Nevermind

Message

LtTPfan · #1 Post by **LtTPfan** » Fri Dec 09, 2005 6:50 pm

LtTPfan · #2 Post by **LtTPfan** » Sat Dec 10, 2005 8:09 am

Website · #3 Post by **tfflivemb2** » Sat Dec 10, 2005 10:22 am

I believe that there is a thread here somewhere that discusses the upgrades possibilities for the T30. I know that it will take a 2.4ghz, but the Celeron is what is throwing me off. I am keeping this option in mind for the T30 that I just picked up (currently a 1.8ghz)

LtTPfan · #4 Post by **LtTPfan** » Sat Dec 10, 2005 2:58 pm

LtTPfan · #5 Post by **LtTPfan** » Mon Dec 12, 2005 6:15 pm

a31pguy · #6 Post by **a31pguy** » Mon Dec 12, 2005 7:21 pm

So you want to replace a P4-M with a celeron? Why? I mean you could do it in theory on the 845MP chipset - but the BIOS would have to support it. With a celeron can expect worse performance. The L2 cache is smaller and so the efficiency per clock cycle is worse. So while it's a 2.4 ghz clock - it does less per clock cycles than the equivalent speed of a P4-M.

Cache sizes of processor
-----------------------------
P4 Celeron - 256 Kb L2 cache
P4-M - 512 Kb L2 cache
PM - 1 Mb L2 cache

LtTPfan · #7 Post by **LtTPfan** » Mon Dec 12, 2005 9:30 pm

a31pguy · #8 Post by **a31pguy** » Mon Dec 12, 2005 9:50 pm

That's exactly what I'm saying. Why do you think they are so cheap?

http://www.cpuscorecard.com/all_cpus.htm

In addition to worse performance they don't have all the multimedia instructions like SSE2 (128-bit instructions) that the pentium 4-M have.

EDITED - apparent has SSE2 instructions.

But if you're seeking a bump in performance - try upgrading the RAM and hard drive first. They are the best bang for the buck. 5400 rpm or 7200 rpm disk and at least a gig of ram.

My wife has the T30 and there was a significant bump in performance from 256 mb to 1 gb.

LtTPfan · #9 Post by **LtTPfan** » Tue Dec 13, 2005 8:23 am

a31pguy · #10 Post by **a31pguy** » Tue Dec 13, 2005 12:48 pm

My apologies. I stand corrected. In the middle of a project and trying to post on the breaks. You know you could just buy the cpu and give it a shot and post the benchmarks.

LtTPfan · #11 Post by **LtTPfan** » Tue Dec 13, 2005 10:28 pm

R51-Smashedbanana · #12 Post by **R51-Smashedbanana** » Wed Dec 14, 2005 10:08 am

Well I'd have to agree with what's been said about Celeron being slower than P4. Desktop Northwood P4 vs. Celeron performance reviews are abundant.

That said I'd also point ot that 512k P4-M are super cheap on E-bay, not necessitating a Mobile Celeron to save money.

Here's an example, a 2.2ghz P4-M: http://tinyurl.com/9ul3e

I'm not affiliated with the seller, just using his auction as an example.

Ed

#13 Post by **underclocker** » Wed Dec 14, 2005 3:07 pm

FYI-

The example listed is actually a Pentium 4 NON-mobile cpu. The seller has made a mistake - here is the Intel info based on the spec listed by the seller - http://processorfinder.intel.com/script ... CorSpd=ALL Being careful on eBay is the number one rule.

Mobile P4 2.2's usually sell for much more than this. The thought of using a Mobile Celeron is a good one, if it works, it should be quite cost effective. The Mobile Celeron 2.4 is quite fast.

The biggest down side would be lower battery life as power saving is crippled on Celerons.

My R51 came as a Celeron M 1.3 machine and I was able to pop in a Pentium M 1.5 with no trouble at all. Yes it's a different beast, but the concept is similar.

Plesae post results, it will be very interesting. Hopefully, you won't blow the system board, or "planar", as IBM is fond of calling them.

Good luck.

underclocker

LtTPfan · #14 Post by **LtTPfan** » Wed Dec 14, 2005 4:01 pm

a31pguy · #15 Post by **a31pguy** » Thu Dec 15, 2005 4:51 pm

My R51 came as a Celeron M 1.3 machine and I was able to pop in a Pentium M 1.5 with no trouble at all. Yes it's a different beast, but the concept is similar.

How were you able to get a M into a R51?

EDIT: Nevermind - the M was an optional processor for the R51. So the Bios and chipset would support it.

wolfman · #16 Post by **wolfman** » Thu Dec 15, 2005 8:28 pm

Interesting discussion above regarding the Pentium 4-M versus the Pentium 4-M Celeron (512kb cache vs 256kb L2 cache). I think the performance estimate above is off (a 1.6 gigahertz P4-M will outperform a 2.4 gigahertz P4-M celeron). Here is a thread from a long while back in which you can see the P4-M 2.4 gigahertz (512kb L2 cache) benchmarks vs a P4-M Celeron 2.0 (256kb L2 cache) vs a P4-M 1.7 gigahertz (512kb L2 cache). The interesting thing is that the 2.0 gigahertz P4-M Celeron falls pretty much right where it should based on megahertz.

