Which is better, IYHO
Philip C. Cavanagh
When I eventually get back to work, I'll be looking into investing in a
motherboard to play chess and do work on.
WHICH IS FASTER??? 486DX4/80 mHz or P-60???
Both will run a minimum of 8 megs, but would like to run a base memory
set of 12 megs (I've heard Chess Assistant requires an ideal 12 meg
environment).
I plan to run Fritz 3 or the new Rebel 7.00 (I'd like to run CM5000, but
why do I have to run Windows or OS/2 just for 1 or 2 programs; the other
is Chessmaster 4000 Turbo).
Appreciate any help, thanks!
--
Philip Cavanagh: ui...@freenet.victoria.bc.ca Support Free Chess Servers!!
---------------------------------------------------------------------------
EICS: daimi.aau.dk 5000 FICS: ics.onenet.net 5000 FICS: getnet.com 4000
Joe Stella
>WHICH IS FASTER??? 486DX4/80 mHz or P-60???
I remember seeing a benchmark test which showed that a P-60 was slightly
faster than a DX4/100.
Also, Pentium-optimized chess programs will probably come out soon, so I
think the Pentium is a better choice.
I don't know how much more expensive a P-75 is over a P-60, but the P-75
uses 3 volt technology and it is worth it if you can get one.
Joe S.
Al Cargill
ui...@freenet.Victoria.BC.CA "Philip C. Cavanagh" writes:
> When I eventually get back to work, I'll be looking into investing in a
> motherboard to play chess and do work on.
>
> WHICH IS FASTER??? 486DX4/80 mHz or P-60???
>
I don't think there is such a chip as the DX4/80 - there is however
an AMD DX4/120 which I would have thought would be a very nice chip!
I personally wouldn't go for a Pentium 60 - hot 5V chip with most m/b
which take it not upgradable to the newer Pentiums. Again I would aim
a little higher with the P75 or P90. I would have thought in the US
that the difference between a P60 and the other two would be low
enough to go for the higher spec one. NOW! As to which would be
better for chess between a DX4/120 and a P75 - no help here I am
afraid - no Pentium to test on :-(. My gut feeling is that there
would not be a lot in it and the 486CPU plus motherboard is going to
be significantly cheaper than the Pentium combo so I would go for the
486 system and spend the difference on extra RAM, graphics card or CA
(or even a bigger hard drive for all the games you get off the
Internet!)
Al
Halibut
: When I eventually get back to work, I'll be looking into investing in a
: motherboard to play chess and do work on.
: WHICH IS FASTER??? 486DX4/80 mHz or P-60???
When the Pentium 60 came out, it was hyped by Intel and the
magazines as being about double the effective speed of the 486. This was
an exaggeration, since even the Pentium 66 is not twice as fast as a
dx2/66. However, I imagine that the P-60 is faster than a 486/80.
: Both will run a minimum of 8 megs, but would like to run a base memory
: set of 12 megs (I've heard Chess Assistant requires an ideal 12 meg
: environment).
Good choice. Actually, Chess Assistant does not require anywhere
near that. However, for maximum speed and strength of ches playing
programs, having lots of RAM is useful.
:
Halibut
: Also, Pentium-optimized chess programs will probably come out soon, so I
: think the Pentium is a better choice.
Where did you hear this? For which programs?
And how does something become optimized for the Pentium?
Joe Stella
>Joe Stella (jo...@ultranet.com) wrote:
I didn't exactly *hear* it. I am just going by past experience. When
the 386 came out, the best programs were optimized for the 386 and didn't
run anymore on the 286. This will probably happen with the Pentium. The
difference between the P5 and the 486 is not as great as the difference
between the 386 and 286, so I expect the P5 optimized programs will
still run on the 486 but not as fast.
A program can be "optimized" for a given processor by someone who has a
working knowledge of how that processor schedules and executes instructions.
Instructions can be placed in the correct order so that parallel execution
can be utilized. That's just one thing that can be done; there are others.
If a processor has new instructions that the old one does not have, then
it is possible to write a program that won't run on the old processor. If
the new instructions increase the efficiency of the program, they will
probably be used and then the program won't run on the 486 at all (which was
the case for 286->386).
Joe S.
Jim Thevenot
Philip C. Cavanagh <ui...@freenet.Victoria.BC.CA> wrote:
>
>When I eventually get back to work, I'll be looking into investing in a
>motherboard to play chess and do work on.
>
>WHICH IS FASTER??? 486DX4/80 mHz or P-60???
>
a P60 depending upon the motherboard. The AMD 486DX4/120 is pretty much
the same as a P-75. Now this is for integer math only (I don't think you
need floating point for chess anyway). I would guess that a DX2/80 (it's
only doubled(dx2) not trippled (dx4)) would be slower than a P-60 or 66 for
most applications. The only way it would be faster is if the program didn't
read data/instructions from outside it's cache very often. So it the DX2/80
didn't have to slow down to read from ram then it would appear to be a DX80.
which would be faster than a P-60. But as has been previously mentioned I
wouldn't recomend a P-60/66 to anyone. It runs hot and slow. Spend the
extra money and get a 75/90/100 or whatever pentium or go with a dx4/100.
