So regular Fury's in CF beats 980 Ti's in SLI, eh? O.o

[KarateHottie93]

I found some pretty old benchmarks here that are kinda somewhat pretty goddamn interesting to me. It shows two Fury Tri-X cards beating two slightly OC'd 980 Ti's in CrossFire at 4K. The best the 980 Ti's did at 4K was tie it on one game.

 

If a true 4K experience pretty much requires a multi-GPU setup, why in the blue hell would I want two 980 Ti's over two air cooled, cheaper Fury cards?

 

http://www.pcworld.com/article/2947547/components-graphics/amd-radeon-fury-crossfire-review-2-fast-2-furious.html

 

Also, before you count the overclocking ability of the 980 Ti, keep in mind that unlocking the extra units on the Fury has also shown to be possible and most that try have documented some decent success, so don't factor in something contributed to luck of the draw.

[Dabombinable]

I found some pretty old benchmarks here that are kinda somewhat pretty goddamn interesting to me. It shows two Fury Tri-X cards beating two slightly OC'd 980 Ti's in CrossFire at 4K. The best the 980 Ti's did at 4K was tie it on one game.

 

If a true 4K experience pretty much requires a multi-GPU setup, why in the blue hell would I want two 980 Ti's over two air cooled, cheaper Fury cards?

 

http://www.pcworld.com/article/2947547/components-graphics/amd-radeon-fury-crossfire-review-2-fast-2-furious.html

AMD's Crossfire isn't crippled by a bridge unlike Nvidia's SLI, so in other words the scaling is far better, with up to 99% scaling using the Fury X.

[Enderman]

look at the minimum fps

CF is way worse than the 980ti sli

 

also the 980ti SLI wins in like 75% of the benchmarks, so if you really think the fury X is better you need to look at more than one image

[bobhays]

Paul's Video says a slightly different story with Fury X's.

[KarateHottie93]

 

look at the minimum fps
CF is way worse than the 980ti sli

Gonna play the "minimum fps" card, eh?

 

I'd take higher FPS 99% of the time over a slight split second drop that you don't even notice ever 10 minutes.

 

 

Paul's Video says a slightly different story with Fury X's.

Nobody likes the Newegg guy anyways.

[Kinda Bottlenecked]

Nobody likes the Newegg guy anyways.

 

WHAT?????????>????":???????????????

[79wjd]

AMD's Crossfire isn't crippled by a bridge unlike Nvidia's SLI, so in other words the scaling is far better, with up to 99% scaling using the Fury X.

Isn't it the other way around? Isn't the reason that Nvidia stuck with SLI bridges, rather than doing it over the PCIE bus, that SLI bridges are still the superior alternative. 

[Enderman]

Gonna play the "minimum fps" card, eh?

 

I'd take higher FPS 99% of the time over a slight split second drop that you don't even notice ever 10 minutes.

actually fps drops are what ruin a gaming experience

linus has talked about this a lot, idk how you could miss it

 

stuttering and fps drops are freaking annoying and thats why gsync and freesync were made

because fps drops happen

 

it doesnt matter if you have 65 or 75fps if your monitor can only do 60, but if your fps drops below 30 for even a tenth of a second its really obvious

 

 

oh and btw the 980ti SLI is better in three quarters of the benchmarks, so even if you care about avg fps more then the 980ti sli is still better...

[KarateHottie93]

actually fps drops are what ruin a gaming experience

linus has talked about this a lot, idk how you could miss it

 

stuttering and fps drops are freaking annoying and thats why gsync and freesync were made

because fps drops happen

 

it doesnt matter if you have 65 or 75fps if your monitor can only do 60, but if your fps drops below 30 for even a tenth of a second its really obvious

Maybe it's just a personal thing but a split second drop never really bothered me as long as it wasn't something that happened often.

 

[D2ultima]

Well AMD cards have done better at higher resolutions than nVidia cards for quite some time.

 

Kepler was made for 1080p and even 1440p saw AMD cards catching up/pulling ahead. 4K was a mess for nVidia.

 

Maxwell can do 1440p well, but 4K is still not in the scope of their cards' designs. AMD cards which lose at 1080p/1440p still easily catch up at 4K or beat them.

 

The problem with AMD's multi-GPU is that it only works in fullscreen and is supported in less games. When CrossfireX is supported, it's great. Beautiful scaling. But frametime variances are still an issue (especially with new cards in multi-GPU) and while they may eventually get fixed properly, it doesn't mean at all that AMD is automatically the best choice.

 

Then further that, remember that AMD cards are also bad at tessellation. In games where tessellation is barely used, they're fine. But there's quite a few games out there (not on that list) that have tessellation and will falter even with AMD's 4K advantage.

 

Then to go beyond that, they're using an i7-5960X at an unspecified clockspeed to remove CPU bottlenecks, which is not something most people would have. AMD's DX11 drivers definitely have more of a CPU hit than nVidia's, even at the CPU-relaxing GPU-bottleneck-heavy 4K. Someone with an i5 may very well experience rather different results; the fact that their average GTA V FPS was so high also means the clockspeed was quite high on their CPU itself.

 

Finally, remember that nVidia's drivers there were the beginning of THE problematic driver streak (353.06 and 353.30 are *IDENTICAL* drivers, except for the Batman Arkham Knight SLI profile that nVidia didn't bundle with a GFE update and decided to release a whole new driver for) and such performance isn't indicative of current performance today (testing launch cards with launch drivers is never a good idea to link to months after it took place).

 

Basically, you need to take these things with a word of salt. I'm not saying AMD cards are bad, but the truth is that they're not the best in every situation, and for most purposes, 980Ti is the best single GPU and the most desirable card to use for multi-GPU configs. If you want a couple fury cards and wish to play fullscreen (probably single monitor-only) at 4K, then maybe AMD is for you. But that test doesn't have enough data to satisfy me. I require data to change my views.

[Castdeath97]

Nobody likes the Newegg guy anyways.

Apparently 184 thousand people disagree with you. 

[Lethal Seraph]

1080 1440p @ 980Ti

4k+ Fury X

 

That's the trend that most benchmarks shows.

[Enderman]

Maybe it's just a personal thing but a split second drop never really bothered me as long as it wasn't something that happened often.

either way, the 980ti sli wins in more benchmarks...so...

[KarateHottie93]

Well AMD cards have done better at higher resolutions than nVidia cards for quite some time.

 

Kepler was made for 1080p and even 1440p saw AMD cards catching up/pulling ahead. 4K was a mess for nVidia.

 

Maxwell can do 1440p well, but 4K is still not in the scope of their cards' designs. AMD cards which lose at 1080p/1440p still easily catch up at 4K or beat them.

 

The problem with AMD's multi-GPU is that it only works in fullscreen and is supported in less games. When CrossfireX is supported, it's great. Beautiful scaling. But frametime variances are still an issue (especially with new cards in multi-GPU) and while they may eventually get fixed properly, it doesn't mean at all that AMD is automatically the best choice.

 

Then further that, remember that AMD cards are also bad at tessellation. In games where tessellation is barely used, they're fine. But there's quite a few games out there (not on that list) that have tessellation and will falter even with AMD's 4K advantage.

 

Then to go beyond that, they're using an i7-5960X at an unspecified clockspeed to remove CPU bottlenecks, which is not something most people would have. AMD's DX11 drivers definitely have more of a CPU hit than nVidia's, even at the CPU-relaxing GPU-bottleneck-heavy 4K. Someone with an i5 may very well experience rather different results; the fact that their average GTA V FPS was so high also means the clockspeed was quite high on their CPU itself.

 

Finally, remember that nVidia's drivers there were the beginning of THE problematic driver streak (353.06 and 353.30 are *IDENTICAL* drivers, except for the Batman Arkham Knight SLI profile that nVidia didn't bundle with a GFE update and decided to release a whole new driver for) and such performance isn't indicative of current performance today (testing launch cards with launch drivers is never a good idea to link to months after it took place).

 

Basically, you need to take these things with a word of salt. I'm not saying AMD cards are bad, but the truth is that they're not the best in every situation, and for most purposes, 980Ti is the best single GPU and the most desirable card to use for multi-GPU configs. If you want a couple fury cards and wish to play fullscreen (probably single monitor-only) at 4K, then maybe AMD is for you. But that test doesn't have enough data to satisfy me. I require data to change my views.

Good and logical explanation, makes sense and I appreciate you taking the time to type that

[KarateHottie93]

either way, the 980ti sli wins in more benchmarks...so...

At 4k where cards like these matter it's pretty close in most scenarios.

[Grrizz]

Maybe it's just a personal thing but a split second drop never really bothered me as long as it wasn't something that happened often.

Minimums tend to happen during high action times which tend to be the times when frame rate really matters

[KarateHottie93]

Minimums tend to happen during high action times which tend to be the times when frame rate really matters

I personally just want to enjoy my brisk stroll through the forest at the maximum possible frames

[FLUFFYJELLO]

WHAT?????????>????":???????????????

We can officially disregard this guy as a troll now.

/s

Everybody likes the newegg guys.

[don_svetlio]

I like to make a correction to a statement a few posts back - regarding AMD's tessellation performance.

The issue with Tess and AMD is mostly in GCN 1.0 - the 7000 and R9 200 series are the ones who take the hit mostly. GCN 1.1 (R9 *90(x)) series or GCN 1,1 have been improved and GCN 1.2 (the 380 and Fury series) have Tessellation on par with Nvidia's - that was one of the main focuses of GCN 1.2 - improving Tess since Nvidia has been pushing it heavily.

As for Fury X vs 980 Ti argument - 980 Ti wins at 1080p and 1440p but is a total mess ar 4K. The Fury X is simply the better option there and has been for quite some time. Ever since the 290X launched Nvidia has been the lesser choice for 4K with even the Titan and 780 Ti losing to the 290X @ 4K back in the day.

[Dabombinable]

Isn't it the other way around? Isn't the reason that Nvidia stuck with SLI bridges, rather than doing it over the PCIE bus, that SLI bridges are still the superior alternative. 

Nope, the bandwidth is extremely low over that SLI bridge.

[Dabombinable]

I like to make a correction to a statement a few posts back - regarding AMD's tessellation performance.

The issue with Tess and AMD is mostly in GCN 1.0 - the 7000 and R9 200 series are the ones who take the hit mostly. GCN 1.1 (R9 *90(x)) series or GCN 1,1 have been improved and GCN 1.2 (the 380 and Fury series) have Tessellation on par with Nvidia's - that was one of the main focuses of GCN 1.2 - improving Tess since Nvidia has been pushing it heavily.

As for Fury X vs 980 Ti argument - 980 Ti wins at 1080p and 1440p but is a total mess ar 4K. The Fury X is simply the better option there and has been for quite some time. Ever since the 290X launched Nvidia has been the lesser choice for 4K with even the Titan and 780 Ti losing to the 290X @ 4K back in the day.

The 380 is the R9 285 rebadged, and the Fury X just uses the same architecture (Tonga) except its been beefed up a lot.

[don_svetlio]

The 380 is the R9 285 rebadged, and the Fury X just uses the same architecture (Tonga) except its been beefed up a lot.

It's definitely not the same - quite a bit beefed up really

Fiji and Tongo are similar but not the same, not by a longshot.

[Dabombinable]

It's definitely not the same - quite a bit beefed up really

Fiji and Tongo are similar but not the same, not by a longshot.

I said the FuryX is beefed up for a reason-they did large tweaks and additions to it-Fiji is still Tonga, just like GM204 and GM104 are both Maxwell.

 

Edit: And the same applies to *cough*Grenada*fuckrebadgesandthemodernAMD* and Hawaii-though that is a blatant attempt to fool customers.

[don_svetlio]

I said the FuryX is beefed up for a reason-they did large tweaks and additions to it-Fiji is still Tonga, just like GM204 and GM104 are both Maxwell.

Both are GCN 1.2 as GM204 and GM104 are both Maxwell

The individual chip names on AMD work slightly differently from Nvidia. Also - Fiji was in development around 3 years prior to it's release alongside HBM - so it's more of a case of Tonga being a modded Fiji

[Dabombinable]

Both are GCN 1.2 as GM204 and GM104 are both Maxwell

The individual chip names on AMD work slightly differently from Nvidia. Also - Fiji was in development around 3 years prior to it's release alongside HBM - so it's more of a case of Tonga being a modded Fiji

Either way the base architecture is the same.