Don't get 48xx series card unless you have fast cpu
I have 3.4GHZ C2D so I am ok but a friend went from 3870 to 4850 on AMD
2.4GHZ cpu and the stress test in HL-2 Lost Coast shows hardly any
difference for him. For him it is not a worthwhile upgrade so think about
that if you are planning to get a 4850 or 4870 and you don't have a 3+GHZ
cpu. Maybe there is somehting wrong with the HL-2 stress test or something
because I saw a guy post at HardOCP that he got a higher score on his old
8800GT than his new 4850.
Re: Don't get 48xx series card unless you have fast cpu
"Dauphin de Viennois" <spamtrap@on.here> wrote in message
news:fKsek.67796$kx.56473@pd7urf3no...
>I have 3.4GHZ C2D so I am ok but a friend went from 3870 to 4850 on AMD
> 2.4GHZ cpu and the stress test in HL-2 Lost Coast shows hardly any
> difference for him.
This is why traditional timedemo benchmarking is getting silly. HL2 Lost
Coast is from the Radeon X800 era. A 3870 would be already stupid-fast in
this benchmark with every in-game option maxed out. Of course a 4850
wouldn't show much of an improvement.
> For him it is not a worthwhile upgrade so think about that if you are
> planning to get a 4850 or 4870 and you don't have a 3+GHZ cpu. Maybe there
> is somehting wrong with the HL-2 stress test or something because I saw a
> guy post at HardOCP that he got a higher score on his old 8800GT than his
> new 4850.
So what would be a worthwhile upgrade? How is getting, say, 120 fps in HL2
Lost Coast going to improve his experience in this game compared to getting
80 fps?
--
"War is the continuation of politics by other means.
It can therefore be said that politics is war without
bloodshed while war is politics with bloodshed."
Re: Don't get 48xx series card unless you have fast cpu
"rms" <rsquires@REMOVEflashMOO.net> wrote in news:ibuek.4900$np7.2837
@flpi149.ffdc.sbc.com:
>
> Oh please. And people complain about 3dmark06?
>
> rms
>
Oh please, what? It's a good guage of overall system performance, including
video card. Tweak one part and it will show up in both HL-2 Stress Test and
3DMark.
Re: Don't get 48xx series card unless you have fast cpu
"First of One" <root@127.0.0.1> wrote in
news:J7WdnW5V_rYnC-fVnZ2dnUVZ_r3inZ2d@giganews.com:
> This is why traditional timedemo benchmarking is getting silly. HL2
> Lost Coast is from the Radeon X800 era. A 3870 would be already
> stupid-fast in this benchmark with every in-game option maxed out. Of
> course a 4850 wouldn't show much of an improvement.
Why not? My 4870 shows AVG 154.91. HL-2 uses a lot of advanced graphics
features so who cares if it was made a few years back? It's still a good
and valid benchmark.
> So what would be a worthwhile upgrade? How is getting, say, 120 fps in
> HL2 Lost Coast going to improve his experience in this game compared
> to getting 80 fps?
That's not the point. His 3870 scored 80 and 4850 only improved it by 9fps.
Review benchmarks show 4850 being almost twice as fast as 3870 so obviously
his cpu is holding the 4850 back from being able to performa as well as it
can. That's the whole point of my post and you and the other dumbass just
didn't get it.
Re: Don't get 48xx series card unless you have fast cpu
"Dauphin de Viennois" <spamtrap@on.here> wrote in message
news:bizek.368$nD.297@pd7urf1no...
> Why not? My 4870 shows AVG 154.91. HL-2 uses a lot of advanced graphics
> features so who cares if it was made a few years back? It's still a good
> and valid benchmark.
The whole practice of running canned benchmarks/timedemos and comparing
triple-digit framerate "scores" had been picked apart years ago, simply
because more demanding games don't necessarily scale the same way.
Looking in the other threads, your 4870 allows you to run Crysis at higher
settings and still maintain good performance than your previous 8800GT
could. This is how a faster card benefits you, by *enabling* higher visual
quality. There is only one meanful way to compare cards: crank up the
resolution/AA and game settings until one card can't deliver playable
framerates. If both cards can deliver good performance at maximum settings,
then the game is useless as a benchmark.
> That's not the point. His 3870 scored 80 and 4850 only improved it by
> 9fps.
> Review benchmarks show 4850 being almost twice as fast as 3870 so
> obviously
> his cpu is holding the 4850 back from being able to performa as well as it
> can. That's the whole point of my post and you and the other dumbass just
> didn't get it.
