1440p and vram

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Robster

Member
Jul 16, 2005
126
0
0
Games @ 1440p

All playable settings for my setup

Metro 2033 - AAA - Maxed out - 1557MB
Crysis 3 - FXAA on Very High - 1950MB
BF4 SP - 2XMSAA on Ultra - 2153MB
GTA 4 - Modded textures and ENB - Maxed out - 1500MB
Batman AO - Ultra with Max TXAA - 2100MB
Metro LL - AAA - Maxed settings - 1200MB
Skyrim - with 2K and 4K Texture mods with 2XMSAA - 2300-2500MB
Assetto Corsa - 4XMSAA - Ultra - 2450MB

Interesting! Metro 2033 maxed for me around 2000 MB
 
Last edited:

SlowSpyder

Lifer
Jan 12, 2005
17,305
1,001
126
My 7970 came with 3GB when I got it 2.5 years ago. I wouldn't buy a card for 1440P that came with less, especially since the new consoles just launched. But, I don't think I'd spend the extra for 4GB on a 770 either, at least not at those prices. If it were me, I'd probably either sit tight with the card I have and wait till 20nm in a year or so, or think about unloading the 770 and looking at the 'big' GPU's that are available from Nvidia/AMD.
 

OCGuy

Lifer
Jul 12, 2000
27,224
36
91
Interesting! Metro 2033 maxed for me around 2000 MB

It will cache what it can, but this is what I found regarding usage on Metro: LL, which does not show a trend that should worry you:


Above you can see a wide variety of in-game selectable quality modes @ 2560x1600. We switch in-between the image quality modes (Low to Very HIGH) to monitor frame buffer utilization. As you can see we are hovering at the 1 GB of graphics memory even with SSAA enabled starting at High quality mode settings. The measurement shown above is fairly similar and applicable for 1920x1080/1200 and 2560x1600



http://www.guru3d.com/articles_pages/metro_last_light_graphics_performance_review_benchmark,8.html




I also recommend reading this review that tested specifically the 2GB v 4GB in an array of games. This link is to the performance summary:

http://alienbabeltech.com/main/gtx-770-4gb-vs-2gb-tested/3/
 
Last edited:

bystander36

Diamond Member
Apr 1, 2013
5,154
132
106
OP do your own research on VRAM usage, or read an entire thread that has actual statistics.

At 1440p you do not need more than 2GB now, or in the near future. Developers don't write games that leave them 20% or less of the PC market intentionally. (Although some poorly-coded games such as Skyrim and the Titanfall Beta did fail to properly utilize memory).



http://store.steampowered.com/hwsurvey

Most people you find giving VRAM utilization numbers are actually giving you the amount that the game allocates based on the available amount on the card. That is why you can find someone with a 3GB card claiming 2900MB "usage" on their system, and someone with identical monitors and settings on the same game seeing 1900MB usage on a 2GB card.
Skyrim doesn't even have an issue. It is some mods that can have an issue, usually related to super high res textures.
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76

I don't think anyone is saying 2GB is insufficient for old-gen console ports at 1440p out of the box. Though Skyrim with enough mods may require more.

But the problem with the above URL is captured by the first three comments on the Conclusion page:

1. Next-gen games may require more VRAM, with or without DirectX 12, due to how next-gen consoles have 8GB of shared memory and can utilize larger textures, more AA, and more memory-hungry game engines
2. Minimum framerates (and also frametimes) are what matter--NOT average fps
3. Modded games can exceed stock game requirements. Skyrim is the poster child for this.


herathrig • 14 days ago

You should revisit vram in the future with newer games like watch dogs, bf4, star citizen and maybe gta5, if ever it comes out on the pc. I think it would be also nice to test it with the future direct x 12 once its out.

•
Reply
•
Share ›

Avatar
dave • 2 months ago

How are you going to do this comparison without posting minimum frame rates...that is where you will see the difference. Check out anandtech or toms hardware for people that actually know what they are talking about.

3
•
Reply
•
Share ›

Avatar
Shahriyar • 3 months ago

Please check GTAIV too. It's the most memory consumer game that i know.

•
Reply
•
Share ›

Avatar
nashathedog • 5 months ago

What about people who mod there games with high res textures and do pass 2gb's of ram at 1080p? I do and I no longer get the issues it caused since upping my vram, this review doesn't cover all the bases and for that reason it isn't a very good review in my opinion. Sorry, I realise it's a few months old now but i was passing 2gb's when modded a year ago, And now if Battlefield 4 does in fact hit 2gb's as claimed (I haven't tested that myself) then this is already outdated and inaccurate and giving bad advice. I know you say how you don't take possible future changes into account in the review but it's not a possible change it's something that is guaranteed and the industry knows and accepts that, Thumbs down.
 

