OS Name Microsoft Windows XP Professional
Version 5.1.2600 Service Pack 1 Build 2600
OS Manufacturer Microsoft Corporation
System Model KT333CF-8235
System Type X86-based PC
Processor x86 Family 6 Model 10 Stepping 0 AuthenticAMD ~1994 Mhz
BIOS Version/Date Phoenix Technologies, LTD 6.00 PG, 8/25/2003
SMBIOS Version 2.2
Windows Directory C:WINDOWS
System Directory C:WINDOWSSystem32
Boot Device DeviceHarddiskVolume1
Total Physical Memory 768.00 MB
Available Physical Memory 496.43 MB
Total Virtual Memory 1.56 GB
Available Virtual Memory 1.08 GB
Page File Space 826.86 MB
Page File C:pagefile.sys

--- - 2004-09-05 23:58:52 - Benchmark results:
--- - 2004-09-05 23:58:52 - Number of CPUs: 1
--- - 2004-09-05 23:58:52 - 2410 double precision MIPS (Whetstone) per CPU
--- - 2004-09-05 23:58:52 - 4253 integer MIPS (Dhrystone) per CPU

Average time per WU: 3:15

[url=http://www.setiathome.ssl.berkeley.edu/banners/5setibanner_bl.jpg]
Processing for the Planetary Society since July 5, 1999

> --------------------------------------------------------------------
>
> CPU type ARM (tm) *nixnix* SNAP 38000 Home Edition
> Number of CPUs changing dayly, some are still in assembly, handmade in india
> Operating System Microsoft Windows 98 freeware Edition, bug level *overflow*
> (01.01.1950)
> Memory 1.49 MB
> Cache 00.08756 KB
> Swap space -4 MB
> Raid level 10 on c64 dataset
> Extension Box Hypertransport for multi bug support
> Measured floating point speed -0 million ops/sec
> Measured integer speed -0 million ops/sec
>
> sorry could not execute Benchmarks due frozen system
>
>
> --------------------------------------------------------------------
>
>

LOL

I like it. Can I have one please!!!! ;-)

Hi all,

--- - 2004-09-06 11:30:25 - Benchmark results:
--- - 2004-09-06 11:30:25 - Number of CPUs: 1
--- - 2004-09-06 11:30:25 - 1474 double precision MIPS (Whetstone) per CPU
--- - 2004-09-06 11:30:25 - 2647 integer MIPS (Dhrystone) per CPU
--- - 2004-09-06 11:30:25 - Finished CPU benchmarks

CPU Properties IBM Thinkpad X30
CPU Type Mobile Intel Pentium III-S, 1200 MHz (9 x 133)
CPU Alias Tualatin, A80530
Original Clock 1200 MHz
L1 Code Cache 16 KB
L1 Data Cache 16 KB
L2 Cache 512 KB (On-Die, ATC, Full-Speed)

I really don't suffer from any 'CPU envy', but those AMD results are really 'WOW'... :-)

On the other hand, how come all those Intel Pentium 4 2.4 & 2.8 3.06 (even some 3.20 have a lower 'integer speed') results are worse than my 'measly' Pentium 3 @1200? Not to forget my reeeaaally slow memory (albeit 1024MB)?

Any explanation anyone???

Thanks a million, and don't forget to join our
team; 'a TINY elite force of Mossad operatives - http://groups.yahoo.com/group/theisraaliens/' ;-)

<a> [/url]

CPU type Intel(R) Pentium(R) 4 CPU 3.20GHz Pentium
Number of CPUs 2
Operating System Microsoft Windows XP Home Edition, Service Pack 2, (05.01.2600.00)
Memory 1022.72 MB
Cache 976.56 KB
Swap space 3527.47 MB
Total disk space 138.48 GB
Free Disk Space 101.26 GB
Measured floating point speed 1931.33 million ops/sec
Measured integer speed 2118.77 million ops/sec

Average Work Unit= 3hr. 20Min. (running two simultaneously).

4.05
***********************


CPU type Intel(R) Pentium(R) 4 CPU 2.40GHz Pentium
Number of CPUs 1
Operating System Microsoft Windows XP Home Edition, Service Pack 2, (05.01.2600.00)
Memory 767.07 MB
Cache 976.56 KB
Swap space 4096 MB
Total disk space 44.66 GB
Free Disk Space 18.11 GB
Measured floating point speed 1442.71 million ops/sec
Measured integer speed 3042.13 million ops/sec

Average Work Unit= 3Hr. 51Min.