Just because the CPU is the bottleneck in this game doesn't mean it will
still be a bottleneck in a more modern, shader-heavy title like Oblivion or
Crysis. Think for a moment before you say something foolish next time.
--
"War is the continuation of politics by other means.
It can therefore be said that politics is war without
bloodshed while war is politics with bloodshed."
Re: Don't get 48xx series card unless you have fast cpu
"First of One" <root@127.0.0.1> wrote in
news:kuOdncyIM8NkdufVnZ2dnUVZ_h_inZ2d@giganews.com :
> Just because the CPU is the bottleneck in this game doesn't mean it
> will still be a bottleneck in a more modern, shader-heavy title like
> Oblivion or Crysis. Think for a moment before you say something
> foolish next time.
>
Oblivion is more shader dependant than HL2? And you dare to call me a fool?
Oblivion uses speedtrees which is known to be an inefficient method of
doing trees, that's why Oblivion has so much stutttering. Oblivion runs
faster on my 4870 but the stuttering as it loads new textures is just as
bad as it was on my 8800GT. My first post is still valid and you still
don't get the point I was making. If you want to take advantage of your
shiny new 4850 or 4870 then you better make sure you have a 3.0+ GHZ cpu or
else you may be in for a bit of a dissapointment compared to your last
card. Your cpu may not be enough to feed these new video cards efficiently.
That's what the HL-2 stress test tells me and it is confirmed in many of
the reviews of these cards around the web. I have been running benchmarks
on video cards since the days of Hercules Stingray, Diamond Stealth etc. so
screw you.
Re: Don't get 48xx series card unless you have fast cpu
"Dauphin de Viennois" <spamtrap@on.here> wrote in message
news:ceSek.99329$gc5.70742@pd7urf2no...
> Oblivion is more shader dependant than HL2? And you dare to call me a
> fool? Oblivion uses speedtrees which is known to be an inefficient method
> of doing trees, that's why Oblivion has so much stutttering. Oblivion runs
> faster on my 4870 but the stuttering as it loads new textures is just as
> bad as it was on my 8800GT.
What's considered common knowledge apparently eludes you. You seem to lack a
basic understanding of what shaders do. Oblivion relies on shaders to do all
the dynamic shadows, HDR lighting and heat shimmers.
Take a look here for example: http://www.anandtech.com/video/showdoc.aspx?i=2746&p=4
The X1900XT has identical fill rate and memory bandwidth as the X1800XT, but
triple the number of pixel shader units. As a result the X1900XT is 35%
faster than the X1800XT in the Oblivion Gate test.
Incidentally shader performance is the biggest jump in going from the HD3870
to HD4870. Number of unified shaders (aka stream processors) went from 320
to 800. The reason ATi/AMD made this architectural decision was because most
modern games are shader-constrained, not CPU-constrained. This has been a
trend since the F.E.A.R. days.
HL2 Lost Coast was popular a few years ago because it did a "half-assed" way
of getting HDR lighting with integer blends (rather than FP blends). This
allowed DX9.0b cards like the Radeon X800 to draw the HDR effects, and IIRC
allowed Geforce7-series cards to do it with AA.
> My first post is still valid and you still don't get the point I was
> making. If you want to take advantage of your shiny new 4850 or 4870 then
> you better make sure you have a 3.0+ GHZ cpu or else you may be in for a
> bit of a dissapointment compared to your last card. Your cpu may not be
> enough to feed these new video cards efficiently. That's what the HL-2
> stress test tells me and it is confirmed in many of the reviews of these
> cards around the web. I have been running benchmarks on video cards since
> the days of Hercules Stingray, Diamond Stealth etc. so screw you.
Yep, and you seemed to be stuck in the old ways where frames-per-second was
all that mattered. That Hardforum poster "CamaroZ28" is just as misguided in
trying to "improve" his Lost Coast benchmark scores when it was already at
85 fps. The HardOCP editors don't even test video cards that way anymore. In
fact, if you look a few posts down there was a voice of reason: http://www.hardforum.com/showpost.ph...8&postcount=19
At the resolutions where most people like to play games at, the CPU just
doesn't matter as much.
--
"War is the continuation of politics by other means.
It can therefore be said that politics is war without
bloodshed while war is politics with bloodshed."
Don't get 48xx series card unless you have fast cpu 
Linear Mode
