Challenging problem-Tosh A30+P4 Northwood?

[gofasterstripes]

Dear Community,

do you think you can help me?

I have a Toshiba Satellite PSA30E-1CVQC [European version]

Northwood Celeron 2.6GHZ, 128kb SL6VV

The rest of the machine is decent - 160GB HDD [just added], 1.5Gb RAM, OK Battery, new PSU

It just could really do with a CPU that doesn't suck. I Wanted a P4 Northwood, as fast as possible.

Today I tried it with: 2.8GHz 533Bus Chip
2.6GHz 533Bus Chip
3.0GHz 800Bus Chip
all with no joy. Power on, for 2 seconds, powers off. Keeps doing that.

Putting the original chip back in results on successful functioning again. So
I didn't break it.

The problem is how slow it is. I'm running Ubuntu 12.04, and because the machine has Intel 855 integrated graphics, it has to use software rendering. Running Firefox with 4 tabs, 4 open PDFs in the background and two file system windows open, the CPU is stuck at 100% and the machine noticeably lags, even typing this!

I'm sure that a full-fat P4 will improve things a lot, I just need to find the right one. The fact that this chassis was available with a P4 gives me a lot of hope.

I can buy a 2.8GHz, 400MHz Bus P4 SL7EY,

http://ark.intel.com/products/27446/Intel-Pentium-4-Processor-2_80-GHz-512K-Cache-400-MHz-FSB

the voltage range is damn close 1.475V-1.55V but that is higher and gives a max TDP of 68.4W -- I guess that's too much in a laptop designed for 62.6! Shame, as it's the fastest 400MHz bus chip, I think.

So I guess it's better to buy an SL6PP for £10, which appears to be the same thing as my Celeron, but with 4 times the cache? Voltages and wattages are within range...

Can you, or anyone reading this help me? I Found the machine in a skip, and managed to trade the two new parts for favors and technical assistance, so if I can make it run somewhat fast for £10, I'll be very very happy.

Thanks in advance!

GfS

 

[WR2]

The voltage and the thermal design power (TDP) differences would worry me. 'close' isn't quite good enough IMO.

-> Have you found any example of someone that's already made that swap successfully? And what did they do to make it successful?
I think you'd want to know for sure that someone else did it - and it worked - before you do that type of testing.

As for the SL6PP £10 upgrade. Going from the SL6VV to the SL6PP isn't going to change the speed the system run at enough to be worth it (IMO). It will be 'faster' but you'll need to run benchmark tests or use a stopwatch to detect the difference. It won't be noticeable as you run your normal workloads.

 

[gofasterstripes]

Thanks very much. I wasn't clear enough, sorry.

The machine has a 2.6GHz Celeron, and I want to spent £10 buying one of the two P4s. Are you agreeing with me that the 2.6GHz chip, which has the exact same TDP and the same Max Voltage is the better bet?

Either of them will be a great £10 upgrade, if they work.

Thanks again,
GfS

 

[gofasterstripes]

There doesn't seem to be any precedent. A couple of people asked online, here for example, but no answers that I can see. I have looked for a couple of hours.

GfS

 

[WR2]

Without a known example of a successful upgrade you're in 'experimental territory.
It could be all that's necessary is a custom BIOS that supports the slight differences in chip specs. But custom edited BIOS mods can get expensive.

Going from a 2.6Ghz Celeron to a 2.6Ghz P4 isn't going to get you a more than a miniscule performance increase.

 

[gofasterstripes]

Thanks again for your help.

It seems that the 2.6GHz 128k chip is massively cache starved, and the 2.6GHz P4 512k will be about twice as fast for common mixed tasks. This Celeron is usually considered the worst family of Celerons, perhaps tied with the 266 or 300 with 0KB of cache [but the P4 has a muuuuuch longer pipeline and suffers massively from cache misses]

Hence the desire to upgrade for £10

 

[WR2]

An upgrade @ £10 it's probably worth it to find out.

Let us know how much it improved things for you.

 

[WR2]

and the 2.6GHz P4 512k will be about twice as fast for common mixed tasks.

I'm going to guess, if you test with the same benchmarks, common task improvements are going to be 1-2% at most.

 

[gofasterstripes]

I'll tell you if i get it to work.

My expectations are based on this sort of information:
http://www.overclockers.com/forums/showthread.php?t=247485

a 1.7*256K P4 showing 10%-50% CPU bound improvement over the 2.6GHz 128k Celeron. I'm looking to install a 2.6GHz*512k chip, that's go to be 15-75% difference, and on this machine the CPU is maxed out ALL the time. Hells bells, if it just switches tabs 50% faster I'm gonna be happy

GfS

 

[Chainzsaw]

To be honost here, I don't know why you would want to waste so much time on an outdated platform for very miniscule performance gains (if you can even upgrade the processor at all).

