Measure your Notebook CPU Speed

One more thing.

There are many more comprehensive benchmarks, SisSoft Sandra, PCMark2004, and AIDA32 contain some. Free to downlad (base versions) some of these will compute cache and SDRAM memory access rates in MB/sec and also integer arithmetic performance. Such numbers would be very useful to this discussion.


-Steve

 

Hi Highlandsun ....

>I'm still a bit perplexed at why my 2GHz Pentium M is yielding >identical run times to the 1.7 and 1.8GHz numbers posted so far.

FWIW I've seen similar results across several benchmarks of the varius Pentium-M( article at Tom's hardware page I think). The 2Ghz -Ms aren't really keeping up on a performance to clockrate basis. I don't know why either.

>My only explanation is that this test is memory-bound.

The 400Mhz memory bus of the bridge (is it i855 ? can't recall the number) used w/ the P-Ms sees to be very inefficient but that problem is masked by the massive cache.

So I agree that it is possible that this test is memory bound on trhe penium-m architecture with it's slow DDR333 memory and very inefficient use of the memory bus at that. The extra CPU power won't help then. The Super_PI test creates severy (5 or 6 as I recall) ~1MB files which contain the several calculated variables and the processor is performing extended precision arithmetic arcoss these files.

> I haven't run it on Linux yet; I'm surprised that the Linux version > only uses a 1MB buffer to calculate the same 2M digits. Anyway,
> since the program allocates a 16MB buffer on Windows, one can only
> assume that its working set exceeds the 2MB L2 cache of the Dothan
> chip.

The 2M digit variables are kept in files as part of the IO system on both Windows and Linux - that memory (or disk space) is not part of the process. The process is performing multidimensional arithmetic across the files while apparently using some small amount memory for buffers only.

The reason why this process requires an extra 15MGB on windows is probably the same reason why a little "hello world" program takes several megabytes on Windows. Even the tiniest app pulls in huge sets of library and system calls which all instantiate large amounts of data space when executed. For reasons I will never understand the Redmond crew has neither the discipline to segregate libries into small incremental bits, nor to instantiate only the resources needed (as the MIT Athena team did the X11). I guess Windows is the sort of OS you get when you you pay programmers minimum wage

>Since I'm using DDR333 CL3 memory, I'm curious about whether some of >those faster times were reported using DDR400 memory.

I've searched high and low and no vendor is using DDR400 on a Pentium-M. Even the Pentium-M motherboards produced in Japan use only PC2100/DDR333 memory. I asked this question on the hardware forum and got a rather uninformative answer from a moderator that DDR400 doesn't work with the -M. No explanation as to why not. It's a very good question Highlandsun and I wish I had an answer.

===

> That tells me there's too much difference between the Linux
> and Windows binaries to really compare them, but it's
> interesting nonetheless.

If I can modify your conclusion - it is a fair comparison as they are performing the same functional computation, but we are no longer just comparing hardware differences but software differences too.

-Steve

 

1:47 here on full cpu speed [ ]

Sony A170 2.0 Dothan
2x512MB PC2700 2.5-3-3-7
80GB 5400rpm HDD

 

<blockquote id='quote'> quote:<hr height='1' noshade id='quote'>
>>Since I'm using DDR333 CL3 memory, I'm curious about whether some of
>>those faster times were reported using DDR400 memory.

>I've searched high and low and no vendor is using DDR400 on a
>Pentium-M. Even the Pentium-M motherboards produced in Japan use
>only PC2100/DDR333 memory. I asked this question on the hardware
>forum and got a rather uninformative answer from a moderator that
>DDR400 doesn't work with the -M. No explanation as to why not.
>It's a very good question Highlandsun and I wish I had an answer.
<hr height='1' noshade id='quote'></font id='quote'></blockquote id='quote'>

They use PC2700 because that's all the 855x chipsets support, and last time I checked nobody else was making a Centrino chipset. []

 

2:01 with AC Power

Pentium M 1.7 Dothan
512 MB Ram
5400 RPM HD
ATI Mobile Radeon 9200 64MB
HP ZT3000

I'm very happy with my results...

 

HP Zv5000z.