OCGuy

Lifer
Jul 12, 2000
27,224
36
91
^ As soon as there is some data showing that is the case, please do post it.
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
^ As soon as there is some data showing that is the case, please do post it.

In this and other posts, you are trying to prove 2GB is enough.

I'm simply saying 2GB is enough for now at 1440p for games out of the box, but there could be changes going forward.

I'm not trying to prove anything, you are. Thus, you have the burden of proof, not me.

P.S. My card has 4GB of VRAM. Yours has 2GB. Just an observation.
 
Last edited:

SlowSpyder

Lifer
Jan 12, 2005
17,305
1,001
126
To me buying a 2GB card for 1440P is kind of like buying an i3 over an i5 or i7 for a gaming system at 1440P. It will work, and many games will even play quite well. But there are obvious disadvantages and zero advantages that I can see.
 

Robster

Member
Jul 16, 2005
126
0
0
It will cache what it can, but this is what I found regarding usage on Metro: LL, which does not show a trend that should worry you:






http://www.guru3d.com/articles_pages/metro_last_light_graphics_performance_review_benchmark,8.html




I also recommend reading this review that tested specifically the 2GB v 4GB in an array of games. This link is to the performance summary:

http://alienbabeltech.com/main/gtx-770-4gb-vs-2gb-tested/3/

Posted the 2 vs 4 GB link in my original post

I've already seen the Guru3D link(s), they provide similar analysis of other games as well. Really interesting read and comparison between game engines. Thanks anyway (if they were intended for me D):thumbsup:
 

SlowSpyder

Lifer
Jan 12, 2005
17,305
1,001
126
Just out of curiosity, does anyone know of any articles that review some cards from yesteryear in today's games? I think a GTX580 with 1.5GB and 3GB benched with current games would be an interesting read. It'd be nice to see if the extra memory ends up providing any tangible benefit even with the fairly mid range level GPU (by current standards), or if the GPU really is too weak to use more than the amount of memory found on the reference card.
 

OCGuy

Lifer
Jul 12, 2000
27,224
36
91
Just out of curiosity, does anyone know of any articles that review some cards from yesteryear in today's games? I think a GTX580 with 1.5GB and 3GB benched with current games would be an interesting read. It'd be nice to see if the extra memory ends up providing any tangible benefit even with the fairly mid range level GPU (by current standards), or if the GPU really is too weak to use more than the amount of memory found on the reference card.

This is a fairly extensive card selection, which shows a 1GB v 2GB 7790. It doesn't show that the memory bandwidth can't keep up, it shows that either the game doesn't use over 1GB, or that it does and there is no performance hit for cache swapping.

The 1GB and 2GB framerates are identical to each other within both the VHQ @ 2560x , and the Medium @ 1080p settings. So while the bottleneck is the GPU, it isn't the memory.

http://kotaku.com/metro-last-light-pc-performance-benchmarked-509562323


 

Majcric

Golden Member
May 3, 2011
1,388
52
91
There is nothing wrong with buying a 2gb card, however, more than one could be a problem. I don't know of a single 2gb card that has enough thump to choke itself on Vram.
 

lehtv

Elite Member
Dec 8, 2010
11,897
74
91
When you do run into Vram issues, turning down AA almost always fixes it, but of course games are starting to use more vram, so what happens in the next year could change.

On 2560x1440, I'd probably just use SMAA with SweetFX. 1440p shows less aliasing than 1080p, and personally, SMAA is all I ever need for any game on 1080p. I only use AA in games that are stupidly easy to run with modern hardware, if supported.
 

moonbogg

Lifer
Jan 8, 2011
10,637
3,095
136
I used to argue about this, but it won't matter pretty soon since this gen is almost dead and new cards will have more. Then we will get to argue about how 4gb is barely enough for new games. I fully expect new games to surprise us in the next 1.5 to 2 years as they plow right through the 3gb barrier.
 

KingFatty

Diamond Member
Dec 29, 2010
3,034
1
81
General rule of thumb is to buy a video card for the games you already own, and not the games you plan to buy year(s) into the future.
 

Throckmorton

Lifer
Aug 23, 2007
16,829
3
0
I read that modern games like to fill available memory, like Windows Vista-8 do with RAM. So just because a game will use 2gb+ on a 4gb card doesn't mean it will bottleneck on a 2gb card.

I bought a 1gb 8800GT back in 2008, and it helped when I was playing games in 2011, but I could have also saved the money and put it into a newer card in 2011...


Either way it's not the resolution that determines how much VRAM gets used, so saying "You need xxxGB for 1440p gaming" is a non sequitur
 

bystander36

Diamond Member
Apr 1, 2013
5,154
132
106
I read that modern games like to fill available memory, like Windows Vista-8 do with RAM. So just because a game will use 2gb+ on a 4gb card doesn't mean it will bottleneck on a 2gb card.