Here is an abbreviated comparison:

P4-M 2.4 gigahertz w/ 512kb L2 cache (Intel i845MP chipset)
Addition/Multiplication (SSE) : 3673.62 MFlops
Division (SSE) : 629.82 MFlops
Square Root (SSE) : 630.17 MFlops
Addition/Multiplication (FPU) : 1699.32 MFlops
Division (FPU) : 54.81 MFlops
Square Root (FPU) : 55.66 MFlops
Whetstone : 380 KWPS

P4-M Celeron 2.0 gigahertz w/ 256kb L2 cache (Intel i845MP chipset)
Addition/Multiplication (SSE) : 3080.11 MFlops
Division (SSE) : 524.36 MFlops
Square Root (SSE) : 524.12 MFlops
Addition/Multiplication (FPU) : 1411.82 MFlops
Division (FPU) : 45.61 MFlops
Square Root (FPU) : 46.27 MFlops
Whetstone : 316 KWPS

P4-M 1.7 gigahertz w/ 512kb L2 cache (Intel i845MP chipset)
Addition/Multiplication (SSE) : 2601.38 MFlops
Division (SSE) : 444.33 MFlops
Square Root (SSE) : 446.07 MFlops
Addition/Multiplication (FPU) : 1201.31 MFlops
Division (FPU) : 38.72 MFlops
Square Root (FPU) : 39.4 MFlops
Whetstone : 264 KWPS

http://forum.thinkpads.com/viewtopic.ph ... +r40#69796

I think the Celeron with 128 kb L2 cache is the real dog and I'd agree that the P4-M 1.6 gigahertz would smoke a 2.4 gigahertz celeron with 128kb L2 cache.

On each benchmark I did with my 2.0 gigahertz celeron it came in around 1-2% and at most 5% under the performance of a 2.0 gigahertz P4-M with 512kb L2 cache. YMMV...I'd have no reservations recommending the 2.4 gigahertz P4-M celeron w/ 256kb L2 cache if it's 1/2 the price of the 512kb model...

#17 Post by **underclocker** » Thu Dec 15, 2005 9:18 pm

This is a good thread that will hopefully yield some nice test info. It would be nice if the A31 was able to accept the Mobile Celeron, as we are hoping the T30 will.

As far as Intel reports, all Celerons above 933MHz have 256KB caches (which would make a 2.4GHz with 128KB cache slow AND very rare!) See this link -> http://processorfinder.intel.com/script ... kgType=ALL

I think a Mobile Celeron 2.4GHz cpu (w/256KB cache) would be quite fast...just not battery efficient.

LtTPfan · #18 Post by **LtTPfan** » Fri Dec 16, 2005 5:32 am

wolfman · #19 Post by **wolfman** » Fri Dec 16, 2005 9:49 am

Thanks for the link to the Mobile Celeron models at Intels site.

Intel did, however, make a desktop celeron 2.4 gigahertz with 128kb L2 cache that found it's way into desktop replacement laptops and was a dog. I had that one in mind, but wasn't clear in my post, sorry for the confusion

. Here is the link to the product:

http://processorfinder.intel.com/script ... CorSpd=ALL

Steve

a31pguy · #20 Post by **a31pguy** » Fri Dec 16, 2005 8:59 pm

The benchmarks we posted back then are interesting and brings up an interesting point.

The relationship between cache and performance. The Pentium M vs. Pentium 4-M vs. Celeron-M.

The pentium Ms measured were not clocked at full speed, rather on their speedstep settings which were set to adaptive. So while their floating point and integer measurements were low - the whetstone numbers showed better performance then the other metrics showed.

It would be interesting to see a pentium 4-m at 2.0 ghz vs a 770 pentium M at 2.0 ghz vs. a celeron at 2.0 ghz.

Sidenote - the difference between a P4-M 2.4 and a 2.6 P4-M is about 6%
so a the numbers of a Celeron and P4-M of significant speed of 2-5% is still significant enough to say it performs worse IMHO - but probably not to usability.

The way I've been testing CPUs of recent is to compile code in C using gcc that computes PI to the Nth digit. Run the code on both and compute the value to 100,000 digits and compare the run times.

LtTPfan · #21 Post by **LtTPfan** » Mon Jan 09, 2006 3:58 pm

LtTPfan · #22 Post by **LtTPfan** » Mon Jan 09, 2006 7:51 pm

wolfman · #23 Post by **wolfman** » Mon Jan 09, 2006 10:18 pm

Excellent, glad to hear that worked in your T30. The benchmarks you show are pretty much in line with what I ran myself before when I was considering replacing my P4-M 2.0 celeron w/ 256k L2 cache with either another celeron @ 2.4 or a straight P4-M 2.4 w/512kb cache. For me the extra 256k cache didn't warrant the price difference...however I'm probably going to hold off another year till my warranty runs out before I switch...LOL

Nevermind

Nevermind

Who is online