Measure your Dual/Multi Core Notebook CPU Speed - Bye Bye to SuperPI

I am developing a parallel benchmark suite as part of my research.

 

HP 8710w Mobile Workstation
Intel Core 2 Duo T7500 2.2GHz
4096MB DDR2-667
Windows Vista Business

 

Lenovo T61
Intel Core 2 Duo T7500 2.2GHz
4096MB DDR2-667
Vista Premium 64-bit

 

hey chaz i hope thats not you trying to steal my slot in the charts

i worked hard to get to where i am

oh by the way my brother emailed this to me its a score from his laptop he had loaned to him from his boss (he works in a fujitsu/uk shop selling laptops) tho he said it an engineer sample

also im not sure i believe him as i have not seen any amd tl64 or even tl66 yet available yest only advertisments for pre-order. also he said his ddr2 667 was cl3 i think hes talking $"£"$" as again i havent even seen any cl3.

so dont use this but look anyway and make your own mind up.

then again i might be wrong but as i always say seeing is believing..

which after seeing this laptop i might start 3 hardrives and sli graphics

http://www.pcmicroworks.com/blackhawk.htm

 

Yeah, that screen shot certainly appears to be edited...

 

I do believe that that score is not genuine.

No mobile CPU has broke the 38.000s mark.

So I would not post any scores that are not reasonable.... as well as the screenshot looking shady.

If that is a real score, please Submit it, and link it back to me.

 

i knew it was dodgy when i seen it but i thought id ask first

anyway as i stated i looked around and couldnt locate a tl64/66 or cl3 ddr2 667 memory so i knew it wasnt real.

ill get my own back on him later

cheers Gophn and odin for clearing that up but it looked real to me and i didnt notice any alteration on the jpg.

 

What about the score i posted above Gophn?

http://forum.notebookreview.com/showpost.php?p=2215256&postcount=303

 

Oh yeah, you're right. I havent updated the charts yet.

Nice score BTW

I stand corrected.

No mobile CPU has broke the 37.000s mark

 

Untill next week then itll be 36000s

 

Thanks.

Not trying brag or anything, but i probably could have posted a slightly better score but i didnt want to disable aero and run it in Vista Basic or Windows Classic mode.

Edit: Also, my scores were double what they are now when my CPU Maximum Frequency was mysteriously capped at 54% due to a logical error with Vista's, and or, Lenovo's Power Management when running only AC outlet.

 

tho im still impressed with mine doing 38

do you think itll get better score wise when i take the plunge with v64 next week?

been holding off doing stuff like downloading and other things untill the moment was right

 

Okay something wierd.....
32M test 4 threads, = this score of 27.58s.

32M test 2 threads, = this score of 28.5s.

1024 stress test with 2 threads is 908.5s, here.

1024 stress test with 4 threads is 904.47s, here.

Why does going from 2 threads to 4 threads give me almost a second decrease in time?

Using Vista x64 for my operating system.

 

Interesting.... that you can use more threads than your CPU has to offer.

the score increase is very slight... which makes me think that forcing more threads to be calculated on a dual-core.... squeezes as much number-crunching juice out of the CPU.

I would be interested to see if a mobile CPU would be able to do the same.

I know some that got BSOD's or crashes when they tried this.

 

Yeah, a bit strange that why I was asking if it was normal. It does better on 4 threads than 2. It says in CPU-Z that its a X6800 chip and not a QX6800 and my laptop hasn't melted yet so.....

And no BSOD when running it with more threads.

 

Try running the test several times. Compare the averages for each test and also look at how much the different runs differ by.

Has Intel started producing dual core procs with hyperthreading yet? If so, that could explain it.

 

My guess = more threads = more cpu priority. Its not like there is nothing else running in the background.

Could be a flaw in the program too tho, letting the cpu get away with less work.

 

Intel Core 2 Duo CPU T7300@ 2.00GHz
Sager 5790
Win XP Pro

The lowest I've gotten so far tonight was 45.901s for the 32M test...

I'm still working this computer out as it has a lot of minor issues right now, and I'm a big laptop newbie.

 

Same thing Gophn, if I use 2,3,5,6,7,16,32,64 I get the 28.5s score. on the 32M test.
If I use the 4 or 8 thread count I get almost a seconds difference, i.e. the 27.5s score.

