mess about with, although I don't think in reality a p4 can be slowed to say p100 speed, which for me is I think where it gets interesting.

There was something like an engineering sample celeron 1.4 willamette in s478 listed on the cpu-world specs page. That might I guess go pretty slow, and if cache could be disabled that might do it.

Then again a 1.4 cpu will severely limit the top end of the board without swapping the cpu. So the practical uses of all this for me maybe a bit limited 🤣, it won't replace both a k6/3 and an athlon xp system even if a 1.4 could be sufficiently "crippled", without swapping cpu's. Nonetheless it's piqued my interest.

Can the p4 cache be disabled? I came across some benchmarking utility that seemed like it might do it but it needed xp/2000 whereas I'm running 98. Wondered if anyone might be aware of anything either way on this.

Iirc Qualification Samples were send to hardware sites and such so they could have a first glance at it, and maybe motherboard manufacturers while Engineering Samples were for internal use??

And that Engineering Samples are usually unlocked while Qualification Samples are sometimes unlocked, that's kinda all I know about these chips.

SavantStrike wrote:

Or jack the FSB up to something crazy. Though trying to run a chip with a 400mhz FSB at say... 8-900mhz might not work so well e […]
Show full quote

Tetrium wrote:

Old Thrashbarg wrote:

The Speedstep on those things allows only two fixed settings... the low mode, which is 12x (or 9x on some of the later ones, I think), and the high mode, which is whatever the full speed of the chip. Nothing in between. Furthermore, the mode is switched by a signal pin on the CPU, a pin which isn't hooked up on desktop motherboards, and that signaling is controlled by the EIST support in the BIOS, which isn't present on desktop boards. Without that hardware and BIOS support, you're stuck at the lowest multiplier. Quite a few people tried to find ways around that, and AFAIK nobody ever had any success.

Ouch...well then the only route for a free multi seems to be the ES.

Or jack the FSB up to something crazy. Though trying to run a chip with a 400mhz FSB at say... 8-900mhz might not work so well either. I can see someone trying this for overclocking purposes, but industrial mobos don't usually have that kind of head room.

I hated this socket 479 crap. I was happy with the desktop S478. Call me crazy, but despite it's lower per clock performance, the P4 had a pretty significant clock lead over the Athlon up until the A64 came out. Add in significantly higher memory bandwidth and they were very attractive for gaming (and crap for everything else in retrospect). Of course then when the A64 came out they were crap in memory bandwidth by comparison, and were outclassed in every way imaginable.

My mom's P4-M in her laptop was a train wreck though. It came with a 1.8ghz Celeron and I upgraded it to a 2.4ghz P4-M. I swear it probably only used like 10 watts less energy than the 2.4ghz part I had in my desktop a year or two before. A laptop CPU with a 50-60W TDP is just crazy. For the record, it was socket 479, or I'd be putting it in a desktop board and torturing the poor thing. 60W is a good 20W away from the desktop counterpart.

Clock speed only matters when comparing chips with otherwise identical architectures. It's the number of ops per second that a CPU can carry out that's more important. Look at how much faster the PPC G3 and G4 were over similarly clocked PII's and PIII's. AMD didn't have to make chips with higher clocks because theirs were faster even at lower clocks the same as the G3 and G4 were. I always hate it when Intel people (not saying you're one) try to make a point about Intel having a lead in clock speed over AMD. High clocks have little relevance to performance if your chip design is inefficient in every other way that matters and P4 was extremely inefficient. The Tualatins were faster than the early P4's clock for clock which was why they became the basis for the chips that came after the P4. The Pentium M and Core architectures derive more from the PIII than the P4.

Tetrium wrote:

MatthewBrian wrote:

My laptop uses.....

I mean the two systems listed on the bottom of the post:

MatthewBrian wrote:

As a comparison, a Core 2 Duo Wolfdale at the same room runs around 39 degree Celcius in idle mode, and Intel Atom 230 runs around 36 degree Celcius.

C2D Wolfdales and Atom 230 are desktop CPUs 😀

MatthewBrian wrote:

I mean the two systems listed on the bottom of the post: […]
Show full quote

Tetrium wrote:

MatthewBrian wrote:

My laptop uses.....

I mean the two systems listed on the bottom of the post:

MatthewBrian wrote:

As a comparison, a Core 2 Duo Wolfdale at the same room runs around 39 degree Celcius in idle mode, and Intel Atom 230 runs around 36 degree Celcius.

C2D Wolfdales and Atom 230 are desktop CPUs 😀

Was just saying you still can't compare temps if the temps are measured in totally different systems with different cooling solutions 😉