3DMark Vantage CPU score - how reliable for a speed rating?

  • Thread starter Thread starter RayLopez99
  • Start date Start date
R

RayLopez99

The below chip has a 3DMark Vantage CPU score 3x my present Core 2 Duo
chip. Does this mean that my system will be three times faster for
everyday work? So instead of waiting three seconds I only wait one
second? Obviously let's ignore network latency issues.

RL

Intel Core 2 Extreme QX9650 $1069.00 Product details Tom's Hardware
3DMark Vantage - CPU Score 12593
 
The below chip has a 3DMark Vantage CPU score 3x my present Core 2 Duo
chip. Does this mean that my system will be three times faster for
everyday work? So instead of waiting three seconds I only wait one
second? Obviously let's ignore network latency issues.
Click to expand...
No as its a synthetic artifical benchmark.

Far better is to get onto Tomshardware.com and look at their CPU charts
which do REAL WORLD APPLICATION tests.

Also remember it'll only speed up the processing. It won't speed up
reading from the hard drive.
 
RayLopez99 said:
The below chip has a 3DMark Vantage CPU score 3x my present Core 2 Duo
chip. Does this mean that my system will be three times faster for
everyday work? So instead of waiting three seconds I only wait one
second? Obviously let's ignore network latency issues.

RL

Intel Core 2 Extreme QX9650 $1069.00 Product details Tom's Hardware
3DMark Vantage - CPU Score 12593
Click to expand...

A better metric, is to compare SuperPI benchmarks for the two processors.

SuperPI is single threaded, and will give a pessimistic estimate of the
speedup.

An even easier way to compare, is to take the clock speed ratio
(since both of your examples share architectural characteristics).
If the original Core2 ran at 2.4GHz and the new one runs at
3.0Ghz, then roughly speaking, the *minimum* improvement to be
expected is 3.0/2.4 . And for some people, they can get there
for free, by overclocking the core of their old processor.

Some modern multimedia software is multithreaded, but a lot
of the other things you do with a computer are single threaded.
Consequently, for those, you would not expect to see a big speedup.

Multithreading is the key. Without it, the $1069 processor would be
wasted.

You really need to consider two kinds of benchmarks, the single
threaded ones, and the multithreaded ones. Tomshardware has only
the latter type. Computing activities can fall into either camp.
On my machine, the majority of applications do not peg all the
cores at the same time, so I rely more on single threaded benchmarks
for performance estimates. While certain background activities can occupy
a second core, I don't count that as significant.

You cannot compare clock rates, between processors with different
architectures. For example, I have a P4 running at 3.1GHz and
also an AthlonXP 3200+ running at 2.2GHz, and they do the following
benchmark in 45 to 50 seconds. The Core2, using a single core,
blows away those results. The processor used here is an E4700
at 2.6GHz nominal. The core clock is lower than the P4 at
3.1GHz, and yet it is significantly faster.

-------------------------------------------------------SuperPI Memtest 1.65
-------------------------------------------------------1M (sec) Bandwidth
-------------------------------------------------------lower is (MB/sec)
------------------------------------------------------- better

200 x 13 = 2.60GHz, FSB800, DDR2-533, Single channel 24.05 2203
(stock) Dual channel 22.87 2668

266 x 10 = 2.66GHz, FSB1066, DDR2-533, Single channel 23.47 ----
Dual channel 22.52 ----

266 x 13 = 3.46GHz, FSB1066, DDR2-533, Single channel 19.37 2419
Dual channel 18.42 3305

The SuperPI benchmark is available here. (The originating web site
is sending RST packets right now, so use the archived copy, and
download the super_pi_mod-1.5.zip file.)

http://web.archive.org/web/20070823114006/www.xtremesystems.com/pi/

If you identify your processor model, there is a web site that
stores benchmark results from all over the world. And is a good
source for quick estimates.

HTH,
Paul
 
Yes, "some" things will improve, but unless you are pushing some heavy
games, crunching some severe numbers, or doing video transcoding, you
probably won't see much more than a slight improvement. For most current
daily apps like email,surfing, or office suites, anything faster than a
1.5 Ghz single core CPU looks about the same at this time.
Click to expand...

What about compiling computer code? I wonder if this is anything more
than text processing however...most of my stuff compiles pretty fast
as it is.

Thanks for your input. I guess I'll stick to my 'old' (2 years old) h/
w for now.

RL
 
A better metric, is to compare SuperPI benchmarks for the two processors.

SuperPI is single threaded, and will give a pessimistic estimate of the
speedup.
Click to expand...

Most programs are single threaded though, so pessimistic is realistic.
Some modern multimedia software is multithreaded, but a lot
of the other things you do with a computer are single threaded.
Consequently, for those, you would not expect to see a big speedup.

Yes.

Multithreading is the key. Without it, the $1069 processor would be
wasted.

You really need to consider two kinds of benchmarks, the single
threaded ones, and the multithreaded ones. Tomshardware has only
the latter type. Computing activities can fall into either camp.
On my machine, the majority of applications do not peg all the
cores at the same time, so I rely more on single threaded benchmarks
for performance estimates. While certain background activities can occupy
a second core, I don't count that as significant.
Click to expand...

Yes, seems that way. When the 64 bit Word comes out, in 15 years, it
will be a different story.
You cannot compare clock rates, between processors with different
architectures. For example, I have a P4 running at 3.1GHz and
also an AthlonXP 3200+ running at 2.2GHz, and they do the following
benchmark in 45 to 50 seconds. The Core2, using a single core,
blows away those results. The processor used here is an E4700
at 2.6GHz nominal. The core clock is lower than the P4 at
3.1GHz, and yet it is significantly faster.
Click to expand...

I would guess it's because of the process node size: 90 nm vs 65 nm
vs 45 nm--the electrons have shorter distances to travel before they
switch, regardless of the clock.

If you identify your processor model, there is a web site that
stores benchmark results from all over the world. And is a good
source for quick estimates.
Click to expand...

Where is that website? Without the website I am reluctant to download
SuperPi.

Tx,

RL
 
RayLopez99 said:
I would guess it's because of the process node size: 90 nm vs 65 nm
vs 45 nm--the electrons have shorter distances to travel before they
switch, regardless of the clock.
Click to expand...

If distance traveled makes a difference, it is (or would be) reflected
in the clock speed.
 
RayLopez99 said:
What about compiling computer code? I wonder if this is anything
more than text processing however...most of my stuff compiles
pretty fast as it is.
Click to expand...

I think that was the point. If there is little room for improvement,
then there will be little improvement at best.

If you want a fact-based decent estimation, use Windows XP
Performance Monitor, Windows Task Manager (the Performance tab), or
some other CPU performance measuring utility. Look to see if CPU
usage is maxed out during a task. If your CPU is overworked (like
stuck at 100% usage for seconds at a time) and if the rest of your
system can handle a faster CPU, then a faster CPU will provide a
significant or major boost.

The performance increase is very difficult to figure.

If you have lots of money or if you can use a very small primary
hard drive, consider an SSD (solid-state) hard drive too.

Good luck and have fun.
 
Back
Top