 

So, I re-tested my G1s (T7500) and got a score of 38.468s (More than a whole second faster than my first test).

I wonder why this is? Since the first test, all I've done is update my graphics driver to 163.11, but that shouldn't have any effect.

 

I wanted to throw my name into the hat. My Pentium M gets 98.702 on the speed test.

</br>

</br>

I'll test my Pentium D server tonight!

 

He,

want to throw my numbers into the battle.




exxxer

 

So I hope the charts can be updated because I worked all day to achieve this number lol. (even though this is going to be the last on the charts!)

Also, I understand why we need the Core architecture. Pentium 4 just don't cut it.

Ok so when I saw this poll and the section for the desktop processors, I thought about running my PD 940 for the gold. To see if it came any close to the Core design.

At stock my PD (3200mhz) gets 61.265s on the 32M test. I thought that was decent for a Pentium D, so now time for the good stuff.

First off, I have never overclocked before, so this was a cool way to learn how to o'clock. Just like any standard overclocking, I started cranking on the fsb of the processor, dropping the dram freq. to keep the ram speed from getting out of hand. After I got to 4ghz, I thought about quitting. I nearly did because I got the blue screen during boot. I really didn't want to do it, but I raised the vcore to 1.3v. That breathed new life in my venture for the max overclock. On a side note, you have to increase the vcore alot after the 4.0ghz level.

So at the end of my overclocking journey, I achieved a stable 4389mhz at vcore 1.3875V . The 32M test score dropped to an impressive 43.562s.

The facts: </br>
3200mhz to 4389mhz (a 1.189ghz increase!) </br>
61.265s to 43.562s (17.703s decrease!) </br>

There you have it. No match for the Core2 duo, but it still has a running for its own glory!

 

Wow, that's quite an increase. That's comparable to a 2 GHz Core 2 (Pentium D is dual-core, right?). Although I'm sure the power consumption and heat production are much higher after the OC. And you can't buy a 4.389 MHz P-D in the store. Still, really nice performance!

I'm still waiting for someone who has a P-4 3.8 or a Gallatin 3.73 overclocked. To get this out of a 3.2 is amazing!

Tried to run wPrime on my new Vista machine (see sig) but got an error. Will post error once I get my new machine out again.

 

you have to run it as admin to get it working otherwise you will get an error message.
right click on the link or shortcut and click run as administrator.

 

Yup, it's dual core!

At vcore 1.4125V, I finally hit the 4.4ghz mark. http://valid.x86-secret.com/show_oc.php?id=222073 is the picture of the 4.4ghz mark. However, I didn't run the 32M test because I was scared to what would happen (Intel says the max vcore before damage is 1.4V)!

You are right about the temps though. During the 4.389ghz test, I idled around 52C and topped out at 68C... The Pentium D max temp is 68.8C before throttling back. I came close lol. My little Zalman CNPS9500 can't cool enough. Maybe it's time to go water.

Question though: Is running a processor at 60C full load going to hurt the processor in the long run? I can run 3.7-3.8ghz at 60C full load.

 

Strangely enough, I tried running WPrime with 3 threads on my Turion X2 TL-52, and it was consistently quicker than running WPrime with 2 threads. Attached is a screenshot. 4 threads was consistently slower.

 

@ ccbr01 - your scores seem high for a processor overclocked to 4.4GHz . . . did you do the multicore hotfix for XP?

Perhaps it's just the Pentium D.

 

Patch and Pentium D. I did the reg hack a couple of months ago on this thing for the load balancing. The times did improve with the patch. 32M is now at a faster 43.297s. Just 3 tenths better. That is at 4.389ghz with a 1.4V vcore (room temps are higher today).

Edit: So playing around again today, I underclocked the processor to 1.6ghz (the slowest bus speed I have!). It kinda reminded me of the rdram days of the Pentium 4. On 32M, I achieved 124.984s.

 

Ok folks. The table is getting larger and larger. I appreciate all of the participation.

