Dayaks
[H]F Junkie
- Joined
- Feb 22, 2012
- Messages
- 9,774
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Fable Legends DX12 benchmark
- Thread starter Quartz-1
- Start date
And yet the 750 ti still has lower sales numbers on the amazon chart, and makes up a smaller number of steam builds.
http://store.steampowered.com/hwsurvey/videocard/
The 970 easily tops it. The new "volume" segment for gpus has shifted to the 300-350 dollar price point. This is the price segment most people target in higher numbers for their game builds, and the 970 no longer deserves the catbird seat there.
The only way one can make that case, is if they define "better card" as "nvidia makes it"
If that is your definition, then sure, that statement holds. For everyone else, it's a false statement. You go ahead and turn nvidia into the iphone brand of your life, but to expect that crippled state of affairs to infect the gpu gaming market as well makes me hurl.
Tup3x
[H]ard|Gawd
- Joined
- Jun 8, 2011
- Messages
- 1,942
I had 290 clocked to decend speeds last year and to be honest 970 is definitely a better card. The massive difference in heat and noise is already enough to make it better buy but at least the games that I play definitely do not run worse on my GTX 970 @ 1,5 GHz (no voltage increase and I even manually made the fan curve less aggressive so it's way quieter than at stock and status in 60's). It felt like the driver overhead was problematic in some MMOs.
For games like wow the 970 will be better since it's just dx9/11.
Even the newer variants like the 390 that do NOT have the heat/noise issues you mentioned will probably lag a 970 a bit. See there people? This is a real example for the type of game where the 970 would be better. a cpu bottlenecking game like wow where amd has worse cpu throughput performance in dx11.
That said, I find it kind of disgraceful and LAZY for a company like blizzard/activision to not be making a dx12 code path for wow. Still a wildly profitable enterprise, and one of the types of games that would benefit MOST from jumping to dx12 due to the cpu overhead issues.
TaintedSquirrel
[H]F Junkie
- Joined
- Aug 5, 2013
- Messages
- 12,691
There's a new expansion coming, maybe it'll support DX12.
They could add it with a new patch someday. They jumped to DX11 after all.
http://www.pcgameshardware.com/aid,...lysm-Experimental-DirectX-11-support/Reviews/
No its not, OEM's sell more graphics cards than consumers buy cards at a retail or e-tailer.
This is why 3dfx died when they stopped OEM sales. 300+ dollar cards only account for maybe at most 10% of total sales in the discrete market.
On reflection, the 300-350 price point seems like the most popular target for a single gpu popularity (the 970). You are right that the segment as a whole for sub 300 dollar cards is higher but below a 750 ti class card should we care? Does it really make a big difference that some dell came with an nvidia 730? Most of the time when a gpu comes inside a tower or laptop it's a pretty anemic gpu unless it's explicitly billed as a gaming computer.
People stuck with that are mostly relegated to ultra low settings in everything but ancient games or light games like lol.
dragon age inquisition and the witcher and bf4 is not targetting that low segment, I'd say the minimum is something like a 750ti for a halfway decent experience at 1080p.
for people that buy in add in cards I don't think the 750 ti is what people will be looking for, but for people buying an off the shelf system the graphics card isn't that important to them, but if they get one that doesn't increase the price of the computer much they probably will. A decent $600 system can come with a 750 ti.
How many PC's do you think Dell, Lenovo, and HP sell? I would gander some where over 50% of the market cpu's and gpu's.
How many PC's do you think Dell, Lenovo, and HP sell? I would gander some where over 50% of the market cpu's and gpu's.
It's the opposite now, gpu makers are selling far fewer entry level gpus because integrated graphics has gotten good enough for most people. I've never seen a ti variant of anything in an electronics store, let alone a 750 ti. I think best buy has a couple gaming notebooks with a 860/960 but that's about it in stores. I've seen a smattering of entry level cards from time to time, but it's extremely rare. The discreet gpu market has shifted upward, and something like a 750 ti is near the bottom of what anyone ought to want. That's what you stick in an ultra budget build for sub 500 dollars, and built yourself as you'd never find it in store.
Some of you guys act as if the 390 blows the 970 out of the water in terms of performance, at least. We got the review from H, and from TPU (4k: 970-93: 390-99/100; 1440p 970-98: 390:99/100: 1080p 970-103: 390-99/100). Pretty close performance, so from where most consumers start to look at other metrics.
So, going by the same train of thought, aside from being more of a space heater (than the 970) and having 8Gb (that you'll never fully utilize, even in CF unless you like slideshows), is there any reason to buy a 390?
Cases with bad airflow is one.
Cases with weak power supply is two.
G-Sync is three.
3DVision is four.
SFF on a budget is five.
