[gamegpu] Far Cry 4 performance

[3DVagabond]

But, I disagree with the claim that R290X is ~= 780ti on launch, that never occurred outside highly cherry pick reviews. Overall it was ~10% slower over many reviews, and more slower compared to custom 780ti that boost to 1.25ghz out of the box.

The first AT review had them pretty much even. Actually, ever so slightly faster for Hawaii) That's when all of a sudden we saw the switch to quiet mode being the default mode. [H] has had them pretty even, at least in gameplay, at HiRes as well. To be honest, I'm not as concerned which is faster when they are running 1080 because both have been plenty fast enough. That was until the latest release of console ports came out, but now any advantage seems to have disappeared again. If you look at reviews where they warm the cards up and play games, [H] and Hardware.fr for example. In crossfire vs. SLI Hawaii is pretty dominant over GK110, and has been. Again, at HiRes where that GPU horsepower can be most useful.

Running 30sec. benches? Sure GK110 is faster because it's running max boost over that time. The only thing I'm cherry picking is using real world conditions instead of 30sec benches that nVidia's boost tech is highly effective at.

 

[psolord]

We've known that for years now, Tahiti>GK104. nVidia marketing though managed to keep that a secret to most all that time. Go figure?

You check any review that uses games and warmed up cards and Hawaii has been as fast or faster than the 780ti from day one. Maybe nVidia has shifted their marketing dollars to Maxwell now and the sites are free to give us real numbers without fear of reprisal?

Is that really the case though or maybe just some gcn console code ends up on the PC versions, thus helping gcn cards, as many speculated would happen due to the AMD chips inside the consoles?

 

[96Firebird]

Yay, more conspiracy theories! When you don't have anything else to talk about, might as well make stuff up!

 

[3DVagabond]

Is that really the case though or maybe just some gcn console code ends up on the PC versions, thus helping gcn cards, as many speculated would happen due to the AMD chips inside the consoles?

Newer games could well be better optimized for GCN. That doesn't change what's been going on for the entire time though.

 

[RussianSensation]

^ Let's be honest though the after-market 780Ti versions were much better than the reference card. Gigabyte 780Ti edition for instance.
http://www.techspot.com/review/738-gigabyte-geforce-gtx-780-ti-ghz/page9.html

It doesn't matter though now as 780Ti doesn't perform anywhere near its $700 GPU price tag. Considering you can now get R9 290 for $185, R9 290X for $255, 295X2 for $650, and these cards were $399, $549 and $1499 1 year ago, this proves time and time again that buying a brand new high end GPU at launch is walking into a landmine. GPUs are probably the most depreciated PC parts, along with SSDs.

 

[3DVagabond]

Yay, more conspiracy theories! When you don't have anything else to talk about, might as well make stuff up!

Or just label and dismiss things when you have no real argument.

 

[96Firebird]

Or just label and dismiss things when you have no real argument.

Its not hard to dismiss nonsense.

 

[3DVagabond]

^ Let's be honest though the after-market 780Ti versions were much better than the reference card. Gigabyte 780Ti edition for instance.
http://www.techspot.com/review/738-gigabyte-geforce-gtx-780-ti-ghz/page9.html

It doesn't matter though now as 780Ti doesn't perform anywhere near its $700 GPU price tag. Considering you can now get R9 290 for $185, R9 290X for $255, 295X2 for $650, and these cards were $399, $549 and $1499 1 year ago, this proves time and time again that buying a brand new high end GPU at launch is walking into a landmine. GPUs are probably the most depreciated PC parts, along with SSDs.

Yes, the aftermarket cards are/were much better at holding their boost clocks. Not that many review sites, and no English speaking ones that I've seen, pointed out that the reference cooler allowed GK110 to throttle.

You must get so tired of pointing out the relative value of AMD cards because nobody cares. They just drone on about whatever metric they can claim an advantage in. Right now it's avg. FPS, but only on the top single GPU card (gtx 980) because that's the only one that wins, and efficiency. Nothing else matters and it's worth paying 2x as much for.

You know how it works.

 

[3DVagabond]

Its not hard to dismiss nonsense.

Or post it.

 

[Lepton87]

The low performance of GK110 cards might be due to throttling, when I leave everything at default it can get as much as 30% slower than when I set temperature target, power target, and fan speed. That's probably the reason it's close to 280X in that particular test I guess it's done at completely default settings and allowed to throttle which doesn't make any sense to operate it like that. A Titan should be about on par with 970(maybe 5% slower), I had one and tested it, essentially those two cards performed identically. That's how it should look when GK110 doesn't throttle

Still, not especially impressive performance of NV's cards.
Only throttling explains this results, I don't care about the results in the OP as I don't let my cards throttle.
BTW. the huge difference between 7970 and 280X also doesn't make any sense, they are for all intents and purposes clocked identically and should perform identically. I don't trust their methodology.
The only game that behaves very strange is DA:inquisition with 280X, 290 and Titan all performing very close to one another, why Hawaii can't convincingly outperform tahiti? Something is not right in that game.

 

[skipsneeky2]

A i5 2500 delivers more frames then a i7 2600k? Why is that? I know this is the the gpu thread focusing on gpu performance but personally i can't get over those numbers.

 

[raghu78]

The low performance of GK110 cards might be due to throttling, when I leave everything at default it can get as much as 30% slower than when I set temperature target, power target, and fan speed. That's probably the reason it's close to 280X in that particular test I guess it's done at completely default settings and allowed to throttle which don't make any sense to operate it like that. A Titan should be about on par with 970(maybe 5% slower), I had one and tested it, essentially those two cards performed identically. That's how it should look when GK110 doesn't throttle

it looks like Farcry 4 likes higher core clock speeds. The performance of GK110 Kepler cards is better with higher clocks.

http://www.purepc.pl/karty_graficzn...e_test_kart_graficznych_i_procesorow?page=0,5

http://www.techspot.com/review/917-far-cry-4-benchmarks/page3.html

http://www.hardocp.com/article/2014/11/21/far_cry_4_video_card_performance_iq_preview/3#.VHfZxmes98E

http://www.notebookcheck.com/Far-Cry-4-Benchmarks.130346.0.html

Still AMD cards beat the respective Kepler cards at Ultra SMAA with Godrays off. R9 290X matches the faster GTX 980 and R9 290 is on par or slightly ahead of GTX 970. Only with Godrays turned on AMD cards get hit almost by 25% on framerates whereas Nvidia cards don't get hit much at all.

BTW. the huge difference between 7970 and 280X also doesn't make any sense, they are for all intents and purposes clocked identically and should perform identically. I don't trust their methodology.

As I said above core clock speed is a big factor in this game.

 

[Lepton87]

A i5 2500 delivers more frames then a i7 2600k? Why is that? I know this is the the gpu thread focusing on gpu performance but personally i can't get over those numbers.

Because the numbers in the OP are non-sense.

ps. Yes at default Titan and 780TI throttle after warming up but it's extremely trivial to keep them under full boost at all times and the fan is still silent. And then again there are non-reference 780TI that don't throttle out of the box and are even overclocked.

 

[Lepton87]

As I said above core clock speed is a big factor in this game.

oh sure, the only problem is 280X is not clocked any higher than 7970. I won't trust those benchmarks unless they explicitly say which cards throttled and which didn't the difference can be huge and it's trivial to stop GK110 from throttling.,

 

[MTDEW]

A i5 2500 delivers more frames then a i7 2600k? Why is that? I know this is the the gpu thread focusing on gpu performance but personally i can't get over those numbers.

It's true though.
If I disable HT on my 2600k I get a FPS/performance increase.
It just seems this title plays better without HT that's all.

 

[skipsneeky2]

It's true though.
If I disable HT on my 2600k I get a FPS/performance increase.
It just seems this title plays better without HT that's all.

Trying to figure out the magic involved with the 980 sli numbers,a single 980 pulls a minimum of 71 with a stock clocked 5960x,a 4.6Ghz 5960x pulls 84fps and then theres 176 for 980sli.

Been staring at those charts like you wouldn't believe,attempting to make sense of those insane numbers.Bats**t numbers.

All i can get is that you take a $1,000 8 core chip,take it from 3GHz to 4.6Ghz and suddenly magic happens and 980 sli and the minimum framerate is higher then double a single 980....86 vs 176.Magic?

 

[RaulF]

^ Let's be honest though the after-market 780Ti versions were much better than the reference card. Gigabyte 780Ti edition for instance.
http://www.techspot.com/review/738-gigabyte-geforce-gtx-780-ti-ghz/page9.html

It doesn't matter though now as 780Ti doesn't perform anywhere near its $700 GPU price tag. Considering you can now get R9 290 for $185, R9 290X for $255, 295X2 for $650, and these cards were $399, $549 and $1499 1 year ago, this proves time and time again that buying a brand new high end GPU at launch is walking into a landmine. GPUs are probably the most depreciated PC parts, along with SSDs.

Nothing but a farce.

I owned one of those cards and it would crash games left and right. If i would downclock it by 50 or so hz it would be stable.

Go read the reviews on Newegg and see how many people had problem with those Gigabyte GHZ edition 780Ti.

 

[escrow4]

^ Let's be honest though the after-market 780Ti versions were much better than the reference card. Gigabyte 780Ti edition for instance.
http://www.techspot.com/review/738-gigabyte-geforce-gtx-780-ti-ghz/page9.html

It doesn't matter though now as 780Ti doesn't perform anywhere near its $700 GPU price tag. Considering you can now get R9 290 for $185, R9 290X for $255, 295X2 for $650, and these cards were $399, $549 and $1499 1 year ago, this proves time and time again that buying a brand new high end GPU at launch is walking into a landmine. GPUs are probably the most depreciated PC parts, along with SSDs.

Not here. My 780Ti GHz boosts to 1215MHz in game factory overclock and cost $929 back in February here in Australia. February 2015 I'll give Nvidia another $929 for a 980 Ti if its at least 30% faster and isn't a gimpy mid range card.

 

[Silverforce11]

Not here. My 780Ti GHz boosts to 1215MHz in game factory overclock and cost $929 back in February here in Australia. February 2015 I'll give Nvidia another $929 for a 980 Ti if its at least 30% faster and isn't a gimpy mid range card.

NV hardware has always priced crazy in AUS/NZ.

Back in late 2013, I got 2x R290 for the price of your one 780ti. It also then made me millions of Dogecoins which I converted to Bitcoins at its highest. Which I then spent to buy heaps of R290s to make even more coins.

Whenever the next-gen big boys come (GM200 or R390X), I'll grab a few with Bitcoins. Free upgrade for life (or as soon as bitcoins become worthless heh)!

 

[Erenhardt]

oh sure, the only problem is 280X is not clocked any higher than 7970.

But it is.
From the graph you posted:
280X - 1000MHz Core / 1500 MHz mem
7970 - 925MHz core / 1375 Mhz mem

That is +8,1% on core speed and +9,1% memory speed increase

280X - 47,7 FPS vs 7970 - 43,3 FPS
10% faster for 8% core speed bump and 9% mem speed bump? Seems about right.

 

[guskline]

Just downloaded Far Cry4 last night and played for @ 1 hr this morning. I have settings at 2560 x 1440 and Ultra. Runs very smooth and beautiful on my BenQ BL3200PT

 

[skipsneeky2]

Just downloaded Far Cry4 last night and played for @ 1 hr this morning. I have settings at 2560 x 1440 and Ultra. Runs very smooth and beautiful on my BenQ BL3200PT

Screen must be a dream to use, i had to google the specs and it just sounds like a overall awesome screen.Had no idea they made 1440p panels above 27''.

Ppi of a 20'' 900p or 24'' 1080p screen pretty much eliminates any arguments on the usability side and just allows a massive screen up close which is nice.

 

[guskline]

Screen must be a dream to use, i had to google the specs and it just sounds like a overall awesome screen.Had no idea they made 1440p panels above 27''.

Ppi of a 20'' 900p or 24'' 1080p screen pretty much eliminates any arguments on the usability side and just allows a massive screen up close which is nice.

Yes it is nice; expensive but nice. It will be a lot nicer when AMD releases the CrossFire profile for this game.

This weekend, I'm going to run it on my 3770k rig with 2 GTX670s and a 2560 x 1440 monitor and AMD 8350 rig with GTX780 Classified and 1920 x 1080 monitor to compare.

When I get time, I'll run fraps to get an idea of my fps from my gpus below. The Sapphire Tri-X OC 290 is rated 1000 core 1300 memory. I had them OC'd but decided to run them stock. Since Far Cry 4 doesn't yet support CrossFire I'm sure the fps will be lower. As I said, I have the settings at Ultra and game play is good. I had some screen tearing during the intro but in the game itself, it has been smooth.

 

[Carfax83]

It will take months before final performance in Far Cry 4 can be determined, as the game will get several patches and driver optimizations require time to peak..

So all this talk about whether AMD or NVidia has the better performance at this early stage is premature..

In the end, it will likely be Nvidia though if I had to bet, because several of the settings use heavy amounts of tessellation, and tessellation was never AMD's strong point unlike Nvidia.

 

[guskline]

It will take months before final performance in Far Cry 4 can be determined, as the game will get several patches and driver optimizations require time to peak..

So all this talk about whether AMD or NVidia has the better performance at this early stage is premature..

In the end, it will likely be Nvidia though if I had to bet, because several of the settings use heavy amounts of tessellation, and tessellation was never AMD's strong point unlike Nvidia.

I agree with you Carfax83. Since I own a 3930k rig at 4.6Ghz with 2 R9 290s in CF and you have a 4930k rig at 4.5Ghz with 2 GTX970s in SLI we should both run a segment of Far Cry 4 to compare performance.:thumbsup: