3dmark06 low score with new GTX 460

[Gabb]

So I just got my new GTX 460 768mb for a good $145 deal and swapped my 9800 GTX+ out ( which I gave to a friend ). I know the performance between the two isn't that huge but I was shocked to see my 3dmark06 score only increased from 14538 to 15904 ( approx a 10% increase ). Does this accurately convey the real performance difference between the two or is something up with my 3dmark06 score or possibly gfx card?

I'm running 4 gigs of ram ( 3.25 recognized due to win xp 32 bit ) and a q6600 OCed to 3.0 ghz, the GTX 460 card is left at stock which is apparently boosted to 720/900

[Jarnis]

Don't use 3DMark06 to benchmark DX10/DX11 hardware. It isn't very good at it, being CPU limited in most setups on any card that performs around the level of 9800GTX or thereabouts. It is old and the age is starting to show...

3DMark Vantage would be a much better tool for your purposes.

Edit: Or you can look at this handy list for a rough idea; http://service.futuremark.com/hardware/

(those scores are estimated graphics scores off 3DMark Vantage, and yes, there is a massive difference between those two cards)

[mapesdhs]

I've researched this extensively. Newer cards only help if:

- You're going to be playing newer games which use newer features,

- You're playing older games but will be playing them at a higher resolution and
higher detail settings (eg. old card: 1280x1024, no AA, 4X AF; new card: 1920x1200,
4X AA, 8A AF).

- Both the above! 8)

See my results pages:

http://www.sgidepot.co.uk/misc/pctests.html
http://www.sgidepot.co.uk/misc/stalkercopbench.txt
http://www.sgidepot.co.uk/misc/x3tcbench.txt
http://www.sgidepot.co.uk/misc/uniginebench.txt

3DMark06 is also heavily affected by CPU performance. Look carefully at the first
of the above pages, compare 1 vs. 2 older cards at a low clock, vs. 1 vs. 2 older
cards at a high clock (eg. 4890), look how these data compare to the same
comparisons of 1 vs. 2 newer cards at low vs. high clock (GTX 460).

More data being added this evening, including my own GTX 460 results, which, btw,
might surprise you. ;)

Ian.

PS. Forgot to mention: for modern Intel CPUs, HT does not help. It can result in
higher overall numbers (because of how strongly the CPU result affects the overall
score) but using a higher clock without HT gives better frame rates, eg. examine:

http://3dmark.com/compare/3dm06/15084731/3dm06/15085763

I've checked this with Stalker COP, X3TC and Unigine: none of them benefit from HT.

[treinier]

Im having a similar problem, but then ... worse. I bought a msi n460gtx today and i got a 10400 as a score on 3dmark06. Im using P5K motherboard and e6750 chip 2GB ram. Is this expected with my cpu..? 10k seems a bit to low to me ... btw i have the 259. drivers

[AgentSmith5150]

So I just got my new GTX 460 768mb for a good $145 deal and swapped my 9800 GTX+ out ( which I gave to a friend ). I know the performance between the two isn't that huge but I was shocked to see my 3dmark06 score only increased from 14538 to 15904 ( approx a 10% increase ). Does this accurately convey the real performance difference between the two or is something up with my 3dmark06 score or possibly gfx card?

I'm running 4 gigs of ram ( 3.25 recognized due to win xp 32 bit ) and a q6600 OCed to 3.0 ghz, the GTX 460 card is left at stock which is apparently boosted to 720/900

Im having a similar problem, but then ... worse. I bought a msi n460gtx today and i got a 10400 as a score on 3dmark06. Im using P5K motherboard and e6750 chip 2GB ram. Is this expected with my cpu..? 10k seems a bit to low to me ... btw i have the 259. drivers

Both you guys are being held back by your CPUs. The OP is also being held back by Windows XP (don't know what the second fellow has for OS). I just did an '06 run and got a score of 17,265. And that's with a 460 1gb, 4 gigs RAM, and an E8400 OC'd to 3.6ghz. The E8400 will beat the Q6600 every time in games and benchmarks, due to the newer architecture and much higher clock ability.

Time to upgrade!

[Thierry14]

If you want higher scores a cpu upgrade is indeed what you need.

This what I got with a GTX460 and a i7 930..http://hwbot.org/community/submissio...it_28483_marks

[treinier]

Thanks for the replies.

What cpu is a good choise for my motherboard (775 socket)? (i dont want to upgrade my motherboard)

[mapesdhs]

Here is a larger table of results I've accumulated which I hope you may all find useful:

http://www.sgidepot.co.uk/misc/pctests.html

Real games may behave differently; never, ever confuse overall 3DMark06
scores with how well a system might do for a real game with respect to frame
rates - I can get higher '06 fps scores with my CPU at a higher clock without
HT, even though the overall score may be lower. Thus, I have run numerous
other benchmarks show these effects in greater detail:

http://www.sgidepot.co.uk/misc/stalkercopbench.txt
http://www.sgidepot.co.uk/misc/uniginebench.txt
http://www.sgidepot.co.uk/misc/uniginebench2.txt
http://www.sgidepot.co.uk/misc/uniginebench3.txt
http://www.sgidepot.co.uk/misc/x3tcbench.txt
http://www.sgidepot.co.uk/misc/cinebench.html
http://www.sgidepot.co.uk/misc/viewperf.txt
http://www.sgidepot.co.uk/misc/ptboats.txt
http://www.sgidepot.co.uk/misc/callofjuarez.txt


More data coming soon, including i3 540 and i5 760 results. For example,
it is very possible the i3 540 will give better results for the X3TC test (it seems
to be CPU bottlenecked but doesn't benefit much from multiple cores), while I
expect the 760 to do well for fps numbers in '06 even if the CPU and overall
scores are not as high as the 870.


Summary If you have an older card:

1. A newer card will help if you switch to a higher res, more detail, and/or use
newer games that exploit newer features (DX10/DX11).

2. If you continue playing older games and/or use lower resolutions/detail,
adding a 2nd older card can be a very effective solution and will be a lot
cheaper (a newer card will often be barely any better), but the setup may
suffer if you suddenly decide to play a newer game and/or switch to a much
higher res, etc. - this can be mitigated though if you use or can find an older
card that has 1GB RAM instead of the more common 512MB, though this
may offset the cost saving somewhat.

3. 3DMark06 is severely affected by CPU speed. HT on current Intel CPUs
gives higher CPU scores (and thus often higher overall scores) but frame
rates will be better with HT off and a higher clock - attainable since temps
drop by about 10C without HT in use.

4. Some cards are unexpectedly strong in certain situations, eg. the 4890
for the Stalker COP SUN test at high res. Other times, it's the reverse.
Depends on the game/task. If you're very fond of a particular game, research
review sites to see what card models favour your chosen game.

I'll be adding data for a different model of GTX460 later (Palit with an 800 core)
and will also investigate oc'd results later.


Thierry14 writes:
> This what I got with a GTX460 and a i7 930..

Impressive! What is the Compare URL though please? Is that with a single
GTX460? I'd be most interested to see the fps scores and how they relate
to the CPU scores, compared to my results.

Ian.

[Sendog2011]

Hey Ian, just wondering if my scores sound right... I've looked around all over the place to compare specs and scores and there's huge differences... Kinda thought I'd score higher with my card but it must be because of the processor?

3dMark06 - 12,155
3dMark Vantage - 7510
3dMark11 - 1966

MSi G-Series Laptop
Core i5 2.33 - 430M
4GB DDR3
Radeon HD5870 - 1GB DDR5
Windows 7 64bit home premium

Thanks!
Jason

[mapesdhs]

Sendog2011 writes:
> Hey Ian, just wondering if my scores sound right...

Yes, looks ok. CPU-limited again.


> ... Kinda thought I'd score higher with my card but it must be because of the
> processor?

You should compare the individual results, not the overall score. See how
the game test fps rates compare to similar hardware.

Ian.