Those are all very niche use cases so yeah, for the majority of users it's a good thing to go with team red on this one.
The 970 has only 3.5 GB of usable memory, and it shows when you look for stuttering:
(Souce: Tech Report)
The 970 has only 3.5 GB of usable memory, and it shows when you look for stuttering:
(Souce: Tech Report)
In you img. there is a considerable difference between the 390x vs 390. Both have 8GB. Care to explain the difference of almost 3 times as much stuttering, as you put it?
Oh look i can cherry pick too (by the almost ~2,x difference here the 390 must be awful stuttery ). (Edit: just seen this, here the 980 is more stuttery than the 970. OMFG OMFG the plot thickens!!! )
I'd recommend you guys to wait for the game to actually show up and then cheer all you want. Until then we're talking on tech demos and alphas.
Cherry Picking? I chose 16.7 ms because the TR article says that this is the most important value:
If you go to 8.3 ms the difference is still there (though at that point, this is less caused by stuttering and more by raw fps).Tech Report said:
The chart that you posted doesn't tell much, the difference between 22 ms and and 26 ms can be variation due to other factors than GPU performance. Which is more or less substantiated by the fact that the 970 beats the 980.
I think you would be surprised in how many people buy a low end discrete,
Just look at steam hardware survey
http://store.steampowered.com/hwsurvey/videocard/
6 of the top 7 Graphics cards (didn't count the integrated or mobile) are low end 760 and lower. What does that tell ya, $200 cards and lower.
I don't know where you get this crap from but it's a real issue. The OS will always see GTX 970 as a 4 Gb card so any game that needs more then 3.5Gb is going to try and use it which will result in horrible stuttering and fps drops.
thedocta45
[H]ard|Gawd
- Joined
- Oct 10, 2007
- Messages
- 1,325
The 970 was released with a price around 320 while the 290 had a price point of 400.
The card even with 3.5GB of RAM offered a very good price to performance ratio which is why it sold so well.
You might get some stuttering at some higher resolutions, but for the majority of games you will never see it.
Why was this issue not apparent during the reviews, if it actually effected performance it wouldn't have gotten such rave reviews from all the major hardware sites.
Derfnofred
Gawd
- Joined
- Dec 11, 2009
- Messages
- 606
Is this a thread about Fable Legends DX12 benches or am I lost?
TaintedSquirrel
[H]F Junkie
- Joined
- Aug 5, 2013
- Messages
- 12,691
I'm going to put this here since I don't think it's important enough to warrant its own thread.
Nvidia - DX12 Do's and Don'ts
https://developer.nvidia.com/dx12-dos-and-donts
Nvidia - DX12 Do's and Don'ts
https://developer.nvidia.com/dx12-dos-and-donts
http://hexus.net/tech/news/graphics/86783-nvidia-publishes-dx12-dos-donts-checklist-developers/
its just a view of good programming for Dx12.
The first part of your quote, if not using fine grade context switching (Maxwell 2 lacks this anyways, and GCN has it but with a minimal performance hit so better not to use it in a real-time app) so the programmer has to be aware of what are the best approaches to what compute tasks run well with the graphics work load and this varies with different hardware.
The second part of the quote is specific to context switching (Maxwell 2 again should not use this major performance hit, and for GCN same as above), which isn't recommended to be used in async shaders because it isn't necessary and is very slow.
Yes cherry pick, 60 FPS is goal but 58 fps is by no means a stuttering, more of a just not strong enough issue.
On other hand, the not even getting 30 fps is a much bigger issue, it's stuttering no excuse and you will notice them a lot more.
D
Deleted member 82943
Guest
what is the point of this thread? shouldn't we forgo speculation until we have more reviews to compare? Say you speculate on this thread and the reviews come out and your speculation is proven right, what have you really gained ? E-Peen points?
What you propose would reduce 90% of all forum traffic and cause wccftech to implode. Reality is too slow in coming to comment on, we need to fill the gaps with SOMETHING!
So sayeth the wise Alaundo/forum troll
I will defer to TR who say that 16.7 ms (60 fps) is the golden mark to achieve, and 50.0 ms (20 fps) is where slowdown is perceptible.
You will be hard pressed to tell the difference between 22 ms and 57 ms above 33.3 ms, but 1785 ms vs 5397 ms above 16.7 ms is a whole other matter.
It was only sent to selected members of the press from what I've read, but apparently the game will be out on October 13th. 6 more days.
play the battlefield beta until then, and then play this.
where did you get the 1785ms?????
Reality
[H]ard|Gawd
- Joined
- Feb 16, 2003
- Messages
- 1,937
read the graph that was posted, it's the R9 390 result
difference between 22 ms and 57 ms above 33.3 ms????????
Maybe help put this in English?