I bought a 1gb 8800GT back in 2008, and it helped when I was playing games in 2011, but I could have also saved the money and put it into a newer card in 2011...


Either way it's not the resolution that determines how much VRAM gets used, so saying "You need xxxGB for 1440p gaming" is a non sequitur
It has been proven, even on this thread, that higher resolutions do increase Vram usage, but it isn't as much as many like to believe. However, you cannot ignore it either.
 

BrightCandle

Diamond Member
Mar 15, 2007
4,762
0
76
A lot of people seem to get the causation and correlation between computer hardware the wrong way around. They assume that software will require the thing therefore the manfacturers make it. It doesn't work that way at all, its the exact opposite. The manufacturers of hardware make it and then the software developers utilise it. They have to do a bit of prediction (which is how games like Watch dogs end up getting delayed for performance problems on the consoles) for the near future but basically they don't write games for hardware that isn't very common.

Lets look at the history a bit to say something about the future. So if we look at the 7970 v 680 battle that took place two years ago there is 50% more VRAM on the 7970. How many games actually used it? Very very few, maybe 6 in all. All the big games played well on both pieces of hardware for obvious reasons, because their job is to sell games. So if one card had better tesselation or the other more VRAM what happened is you basically got the worst performance of the pair of the cards. The reason for that is that as a game manufacturer you are never going to choose one company over the other exclusively. Sure some games performed better depending on which company helped with development, you had 3D vision and Nvidia AA and HDAO+ and AMD did similar things with other games but its not like a game came out that only ran well on the 7970 because it needed more VRAM. In actuality what happened was the games developers determined they had 2GB of VRAM to play with because of the 680 and they could only tesselate as much as the 7970 could cope with (which performed worse on that aspect that the 680), that they could only use so much compute performance because the 680 was a bit hampered there, that they also couldn't use many double calculations again because of the 680. The extra VRAM on the AMD 7970 was mostly going to waste but so is much of the tessellation hardware on the 680.

This is just how the industry works, even at the top end of performance you still have to choose a relatively wide market to make developing those graphic features worth the development cost, so high end graphics are targeted at both cards. Thus game for 680/7970 class hardware (they perform pretty similarly) is always going to be limited to each others faults. The causation and correlation is always hardware comes first and the software guys use as much as they can get away with without completely screwing anyone. I know of only one game that even remotely intends to break that and its Star Citizen and yet even they aren't intending to make a game that no one can run, they just adding features exclusive to both AMD and Nvidia in the same game, which hasn't IIRC happened before.

We see the same thing on CPUs. There is a tonne of software that can't run on the desktop because its not fast enough yet. Its not that all software ideas have been exhausted its that the performance isn't really climbing so we can't add features and whole new types of applications. The hardware leads and the software follows. The software can never lead because if we don't have hardware capable of running it we can't test it and no one would buy it because they can't run it.
 
Last edited:

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
A lot of people seem to get the causation and correlation between computer hardware the wrong way around. They assume that software will require the thing therefore the manfacturers make it. It doesn't work that way at all, its the exact opposite. The manufacturers of hardware make it and then the software developers utilise it. They have to do a bit of prediction (which is how games like Watch dogs end up getting delayed for performance problems on the consoles) for the near future but basically they don't write games for hardware that isn't very common.

Lets look at the history a bit to say something about the future. So if we look at the 7970 v 680 battle that took place two years ago there is 50% more VRAM on the 7970. How many games actually used it? Very very few, maybe 6 in all. All the big games played well on both pieces of hardware for obvious reasons, because their job is to sell games. So if one card had better tesselation or the other more VRAM what happened is you basically got the worst performance of the pair of the cards. The reason for that is that as a game manufacturer you are never going to choose one company over the other exclusively. Sure some games performed better depending on which company helped with development, you had 3D vision and Nvidia AA and HDAO+ and AMD did similar things with other games but its not like a game came out that only ran well on the 7970 because it needed more VRAM. In actuality what happened was the games developers determined they had 2GB of VRAM to play with because of the 680 and they could only tesselate as much as the 7970 could cope with (which performed worse on that aspect that the 680), that they could only use so much compute performance because the 680 was a bit hampered there, that they also couldn't use many double calculations again because of the 680. The extra VRAM on the AMD 7970 was mostly going to waste but so is much of the tessellation hardware on the 680.

