My HD6970 Multi-GPU scaling Eyefinity review

[minaelromany]

Hello guys

This is my review that 2 months to complete, hope that you like it :


























The rest of the review is here :

http://www.egypthardware.net/vb/t589.html

It is in Arabic but you can use google translate

[almighty15]

You're CPU imited with 4 cards.

Sandy Bridge is needed to give it that little extra boost as the i7 920 doesn't quite have enough power to handle the driver overhead of 4 cards.

[minaelromany]

You're CPU imited with 4 cards.

Sandy Bridge is needed to give it that little extra boost as the i7 920 doesn't quite have enough power to handle the driver overhead of 4 cards.

That's why I included 5760x1080 benchmark , CPU bottleneck is minimal at such high resolution

[minaelromany]

It seems that Crysis 2 benchmark was flawed , retested and updated charts


Please tell me if there is any other "fishy" result

[PatRaceTin]

good review

[almighty15]

Could you do Crysis 1 benchmark?

Using the bencmark tool can you do a run of Assault Harbor?

[mattkosem]

You're CPU imited with 4 cards.

Sandy Bridge is needed to give it that little extra boost as the i7 920 doesn't quite have enough power to handle the driver overhead of 4 cards.

Eeh, I don't think you can really make a blanket statement like that. [H] found that a SB was needed over their nehalem i7, but their nehalem chip was clocked a paltry 3.6ghz or something. This rig (I'm assuming looking at the sig) is a 4.25Ghz I7 with DDR3 at 2025.

--Matt

[almighty15]

I have a screen grab of a 5+Ghz 2600K holding back Tri-SLI GTX 570's...

So yea, I can says He's CPU limited

[mattkosem]

I have a screen grab of a 5+Ghz 2600K holding back Tri-SLI GTX 570's...

So yea, I can says He's CPU limited

I'll bite. Please show us a screenshot depicting a 5Ghz 2600k as a bottleneck for a tri-sli GTX570 setup. Also, since it wouldn't be relevant otherwise, show me proof that a 4.2Ghz Nehalem chip with 1012mhz ram is a bottleneck for a 6970 quadfire setup. I can see 3 games where scaling stops past 3 cards in the charts above, but I'm interested in seeing how a single screen grab can confirm a cpu bottleneck as the cause.

--Matt

[almighty15]

I'll bite. Please show us a screenshot depicting a 5Ghz 2600k as a bottleneck for a tri-sli GTX570 setup. Also, since it wouldn't be relevant otherwise, show me proof that a 4.2Ghz Nehalem chip with 1012mhz ram is a bottleneck for a 6970 quadfire setup. I can see 3 games where scaling stops past 3 cards in the charts above, but I'm interested in seeing how a single screen grab can confirm a cpu bottleneck as the cause.

--Matt

If the GPU's aren't running at a constant 99% then you're CPU bottlenecked

Sent from my ZTE-BLADE using Tapatalk

[mattkosem]

If the GPU's aren't running at a constant 99% then you're CPU bottlenecked

Sent from my ZTE-BLADE using Tapatalk

That is not an accurate test. Not all conditions produce a 99% constant GPU load.

--Matt

[zalbard]

That is not an accurate test. Not all conditions produce a 99% constant GPU load.

Agreed.
Some games are just poorly coded. Not all of them can actually achieve 99% GPU load in multi-GPU configs (most can't) no matter how good your CPU is.