Pentium II vs. Pentium Pro 
Author Message
 Pentium II vs. Pentium Pro

Hi all,

While this isn't strictly fortran, I figured someone here would know the
answers:

1)  Is the Pentium II 266 a faster pure "number cruncher" than the Pentium Pro
200?  I've heard arguments that the 266 is not as good a number cruncher as
the 200.  I didn't see a Pentium II listing in the TEST_2 benchmark.
2)  Is the Pentium optimization option in MS 4.0 also good for the Pentium Pro
or Pentium II architecture?

The answers are kind of important (aren't they all), as I'd like to buy the
faster Pentium computer around (and then optimize for it properly).

Thanks in advance,

Morris



Fri, 10 Dec 1999 03:00:00 GMT  
 Pentium II vs. Pentium Pro

Quote:

> 1)  Is the Pentium II 266 a faster pure "number cruncher" than the Pentium Pro
> 200?  I've heard arguments that the 266 is not as good a number cruncher as
> the 200.  I didn't see a Pentium II listing in the TEST_2 benchmark.
> I'd like to buy the
> faster Pentium computer around (and then optimize for it properly).

I was wondering the same thing. I have been tolds by some in the know
that your conjecture is correct for programs small enough to fit in
the PPro's internal cache. The PII moved the cache to the motherboard
so the cache access is slower, enough to outweigh the additional
66 MHz of clock.

For programs that are too big to fit entirely in the PPro's on-board
cache, the faster clock dominates.

Like I said, I have no solid test data. This is heresay from some
people who have a 10-year history of being right most of the time.
I would love to see some hard numbers if anyone has them.

--
Glen Clark



Sat, 11 Dec 1999 03:00:00 GMT  
 Pentium II vs. Pentium Pro

Quote:


>> 1)  Is the Pentium II 266 a faster pure "number cruncher" than the Pentium
> Pro
>> 200?  I've heard arguments that the 266 is not as good a number cruncher as
>> the 200.  I didn't see a Pentium II listing in the TEST_2 benchmark.

>> I'd like to buy the
>> faster Pentium computer around (and then optimize for it properly).

>I was wondering the same thing. I have been tolds by some in the know
>that your conjecture is correct for programs small enough to fit in
>the PPro's internal cache. The PII moved the cache to the motherboard

                                                           ^^^^^^^^^^^

Quote:
>so the cache access is slower, enough to outweigh the additional
>66 MHz of clock.

That's not technically correct.  For the Pentium II, Intel moved the
cache off of the chip package (as it is for the Pro) but put it on the
new Single Edge Connector board of the II and not on the motherboard.  
The Pentium Pro runs its L2 cache at full processor speed (IIRC), while
the Pentium II runs its L2 cache at 1/2 processor speed, and Pentiums
run their L2 cache at bus speed (usually 50 or 66MHz).

Quote:
>For programs that are too big to fit entirely in the PPro's on-board
>cache, the faster clock dominates.

Even for smaller programs, the faster clock may dominate.  Don't forget
the effect of the L1 cache as well.  It gets complicated.

Quote:
>Like I said, I have no solid test data. This is heresay from some
>people who have a 10-year history of being right most of the time.
>I would love to see some hard numbers if anyone has them.

One of the best places to find this kind of information is Tom's
Hardware site at:      

    http://sysdoc.pair.com/  

Unfortunately, I'm having trouble getting through right now so I can't
get any solid numbers.  IIRC, the 266MHz Pentium II beat the 200MHz
across the board; including some floating point benchmarks.

--

All opinions expressed are my own and not my employer's.
--



Sat, 11 Dec 1999 03:00:00 GMT  
 Pentium II vs. Pentium Pro

You could check "TOM'S HARDWARE GUIDE" on the
Internet.

Sorry, I don't know the Internet address of this site
by heart (but I guess you will be able to find it with
using a search engine)



Sun, 12 Dec 1999 03:00:00 GMT  
 Pentium II vs. Pentium Pro

On 25 Jun 1997 19:24:53 -0400, Nicholas Fitzpatrick

Quote:



>>Hi all,

>>While this isn't strictly Fortran, I figured someone here would know the
>>answers:

>>1)  Is the Pentium II 266 a faster pure "number cruncher" than the Pentium Pro
>>200?  I've heard arguments that the 266 is not as good a number cruncher as
>>the 200.  I didn't see a Pentium II listing in the TEST_2 benchmark.

>I'd go check out the lastest SPECfp numbers.  I saw some early numbers
>on the PII that showed that the PII 233 was slower than a PP 200 but
>that the PII 266 was moderatly faster.

>Ah yes, here it is, from the May 12th version of the file that is
>available via anonymous ftp from ftp.cdf.toronto.edu as
>/pub/spectable.

>SPECfp_base95

>-------------------------------------------------------------------------------
>System            CPU        ClkMHz  Cache      SPECint SPECfp  Info  Source
>Name              (NUMx)Type ext/in  Ext+I/D    base95  base95  Date  Obtained
>================= ========== ======= ========== ======= ======= ===== =========

>Intel Alder       PentPro    150     256+8/8      6.08    4.76  Jan96 www.intel
>Intel Alder       PentPro    166     512+8/8      7.11    5.47  Jan96 www.intel
>Intel Alder       PentPro    180     256+8/8      7.29    5.40  Jan96 www.intel
>Intel Alder       PentPro    200     256+8/8      8.09    5.99  Jan96 www.intel
>Intel PD440FX     PentiumII  66/233  512+16/16    9.49    5.91  May97 www.intel
>Intel PD440FX     PentiumII  66/266  512+16/16   10.8     6.43  May97 www.intel

>Now there are tons of disclaimers, and these are a function of
>system, but the bottom line appears to be that there is no gain
>from the 233 for floating point and maybe a 7% gain with the 266 over
>the 200.

>Our decision for numerical modelling, is that any gains in the 266 are
>in no way related to the high cost, and we are trying to obtain enough
>PP 200's, before some joker at Intel decides to add an MMX processor
>to it as well!!

Indeed. I'd love to see SPEC numbers for the PPro 200 with 512 kB L2
cache ... Too bad Intel doesn't like it enough (or doesn't know how)
to give it another 100 MHz or so .... Stefan

--
============================================================================
Stefan A. Deutscher                     |  (+1-423-)   voice      fax
The University of Tennessee, Knoxville  |  UTK  :     974-7838  974-7843
Department of Physics and Astronomy     |  ORNL :     574-5897  574-1118
401, A. H. Nielsen Building             |  home :     522-7845  522-7845

============================================================================
Dictated with OS/2 Warp 4 VoiceType dictation. Using slrn 0.9.3.2 offline.



Mon, 13 Dec 1999 03:00:00 GMT  
 Pentium II vs. Pentium Pro

For those who are concerned with the very best in Wintel
performance, check out "Tom's Hardware Guide" located at

http://sysdoc.pair.com/

Bottom line:  the motherboard has something to do with
performance, too.  Intel is not necessarily the fastest W95/NT
platform.  (To build the fastest, you might want to get cpu,
memory, MB, etc from different sources, and build it yourself.)

After one visit, my biggest surprise was that *some* MBs
only cache the first 64MB of core.  So "memory slots" /=
"cachable memory slots" !?!?  There are too many
variables (read: benchmarks) for me to get them all in one visit.

--

Cheers!




Mon, 13 Dec 1999 03:00:00 GMT  
 Pentium II vs. Pentium Pro

Far more interesting than the overall SPECfp95 numbers are the numbers for
the individual benchmarks. The relative performance of the Ppro 200 and
PII 266 vary a lot on these. You can find them at
http://www.specbench.org, incidentally.

For example, on 101.tomcatv, a PPro 200 (Intel Alder) achieves a peak
ratio of 11.6. A PII 266 (PD440FX) can do no better than 10.3, despite its
higher clock speed. There are several of the individual benchmarks on
which the Ppro 200 outperforms the PII 266, often quite handily.

At the other extreme, on 141.apsi, the Ppro manages only 6.74 and the PII
achieves 8.47.

