hello i have bought gtx 970, but now when i started using it i came to know about bad performance above 3.5GB vram
so i decided to return it back…
now i am thinking to get radeon’s r9 290x crossfired.
is it gonna be great in performance over gtx 970 for unreal engine?
I am getting one r9 290x under 300$ and 2 for around 590$.
and gtx 970 was about 500$ here in pakistan.
The with the GTX 970 is not really going to make much of a difference, rarely would you ever get to a situation where you’ would use between 3.5-4.0GB of GPU memory, it’s going to be higher or lower than that.
As for the other option, UE4 does not support SLI or Crossfire, so it would only be able to take advantage of a single graphics card, get the best one in your budget.
So far UE4 doesn’t use more than 2 graphics card but with that being said
Go with the 290x or if you can get even better deal with the 290 go with that. Even if the 290x was around the same price I would say go with that, but with that price difference the 290x is the obvious choice. Just make sure it’s not the reference design which ran loud and hot, where as the aftermarket cards all ran silent and cool( something like the triX or lightning.)
I would be very wary about the 970 going forward with what NV has done with driver updates on previous generation cards, and how their cards performance has aged vs AMD 7xxx series on.
Edit: So what I would suggest is a single 290x or if you get better deal the 290. They will be great performance for the price for quite a while. New cards are coming out in a few months, big maxwell, and R9 3xx, but even then I don’t expect the price of these current cards to change much, these will just have a higher price than where these are at. Then if you want more performance you should be able to sell your 290x without losing much money at all, and go with one of the new cards.
Just going to add from another recent post about the 970, it’s not really an. Nvidia made a mistake on the tech specs. The following is a copy of another post:
You will never notice any difference unless you compare a 980 to a 970 in 1440p or higher resolutions while using greater than 3.5gb of VRAM. Even then it is an FPS drop of at most 5% (un-tested though, a range of 0% to 3% is the best data out there right now). It is not actually an with the card at all, it was a mistake for them to release the tech specs they did for sure, but that doesn’t change anything relating to the card’s speed in the real world.
It is an overblown reaction to a perceived injustice, nothing more. But if you still think it is a problem, get a 980.
It’s not much of a problem NOW because they can spend time optimizing the 970 specific driver for specific games, and most games are built to use less than 3 or 2gb. But for him doesn’t even really matter considering the price difference between the 970 and 290x there would be no reason to go with the 970.
Edit: There is no way was just a mistake, you know at the very least the driver team and engineers knew about the. The 970 has been out for months, and not a thing was said about until the was found by users.
The main reason is efficiency, the 290x uses 2-3x more power than the 970 for the same result. The rest is personal preference really, not much difference.
I do agree with you on , not trying to say otherwise. Either someone didn’t report the error to the marketing team, or like all big corporations do, they found out but didn’t want to say anything. Either way it was wrong and they should have been more upfront with all of , but it doesn’t really affect the card itself. It’s such a small difference you wouldn’t be able to detect it on your own unless you know about , or are really into benchmarking.
I don’t know. Seing requirements for games rise upwards(and using more VRam) it might not be a problem NOW but maybe in 2 years? And when you buy a new GPU you’re usually not calculating for just 2 years (without a very specific reason).
For example my gtx580 still performs quite well, but has only 1.5GB of VRam. Most games start to scratch the 2GB barrier for mid-settings and as a result of not enough VRam i sometimes need to scale down. (From measuring the GPU doesn’t seems to be at 100% for all of those games, so the VRam seems to be the bottleneck) That was a mistake i did in the past, not looking if there was a better one(and there was a 2GB version!). SO for my next card(and i looked for the 970GTX) i am looking for as much VRam as possible to avoid future bottlenecks by . And it sounded quite well to have a GPU at price tag with such a good performance and 4GB of VRam. But seems to be something that’ll get more noticeable in the (more or less) near future. and i don’t want to be blocked by not enough VRam again. SO 970GTX is currently out, and i am waiting for the next Cards.
I feel a little annoyed/cheated with …I know I won’t notice it and I know the card is still a high performer but the fact that it was mis-sold to me makes me angry :mad:
Sure it is, it’s going to be an when ever you need more than 3.5gb in a game. It will cut it’s useful life, and will mean lowering settings to stay below 3.5 gb, compared to a 4gb card. It’s easy to see looking back at previous cards performance and how it ages with vram.
Above 3.5 but below 4? Not likely at all, stuff these days stuff doesn’t scale like it used to, 0.5GB is a big increase compared to like 1GB, but for games what’s most likely to happen is that one setting might use 3GB of memory and the next might need 6GB, it’s not like 0.5GB of memory is the difference between High and Very High settings.
There really isn’t any hard evidence the last partition of 512mb will be any slower than the first 3.5gb. The memory is being routed through in a different manner, but there is no data out there right now saying it is actually impeding the performance of that partition.
There will tests performed in the coming weeks to see if there is any difference, but most likely it will be something like 0-2% slower than the first partition (that’s my best guess based on all info out there right now). It is a blunder on their part, no doubt about that, but the actual doesn’t really look like an.
There really isn’t any hard evidence the last partition of 512mb will be any slower than the first 3.5gb in real world performance usage scenarios.
As far as tech specs yes the details are known, but there is no data yet on whether the bandwidth difference will actually affect performance. It is pretty clear it will have some effect but I highly doubt it will be anything noticeable. It is easy to see how all came about though, since the 970 does have a total of 224 gb/s when added together. It’s obviously not the same as the 980, but I can see how might have slipped by on launch.
Anyways I have an ASUS GTX Strix 980 and absolutely love it, beast of a card. I personally wouldn’t let stop me from buying the 970 if I was looking to get a new video card. It doesn’t change the fact that the 970 is one of the best performance per dollar video cards on the market. I would still recommend it to anyone without hesitation.
I will be interested to see more frametime tests for higher vram usage on settings where the extra vram is actually needed. The vram isn’t the only spec that was incorrect.
But seems we got a bit off topic I am sure everyone can agree for the prices he has for the 970 and 290x, the 290x is obviously the way to go. Especially with the 3xx series and big maxwell will both be coming in a few monhts.
Basically performance wise a single R9 290x and 970 are similar performance, 980 about 15% better performance, and crossfire 290x much better than single 980 in anything that supports crossfire. There is no reason to spend more on a 980 than you would on crossfire 290x especially since you already have a PSU that has enough power to handle them.
But my suggestion would be single R9 290x or 290x crossfire depending on how much you want to spend now. They will give you great performance now, and if you decide you want more in a couple months the new top end cards will be coming out and the 290x should keep it’s value very well especially at the price you can get it.
Exactly. Just as I stated above, the is very minor at best, and won’t really affect you unless you are using a 4k setup.
A 1% to 3% difference in framerate is the best data out there so far for certain games, others are reporting no difference at all. We need more testing to be done before coming to a conclusion on how much, if at all, will affect the average person using the card.
Yeah without seeing frametime graph or percentile of frametime, difference in average means about nothing. Is just a small normal fps drop which would be totally playable, or is it stuttering really hurting playability.