This is just how the industry works, even at the top end of performance you still have to choose a relatively wide market to make developing those graphic features worth the development cost, so high end graphics are targeted at both cards. Thus game for 680/7970 class hardware (they perform pretty similarly) is always going to be limited to each others faults. The causation and correlation is always hardware comes first and the software guys use as much as they can get away with without completely screwing anyone. I know of only one game that even remotely intends to break that and its Star Citizen and yet even they aren't intending to make a game that no one can run, they just adding features exclusive to both AMD and Nvidia in the same game, which hasn't IIRC happened before.

We see the same thing on CPUs. There is a tonne of software that can't run on the desktop because its not fast enough yet. Its not that all software ideas have been exhausted its that the performance isn't really climbing so we can't add features and whole new types of applications. The hardware leads and the software follows. The software can never lead because if we don't have hardware capable of running it we can't test it and no one would buy it because they can't run it.

I generally agree with the above, but consider:

- The upper limit for image quality is near infinite, so a program could have settings that only the most powerful, VRAM-filled cards can use, and lower settings for more-common, less-powerful, less-VRAM-laden cards to use. When Crysis came out, all but the fastest cards choked on highest settings, so the common peasantry had to use lower settings. Just as an example.

- Next-gen consoles have 8GB combined RAM. Some for OS, some for the actual game program, and some for VRAM. It is possible that, once devs get more comfortable with the limits of the new consoles, games will be programmed with this 8GB combined RAM in mind. So when it comes time to port the game to PC, there may be a "full" mode that requires more resources and a "common peasantry" mode for those with 1GB, 2GB, VRAM, or whatever they decide is the lowest target for the game.

And lastly:

- Mods (see e.g. Skyrim with enough mods eating >2GB VRAM). Skyrim runs ok on 1GB VRAM for all but the highest resolutions if you don't go crazy with resolution/settings. But mods can and do eat up more VRAM. More VRAM thus gives you the flexibility to install bigger and more mods. Some people don't care about mods, but many people do.

2GB is enough at 1440p for future games at some given quality level. Whether it will be enough for the quality level you want, I don't know. As some have pointed out, you can worry about this once we get there and upgrade your card only when forced to. That's a valid strategy.
 
Last edited:

toyota

Lifer
Apr 15, 2001
12,957
1
0
its real simple. does 770 sli have the gpu power to run games at 2560 using settings that can utilize more than 2gb in some cases? yes
 

Enigmoid

Platinum Member
Sep 27, 2012
2,907
31
91
its real simple. does 770 sli have the gpu power to run games at 2560 using settings that can utilize more than 2gb in some cases? yes

SLI 770 most definitely more than 2 GB at 2560. In a few years it will be just like SLI 580 now. Or like 560 Ti SLI.

I was in a similar situation for my laptop. 1 GB 660m or 2 GB 660m. Initially and for many games it makes no different but it allows me to play modded skyrim (1100 MB), Crysis 3 with ultra textures (1200-1300 MB), and some other couple games.
 

moonbogg

Lifer
Jan 8, 2011
10,637
3,095
136
Only reason I got 670's was because my 570's were Vram choked in BF3. Skipps, stutters and all that went away with more Vram. I'm in the EXACT same boat now with BF4, but this time I lowered texture settings because screw paying a grand just to turn that one setting up the last notch.
 

Bateluer

Lifer
Jun 23, 2001
27,730
8
0
Remember that the new consoles have 8GB of VRAM. Developers are going to be hitting for that and then tunning for PC.

Correction, they have 8GBs of shared RAM. With 1-3GBs reserved for the OS, leaving 5(XB1) to 7GBs (PS4)for developers.


I am tired of the myth that VRAM has something to do with resolution. It just doesn't. The frame buffer is a very small portion of graphics card memory.

2560*1440*24 = 88473600 bits = 10 megabytes

Even if you anti-alias every pixel (supersampling) 8x that's 80 megabytes.

It's textures and geometry that use most of your graphics card memory! For resolution what matters is GPU computing power

. . . so much fail. By this logic, a video card would never need more than 4MBs for 1080p resolution because FHD is only 2 MegaPixels.

You can prove this yourself. Use a program to monitor GPU mem usage and switch between resolutions. You'll see very little change

I've done exactly that in Skyrim, trying to mitigate the CTD issues with its system RAM usage when it breaks 3.1GBs. So I always have monitoring programs open on my second display. VRAM usage varies by several hundred megabytes between 1080 and 1440. Some what inprecise, I know, but I'm not going to spend my limited weekly gaming hours making sure I record Skyrim VRAM usage at two different resolutions to prove what I've already observed.

Only reason I got 670's was because my 570's were Vram choked in BF3. Skipps, stutters and all that went away with more Vram. I'm in the EXACT same boat now with BF4, but this time I lowered texture settings because screw paying a grand just to turn that one setting up the last notch.

Or just pay 500-600 for a 290 or 780, get twice the VRAM and more computing power to go along with it?
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |