Recommend

[Nende21]

Hiya guys,

I am new to the forums and since I decided to 'upgrade' my ol' beast. I thought i would ask you guys for advice. On which Graphics card to buy that would be still good for next gen games possibly. My max spending bracket is about 270 pounds. So that would be $412 after the conversion from Pounds to Dollars. Keep in mind this Graphics card has to be sold in UK as i do not wish to wait for my graphics card to arrive after a year xD (That was a joke for some that did not get it.)

I was looking into;

Sapphire AMD R9 390 Nitro Graphics Card
and
MSI NVIDIA GTX 970 Gaming Twin Frozr


Read and watched a couple of a reviews about them but they still have not given me a sure go. Came here just for the advice before i make my final decision as some say that could be a waste of money and so on. I will mostly be playing GoW2 (Guild of wars 2) but would want to be in a mind set if a game comes out this year which i like i would not need to check my GFX requirements if it is going to run on min settings xD

Thanks a lot guys and please give me information on the products you recommend not just because 'it is better'...

P.s feel free to suggest a cheaper card that could potentially give the same results. Does not have to be THAT price. It is just the limit i have set my self.

tl;dr
Need advice on which graphics card to buy with a budget of around $412. ty.

[BicycleMafioso]

The 390 is a stronger card than the 970 by a bit, but none are really a bad choice.

[Akainakali]

In terms of how powerful a card you need. That is really determined by the resolution that your monitor(s) has. Most $200+ cards will handle 1080p with all the eye candy on most games. For 2560x1440 you're going to want at least a 390 or 970 and those are the sweet spot in terms of performance.

Do you mean Guild Wars 2?

The big question mark at the moment is what effect Dx12 (windows 10) will have on things. Early indications are that there may be architectural issues with the current (and at least some prior) nVidia cards. If you plan to stay on W7/Dx11, then on average the nVidia cards are better.

In general what you are really buying a card on is the cooler. With rare exceptions, usually just after a new product release, AMD and nVidia cards in a similar price range perform similarly. The exception being the top end, where one or the other may have a significant advantage, since those are the maximum the current products allow, but you aren't looking at those regardless.

So IME, what really distinguishes the cards is the coolers. That controls the overclocking limits (due to heat being what causes current cards to throttle and reduce performance) and how much noise it generates. nVidia has a significant advantage here on most current cards since their power draw is much lower than most of the AMD cards. So they are less likely to throttle and fans can run slower (or not at all) and so make less noise.

Most of the non-reference cards have better and quieter coolers than the reference cards. It's not hard to find one that should suit you.

http://www.techpowerup.com/reviews/?...=25&order=date

[Vash The Stampede]

Right now with the way Windows 10 and DX12 is going, the AMD R9 390 is the better choice. Also cause 8GB vs 3.5GB.

[Akainakali]

Right now with the way Windows 10 and DX12 is going, the AMD R9 390 is the better choice. Also cause 8GB vs 3.5GB.

The Dx12 may well be important, but the ram is not.

Adding lots of ram to graphics cards isn't an advantage in most situations and is one of the biggest scams in computers. It is specifically designed to con the ignorant who think "GTX 950 wt 4gb of ram, it must be as good as the 970 and it's half the price".

The AMD would do just fine with half the ram, just like the previous versions using the same chip did. They doubled the ram, because it was a way of making the cards look better relative to the nVidias.

At the resolutions either card is intended to run at they both have more than sufficient ram and people are generally either replacing cards long before it would hurt them and/or they lack the horsepower to take advantage of it.

[BicycleMafioso]

The Dx12 may well be important, but the ram is not.

The real issue with the RAM is actually memory bandwidth, it having a 192-bit memory bus with a 64-bit helper. The membus isn't wide enough. Can run at 1440p? Yeah. But its performance relative to its comptetion drops off even further, meaning the gap up widens more and more.

[Akainakali]

The real issue with the RAM is actually memory bandwidth, it having a 192-bit memory bus with a 64-bit helper. The membus isn't wide enough. Can run at 1440p? Yeah. But its performance relative to its comptetion drops off even further, meaning the gap up widens more and more.

167 FPS in Wow at 2560x1440 seems adequate to me.

http://www.techpowerup.com/reviews/M...Gaming/24.html

Beyond that, yeah it starts running out of steam, but even SLI/Crossfire top of the line cards struggle at 4k+ resolutions.

