Cache is King, Pentium-M CPU is the fastest!!!

Message boards : Number crunching : Cache is King, Pentium-M CPU is the fastest!!!
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · Next

AuthorMessage
Heaphus
Volunteer tester

Send message
Joined: 1 Apr 03
Posts: 96
Credit: 4,148,549
RAC: 0
United States
Message 104972 - Posted: 29 Apr 2005, 0:27:21 UTC

Here is a nice comparison of two of my laptops, using Boinc 4.19.

1.) Dell 600m: 1.6Mhz Pentium M(1MB cache, 400Mhz fsb), 512MB DDR - 2:40/wu

2.) Dell 6000: 1.6Mhz Pentium M(2MB cache, 533Mhz fsb), 512MB DDR2 - 2:20/wu
ID: 104972 · Report as offensive
Profile Steve Cressman
Volunteer tester
Avatar

Send message
Joined: 6 Jun 02
Posts: 583
Credit: 65,644
RAC: 0
Canada
Message 104975 - Posted: 29 Apr 2005, 0:32:36 UTC - in response to Message 104972.  

> Here is a nice comparison of two of my laptops, using Boinc 4.19.
>
> 1.) Dell 600m: 1.6Mhz Pentium M(1MB cache, 400Mhz fsb), 512MB DDR - 2:40/wu
>
> 2.) Dell 6000: 1.6Mhz Pentium M(2MB cache, 533Mhz fsb), 512MB DDR2 - 2:20/wu
>
Lower the bus speed on the dell6000 to 400Mhz and try it again, see what you get then.

98SE XP2500+ @ 2.1 GHz Boinc v5.8.8

And God said"Let there be light."But then the program crashed because he was trying to access the 'light' property of a NULL universe pointer.
ID: 104975 · Report as offensive
Steve MacKenzie
Volunteer tester
Avatar

Send message
Joined: 2 Jan 00
Posts: 146
Credit: 6,504,803
RAC: 1
United States
Message 105034 - Posted: 29 Apr 2005, 3:23:51 UTC

Ulli,

Also only for Info
My P4 3.6 with HT and 1GB RAM does a Seti WU in
2:45 for two WUS
Hmmmmmm Why is your's 2:35 for the 2 ?
I would have expected just slightly better.

Im think as an experiment, I'll change a preference
and see what happens.

First. Increase disk usage from 60 sec. to ~1200 ( 1/20th )

I'll keep you posted.

Steve


ID: 105034 · Report as offensive
Steve MacKenzie
Volunteer tester
Avatar

Send message
Joined: 2 Jan 00
Posts: 146
Credit: 6,504,803
RAC: 1
United States
Message 105036 - Posted: 29 Apr 2005, 3:27:27 UTC - in response to Message 105034.  


Bummer.
No schedulers available.
I thought that was supposed to be earlier today ?
Oh well.
Will try the experiment tomorrow.

S



ID: 105036 · Report as offensive
Matthew Williams

Send message
Joined: 7 May 99
Posts: 3
Credit: 148,519
RAC: 0
United States
Message 105074 - Posted: 29 Apr 2005, 5:34:09 UTC
Last modified: 29 Apr 2005, 5:34:34 UTC

If it's been mentoined I didn't see it,

There are two different Pentium M cores now. The older Banias core and the newer Dothan core. Banias (130nm) only comes with the 1MB cache but the Dothan can be had with a 2MB cahce in 90nm form. I am still not convinced that cache is as big a factor as some of you think but I have noticed that the Dell 1.6GHz Latitude D600s (90nm Dothan 2MB cache) do run about 30 mins faster than the IBM 1.5GHz T40s (130nm Banias 1MB cahce). The reasoning behind this is inconclusive though because there are other contributing differences between these machines (memory speeds, etc.). Both are running the same 400MHz BUS (CPU to North bridge)however.
http://www.intel.com/products/processor/pentiumm/


One consecutive isse I see with this Intel SpeedStep is that is doesnt always work properly. Not even when you have the correct applets installed. Many of the machines will run at a reduced speed even when they're set to max performance and running 99% CPU load. The IBM T40s allow for the speedstep to be completely disabled in the BIOS which solves that problem. But the Dells (D600) supposedly run at a fixed reduced speed if you disable speedstep in them.
ID: 105074 · Report as offensive
Matthew Williams

Send message
Joined: 7 May 99
Posts: 3
Credit: 148,519
RAC: 0
United States
Message 105076 - Posted: 29 Apr 2005, 5:44:01 UTC - in response to Message 104972.  