Athalon 3000+ 1.8 Ghz 64 bit CPU
512 RAM (2 x 256)
40 Gig HDD
1600 x 1050 display
54G Wireless.

With Power: 1.53 Minutes
With Battery: 2.04 Minutes

Even Linux has started crashing after they started putting fancy GUI. Its a tradeoff between functionality and security.

 

Sony Vaio Z1, 1.5Ghz, Pentium-M, 768MB Ram, 60GB HDD, WinXP

AC :2Min 32Sec
Batt:2Min 35Sec

Power Management Disabled, so both result are from 1.5Ghz

There are 10 kinds of people in the world, understanding binary, and not

 

I'd just got a Samsung M40 (from Asia) and try to execute the tools with strange result ...

Samsung M40 P-M 725 1.6GHz, 512 DDR333 Ram, 60GB 5400rpm

AC powered: 2min5sec for 2million digits

 

ITC Millennia 7200 (Australian branded Clevo)
Pentium 3.4 Northwood 800MHz FSB 512 L2 cache
2x60 Gig 7200 RPM hards @ RAID0
1 GIG DDR400 RAM
256MB Radeon 9700 running a 17” 1680x1050 WSXGA Super Glossy
4X DVD±RW RAM Drive
all the trimmins’ (sans sub and TV card)
XP Pro current updates

some intensive services running in background, like, um this page

AC power: 1:45
Battery? Haven’t quite got the nerve yet to unplug it ;-)



DWTSBOL

 

<blockquote id='quote'> quote:<hr height='1' noshade id='quote'>Originally posted by radeon_x

 

Machine @ work - Athlon 2500+ 1.46GHz 512ram = 3:05

Tecra M2-S730 - Dothan 1.6 (725) 512ram = 2:05

quite interesting that I managed to get 2:05 running test on battery power with Full Performance setting. On AC with same setting it gave 2:10 (!?!) , but when turned to Dynamic Switching regime (on AC) it gave 2:05 again. [?][ ]

 

IBM X31 CC0, PM 1.4GHz, 512M momory, 40GHD, 4200RPM, Battery power notebook
RESULT 2'39"

 

Plugged in: 2 m 09 sec
On battery: 5 m 39 sec



Fujitsu-Siemens AMILO M 1420,1.7 dathon,1 gb ram,ati 9600pro 128,crystal bright+15,4 WXGA over 22 million

http://www.notebooksbilliger.de/images/m1420_big.jpg

 

Celeron D 325 @ 2.53Ghz 256Mb DDR33Mhz
3:25

 

HP nc6000 1.4GHz

512MB 2:27
768MB 2:36

For comparable results you can try running "msconfig" from the "run..." in the "start" menu and clicking diagnostic setup. This will disable most of the background processes and after restarting you will have a much better and comparable result.

HP Compaq nc6000

 

After reading all the thread I found some quite interesting questions. I have been oc for a long time and been using super-pi for comparisons between different settings.

1) Super-Pi calculation does not measure only the CPU processing power. Lots of parameters are involved. Memory timings are the most important (you can improve as much as 10% from 2.5-3-3 to 2-2-2), even the motherboard itself. So don't be surprised if your faster cpu does not translate in better results.

2) Try to use the same settings. I suggested running in diagnostic setup so that we are all running with the same background processes.

3) Use CPU-Z utility to check your memory timings etc. VERY IMPORTANT to know your memory timings and any dividers.

4) As soon as we start sending the updated results can someone make a spreadsheet and update it with the new info so we can compare in once?

HP Compaq nc6000

 

Emachines M6811, factory stock.

1:42

 

Just ran the Super Pi 2M test on my old Sony VAIO and got 6:42 (that was with all of my normal processes running and the AC plugged in).

Sony VAIO PCG-FX190K, PIII 850MHz, 256MB PC100 SDRAM, 30GB HD, Win2k OS (this was a state of the art $3400 machine when I bought it back in June of 2001!).

Plan on buying an Inspiron 8600 soon, just trying to see if the 2.0 GHz Dothan is worth the cost over the 1.8 GHz version.