Sorry someone had to say it. You may be saving yourself some coin by trying to do what you want to do, but for the amount of time your "wasting" for the "performance" gain you want, your losing money all around.

I'm not trying to sound mean at all, just trying to get the facts out there.

 

[bucknutty]

look at it as an adventure... if you have the time and you are learning something then its all good.

My guess is the issue is not hardware, My guess is the bios on the board does not recognize your p4 chip. Check to see if there are any bios updates.

Maybe get crazy and see if you can compare the pentium model to the celeron model board. If toshiba has the same part listed for both boards maybe you can use a p4 board bios on your celeron.

I play with old parts all the time, build htpcs, and nas devices out of P4s. They are fun to play with. Normally mixing p4s celerons and P4ms is all about the bios support from the vendore.

 

[gofasterstripes]

Aaah - a good question - why when I have a overclocked i7 920 desktop?

Just had a day off before I start my new job! And having acquired the machine, the PSU and the HDD for free all in a couple of days, separately.

I wanted to mint the deal, allowing me to use this machine to read my daily news and comment in bed in the morning with a coffee, sitting next to my girlfriend who always sleeps in later than me, by about an hour.

I have a netbook, NC10, but i find the 600pix vertical to little.

Also, my friend really wants to buy that machine as me stays on a boat, and power efficiency is key. I'll sell the NC10, gain a small profit, gain a faster machine with a better screen [the Tosh is a little faster as it is, but as mentioned Ubuntu's graphics shell is taxing it] and have a little money left over to put toward the Asus Transformer I will buy in a couple of months, when the wages start to come in.

So, yeah, it's hard to imagine, but it IS worth me taking my day off to mess around with the machine.

I just happen to really want a machine just like this right now....

 

[gofasterstripes]

look at it as an adventure... if you have the time and you are learning something then its all good.

My guess is the issue is not hardware, My guess is the bios on the board does not recognize your p4 chip. Check to see if there are any bios updates.

Maybe get crazy and see if you can compare the pentium model to the celeron model board. If toshiba has the same part listed for both boards maybe you can use a p4 board bios on your celeron.

I play with old parts all the time, build htpcs, and nas devices out of P4s. They are fun to play with. Normally mixing p4s celerons and P4ms is all about the bios support from the vendore.

I have the latest BIOS - But, I wonder if flashing it to the BIOS of a machine that shipped with a P4. That's tricky though, as there's no linux BIOS tool -I have to use Windows...somehow....

 

[gofasterstripes]

CPU SL6PP is on it's way...

Thanks for all the input - it's just fun to see if the free machine can be generally useful! I sure hope it's faster with that chip, 'cos it's slooooow with the Celeron...maybe I should try Mint or something else. I'll let you know if it works.

GfS

 

[gofasterstripes]

WAAAAAYYY!!! It works! The machine booted first time with the SL6PP, which is the P4 at the exact same clock speed, same voltages etc as Celeron SL6VV

POST screen shows P4 logo, so I guess the BIOS was locked out of supporting unknown chips, even if the chipset should support 533bus chips, it's not allowed to. I suspect that SL7EY might work, but I can't afford another gamble

Very rudimentary testing show marked improvement as expected:

Boot times [about 20% faster from power button], rendering 4 tabs of Large webpages from disk then reloading them on timer [assuming data is cached it's roughly a test of cpu's mixed workload power - [about 40% faster],

Lastly - launching Libre Office Write, a second time [from disk cache?], about 100% faster!!

The whole Ubuntu Unity 2D interface seems much smoother, frame rate about 100% better

CPU tracking show large drop in general use, the fan runs less and, one assumes, if the CPU is taxed less the battery life should be better [this thing still hits about 1:45 of the 2 hours it should have from new - maybe the previous owner gave it a new battery?]

Youtube now mostly runs smooth - I think the issue here is the GFX hardware and Ubuntu's problem with this chipset, even the extra CPU power isn't up to this in software.

128Kb Celerons are as I thought, the worst CPU made by Intel, or thereabouts!

I didn't benchmark the CPUs with a benching software, and it's unlikely I'll get 'round to it any time soon as each time I swap them I have to change the thermal paste and I don't have any left now But you never know, as I will order some soon as I'm gonna lap my Core i7 [desktop] and beat the daylights out of of it...

But if I do it I'll let you know how it measures.

Thanks for all the help guys, as ever.

GfS

 

[gofasterstripes]

iDon't know what this is doing in the iPad forum - can someone move it to Laptops?

 

[Brett928S2]

Best answer selected by Brett928S2.

 

[Maziar]

This topic has been closed by Maziar