News - The Order of Play - Radeon vs. GeForce in 3DMark

[HEXUS]

How does your own graphics card compare?

Read more.

[Tabbykatze]

Interesting....

*tin hat on* However a fair test for the most transparent unbiased results I would rather see that data sourced from a third party testing not sponsored by one of the competing companies.

I can hear the "faked/skewed results" shout outs already.

[Biscuit]

For full disclosure, this article is sponsored by AMD.

Well its good that you told us, very good... but it also adds some doubt into the results IMO. Would like to see this test done completely independently.

[DisturbedJim83]

Ha Suck my 290X Team you looooooooooose,just know team green fanboys gonna cry fowl now they can justify the £80+ extra the 780Ti costs over the 290X

[Tarinder]

The tests were 100 per cent run by this, on our hardware, which AMD had no input into.

[kalniel]

Shows why synthentic benchmarks aren't much use No point thinking 'I've got the best value card at 3Dmark' when you're chugging away sub 30fps in a real game.

For us Brits, I've created a version of the 'value' chart based on the UK prices as given later in the article (seeing as Hexus seems to have based theirs on the US ones, oddly):

[Wozza63]

I got a R9 280 and managed to overclock it massively (nearly 25%), it doesn't cause any heat issues and thats at standard voltage, could get it further with a little bit extra voltage. And with that overclock it makes it great value as it outperforms a 280x and 770

[semitope]

Shows why synthentic benchmarks aren't much use

For us Brits, I've created a version of the 'value' chart based on the UK prices as given later in the article (seeing as Hexus seems to have based on on the US ones, oddly):

Actually synthethics help level the playing field. No Gameworks or other developer manufacturer bias. It helps show which is more objectively faster. For games tho you still have to do the game benchmarks.

[kalniel]

Actually synthethics help level the playing field. No Gameworks or other developer manufacturer bias. It helps show which is more objectively faster.

I agree it'll say which is objectively faster at synthetics. But there is afaik no quality of experience adjustment - a proper test will take into account that optimal fps for a game is greater than 30 but shows diminishing returns above a monitors refresh rate.

And synthetics can be optimised for in a way that doesn't help with real world gaming (just as individual games can be). That's why suites of games are useful for reviews - it's work to test on, but that's why we pay (via ad views etc.) someone to do it for us

[nsplock]

i got 10077 on my R9 290 xD

[CeeGee]

Having these scores all in one place is nice as a sanity check for anyone that thinks they have a problem with their setup. You can easily see what ballpark figures your particular card should be putting out.

I've found browsing the 3dmark scores brings up loads of invalid scores and other shenanigans.

[shaithis]

Who cares about 3D Mark scores? If AMD really want to show us what good value they are, they should have benchmarked against the last years top releases.

[KeyboardDemon]

I was really confused by the results, here are the results I got when I tested my PC using my former R9 290X vs my current GTX 780 Ti, as you can see I tested both on the same day in the same rig with no overclocks on the CPU and cards were test as they would run out of the box. My results showed the GTX 780 Ti to be around 13% than the R9 290X the Hexus results show the R9 as being around 0.3% than the GTX 780 Ti.

Could that difference be purely because my GTX 780 Ti is a factory overclocked model?

I can only assume it is, but that would then make the cards very closely matched, at least in terms of synthetics anyway.

[kalniel]

Could that difference be purely because my GTX 780 Ti is a factory overclocked model?

I can only assume it is, but that would then make the cards very closely matched, at least in terms of synthetics anyway.

Most likely - clock speed has quite a large impact on synthetic scores - there's a 3Dmark score thread here somewhere and you can see how overclocking various cards really increases the score.

[Biscuit]

I was really confused by the results, here are the results when I test my PC using my former R9 290X vs my current GTX 780 Ti, as you can see I tested both on the same day in the same rig with no overclocks on the CPU and cards were test as they would run out of the box. My results showed the GTX 780 Ti to be around 13% than the R9 290X the Hexus results show the R9 as being around 0.3% than the GTX 780 Ti.

Could that difference be purely because my GTX 780 Ti is a factory overclocked model?

I can only assume it is, but that would then make the cards very closely matched, at least in terms of synthetics anyway.

I think the point of the article is that when using the most recent drivers, the performance is significantly different. So much so that AMD are back in the lead.

[KeyboardDemon]

I think the point of the article is that when using the most recent drivers, the performance is significantly different. So much so that AMD are back in the lead.

Yeah, which is nice to see, though a 0.3% lead on its own doesn't really seem like much, but add the price advantage to that too and I know I would definitely be thinking about buying from the red team if I were buying today, but I wouldn't be buying a reference GPU, it would have to be something like the Gigabyte or Sapphire overclocked tripple fan models as these are now around £400 which works out around £55 less than a GTX 780 Ti.

I just ran the test on my PC and scored 10301 points, running a 3770K with 8gb, I would be interested in seeing the 3dMark page for the GTX 780 Ti and the PC they tested with just to compare the scores against mine.