 

Well I finally chose an IBM T42, 1.7Ghz Dolthan and
got 1:53' on AC.

 

following tristeno's suggestion, i just reran superpi in diagnostic mode, my previous score was 2.39

new score: 2.25

acer aspire 2012WLmi
pentium-m 1.5 (banias)
512 ram
60 gb hdd
ac on

 

<blockquote id='quote'> quote:<hr height='1' noshade id='quote'>Originally posted by stevea

 

Sorry I messed up, only tested to 1M on previous post.
I have an older Fujitsu P series with a transmeta chip running at 859 Mhz with 368 megs of sd ram.
scores for 1 Million again
plugged in 03m 35s
unplugged 03m 34s
and now for 2 Million
plugged in 08m 02s
unplugged 08m 02s
interesting that it takes an etra 4m 28s to calc the second million, any thoughts?

 

I have an older Fujitsu P series with a transmeta chip running at 859 Mhz with 368 megs of sd ram.
My processor regulates its voltage depending on the task, not whether it is on battery, so my battery test was actually 1 second faster.
scores
plugged in 03m 35s
unplugged 03m 34s

 

Stock m6805 (AMD 64 3000+, 512 MB Ram, 5400 RPM HD) 1 Min 57 Sec AC, 3 Min 55 Sec Bat. 1 Min 57 Sec on Bat with Power Now! tweak...

First PC = Vic-20, 8" 128KB Home-Brew Floppy Drive ... Current Rig = eMachines m6805 ... Quantum Leap?

 

Just tested my m6805 in safe mode and it took 14 seconds longer! (2 M 11 S)

First PC = Vic-20, 8" 128KB Home-Brew Floppy Drive ... Current Rig = eMachines m6805 ... Quantum Leap?

 

<blockquote id='quote'> quote:<hr height='1' noshade id='quote'>Originally posted by vaguej

interesting that it takes an etra 4m 28s to calc the second million, any thoughts?
<hr height='1' noshade id='quote'></font id='quote'></blockquote id='quote'>

Yeah - the extended math operation for multiply will take about 4 times as long as you double the number of digits. Addition and subtraction just increase by a factor of two.

On my P4 system going from 1M to 2M digits takes 2.21 times as long. You system takes 2.25 times as long. Completely normal.

 

Hey - Just a general note to express my surprise at how fast the Dolthan is.

I have Linux installed on my 1.7Ghz Dothan system (ibm 2378fvu) and in comparison also on a 2.6Ghz P4HT i875 chipset server system with some PC3200 memory. The server has faster memory access tho much less cache.

Anyway in comparing Linux kernel builds (the hyperthreading is not in use for this) the Dolthan laptop beats the pants off the 2.6Ghz P4HT. The huge set of compiles and links is about 2.4 times as as fast as the P4 server and the Dolthan uses less than 1/3rd of th power ! Even if hyperthreading the build (easy to do under Linux) cut the execution time in half (and it won't) the Dothan still wins.

I can't fully explain this result except to say there has to be some 'special sauce' in the Dolthan processor architecture.

Too bad you can't (practically) get the Dolthan on a motherboard !

 

Hyperthreading is more Hype than anything else.

old: Sony PCG-GR300P 1.13GHz PIII-M, 512MB
new: Asus M6Ne 2.00GHz P-M, 2GB

 

<blockquote id='quote'> quote:<hr height='1' noshade id='quote'>Originally posted by highlandsun

 

If "make -j2" is giving a huge gain, you should also consider "gcc -pipe" ...

Hm, I started working on Alliant's TCP/IP stack in '91 so you have a few years on me there. By the way, I designed the current "make -j" (job server) support in GNU make and wrote the first implementation of it, also back in '91.

For something like building large software packages, you're not really doing parallel processing in the classical sense. Yes, it works wonderfully well, but it's really treating each CPU as a separate, tightly networked system, not a shared memory multiprocessor. And on a large software package and a true multiprocessor machine, I find processing speed scales linearly with CPUs up the the limit of the system's I/O bandwidth. On an Alliant, that was a pretty big number, so parallel builds were truly 8x faster on an 8 CPU box. The fact that hyperthreading only gets you a 35-65% gain shows the miserable inefficiency they're getting from two execution units on the same core, where they can't even blame a slow inter-board bus as an excuse.



old: Sony PCG-GR300P 1.13GHz PIII-M, 512MB
new: Asus M6Ne 2.00GHz P-M, 2GB

 

super pi results for my Fujitsu Siemens Amilo M 1420 ;


for 512 kb : 00:19 seconds
for 1 mb : 00:49 seconds
for 2 mb : 02m:03 seconds



Fujitsu Amilo M 1420-1.7 Dathon,1gb(333mhz),Ati 9600Pro 128
  15,4 " crystal view WXGA (23 million),7200 rpm,DVD+RW+R    
                                         M 1420 photo

 

Super Pi results for my MBS WinSystem:

Athlon 1800xp on nForce rev. E Mobo, 512MB DDR266 (sharing 32MB to the integrated GeForce2mx graphics), 8GB 5400rpm IDE HD & 160GB 7200rpm ext. FireWire HD, Win2000 Pro SP4 (slipstream install).
On the 8GB IDE HD: 2:49
On the 160GB FireWire HD: 2:51


* Also, I ran Super Pi on a VIRTUAL WinXP computer in VMWare 4 on the above Win2k PC:
All the following hardware spec is, of course, virtual --> 1533MHz CPU on an Intel Mobo, 384MB SDRAM, 4GB 5400rpm IDE HD, WinXP Pro SP1 (slipstream install).
On the virtual 4GB HD: 3:09


That's not half bad, I guess, for a PC running INSIDE a PC


When my zv5000z arrives in a few days, I'll post the results from that as well. I'm interested to see how an Athlon64 lappy with a 5400rpm HD fares on this test.


Cheers,

Stevo

zv5000z: Ath64_3200+ (2.0GHz) - 1GB DDR - 80GB 5400rpm - 64MB 440 - 15.0" XGA - Win2k (tentatively)

 

Interesting thread. I'm about to order an Inspiron 8600 to replace my old Gateway 9100, P2, 128MB, 6G HD, W98.

So for history buffs:

Gateway:
1M - 9min, 47sec
2M - 21min, 03sec

I'll post the new numbers in a week or so.

Ken

 

Just arrived via FedEx: Sony Vaio PCG-GRT360ZG
Intel Pentium 4-M, 3.06 Ghz
HD: 80GB Enhanced IDE
Memory: 512MB DDRAM
Optical Drives: DVD+-RW / CD-RW
Cache: 512 KB
16.1" SXGA+TFT (REZ: 1400 x 1050)
Graphics Card: NVIDIA GeForce FX Go5600, Giga Pocket w/TV Tuner & a ton of other neat stuff.
Integrated Wireless LAN 802.11b/g
Integrated V.90 Modem

Just pulled out of the box yesterday and will run CPU speed test shortly. Looks to be a racehorse, so I named it "Seabiscuit"[ ][ ]

"IF IT AIN'T BROKE, DON'T FIX IT!!"

 

<blockquote id='quote'> quote:<hr height='1' noshade id='quote'>Originally posted by Andrew

 

Averatec 6130H1
P4 3.0Ghz HT
512 PC2700DDR
40GB (more than likely 4200rpm's)

A/C - 2:28
Battery - 4:15

-Blake

 

Specs are in sig:

2M - 1m:39s

Inspiron XPS | P4HT 3.4GHz EE | 1024MB Corsair XMS PC3200 | 60GB Hitachi 7200rpm | 256MB ATI Radeon 9800 oc 459/378 | TrueMobile 1350 802.11b/g

3DMark03: http://www.geocities.com/mnguyen281/score.jpg

 

I ran the 32M test and finished it in 36 minutes. Anyone care to try? [ ]

Inspiron XPS | P4HT 3.4GHz EE | 1024MB Corsair XMS PC3200 | 60GB Hitachi 7200rpm | 256MB ATI Radeon 9800 oc 459/378 | TrueMobile 1350 802.11b/g

3DMark03: http://www.geocities.com/mnguyen281/score.jpg

 

1min 43sec with a bunch of stuff running in the background and downloading aquamark3... I'll do it again later and see how different it is.

mnguyen281 what did you use to OC your 9800M!!!!! And what are you doing to keep it Cool!? Also do you think the cosair mem makes any difference???

