Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Aug 24, 2016.
Good stuff as usual Hilbert!
Yeah, a mix of medium and high settings is what consoles usually do, review sites like eurogamer agree.
Both 970 and 980 results are surprising.
doesnt really address my point but going by that logic, its an excellent port seeing how the 780ti doesn't drop under 30fps at ultra details in full hd.
e: i feel that most of the "bad port" comments stem from unrealistic expectations from one's own hardware.
It seems consoles are high settings:
So on high settings it should fly even on a 950, but does it?. Poor ports always have been a reality, even consoles get their doses of cheap publishers/devs, imagine on PC.
From that article^...
"Both consoles target 30fps and generally manage to deliver this, but there are trouble spots throughout the game...GTX 970 with a mix of high and medium settings were enough to reach 60 frames per second."
XBOX1 can not run it at 1080
yeah i think people are forgetting that consoles also target 30fps (and do upscaling to 1080 in the case of the xb1) which is considered **** tier performance for a gaming computer. you can argue about how it looks, whether or not you like the aesthetics and whether the performance impact of some of these options is justified but it is objectively not a bad port.
My previous gtx660 was already to run quite a few 30fps consoles games at 60fps, but those were good ports.
I haven't seen the game yet, but it seems that opinion on that is quite split in the internets. Also a shadow resolution from 2k to 8k, or an SSAO search factor of 20pixel instead of 6 will not look that incredibly different while you play, but they will kill performance on everything you turn them on. Hence my comment about actual optimization at a specific level, and how the PC settings are basically post-processing/shadow sliders with little to offer if you are on lower hardware. Most of the Ultra settings in modern games are there for the bragging rights. If you read the NVIDIA performance guides for games, you'll see that for most of them the most reasonable settings are never the ultra ones, even on top notch hardware.
Thanks! The results are less "unexpected" if you get into account the frequencies and the types of cores that each card has. Keep in mind that these are results with the "Game Ready" driver out. The 470 edging out the 980 actually makes sense in a compute-heavy game (like most are/going to be), especially with the frontend job that AMD seems to have done. Once you do the DX12 tests, I would be curious to see the percentage gains for each card. I believe that both will gain, but GCN much more.
My idea of bad port/bad performance is simple if Highest Detail in HR runs 60fps but lowest/medium in Mankind divided look worse and runs worse then HR that something is very wrong.
Mankind divided look nice, from compressed videos i seen it does not look leap and bound better the HR so performance seen on 1080gtx is kinda confusing to me and 1060gtx performance is just as confusing.
Then again I got HR 2 years after the it came out which by that time I had went from 9800gt to 450gtx which it ran 60fps on higest on my 920 i7 and now I have 660gtx. do i dont know how HR really ran on card that were out at the time.
Err, no. Go and check the benchmarks again please cause if you're saying 1fps is edging out the 980, then, at higher res, the 980 edges out the 470 also.
It is literally edging it out on the most popular resolution. That's noteworthy considering the price.
I'm sure I've talked about this "1fps" thing before...and I was ridiculed.
But this just goes to prove exactly what I said.
So, "depending on the resolution the 470 and the 980 are within a couple of fps of each other".
Is that better? Does it change anything in the actual picture?
In 1080p the 470 is 1 fps faster than the 980:
In 1440p the 980 is 1 fps faster than the 470:
This is with "Game Ready" drivers, keep that in mind, and the DX12 patch is about to be released in a couple of weeks. I don't know what's the "1fps" thing you've talked about before, and yes, that probably means that there is no difference when gaming with these cards, but look a bit a the big picture and see how the 1060 and the 980 are providing the same performance while being 35-70 percent more expensive.
I'm really curious about the DX12 results for this now.
But it´s 1080p on ps4, and I can tell you from playing tons of ps4 ports that gtx660/gtx960 are much faster than the ps4, with such cards decent ports do that what you mention with the gtx970. People and devs playing with 780ti´s and titans for years tend to lose ground for comparisons because they forget how weak consoles truly are. Considering a game a good port because it runs well on a 970 on reduced settings would be like believing that the programmers of battlefield, battlefront, f1 2015, Mad Max, MotoGp 15, Lego Force awakens, Alien Isolation, Crysis 3, GtaV, etc, etc, can´t code **** on consoles and end using half of it´s gpu power, which is laughable.
Apologies if this has been answered by I've missed it, but were the leaked DX12 benchmarks fake or did an earlier build still have the option?
I've also heard the benchmark is scripted to favour AMD as well, not sure how true that is though.
Kitguru post the benchs of mankind divided too, even they are disappointed with performance of the game
Well the conversation will diverge to three different fronts moving forward.
1. DX12 changes performance landscape when it releases - unlikely as the game appears to be heavy on the GPU side, with many high-end cards achieving sub 60fps in 1440p which is outrageous really.
2. Driver updates change performance landscape, maybe some gains here but major changes unlikely
3. The benchmark doesn't reflect gameplay experience
Having said that, the benchmark at this point simply seems to favor brute power which gives GCN a lead over it's Maxwell and Pascal counterparts.
I'm not sure what's going on but some websites report different results
computerbase.de used the internal benchmark and the 480 was only 3% faster than a 1060
pcgameshardware.de used their own benchmark run I suppose, and performance was far better and tending towards favoring nvidia.
I supposed the fact that it ran better than the benchmark could give NV the lead because of the poly throughput needed for higher framerates, but it seems like a big difference and needs some explanation.
Maybe there's some game patch that some reviewers tested and others didn't ?
That said Maxwell users that are less in the know may find these benchmarks confusing because many purchase factory overclocked cards and as many here likely know there can up to 20-25% gains on that front, so before selling your 980s and buying 470s consider that with headroom the 980 will pull ahead
Doesn´t look any better than Advanced warfare and I played that at 60fps on a gtx660 :grin:
It looks so average that I can´t stop thinking that this is just a case of "We don´t need optimizations on pc because new hardware will take care of it".
Anyhow you turn it you basically just said how more expensive card had to overclocked to match the other one? Does that makes sense to you?
Also, this is the first time i've seen in a game that 980ti is not even in top 5 cards.