Sign in with
Sign up | Sign in
Your question

Low Cost Celeron Compared to Pentium-M?

Last response: in CPUs
Share
January 20, 2004 1:01:58 AM

My friend is selling his 5 month old 1.8ghz celeron sony vaio notebook for $900. the notebook i was looking at was the ibm x31 1.4ghz pentium-m for around $1500. at 4.4lbs, the sony notebook is a little heavier, but is an extra 600 worth it for a pentium-m & a little less weight? i don't plan on gaming, so graphics isn't an issue, what i'm really wondering is if the bigger cache on the pentium-m is worth extra money over the cheaper celeron in day to day usage.

thanks in advance,


scott.
January 20, 2004 1:15:35 AM

Hm... Pentium-M is a highly sophisticated processor. It has one of the highest - if not the highest - IPC of any processor around (maybe leaving Itanium out).

Celeron is a crippled - and I mean severely crippled - P4. It is very much slower than equivalent P4s, and is actually - clock ignored - one the worst processors around.

So the question is... What will a 1.4Ghz P-M do to a 1.8Ghz Celeron?...

It will rock Celeron's world!
It will disinitegrate, obliterate and destroy the Celeron!
It will eat Celeron for breakfast!
It will torture Celeron until it says "I'm sorry to exist"!
It will cut through Celeron like a hot knife through butter!
It will ...

Well, anyway, you get the point. This 1.4Ghz Pentium-M will have longer battery life, and will be considerably - and I mean considerably - more faster than this 1.8Ghz Celeron. Maybe I've exaggerated... But I don't think so. Pentium-M is a wonderful processor, and Celeron is not a good processor. So this choice is probably for the best. Getting a P-M and selling the 1.8Ghz Celeron. That is, if your friend indeed requires more horsepower... If it's only word and Internet, then, well, the 1.8Ghz Celeron should be enough.

:evil:  <font color=red><b>M</b></font color=red>ephistopheles
January 20, 2004 1:54:30 AM

let me rephrase what i said; my friend is looking to sell me his 5 month old 1.8ghz celeron sony vaio notebook for $900. i was previously looking at a ibm x31 1.4ghz pentium-m at around $1500. so, what i'm looking for is 600 dollars extra for the ibm notebook worth it if my typical usage will be photo editing, internet, & remote desktop access to my pc? or can i save money and take up my friend's offer?
Related resources
January 20, 2004 4:55:29 AM

Well all i can tell is that a Duron 1.6 beats a celeron 2.6 in most apps....and 1.3 P-M Celeron (with 512 kb cache) is faster clock for clock than the duron.. so i think you would need a P4-2.2Ghz/2.4Ghz to match a P-M.
January 20, 2004 6:21:13 AM

As a person that has been using a Pentium M notebook for the past 2 months, my only advice is to get the Pentium M. I wouldn't even consider the Celeron ... way too crappy. Btw, maybe you should check out these <A HREF="http://www.powernotebooks.com" target="_new">guys</A>. I've bought two laptops from them ... EXCELLENT customer service. Btw, the Power Pro C3:16 is a very good choice.
January 20, 2004 8:53:57 AM

The Pentium M will probably win from the Celeron, but not by terribly much. Definitely not noticably in "day to day usage". Personally I wouldn't find it worth the extra money. If the Vaio is in good shape, go for it.
January 20, 2004 10:26:57 AM

Oh, I see... For some reason, I was led to believe that it was your friend that wanted the P-M, not you, sorry.

In that case, you have to consider that this P-M probably is much, much faster than the Celeron, like I said before. So the P-M is highly recommended.

:evil:  <font color=red><b>M</b></font color=red>ephistopheles
January 20, 2004 10:57:06 AM

Quote:
The Pentium M will probably win from the Celeron, but not by terribly much. Definitely not noticably in "day to day usage". Personally I wouldn't find it worth the extra money. If the Vaio is in good shape, go for it.

Since you are expressing an opinion like this, I suppose you have used both systems and that's how you've come to a conclusion like this. Have you or are you simply guessing here?
January 20, 2004 10:58:07 AM

Quote:
The Pentium M will probably win from the Celeron, but not by terribly much.

Don't know about that... Depending on what you do, you'll probably notice the difference, I think... Could be wrong. But I think you will... The problem is that it means a 66% increase in price...

:evil:  <font color=red><b>M</b></font color=red>ephistopheles
Anonymous
a b à CPUs
January 20, 2004 10:58:43 AM

I'm with Mephistopheles. I've worked on many Celeron laptops, and more often than not, they are ASTONISHINGLY slow. REally, really painfully slow (think 486 slow). Don't even mention photoediting, just using the windows UI while switching between a few open apps takes forever.

This is partly because of the Celeron itself, which is regardless of clockfrequency, slower than slow (did you see Anand's review ?), but also, or even mostly ? because Celeron laptops are usually budget laptops where the oem saves on everything by using slooooow harddisks, UMA architecture making even basic screenredraws slow, etc, etc.

PM on the other hand, I had the pleasure of working with one for a few months, and I loved it. Fast, responsive, impressive battery life. But expensive.. Might be worthwhile looking at a few new alternatives with a Athlon-XPM (which is vastly underestimated), or even a P4-M model. Used alternatives might also want to look at P3 (yes, Pee THREE !) models. I'm pretty sure an average 1.2 or even 1 GHz P3-M will smoke a typical P4 Celeron, including >2 GHz ones.

Anyway, instead of asking, why don't you try your friends VAIO, and see for yourself if it is fast enough for your needs ? Maybe Sony didnt skimp too much on the harddisk&memory subsystem and it does perform reasonable as long as you do nothing too CPU intensive..

= The views stated herein are my personal views, and not necessarily the views of my wife. =
Anonymous
a b à CPUs
January 20, 2004 11:02:00 AM

>The Pentium M will probably win from the Celeron, but not
>by terribly much. Definitely not noticably in "day to day
>usage".

You're wrong. you DO notice it, and you notice it much more than what benchmarks will tell you. Benchmarks will already give the PM roughly a 50-100% ? performance advantage over the 1.8 celery, but responsivness, and the speed of the other components may well make the difference even bigger when actually using the laptop. If you've not had a Celeron laptop, you have no idea how slow these things can be.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
January 20, 2004 12:53:28 PM

i have a celeron 1.8 laptop and i must say that it is SLOW as hell. i would prefer 100x to use my p3 550 than this bullshit that lag in windows.

My own beast: Athlon 2700xp+ (oc: 3200xp+ with 200fsb) , Radeon 9800pro (oc: 410/370) , 512mb ddr400. SO MUCH faster than my last computer (pIII 550......)
January 20, 2004 1:35:50 PM

If you are just an app-user and not into gamming then the celeron is ok for you and very much cheaper.

Barton 2500+ @ 2200mhz (10x220 vcore @ 1.8)
Asus A7N8X Dlx 440 FSB
1gb Geil GD pc3500 Dual Channel (2-3-3-6)
Segata 80gb SATA 8.5ms seek
ATI Radeon 9800 Pro(420/720)
January 20, 2004 5:57:45 PM

Quote:
Since you are expressing an opinion like this, I suppose you have used both systems and that's how you've come to a conclusion like this. Have you or are you simply guessing here?

I have a Celeron Pentium III (Tualatin) 1.2 GHz desktop and a Pentium M 1.4 GHz laptop. Although I am very happy with the laptop's performance, I would have rather had a much cheaper Celeron like my desktop. The main reason I chose this laptop is because it was the only one in the price range with a good form factor.

So, no, I'm not guessing much. I don't have a Mobile (Pentium 4) Celeron 1.8 GHz but I'm quite sure it's sufficient for "day to day usage", and the Pentium M is not magically twice as fast. So seriously, it's not worth the extra money.
Anonymous
a b à CPUs
January 20, 2004 6:18:34 PM

> I don't have a Mobile (Pentium 4) Celeron 1.8 GHz but I'm
>quite sure it's sufficient for "day to day usage"

And I tell you its not. I've worked on at least 4 different <b>P4 celeron </b> laptops, and none of them where sufficient to run Solitaire alongside Outlook and Word in a satisfactory way. Just closing an app, switching to desktop and opening "my computer" could take forever, even with a fresh isntall and plenty of RAM.

I've not done enough testing to determine wether this was mainly caused by utterly slow harddisks on the celery laptops, but I assure these things where slower than you'd ever think possible. You're experience with your Tualatin Celeron are not relevant. Those chips where screamers compared to P4 celerons, and actually quite decent budget chips.

>the Pentium M is not magically twice as fast

It is. Really, it *IS*. Making a few simplifications here, but a Pentium M 1.7 performs roughly on par with a 2.4 GHz P4. Agreed ? That would give it an AMD PR rating of around 2400+. Now a Celeron 2 GHz is roughly 50% slower than a XP 1700+ on general purpose benchmarks like Business Winstone 2004, Content Creation Winstone 2004, and most gaming benches (cf anandtech), giving it a rating not far above 1000+. You do the math, a Pentium M gets ~1.5 PR points per Mhz, the Celeron around 0.5. That is THREE times as slow for the same frequency.

Believe me, you don't know how slow those P4 Celerons are, until you've had to work with one.

-----edit----
Wether the P-M is is worth the extra money over the Celeron, is not for me to decide; but I would definately suggest looking at anything else than a Celeron laptop, since the same ammount of money should buy you a (used) P3-M, (used) P4-M or AXP-M one, all of which should blow the Celeron out of the water, if not only in benchmarks, then definately in responsivenes, and feeling (MUCH) faster for the same ammount of money.
---- end edit ---

= The views stated herein are my personal views, and not necessarily the views of my wife. =<P ID="edit"><FONT SIZE=-1><EM>Edited by bbaeyens on 01/20/04 03:26 PM.</EM></FONT></P>
January 20, 2004 9:43:05 PM

Quote:
You're experience with your Tualatin Celeron are not relevant. Those chips where screamers compared to P4 celerons, and actually quite decent budget chips.

I don't see why it's totally irrelevant. The Tualatin Celeron has 256 kB of L2 cache, so has the Mobile Celeron. The Tualatin has a 100 MHz FSB, the Mobile Celeron has a 400 MHz FSB (as does the Pentium M). The ICP differs noticably but the actual core is the same as a non-Celeron Mobile Pentium 4, of which I haven't heard many complaints.
Quote:
It is. Really, it *IS*. Making a few simplifications here, but a Pentium M 1.7 performs roughly on par with a 2.4 GHz P4. Agreed ? That would give it an AMD PR rating of around 2400+. Now a Celeron 2 GHz is roughly 50% slower than a XP 1700+ on general purpose benchmarks like Business Winstone 2004, Content Creation Winstone 2004, and most gaming benches (cf anandtech), giving it a rating not far above 1000+. You do the math, a Pentium M gets ~1.5 PR points per Mhz, the Celeron around 0.5. That is THREE times as slow for the same frequency.

Based on other benchmarks, allow me to give the Celeron 0.6 PR/MHz. Now, the Mobile Celeron is somewhere between a Celeron and Desktop model, so it must have 0.8 PR/MHz. That gives the 1.8 GHz Mobile Celeron 1440 PR points.

I agree the Pentium M 1.7 GHz performs like 2400 PR points. So a 1.4 GHz Pentium M gets 1976 points. Even if I round that upward, it's only 40% faster than the Celeron...

I take your word for it that some Celeron laptops you've worked with gave a bad experience. But I higly doubt that's entirely the fault of the Celeron. As you probably know, the chipset and harddisk are critical points and highly determine what performance is like when running a lot of applications.

Either way, considering the price and intended use, I wouldn't dismiss the Celeron as a viable option that easily. I've heard very good things about Sony Vaio so if the other components are good it might still deliver nice performance. Of course the IBM certainly can't dissapoint much either...
Anonymous
a b à CPUs
January 20, 2004 10:28:54 PM

>I don't see why it's totally irrelevant. The Tualatin
>Celeron has 256 kB of L2 cache, so has the Mobile Celeron.

First of all, you're comparing apples to oranges. The P3 and P4 are so radically different, you really can not extrapolate anything based on just similar L2 caches. Secondly, how many laptops do you think use Mobile Celerons ? Though I've not verified on every laptop I laid hands, I'd guess the vast majority uses desktop chips.

>Based on other benchmarks, allow me to give the Celeron 0.6
>PR/MHz.

Off course a lot depends on what benchmarks you chose, but I could also point to a few where it is much worse. Take Business winstone:
2 GHz Celeron: 12.5
1700+ AXP : 17.2
This would rate the Celeron as a 1200+ for the 2 GHz model, IF, (big IF) you also assume winstone scales lineary with clock speed, which it most definately doesnt. The 2400+ is only 11% faster in winstone than the 1700+ and also the Celeron gains only 10% from a 30% clock increase. A quick regression in excell shows that means it would take a 4.4 GHz Celeron to match the 1700+ !!

Now, if you extrapolate downwards (ignoring the bigger cache Barton) I'd have to give the the Celeron 2 GHz a rating of 650+ while the 2.6 GHz Celeron would get a whopping ~800+ rating.

In Content Creation the situation isnt much better. It would take a 3.3 GHz Celeron to deserve a 1700+ rating, and the 2 GHz one gets a 850+ rating using linear regression through the datapoints.

If we take this number, the mentioned Celeron 1.8 scores a whopping 650+ compared to a guestimated 2000+ for the Pentium M 1.4.

Numbers are to be taken with a buckload of salt, but the point remains: the difference is HUGE.


= The views stated herein are my personal views, and not necessarily the views of my wife. =
Anonymous
a b à CPUs
January 21, 2004 9:47:37 AM

> this sounds idiotic

It does.. 'nuf said.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
January 21, 2004 10:28:29 AM

In case the Celeron is a desktop chip, I must agree. The 128 kB L2 cache is just too small to hold the working set of most applications. But if it's a Mobile Celeron with 256 kB of cache, there's not much preventing it to perform close to non-Celeron expectations.

Besides, you have to sharpen your basic math skills again because some of it doesn't make sense or is even contraditory...
January 21, 2004 10:31:30 AM

Another problem with desktop CPU's in notebooks is that they'll eat your b82rez(that's battery) alive.
Anonymous
a b à CPUs
January 21, 2004 12:39:28 PM

> But if it's a Mobile Celeron with 256 kB of cache, there's
>not much preventing it to perform close to non-Celeron
>expectations.

I've not seen a lot of benches of mobile P4 celerons to prove this right or wrong, but a mobile celeron should perform pretty close to a P4 Willamette, the specs are very close, with the exception of the the prefetch I think (or was it the trace cache ?). That is probably acceptable, though still not exactly stellar especially not in combination with slow, single channel DDR or God forbid, even SDR in some laptops.

Anand also included a 1.8 P4 Willamette, with a dual channel 865PE board, and it got beaten in every non gaming benchmark by a Duron 1.6 and it was ridiculed by the AXP 1700+. I stick to my point, Celeron is a bad choice, and the only excuse to get one, is that you wouldnt be able to find a laptop, new or used with a mobile AXP, P4-M or P3-M that fits your requirements, or a P-M that fits your budget. Even a Duron laptop is most likely a better choice..

>Besides, you have to sharpen your basic math skills again
>because some of it doesn't make sense or is even
>contraditory...

Maybe, but its probably because the benchmark results don't always make sense or seem contraditory. I just extrapolated two straight lines through the AXP (without barton) and Celeron results (x-axis being Mhz/PR and Y axis being winstone/cc), and noted where they would intersect. Feel free to redo the math, but I doubt you will come to significantly different results. Its just had to grasp that a chip with such a relative high clock performs so miserable, but the numbers don't lie: it does. And it just gets worse with the even higher clocker variants like the 2.6 Ghz that still performs well below Duron levels.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
January 21, 2004 5:00:53 PM

Quote:
Anand also included a 1.8 P4 Willamette, with a dual channel 865PE board, and it got beaten in every non gaming benchmark by a Duron 1.6 and it was ridiculed by the AXP 1700+.

