3dmark and game framerates

To simplify a question - Does a 3dmark06 score of 8000 equate to 4 times the
frame rates in games compared to a 2000 score in 3dmark06?

On Thu, 4 Dec 2008 13:40:25 -0000, "GT"
wrote:

To simplify a question - Does a 3dmark06 score of 8000 equate to 4 times the
frame rates in games compared to a 2000 score in 3dmark06?


no, it weighs more than one subsystem, only the slowest one
will determine minimal framerate and even then it depends on
the resolution used and the specific game played.

Generally speaking, those with lesser cards just turn down
the eyecandy till framerate is playable so it's not so much
framerate as it is eyecandy per $.

"kony" wrote in message
...
On Thu, 4 Dec 2008 13:40:25 -0000, "GT"
wrote:

To simplify a question - Does a 3dmark06 score of 8000 equate to 4 times
the
frame rates in games compared to a 2000 score in 3dmark06?


no, it weighs more than one subsystem, only the slowest one
will determine minimal framerate and even then it depends on
the resolution used and the specific game played.

Generally speaking, those with lesser cards just turn down
the eyecandy till framerate is playable so it's not so much
framerate as it is eyecandy per $.

Well I agree with the eyecandy statement, but we know what level of eyecandy
we want and the current card can't provide it. We are trying to work out
what card we need to get in order to achieve a decent level of graphics in a
particular game - STALKER. With the current card (an MSI 7600GS silent), we
get frame rates in single figures in STALKER with anything other than
800x600 with minimum settings. We would like to turn the settings up to at
least 1024x768 and preferably 1280x1024 and have the settings on high. We do
however require a quiet system and the current PSU is only 300w, so I was
considering a 4670 which would only draw about 30w more than the current
card. The 3dMark06 results give the above figures (8000 vs 2000 approx), so
I was trying to relate this to frame rates at the same resolution/candy
level. Is there a better benchmark to compare 3d graphics capability (only)?

On Mon, 8 Dec 2008 10:13:22 -0000, "GT"
wrote:

"kony" wrote in message
.. .
On Thu, 4 Dec 2008 13:40:25 -0000, "GT"
wrote:

To simplify a question - Does a 3dmark06 score of 8000 equate to 4 times
the
frame rates in games compared to a 2000 score in 3dmark06?


no, it weighs more than one subsystem, only the slowest one
will determine minimal framerate and even then it depends on
the resolution used and the specific game played.

Generally speaking, those with lesser cards just turn down
the eyecandy till framerate is playable so it's not so much
framerate as it is eyecandy per $.

Well I agree with the eyecandy statement, but we know what level of eyecandy
we want and the current card can't provide it. We are trying to work out
what card we need to get in order to achieve a decent level of graphics in a
particular game - STALKER. With the current card (an MSI 7600GS silent), we
get frame rates in single figures in STALKER with anything other than
800x600 with minimum settings. We would like to turn the settings up to at
least 1024x768 and preferably 1280x1024 and have the settings on high. We do
however require a quiet system and the current PSU is only 300w, so I was
considering a 4670 which would only draw about 30w more than the current
card. The 3dMark06 results give the above figures (8000 vs 2000 approx), so
I was trying to relate this to frame rates at the same resolution/candy
level. Is there a better benchmark to compare 3d graphics capability (only)?


Your best bet would be Googling for some stalker benchmarks,
then if you don't see any that include a 4670, at least note
what the cards reviewed scored, then find another benchmark
where that card and the 4670 are pitted against each other.

It's not surprising a 7600GS isn't up to the task with
settings on high, it wasn't a very powerful card even in
it's day and now shaders make a lot of difference in newer
games. Here's one Stalker 4670 benchmark, it looks barely
up to the task.
http://www.guru3d.com/article/ati-ra...4670-review/11

While the PSU is a limitation, noise level doesn't have to
be. Taking the 9600GT in those benchmarks as an example,
at stock speeds many 9600GT can have their fan left at a
mere 20% PWM duty cycle which tends to be practically
inaudible once the cover is on the case, or more practically
the other system parts' fans are probably making at least as
much noise.

"kony" wrote in message
...
On Mon, 8 Dec 2008 10:13:22 -0000, "GT"
wrote:

"kony" wrote in message
. ..
On Thu, 4 Dec 2008 13:40:25 -0000, "GT"
wrote:

To simplify a question - Does a 3dmark06 score of 8000 equate to 4 times
the
frame rates in games compared to a 2000 score in 3dmark06?


no, it weighs more than one subsystem, only the slowest one
will determine minimal framerate and even then it depends on
the resolution used and the specific game played.

Generally speaking, those with lesser cards just turn down
the eyecandy till framerate is playable so it's not so much
framerate as it is eyecandy per $.

Well I agree with the eyecandy statement, but we know what level of
eyecandy
we want and the current card can't provide it. We are trying to work out
what card we need to get in order to achieve a decent level of graphics in
a
particular game - STALKER. With the current card (an MSI 7600GS silent),
we
get frame rates in single figures in STALKER with anything other than
800x600 with minimum settings. We would like to turn the settings up to at
least 1024x768 and preferably 1280x1024 and have the settings on high. We
do
however require a quiet system and the current PSU is only 300w, so I was
considering a 4670 which would only draw about 30w more than the current
card. The 3dMark06 results give the above figures (8000 vs 2000 approx),
so
I was trying to relate this to frame rates at the same resolution/candy
level. Is there a better benchmark to compare 3d graphics capability
(only)?


Your best bet would be Googling for some stalker benchmarks,
then if you don't see any that include a 4670, at least note
what the cards reviewed scored, then find another benchmark
where that card and the 4670 are pitted against each other.

It's not surprising a 7600GS isn't up to the task with
settings on high, it wasn't a very powerful card even in
it's day and now shaders make a lot of difference in newer
games. Here's one Stalker 4670 benchmark, it looks barely
up to the task.
http://www.guru3d.com/article/ati-ra...4670-review/11

While the PSU is a limitation, noise level doesn't have to
be. Taking the 9600GT in those benchmarks as an example,
at stock speeds many 9600GT can have their fan left at a
mere 20% PWM duty cycle which tends to be practically
inaudible once the cover is on the case, or more practically
the other system parts' fans are probably making at least as
much noise.

Interesting thanks - I actually though the 4670 outclassed the 9600gt due to
the latter's age but it clearly looks more powerful in Stalker. I had not
researched benchmarks for other cards due to their power requirements and
heat output (noise). I suspect the PSU might be on its last legs anyway (it
takes a few second to accept a power button press after mains power comes on
these days), so we should perhaps look to replace that too. This would
remove the PSU restrutcion, but it also of course increases the budget along
with a resulting time delay as funds are pilfered from other budgets!