> Here is a nice comparison of two of my laptops, using Boinc 4.19.
>
> 1.) Dell 600m: 1.6Mhz Pentium M(1MB cache, 400Mhz fsb), 512MB DDR - 2:40/wu
>
> 2.) Dell 6000: 1.6Mhz Pentium M(2MB cache, 533Mhz fsb), 512MB DDR2 - 2:20/wu
>

Just found this post.

This may actually prove the value of the 2mb cache. Our D600s are configured the same as your 600M aside from the cache size and they're averaging 2:20/wu.
ID: 105076 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13722
Credit: 208,696,464
RAC: 304
Australia
Message 105203 - Posted: 29 Apr 2005, 11:07:34 UTC - in response to Message 104938.  

> > For absolute speed, with maximum efficiency, the Pentium-M's have got
> it.
>
> What about <a> href="http://setiweb.ssl.berkeley.edu/top_hosts.php">Xeon's[/url]?
> They seem to do pretty well too - if you have the cash to spend. :)

Don't know about crunching Seti, but for everything else they get seriously hammered by the Opterons, which have stuff all cache compared to the Xeons & theyr'e a good 1GHz or more slower clock speed wise.
Grant
Darwin NT
ID: 105203 · Report as offensive
Profile Beyond
Avatar

Send message
Joined: 11 Jul 99
Posts: 15
Credit: 3,176,621
RAC: 3
United States
Message 105376 - Posted: 29 Apr 2005, 15:52:07 UTC

> AMD Sempron XP 2500+, is the 2nd fastest with a turn around time of between 4-6 hours.

Hmm, something is wrong here. I'm also not so sure about the "Cache is King" theory (especially notice the 128k cache on the 2600+ Sempron listed below), although more cache definitely does help somewhat. You have to remember that the PM and P4 are very different animals altogether.

Granted a Pentium M is faster in many measures than a P4, and is generally much faster clock for clock. It also is quite weak in other areas. The P4 relies in a large measure on raw clock speed, produces massive amounts of heat in it's 90nm form and as we all know has come to the end of it's road. But I digress...

BTW, if your cache theory IS correct, I would think that the Athlon64 FX (1MB cache) processors would be the fastest for S@H BOINC.

As far as your Sempron 2500+ time, there must be something else going on with that machine, it's much slower than it should be. Here are my times for various AMD processors:

Seti@Home Boinc:

Athlon64 3400+ (socket 939, 512k cache)with 9% overclock: 1:50 - 2:12
Sempron 2600+ (socket 754, 128k cache)with 50% overclock: 2:02 - 2:25
AthlonXP 2500+ (socket A, 512k cache) with 20% overclock: 2:11 - 2:33
Athlon64 2800+ (socket 754, 512k cache)with 10% overclock: 2:14 - 2:39
Sempron 2500+ (socket A, 256k cache)with 10% overclock: 2:43 - 3:06

Einstein@Home Boinc:

Athlon64 3400+ (socket 939, 512k cache)with 9% overclock: 5:13
Sempron 2600+ (socket 754, 128k cache)with 50% overclock: 5:12
AthlonXP 2500+ (socket A, 512k cache) with 20% overclock: 5:50
Athlon64 2800+ (socket 754, 512k cache)with 10% overclock: 6:07
Sempron 2500+ (socket A, 256k cache)with 10% overclock: 6:40

A few notes: These processors are overclocked but are 100% stable and have produced 100% valid results in both projects. The Athlon64 3400+ and AthlonXP 2500+ have dual channel memory. The Sempron 2500+ machine is hampered by a very crappy Chaintech MB that does not OC well and also refuses to run dual channel memory no matter what the brand or configuration. The large OC on the Sempron 2600+ is typical and IMO is largely due to it's 90nm SOI process. Interestingly, it's still the coolest running machine of the lot.
ID: 105376 · Report as offensive
Profile [DP]Ghent96

Send message
Joined: 3 Apr 99
Posts: 18
Credit: 7,228
RAC: 0
United States
Message 105383 - Posted: 29 Apr 2005, 16:04:53 UTC

Jimmy & Paul, I think that explains it pretty well, that the P4-M chips are better raw crunchers, per clock speed, than a normal P4 due to lack of the MMX & etc and difference in architecture. Makes quite a bit of sense. Cache is queen, and architecture king? perhaps ;) I must say tho, I'm still quite happy that my laptop can churn out a WU in about 2.5 hr (~9000s), but makes me desperately want to upgrade my home machine. Awesome insight into that!

Anymore, laptops are mobile presentation devices as much as they are for mobile calendar & wordprocessing/spreadsheeting. I wouldn't be suprised if more laptops start offering better in graphics performance.

iirc, the P4 1.3 ghz wilhamettes have pitiful next to nothing cache, or none at all. Gah, nvrm, I'm just lamenting my old desktop and dreaming of a new one.
ID: 105383 · Report as offensive
Profile [DP]Ghent96

Send message
Joined: 3 Apr 99
Posts: 18
Credit: 7,228
RAC: 0
United States
Message 105387 - Posted: 29 Apr 2005, 16:12:08 UTC - in response to Message 105376.  

> As far as your Sempron 2500+ time, there must be something else going on with
> that machine, it's much slower than it should be. Here are my times for
> various AMD processors:
>
> Seti@Home Boinc:
>
> ...
> Sempron 2600+ (socket 754, 128k cache)with 50% overclock: 2:02 - 2:25
> ...
> Einstein@Home Boinc:
>
> Athlon64 3400+ (socket 939, 512k cache)with 9% overclock: 5:13
> Sempron 2600+ (socket 754, 128k cache)with 50% overclock: 5:12
> ...
> ...The Sempron 2500+ machine is hampered by a
> very crappy Chaintech MB that does not OC well and also refuses to run dual
> channel memory no matter what the brand or configuration. The large OC on the
> Sempron 2600+ is typical and IMO is largely due to it's 90nm SOI process.
> Interestingly, it's still the coolest running machine of the lot.

ok, neat... but that's still a massive OC, it's no wonder your times are better than the other guy's even with a small-ish cache, and very close to the vastly superior Athlon64 3400. The only surprise is that you've managed to keep it the coolest.
ID: 105387 · Report as offensive
Profile Clyde C. Phillips, III

Send message
Joined: 2 Aug 00
Posts: 1851
Credit: 5,955,047
RAC: 0
United States
Message 105441 - Posted: 29 Apr 2005, 19:43:35 UTC - in response to Message 104917.  

> I have been interested in the effect of the L3 cache as I don't yet have a
> processor with anything beyond a L2 yet.

About a year ago I saw somewhere, probably at Tom's, that a Xeon equipped with a lot of L3 cache didn't do too well. Maybe L3 is too far from the processor to do as well as L1 and L2.
ID: 105441 · Report as offensive
Profile Paul D. Buck
Volunteer tester

Send message
Joined: 19 Jul 00
Posts: 3898
Credit: 1,158,042
RAC: 0
United States
Message 105456 - Posted: 29 Apr 2005, 20:45:59 UTC - in response to Message 105441.  

> About a year ago I saw somewhere, probably at Tom's, that a Xeon equipped with
> a lot of L3 cache didn't do too well. Maybe L3 is too far from the processor
> to do as well as L1 and L2.

More likely it was an error in the design concept of the L3 cache. L1 and L2 cache are optimized to do different things. So to must the L3. If they made the wrong choices then the speed will suffer for that.
ID: 105456 · Report as offensive
Profile Beyond
Avatar

Send message
Joined: 11 Jul 99
Posts: 15
Credit: 3,176,621
RAC: 3
United States
Message 105608 - Posted: 30 Apr 2005, 2:24:29 UTC

> ok, neat... but that's still a massive OC, it's no wonder your times are
> better than the other guy's even with a small-ish cache, and very close
> to the vastly superior Athlon64 3400.

The 2500+ is only overclocked by 10% and doesn't even have dual channel memory because of the inferior Chaintech MB.

The 2600+ is the "massive OC". Interestingly both it and the A64 3400+ run at 2400MHz. The Sempron gets a boost from the 300MHz FSB vs 220MHz on the A64. The A64 gets a boost from it's dual memory channels and 512k cache. Pretty much comes out a wash in the end. BTW, the 90nm A64 chips are also great OCers. Then you can have the best of both worlds. Both of my A64s are 130nm.

> The only surprise is that you've managed to keep it the coolest.

Stock retail 2700rpm HS/Fan. Planning to install a bit better cooler to see how far it will go. Pretty amazing chips IMO.

Its interesting that the 90nm AMD parts run so much cooler than the 130nm, when Intel's 90nm P4 parts run much hotter than their 130nm.
ID: 105608 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13722
Credit: 208,696,464
RAC: 304
Australia
Message 105731 - Posted: 30 Apr 2005, 10:33:55 UTC - in response to Message 105608.  


> Its interesting that the 90nm AMD parts run so much cooler than the 130nm,
> when Intel's 90nm P4 parts run much hotter than their 130nm.

Intel & AMD are at different stages in implementing power saving techniqes in their CPU construction.
Intel started much earlier with SOI etc than AMD. So where as Intel had already applied many if the techniques that were being developed with their 130nM CPUs, AMD are only starting to apply them with their 90nM CPUs.
Hence the new Dual Core CPUs from AMD will actually run as cool as the previous generation single core CPUs of a similar clock speed where as Intel Dual Core CPUs will have to run 500MHz or more slower than the previous single core CPUs just to stay within Intels maximum heat dissapation rating of 130W (AMDs is about 97W).
Grant
Darwin NT
ID: 105731 · Report as offensive
Profile jimmyhua

Send message
Joined: 16 Apr 05
Posts: 97
Credit: 369,588
RAC: 0
Guam
Message 105751 - Posted: 30 Apr 2005, 12:33:20 UTC - in response to Message 105376.  
Last modified: 30 Apr 2005, 12:33:54 UTC

> Seti@Home Boinc:
> Sempron 2500+ (socket A, 256k cache)with 10% overclock: 2:43 - 3:06

Beyond,

That is an amazing time for a Sempron 2500+ Socket A with 10% overclock.

I've got the same setup w/out overclock, and completes a WU somewhere between 4:00-4:30, in CPU time.

It takes longer in real life, as this is the machine I sit in front of most of the time.

Mobo can make a real difference, I think your chaintech mobo is WAY WAY better than my ECS mobo.

Jimmy

ID: 105751 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13722
Credit: 208,696,464
RAC: 304
Australia
Message 105911 - Posted: 30 Apr 2005, 22:30:40 UTC - in response to Message 105751.  

> > Seti@Home Boinc:
> > Sempron 2500+ (socket A, 256k cache)with 10% overclock: 2:43 - 3:06

XP2600+, stock speed, single channel memory, Win98SE, 512MB RAM: generally around 3:15 per Work Unit.
Grant
Darwin NT
ID: 105911 · Report as offensive
Profile Beyond
Avatar

Send message
Joined: 11 Jul 99
Posts: 15
Credit: 3,176,621
RAC: 3
United States
Message 105935 - Posted: 1 May 2005, 0:04:12 UTC

> Mobo can make a real difference, I think your chaintech mobo is WAY WAY better than my ECS mobo.

Quite likely. It uses the nForce 2 Ultra 400 chipset, so SHOULD be very fast. Unfortunately the implementation is so bad that the dual channel memory is unusable. That's the most serious problem but there are others. No BIOS updates are available. Dual channel generally helps S@H a lot, E@H not as much. This is the second Chaintech I've used lately. Both were poor. The brands I'd recommend are: Abit, Asus, Biostar, Epox, Gigabyte, MSI. The worst sample of any of these was better than either Chaintech.

The socket 754 90nm Sempron 2600+ is a work in progress. Right now it's running stability tests at 310FSB, 2480MHz. So far so good. Impressive for a 1600MHz processor. It's installed on an EPoX 8KDA3J MB, which needless to say is also VERY impressive for a sub $75 motherboard. Of course memory and HT must be locked at a lower speed. Processor voltage is bumped .05, a very modest increase. Chipset voltage is default.
ID: 105935 · Report as offensive
Profile FalconFly
Avatar

Send message
Joined: 5 Oct 99
Posts: 394
Credit: 18,053,892
RAC: 0
Germany
Message 105944 - Posted: 1 May 2005, 0:49:07 UTC - in response to Message 105935.  
Last modified: 1 May 2005, 0:58:06 UTC

It would be interesting to see how the Pentium M performs with the optimized Linux Clients.

With those, this is about what I get right now (last 10 normal WU's Average) :

AthlonXP 2100+ : 2.85h (2h 51m)
AthlonXP 2500+ : 2.52h (2h 31m)
AthlonXP 2700+ : 2.43h (2h 26m)
AthlonXP 3000+ : 2.29h (2h 17m)
Athlon64 3000+ : 1.91h (1h 55m)
Athlon64 3200+ : 1.87h (1h 52m)
ID: 105944 · Report as offensive
Profile Team Cake Boy's

Send message
Joined: 24 Jul 99
Posts: 22
Credit: 6,923,381
RAC: 0
United Kingdom
Message 114288 - Posted: 24 May 2005, 6:23:33 UTC - in response to Message 103957.  
Last modified: 24 May 2005, 6:27:55 UTC

umm no
the new mobiles
are pentium 3 cpu's with new stuff added sse2 more cache etc

as in they are the same core
all these cpu's use the same origional core !!
but are just updated versions

Pentium Pro
Pentium II
Pentium III
Mobile Pentium-M

oh and guess what

the replacment for pentium 4 is going to be pentium pro based

thank god and good riddance to the awfull netburst junk

I guess the only thing that will convince you it is the cache, is if I had a Pentium 3 with 1 MB cache with same Ghz, and a Pentium 3 with 2 MB cache with same Ghz (running off the same exact Motherboard), and show that it literally runs twice as fast.

I think I am in luck! They actually sell such beasts!!! Too bad it's not worth my time to pony up the money to make a point like this. I just share my data, and my conclusions/opinions based on the data.

I believe the cache helps ALOT MORE (for Seti applications) than a faster BUS. Lets face it. My laptop with Pentium-M is running off a substandard PC100 SDRAM bus. It isn't even interleaved. How can you tell me my Laptop with a strangled memory bandwidth is matching my P4 3.2 Ghz machine with DDR2-400, that Memory Bandwidth matters more than cache?

If this was the case, again the P4 with the modern memory bus should be leaving the Pentium M in the dust. Sure faster memory bus definitely helps, but it isn't the main factor that is slowing us down here.

Although the architecture of the Pentium-M is quite different to a Pentium 4. It is alot closer to a Pentium-4 than it is to a Pentium-3.

Just my opinion.

After doing some calculations. I think the Pentium-M's are probably running as fast as they can go and 1MB cache is optimal. However, the Pentium 4s and Athlons need 4MB to really shine.


Jimmy


ID: 114288 · Report as offensive
Profile jimmyhua

Send message
Joined: 16 Apr 05
Posts: 97
Credit: 369,588
RAC: 0
Guam
Message 114296 - Posted: 24 May 2005, 7:19:20 UTC - in response to Message 114288.  

umm no
the new mobiles
are pentium 3 cpu's with new stuff added sse2 more cache etc

as in they are the same core
all these cpu's use the same origional core !!
but are just updated versions

Pentium Pro
Pentium II
Pentium III
Mobile Pentium-M

oh and guess what

the replacment for pentium 4 is going to be pentium pro based


That's interesting. If my memory serves me correctly. Intel dropped the Pentium-3 line in favor of the Pentium-4, as they could not keep up with the MHz increases with AMD's Athlon.

Also, I am pretty sure Pentium-M's are missing SSE2 instruction set, as there have been several motherboards that tried to use the P-M's CPU only to find bad reviews in places like AnandTech and Tomshardware due to lackluster gaming performance (lack of SSE2 instruction set). They do still have MMX and SSE 1.

I too believe that the Pentium-M's are based off the P3s. However, I'd say they are much better (have been significantly improved). For one thing, when it comes to Seti, a Pentium-M 1.6 Ghz will outcrunch a 2.2Ghz AMD Athlon. Back in the day, Mhz for Mhz, the P3's were actually a little slower than the Athlons.


Jimmy
ID: 114296 · Report as offensive
Previous · 1 · 2 · 3 · 4 · Next

Message boards : Number crunching : Cache is King, Pentium-M CPU is the fastest!!!


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.