>Both will run a minimum of 8 megs, but would like to run a base memory
>set of 12 megs (I've heard Chess Assistant requires an ideal 12 meg
>environment).
>
The processor really doesn't 'require' a certain amount of ram to run, the
programs themselves do. Remember that you will need to add 72-pin simms
in a Pentium in pairs but you can add them singly in a 486 board (assuming
it uses 72 pin simms)
>I plan to run Fritz 3 or the new Rebel 7.00 (I'd like to run CM5000, but
>why do I have to run Windows or OS/2 just for 1 or 2 programs; the other
>is Chessmaster 4000 Turbo).
Unfortunately you have to buy the platform for the programs not the other
way around. That is why people buy Intel vs. Mac based machines in the
first place (or the other way around).
>
>Appreciate any help, thanks!
>
Again, look at either a faster pentium or a faster 486. The price difference
isn't all that great between the slower versions. (486DX4/100 & boards go
for about $220 -$250 or so on the comp.whatever.forsale.motherboards)
Jim T.
Al Cargill
jo...@ultranet.com "Joe Stella" writes:
> >: Also, Pentium-optimized chess programs will probably come out soon, so I
> >: think the Pentium is a better choice.
> >
> > Where did you hear this? For which programs?
>
> > And how does something become optimized for the Pentium?
>
> I didn't exactly *hear* it. I am just going by past experience. When
> the 386 came out, the best programs were optimized for the 386 and didn't
> run anymore on the 286. This will probably happen with the Pentium. The
> difference between the P5 and the 486 is not as great as the difference
> between the 386 and 286, so I expect the P5 optimized programs will
> still run on the 486 but not as fast.
>
I think it will be a long time until home chess computer is optimised
to P5 or P6 etc. such that it won't run on 386/486 CPUs. One of the
huge spped advantages of the Pentium (and above) CPUs is the floating
point maths side is very much faster than the 486 and below - however
I believe this is not of much use to chess programs? I wouldn't pick
your CPU on what be written for it in 3 or 4 years time as the type
of CPU that will be entry level will be far removed from the P5! Go
for the best you can afford and don't get too paranoid about what's
around the corner as it's bound to be a lot quicker!
Seriously the speed of the CPU, while important, doesn't make
a huge difference to the playing standard of most chess software (it
obviously cuts down the time to analyse games accordingly however).
From Eric Hallsworths magazine the average grade difference between
programs running on 486DX2/66 and Pentium 90s is around 80 Elo. This
is probably less than the jump you would get from upgrading your
chess program to the latest version (except for Genius2 to 3!).
Further someone mentioned RAM as being important for chess
playing programs. The ability to set aside RAM for hash tables does
make a difference to the speed of moves but it doesn't seem to work
equally for the software I have - Genius2 uses the extra RAM very
well and gets significant increases in speed, but both MCP4 and
Hiarcs don't really seem to use the extra RAM that well (perhaps the
programs are written to use hash differently). By that I mean that
(quoting figures off the top of my head - may not be totally
accurate) that Genius was something like 25-30% faster with 7Mb hash
whilst MCP4 and Hiarcs were something like 5% faster. CM4000 on the
other hand loves RAM or at least hates running in 4Mb but I guess
that is the case for most Windows programs. Your database programs
will benefit from extra RAM by you being able to allocate more RAM to
caching software such as Smartdrive, but if you are looking through a
large database which is bigger than the cache size it won't help so
all you 500,000 game database owners need loads of RAM to use caching!
Sorry to ramble on - I would like to get a posting together
to try and sort out speed differences due to CPUs soon - maybe in a
couple of days
Al
John G. Alvord
is a second execution stream which will operate in complete parallel,
which could effectively double performance. The actual conditions
which permit this to happen are highly specialized... for example the
second instruction stream has to fit entirely within the I-cache. So
it is a tough, but not impossible job.
The P6 will be another optimization field day, with 4 potential
execution streams.
john alvord
Ian Kennedy
>to P5 or P6 etc. such that it won't run on 386/486 CPUs. One of the
>huge spped advantages of the Pentium (and above) CPUs is the floating
>point maths side is very much faster than the 486 and below - however
>I believe this is not of much use to chess programs?
Correct. Floating point maths is incredibly slow - when I changed the
early prototype of my program Psycho from a floating point to an integer
evaluation score, it speeded up by a factor of 10! Don't be misled by the
evals displayed by commercial programs which are displayed in 1/100ths of
pawn units. You can safely assume that these are all calculated using
integer maths and just displayed as a decimal at the end.
Ian Kennedy