"To Be Great Is To Be Missuderstood" ~Unknown

Dell Inspiron 9100
Pentium4 3.2GHz Desktop
1GB DDR 400MHz
60GB Ultra ATA 7200RPM Hard Drive
ATI Mobility Radeon 9800, 256MB, 256-Bit Graphics Card

 

Plugged in: 1.56
HP nx7010 1.7 Pentium M 768MB 40GB HDD

Plugged in: 18.47 [ ]
HP Omnibook 333 Pentium II 256MB 6GB HDD

 

Both plugged-in and unplugged were the same: 2:13

Using Dell 8600, 1.4MHz with 512M and 40G HD, WinXP-Pro, Radeon 9600.

This compares well with my former laptop - a Gateway 9100 solo, P2, 233MHz, 128M and 6G HD running Win98 which cranked out over 21min. [8D]

Ken

 

AS1861wlmi

Dothan 1.5
512 ram
60 GB (5400 rpm?)
MR 9700

IM running, couple windows open...unplugged (battery power)

2M : 2:10

only 7 seconds slower than 1.7 Dothan...

romulus

 

I don't know what I was smoking...

it's an AS1681 and not 1861...

and it's 11 seconds slower than the 1.7 Dothan...

need to get some coffee.

 

Specs in signature...

Plugged In: 2m 01s

ASUS M6Ne | P-M 1.8Ghz | 1Gb Ram | 80Gb 5400rpm HDD | Radeon 9700 64Mb

 

Specs in Siggy...

1m 56sec plugged in *cheers happily*

4m 52sec on battery power (CPU throttles down to 793 mhz; PowerNow! NOT activated) *hangs his head in shame on the battery test*

Well, that was fun....

HD, hoping Fable comes out on PC...doubts it though...

Proud 0wner of the Emachines 6809 Notebook. Scratch that...Proud 0wner of the Emachines 6811 Notebook, thanks to some cracked hinges and a minor technicality in my three year Best Buy warranty.
Athlon 64 3400+ Proc. @ 2.2 Ghz., 512 RAM, ATI Mobility Radeon 9600, Overclocked @ 425.25 Core/222.75 Memory. 80 Gig. Hard Drive @ 4200rpm. DVD +/- RW Drive (Write Max: 2.4x DVD+R/RW, 2x DVD-R/RW, 6x CD-R, 10x CD-RW; Reads 24x CD, 8x DVD) 6-1 Media Reader, 802.11g Wireless...now your playing with power...super portable power!

 

i got 2:04 on both the laptop and the desktop [amd64 2800, 1 gig pc3200 ram]. didnt feel like turning off desktopx tho.. so basically running with 1/2 ram on both.

--Inspiron 8600, 1.8 ghz, wsxga+, 60 gb 7200 rpm, 2*512mb pc2700, 128 mb radeon 9600--

 

Inspiron 8600; 1.8 Ghz Pentium M(Dothan); 1 GB RAM; 60GB@7200,ATI9600 128 MB
Power On: 1m44sec
Battery :1m44sec
I couldnt change this results tryed with different priority in the Task manager but always the same.
I had the next programs runnig during the test
ICQ, BulletProof FTP server with 4 users connected, Norton Antivirus, Norton Internet Security,Some Dell Support Programs( that i have to get rid of), Logitech Mouse software for my Logitech Laser MX1000, Dell QuickSet
I will install my PC later and will try with brand new instalation and nothing runnig on the background

 

1:50

LG LM50 , Dothan 745 (1.7mhz), 512ram - running 'Super PI' with wine under linux

 

Desktop: 2.6 Intel HT, 512MB, 120GB7200RPM-----> 2:27
Laptop: AMD 2800(XPM), 512MB, 30GB4200RPM----> 2:13 AC plugged in
4:03 battery used

 

Novatech Wildfire (P4 3.2Ghz, 512Mb, 60 Gb HDD)

Managed 2:07 plugged in.....