4.05

400 dollar eMachine.


--- - 2004-09-06 03:27:48 - 1399 double precision MIPS (Whetstone) per CPU
--- - 2004-09-06 03:27:48 - 2924 integer MIPS (Dhrystone) per CPU

> I really don't suffer from any 'CPU envy', but those AMD results are really
> 'WOW'... :-)
>
> On the other hand, how come all those Intel Pentium 4 2.4 & 2.8 3.06 (even
> some 3.20 have a lower 'integer speed') results are worse than my 'measly'
> Pentium 3 @1200? Not to forget my reeeaaally slow memory (albeit 1024MB)?
>
> Any explanation anyone???

The Machines with P4 that display the 'non-impressive' results are running with Hyper-Threading. While those appear to the OS as two CPU's, its still only one CPU that switches execution between the two virtual CPU's. The two virtual CPU's still have to make shared use of the cache and memory bus, so the performance cant be as good as that of a single CPU that doesnt need to share anything :) If you disable Hyperthreading, the Benchmark results should improve quite a bit (although the overall throughput over time compared to Hyperthreaded mode decreases a bit).

On the other hand, the AMD CPU's have a considerably larger L1-Cache (2x64 KB if i'm not mistaken) which partially makes up for their smaller L2-Cache (256 KB on the XP?). Also the execution pipeline on the AMD is much shorter than that of the P4, which lessens the impact of conditional branches in the program flow. So the AMD can perform comparably to the P4 at a much lower clock.

Regarding your really 'slow' memory - the industry likes to market every little improvement as a big step forward (they want to sell it after all :). In fact those FSB800 systems are NO WAY 6 times faster than your old 133 Mhz System (despite what the numbers suggest). While the new systems could theoretically move 6 times the amount of data over the bus, the memory still needs to provide the data. And memory latency times didnt change that much since the introduction of SD-RAM's. In fact for each memory access, most time is spent on negotiating the transfer (look at the BIOS setting for RAS/CAS/Precharge and that stuff). That delay occurs for every memory access made, and i am pretty confident those cycles are counted from the 'real' physical bus clock, not the DDR-double clock.

Same computer, according to sah and CPDN:

sah:
CPU type Intel(R) Pentium(R) 4 CPU 3.00GHz Pentium (Northwood HyperThreading)
L1 Trace Cache 12K Instructions
L1 Data-cache 8 KB
L2 Cache 512 KB (On-Die, ATC, Full-Speed)
Number of CPUs 2
Operating System Microsoft Windows XP Professional Edition, Service Pack 1, (05.01.2600.00)
Memory 511.48 MB
Cache 976.56 KB
Swap space 1247.28 MB
Total disk space 74.55 GB
Free Disk Space 7.51 GB
<B>Measured floating point speed 1619.66 million ops/sec
(or a total of 3239.32 million ops/sec)
Measured integer speed 1845.38 million ops/sec
(or a total of 3690.76 million ops/sec)</B>

CPDN:
CPU type Intel(R) Pentium(R) 4 CPU 3.00GHz Pentium (Northwood HyperThreading)
L1 Trace Cache 12K Instructions
L1 Data-cache 8 KB
L2 Cache 512 KB (On-Die, ATC, Full-Speed)
Number of CPUs 2
Operating System Microsoft Windows XP Professional Edition, Service Pack 1, (05.01.2600.00)
Memory 511.48 MB
Cache 976.56 KB
Swap space 1247.3 MB
Total disk space 74.55 GB
Free Disk Space 7.58 GB
<B>Measured floating point speed 1769.4 million ops/sec
(or a total of 3538.8 million ops/sec)
Measured integer speed 2048.9 million ops/sec
(or a total of 4097.8 million ops/sec)</B>

Measured floating point speed 3033.18 million ops/sec
Measured integer speed 5381.22 million ops/sec





Abit NF7-s v2
Athlon XP2400 mobile 35W @ 216 x 11.5 - 2484GHz - 1.8volts
Corsair Twinx LL 2.2.2.5 @ 216MHz - 3.1volts

Osiris

AMD Athlon(tm) XP 2600+
Windows XP Professional Edition, Service Pack 2
Measured floating point speed 1976.83 million ops/sec
Measured integer speed 4554.78 million ops/sec
WU ~ 3 hours 43 minutes

-----------------------------------------------------

Pentium(R) M 1.80GHz
Windows XP Professional Edition, Service Pack 2
Measured floating point speed 1574.09 million ops/sec
Measured integer speed 4370.48 million ops/sec
WU ~ 2 hours 12 minutes

> > I really don't suffer from any 'CPU envy', but those AMD results are
> really
> > 'WOW'... :-)
> >
> > On the other hand, how come all those Intel Pentium 4 2.4 & 2.8 3.06
> (even
> > some 3.20 have a lower 'integer speed') results are worse than my
> 'measly'
> > Pentium 3 @1200? Not to forget my reeeaaally slow memory (albeit
> 1024MB)?
> >
> > Any explanation anyone???
>
> The Machines with P4 that display the 'non-impressive' results are running
> with Hyper-Threading. While those appear to the OS as two CPU's, its still
> only one CPU that switches execution between the two virtual CPU's. The two
> virtual CPU's still have to make shared use of the cache and memory bus, so
> the performance cant be as good as that of a single CPU that doesnt need to
> share anything :) If you disable Hyperthreading, the Benchmark results should
> improve quite a bit (although the overall throughput over time compared to
> Hyperthreaded mode decreases a bit).
>
> On the other hand, the AMD CPU's have a considerably larger L1-Cache (2x64 KB
> if i'm not mistaken) which partially makes up for their smaller L2-Cache (256
> KB on the XP?). Also the execution pipeline on the AMD is much shorter than
> that of the P4, which lessens the impact of conditional branches in the
> program flow. So the AMD can perform comparably to the P4 at a much lower
> clock.
>
> Regarding your really 'slow' memory - the industry likes to market every
> little improvement as a big step forward (they want to sell it after all :).
> In fact those FSB800 systems are NO WAY 6 times faster than your old 133 Mhz
> System (despite what the numbers suggest). While the new systems could
> theoretically move 6 times the amount of data over the bus, the memory still
> needs to provide the data. And memory latency times didnt change that much
> since the introduction of SD-RAM's. In fact for each memory access, most time
> is spent on negotiating the transfer (look at the BIOS setting for
> RAS/CAS/Precharge and that stuff). That delay occurs for every memory access
> made, and i am pretty confident those cycles are counted from the 'real'
> physical bus clock, not the DDR-double clock.
>
Very nice post, Switzerland. I learn from posts like yours. Thanks.
Robert

Benchamrks? You can't trust no stinking benchmarks:

--- - 2004-09-24 16:05:11 - Number of CPUs: 2
--- - 2004-09-24 16:05:11 - 2104 double precision MIPS (Whetstone) per CPU
--- - 2004-09-24 16:05:11 - 3126 integer MIPS (Dhrystone) per CPU
--- - 2004-09-24 16:05:11 - Finished CPU benchmarks
--- - 2004-09-24 16:05:12 - Resuming computation and network activity
--- - 2004-09-24 16:05:29 - Running CPU benchmarks
--- - 2004-09-24 16:05:30 - Suspending computation and network activity - running CPU benchmarks
--- - 2004-09-24 16:06:31 - Benchmark results:
--- - 2004-09-24 16:06:31 - Number of CPUs: 2
--- - 2004-09-24 16:06:31 - 2105 double precision MIPS (Whetstone) per CPU
--- - 2004-09-24 16:06:31 - 1449 integer MIPS (Dhrystone) per CPU
--- - 2004-09-24 16:06:31 - Finished CPU benchmarks

The above I got in two runs, one after anotehr. The rig is a dual AMD MP 2800+. The Integer results are so off, that they can't be trusted.

Cheers!

-mm-

If you are interested in benchmarks from different processors you may be interested in this: http://www.noggintech.com/boincinfo

I wrote this out of my own curiosity but some of you may find it interesting. I will leave it up for a few days.

--- 10000-12-24 16:06:31 - 1000678486789 double precision MIPS (Whetstone) per CPU
--- - 10000-12-24 16:06:31 - 94499997878997 integer MIPS (Dhrystone) per CPU
-----------------------

[url=http://www.guidowaldenmeier.de]Machmal sind kleine aber durchdachte und verstandene Schritte die bessere Wahl.Guidos Boinc Forum

> If you are interested in benchmarks from different processors you may be
> interested in this: http://www.noggintech.com/boincinfo
>
> I wrote this out of my own curiosity but some of you may find it interesting.
> I will leave it up for a few days.

Thanks ! It's nice to explore.
It does confirm that the Power Mac G5 is the best box by all means.