As for the new champs: (.... for now)

(for notebook with desktop CPU)
Wu Jen - PCMicroWorks BlackHawk (Clevo D900C)

		Processor (CPU)	System Type/Model	Operating System	Score (32M) - Default Test	Score (1024M) - Stress Test
1.)	Wu Jen	Intel Core 2 Extreme CPU X6800@ 2.93GHz	PD: "BlackHawk" (Clevo D900C)	[COLOR="Blue"]Vista Ultimate[/COLOR]	27.58s	904.47s

________________________________

(for notebook with mobile CPU)
hypertrophy - Lenovo Thinkpad T61

9.)	hypertrophy	Intel Core 2 Duo CPU T7500@ 2.20GHz	NB: Lenovo T61	Vista Home Premium	37.705s	1202.666s

 

Oh dear.



I'm using Windows XP.

 

I think I might have to run wPrime on my Athlon 64 Mobile for giggles....


Witch: re Trident Cyberblade - it is without doubt the worst graphics card ever made! My old Toshiba Portege 4010 had the 16MB version (shared PC133 ram) and animated fish background on msn made the cpu (PIII 933MHz ULV) temperature jump by 20 degrees C! Worthless IGP, really awful. I'm sorry you have to deal with it.

 

New score of 38.913 sec.

AMD Turion TL-60 x2 running Vista Premium.

 

The new top score Gophen for Wu Jen that you put up used 4 threads, vs the 28.5 he got with 2 threads. I think you should put the 2 thread score up as thats the standard. Otherwise everybody will start adding more threads to the mix and getting new scores and we wont have a set standard anymore.

Im sure he wont be upset it still gives him #1 spot. That leaves me as #2 on both catagories now :/

 

true... i'll do that.

But it does say Dual/Multi Core (for notebooks)... so it might be fine as it is.

Although I would not recommend more people to try this, because it might cause instabilities.

 

Sympathy appreciated. I remember that it was relatively poor even at the time that I got it in late '01.

But I won't have to deal with it much longer: I'm getting something new in the next few weeks, it'll be very basic but it'll do the job for a few years (probably not nearly 6 years, I delayed this replacement for way too long).

 

97.765. Not bad!

I'm running a Clawhammer Core (130nm) Athlon 64 Mobile 3000+ at 1.8GHz, 800MHz hypertransport and 1MB L2 cache.

 

...and from the Toshiba 128.812.

That's a Dothan Celeron-M 380, 1.6GHz, 400MHz FSB, 1MB L2 cache


Edit: both on XP pro SP2 by the way!

 

1503.516 1024M

Windows XP w/hotfix



EDIT: sorry.... 47.343 for 32M

 

Updated.

Thanks for the posts.

Remember to submit it into hwbot.org for Team NBR

 

Tried it out on my
Lenovo T61
Vista Home Premium
T7300@2ghz

42.231 for 32M

Probably could have done better if I disabled some more of my running tasks but I am lazy (only turned off antivirus for a 5 second gain )

 

Ferrari 1003 TL52-1.6Ghz
32M = 50.33seconds
1024M = 1621.648 seconds.. )

 

Just managed to get this score and my CPU temps were about 80'C

13.3" ASUS W7S-B1B
T7500 @ 2.2GHz
Vista Home Premium 32-bit

32M = 36.862
1024M = 1183.651

 

I agree there should be a column for # of threads used. Also there needs to be a column for the FSB speed. Since the C90 is customizable and you can bump the FSB up to 1278 that will make an impact. My poor old D901C is locked in at 1066FSB

Add in Gophn a Column for # of threads used and a FSB column and that should plug up any holes for standards, at least for now lol.

 

Also a columb for cache? I tink that's getting a bit much, you can just look up things like that if you know the cpu.

 

No, the point was that with the C90s, the FSB is overclockable, so you don't know what FSB was used for that score.

 

If we aren't going to add more column's then there needs to be something done to differentiate it.

Is the program really testing the CPU? If so why does the thread count matter?

Bumping up your Front Side Bus speed will increase your processor speed. I.e. your OC'ing. That needs to be shown also. Laptops are hard to overclock but not impossible.

 

The thread count matters because if there are more threads, the system will give a little extra priority to the tasks, and it will run slightly faster.

 

As does increasing the FSB. But as to the # of threads then why when you go to certain increased thread #'s does it give no gain vs 2 threads. Why going to 4 does? Somethings fishy.