Whether a PII is faster than a Ppro for a specific program has nothing to
do with whether or not the program is floating-point intensive and
everything to do with how the program uses memory. If it runs mainly in
the L1 cache, it should do well on the PII.

--
James G. MacKinnon                       Department of Economics
    phone: 613 545-2293                  Queen's University
      Fax: 613 545-6668                  Kingston, Ontario, Canada



Mon, 13 Dec 1999 03:00:00 GMT  
 Pentium II vs. Pentium Pro


Quote:

>On 25 Jun 1997 19:24:53 -0400, Nicholas Fitzpatrick



>>>Hi all,

>>>While this isn't strictly Fortran, I figured someone here would know the
>>>answers:

>>>1)  Is the Pentium II 266 a faster pure "number cruncher" than the Pentium Pro
>>>200?  I've heard arguments that the 266 is not as good a number cruncher as
>>>the 200.  I didn't see a Pentium II listing in the TEST_2 benchmark.

>>I'd go check out the lastest SPECfp numbers.  I saw some early numbers
>>on the PII that showed that the PII 233 was slower than a PP 200 but
>>that the PII 266 was moderatly faster.

>>Ah yes, here it is, from the May 12th version of the file that is
>>available via anonymous ftp from ftp.cdf.toronto.edu as
>>/pub/spectable.

>>SPECfp_base95

>>-------------------------------------------------------------------------------
>>System            CPU        ClkMHz  Cache      SPECint SPECfp  Info  Source
>>Name              (NUMx)Type ext/in  Ext+I/D    base95  base95  Date  Obtained
>>================= ========== ======= ========== ======= ======= ===== =========

>>Intel Alder       PentPro    150     256+8/8      6.08    4.76  Jan96 www.intel
>>Intel Alder       PentPro    166     512+8/8      7.11    5.47  Jan96 www.intel
>>Intel Alder       PentPro    180     256+8/8      7.29    5.40  Jan96 www.intel
>>Intel Alder       PentPro    200     256+8/8      8.09    5.99  Jan96 www.intel
>>Intel PD440FX     PentiumII  66/233  512+16/16    9.49    5.91  May97 www.intel
>>Intel PD440FX     PentiumII  66/266  512+16/16   10.8     6.43  May97 www.intel

>>Now there are tons of disclaimers, and these are a function of
>>system, but the bottom line appears to be that there is no gain
>>from the 233 for floating point and maybe a 7% gain with the 266 over
>>the 200.

>>Our decision for numerical modelling, is that any gains in the 266 are
>>in no way related to the high cost, and we are trying to obtain enough
>>PP 200's, before some joker at Intel decides to add an MMX processor
>>to it as well!!

>Indeed. I'd love to see SPEC numbers for the PPro 200 with 512 kB L2
>cache ... Too bad Intel doesn't like it enough (or doesn't know how)
>to give it another 100 MHz or so .... Stefan

i have the specs for the PPro 200 MHz 512 KB L2 cpu somewhere around here.
the Pentium II 266 MHz is still faster and costs less. I will post
the cpu spec ratings later.

Dewayne

- Show quoted text -

Quote:

>--
>============================================================================
>Stefan A. Deutscher                     |  (+1-423-)   voice      fax
>The University of Tennessee, Knoxville  |  UTK  :     974-7838  974-7843
>Department of Physics and Astronomy     |  ORNL :     574-5897  574-1118
>401, A. H. Nielsen Building             |  home :     522-7845  522-7845

>============================================================================
>Dictated with OS/2 Warp 4 VoiceType dictation. Using slrn 0.9.3.2 offline.



Tue, 14 Dec 1999 03:00:00 GMT  
 Pentium II vs. Pentium Pro


Quote:

>i have the specs for the PPro 200 MHz 512 KB L2 cpu somewhere around here.
>the Pentium II 266 MHz is still faster and costs less. I will post
>the cpu spec ratings later.