It was a Northwood "A" (1.8A GHz), not Willamette. Now guess how bad Willy is gonna perform.....

----------------
<b><A HREF="http://geocities.com/spitfire_x86" target="_new">My Website</A></b>

<b><A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new">My Rig & 3DMark score</A></b>
Anonymous
a b à CPUs
January 21, 2004 9:21:40 PM

>It was a Northwood "A" (1.8A GHz), not Willamette.

Oh.. ouch.. OUCH !


= The views stated herein are my personal views, and not necessarily the views of my wife. =
Anonymous
a b à CPUs
January 22, 2004 5:53:48 PM

I just started a new thread on this, after I redid my math and cleaned up the results. You may find <A HREF="http://forumz.tomshardware.com/hardware/modules.php?nam..." target="_new">this </A>this interesting when deciding. Basically, a 1.8 GHz Celeron would get a ~400+ rating using AMD rating scheme. A P3-500 is likely faster.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
January 22, 2004 9:09:44 PM

We should be a bit careful now... There are Centrino-Celerons (Celeron M's) out there, with 512K and 1.2 and 1.3Ghz speeds. They should be much more respectable than the usual celerons.

Anyone know anything about those?

:evil:  <font color=red><b>M</b></font color=red>ephistopheles
January 23, 2004 12:20:41 AM

IMO, Celeron-M's won't be as bad as P4 Cellys. Pentium-M is based on P6 core. There wasn't much performance difference between P3-S (512k) and P3-Tualatin (256k). I think similar thing will happen with P-M vs. Cel-M

----------------
<b><A HREF="http://geocities.com/spitfire_x86" target="_new">My Website</A></b>

<b><A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new">My Rig & 3DMark score</A></b>
Anonymous
a b à CPUs
January 23, 2004 7:06:01 AM

>In my theory, Intel would never make any CPU over what the
>CPU's core requires which in Pentium M's case is 1MB L2

Oh, so you have tested this with all possible different cache sizes, hmm ? And the P4 really needs 2 Mb cache like the P4EE ? Or 4 Mb like the Xeons ?

>Having 512Kb will cripple what the core truely needs and
>might to do something with pipeline or latency issue with
>only 512Kb.

Well, it has certainly nothing to do with your understanding of how cpu cores work or perform ;) 

>So in the end I'm still sticking with 'all the Celly are
>crap' rule.

I wouldnt. A Pentium-M with 512 kb cache will be like an improved P3 Tualatin with the double the L2 cache (over the non "s" versions) on a supercharged quadrupled FSB with SSE2 to boot. That just can't be all that bad

>(I mean have we EVER seen any good Celly
>performing up to their rated speed?)

Define 'rated speed', but sure, the tualatin celeries werent that bad at all, and even the venerable CeleronA's werent all that bad, even when you didnt overclock them by 50%. If all they change on the PM Celeron is cutting the cache in half, I'd say that is the chip to get for your laptop.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
Anonymous
a b à CPUs
January 23, 2004 8:40:08 AM

BTW, consider the performance or AMD's A64 3000+ which is a 3200+ with its cache cut in half (like the PM, from 1 Mb to 512 Kb). The performance drop is only a few percent. Same applies if you compare a P4EE with a P4 NW, or even a NW to a willamette.. cache cut in half each time, and just a few percent peformance drop, generally well below 10%.
Duron is ever starker, its got its L2 cache cut by a factor 4 compared to the Tbred, or even a factor 8 (!) compared to Barton. Still a respectable performer though.

The only exception which makes up your rule is the P4 Celeron, which is just too crippled by its tiny cache for clockspeed to matter anymore. I really don't believe this will apply to a Pentium M derived Celeron, as we know the P3 core performns well with only a fraction of that cache.

BTW, IMHO one of the main reasons for including a 1 MB cache to the PM was not performance, but power useage. It costs less energy to keep data in L2 cache, especially an extreme power optimized L2 cache like the PM, than to have to travel over to the northbridge and to the main memory to fetch the data.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
!