>Dewayne

here are the Spec Numbers:

Pentium Pro 200 MHz 512 KB L2 Spec Int 95 8.71 Spec FP 95 6.68
PPro 200 MHz 256 KB L2 Spec Int 95 8.20 Spec FP 95 6.21
Pentium II 266 MHz 512 KB L2 Spec Int 95 10.8 Spec FP 95 6.89
Power PC 604e 200 MHz Spec Int 95 7.41 Spec FP 95 5.81



Tue, 14 Dec 1999 03:00:00 GMT  
 Pentium II vs. Pentium Pro

Hello,

After spending an unfortunately large part of my day yesterday
surfing Tom's Hardware Guide (http://sysdoc.pair.com/), let
me summarize what I've learnt:

In order to understand which PC is faster than the other,
one absolutely _must_ know:

1. Which cpu chip, and at what clock speed; and
   (non-trivial:  apparently, _some_ re-sellers (ahem) re-label chips)

2. How much memory, and how much of each level of cache; and

3. What is the clock of the bus (connect said memory to said cpu); and
   (i.e., there are *two* clock speeds of interest here!)

4. How much of the memory *is actually cachable* on a particular
   motherboard; and

5. Probably a bunch of other stuff (read: I've only read about
   half to a third of the pages at T'sHG); and

6. All the usual benchmark qualifiers (how close to my work is the BM?)

Apparently, the lowly PC is a bit more complex that your
Personal Cray...

--

Cheers!




Tue, 14 Dec 1999 03:00:00 GMT  
 Pentium II vs. Pentium Pro

re: seeing some Pentium II results...

Maybe someone is interested enuf to try loading TEST_FPU.exe
found at:     ftp://members.aol.com/DaveGemini/TEST

on floppy and ask local dealer for permission to run it on a Pentium II..
It will write its results back on floppy file   TEST_FPU.DAT
(total run time shud be about 1 min)...

Then send me the results for entry in table and posting here in c.l.f.
for everyones perusal and comparison with existing P6/200 data base.

Thanks Dave...



Fri, 17 Dec 1999 03:00:00 GMT  
 Pentium II vs. Pentium Pro

On 25 Jun 1997 19:24:53 -0400, Nicholas Fitzpatrick

Quote:



>>Hi all,

>>While this isn't strictly Fortran, I figured someone here would know the
>>answers:

>>1)  Is the Pentium II 266 a faster pure "number cruncher" than the Pentium Pro
>>200?  I've heard arguments that the 266 is not as good a number cruncher as
>>the 200.  I didn't see a Pentium II listing in the TEST_2 benchmark.

>I'd go check out the lastest SPECfp numbers.  I saw some early numbers
>on the PII that showed that the PII 233 was slower than a PP 200 but
>that the PII 266 was moderatly faster.

>Ah yes, here it is, from the May 12th version of the file that is
>available via anonymous ftp from ftp.cdf.toronto.edu as
>/pub/spectable.

>SPECfp_base95

>-------------------------------------------------------------------------------
>System            CPU        ClkMHz  Cache      SPECint SPECfp  Info  Source
>Name              (NUMx)Type ext/in  Ext+I/D    base95  base95  Date  Obtained
>================= ========== ======= ========== ======= ======= ===== =========

>Intel Alder       PentPro    150     256+8/8      6.08    4.76  Jan96 www.intel
>Intel Alder       PentPro    166     512+8/8      7.11    5.47  Jan96 www.intel
>Intel Alder       PentPro    180     256+8/8      7.29    5.40  Jan96 www.intel
>Intel Alder       PentPro    200     256+8/8      8.09    5.99  Jan96 www.intel
>Intel PD440FX     PentiumII  66/233  512+16/16    9.49    5.91  May97 www.intel
>Intel PD440FX     PentiumII  66/266  512+16/16   10.8     6.43  May97 www.intel

>Now there are tons of disclaimers, and these are a function of
>system, but the bottom line appears to be that there is no gain
>from the 233 for floating point and maybe a 7% gain with the 266 over
>the 200.

>Our decision for numerical modelling, is that any gains in the 266 are
>in no way related to the high cost, and we are trying to obtain enough
>PP 200's, before some joker at Intel decides to add an MMX processor
>to it as well!!

Indeed. I'd love to see SPEC numbers for the PPro 200 with 512 kB L2
cache ... Too bad Intel doesn't like it enough (or doesn't know how)
to give it another 100 MHz or so .... Stefan

--
============================================================================
Stefan A. Deutscher                     |  (+1-423-)   voice      fax
The University of Tennessee, Knoxville  |  UTK  :     974-7838  974-7843
Department of Physics and Astronomy     |  ORNL :     574-5897  574-1118
401, A. H. Nielsen Building             |  home :     522-7845  522-7845

============================================================================
Dictated with OS/2 Warp 4 VoiceType dictation. Using slrn 0.9.3.2 offline.



Fri, 03 Mar 2000 03:00:00 GMT  
 Pentium II vs. Pentium Pro

On 25 Jun 1997 19:24:53 -0400, Nicholas Fitzpatrick

Quote:



>>Hi all,

>>While this isn't strictly Fortran, I figured someone here would know the
>>answers:

>>1)  Is the Pentium II 266 a faster pure "number cruncher" than the Pentium Pro
>>200?  I've heard arguments that the 266 is not as good a number cruncher as
>>the 200.  I didn't see a Pentium II listing in the TEST_2 benchmark.

>I'd go check out the lastest SPECfp numbers.  I saw some early numbers
>on the PII that showed that the PII 233 was slower than a PP 200 but
>that the PII 266 was moderatly faster.

>Ah yes, here it is, from the May 12th version of the file that is
>available via anonymous ftp from ftp.cdf.toronto.edu as
>/pub/spectable.

>SPECfp_base95

>-------------------------------------------------------------------------------
>System            CPU        ClkMHz  Cache      SPECint SPECfp  Info  Source
>Name              (NUMx)Type ext/in  Ext+I/D    base95  base95  Date  Obtained
>================= ========== ======= ========== ======= ======= ===== =========

>Intel Alder       PentPro    150     256+8/8      6.08    4.76  Jan96 www.intel
>Intel Alder       PentPro    166     512+8/8      7.11    5.47  Jan96 www.intel
>Intel Alder       PentPro    180     256+8/8      7.29    5.40  Jan96 www.intel
>Intel Alder       PentPro    200     256+8/8      8.09    5.99  Jan96 www.intel
>Intel PD440FX     PentiumII  66/233  512+16/16    9.49    5.91  May97 www.intel
>Intel PD440FX     PentiumII  66/266  512+16/16   10.8     6.43  May97 www.intel

>Now there are tons of disclaimers, and these are a function of
>system, but the bottom line appears to be that there is no gain
>from the 233 for floating point and maybe a 7% gain with the 266 over
>the 200.

>Our decision for numerical modelling, is that any gains in the 266 are
>in no way related to the high cost, and we are trying to obtain enough
>PP 200's, before some joker at Intel decides to add an MMX processor
>to it as well!!

Indeed. I'd love to see SPEC numbers for the PPro 200 with 512 kB L2
cache ... Too bad Intel doesn't like it enough (or doesn't know how)
to give it another 100 MHz or so .... Stefan

--
============================================================================
Stefan A. Deutscher                     |  (+1-423-)   voice      fax
The University of Tennessee, Knoxville  |  UTK  :     974-7838  974-7843
Department of Physics and Astronomy     |  ORNL :     574-5897  574-1118
401, A. H. Nielsen Building             |  home :     522-7845  522-7845

============================================================================
Dictated with OS/2 Warp 4 VoiceType dictation. Using slrn 0.9.3.2 offline.



Fri, 03 Mar 2000 03:00:00 GMT  
 
 [ 13 post ] 

 Relevant Pages 

1. Pentium Pro & Pentium II instruction decomposition (uops)

2. Speed on Pentium Pro vs Pentium?

3. RDTSC on Pentium and Pentium Pro for Timing Accuracy

4. Pentium/Pentium Pro simulator

5. Pentium/Pentium Pro simulator

6. Optimized Pentium code on a Pentium Pro

7. Pentium Pro not faster than Pentium?

8. Comparative perf of Pentium/Pentium II under J?

9. Pentium/Pentium II asm ?

10. Pentium & Pentium II specs

11. Pentium & Pentium II spec

12. FORTRAN on dual CPU Pentium II or Pentium III

 

 
Powered by phpBB® Forum Software