[Deleted]

167 FPS in Wow at 2560x1440 seems adequate to me.

They probably stared at a wall the entire time. Benchmarks like this are very misleading or even worse - there are way too many variables to condense into a single number. It's the same with RTS games. Testing different scenarios: good; just going ahead assigning one number for "general performance": intellectually dishonest (or just stupid, take your pick).

And yes, FPS scaling with load very much depends on hardware/drivers/OS/... .

[BicycleMafioso]

167 FPS in Wow at 2560x1440 seems adequate to me.

Wow is a CPU-game.
I'm not trying to be harsh, but it's the absolute worst example you could've come up with.

[Akainakali]

They probably stared at a wall the entire time. Benchmarks like this are very misleading or even worse - there are way too many variables to condense into a single number. It's the same with RTS games. Testing different scenarios: good; just going ahead assigning one number for "general performance": intellectually dishonest (or just stupid, take your pick).

And yes, FPS scaling with load very much depends on hardware/drivers/OS/... .

I am missing the part where I claimed that this was the be all and end all of testing.

Just that it was relevant to TG's claim that the memory bus crippled the 970 at 1440p resolutions.

Wow is a CPU-game.
I'm not trying to be harsh, but it's the absolute worst example you could've come up with.

Yes, the graphics card has absolutely no effect on FPS in wow. That's why all of the graphics cards report identical frame rates.

Besides how could WoW benchmarks possibly be relevant to a WoW dominated web board?

The memory bus on a 970 doesn't hurt it's performance relative to the 512 bit bus on the AMD cards at 1440p vs 1080p or even going up to 4k. The relative performance differences (differences in frame rates) are still essentially the same at each resolution, even on a game like Crysis 3. I'm sure you can find edge cases of games where it does, but it simply isn't the factor you seem to think it is.

http://www.techpowerup.com/reviews/M...Gaming/14.html

[BicycleMafioso]

Yes, the graphics card has absolutely no effect on FPS in wow. That's why all of the graphics cards report identical frame rates.

Besides how could WoW benchmarks possibly be relevant to a WoW dominated web board?

The memory bus on a 970 doesn't hurt it's performance relative to the 512 bit bus on the AMD cards at 1440p vs 1080p or even going up to 4k. The relative performance differences (differences in frame rates) are still essentially the same at each resolution, even on a game like Crysis 3. I'm sure you can find edge cases of games where it does, but it simply isn't the factor you seem to think it is.

http://www.techpowerup.com/reviews/M...Gaming/14.html

1) Indeed it does not. I realise you are trying to be snide, but it absolutely doesn't matter past a certain point, which both cards listed is well beyond.

2) Lol, WoW. Move on. In seriousness, WoW is not a good game to draw any conclusions from whatsoever. Any kind of 'test' you do with WoW is bound to be flawed in execution or setup simply because of it being the abomination of code that WoW is and what it is trying to accomplish. That itself is disqualifying itself. Why anyone includes it in a benchmark is far beyond me, because any benchmark you do in WoW is completely unrepresentative of performance with the GPU; the GPU beyond like a r7 260x will never be a limiting factor.

3) In tests you provided alone it shows that it severely changes the positioning between the 970 and the 290 (a weaker card than the 390) with 900p, to 1080p, draws out the newfound advantage in 1440p and beyond and thus the 970 should never be considered because its arguably stronger GPU is gimped by its membus.
I am hoping this was your point, because this is what you are proving.

[Lathais]

The memory bus on a 970 doesn't hurt it's performance relative to the 512 bit bus on the AMD cards at 1440p vs 1080p or even going up to 4k. The relative performance differences (differences in frame rates) are still essentially the same at each resolution, even on a game like Crysis 3. I'm sure you can find edge cases of games where it does, but it simply isn't the factor you seem to think it is.

http://www.techpowerup.com/reviews/M...Gaming/14.html

Entirely false.

Look here:
http://www.sweclockers.com/test/2019...an-x/6#content

You can see that at 1080p the 970 is actually ahead of the 290x by 3FPS. At 1440p and 4K, they are about equal with the 290x pulling ahead by 1 FPS. The 290x simply performs better at higher resolutions then the 970. Since the 390x is basically a 290x with faster memory, the 390x will pull ahead even more.

[Vash The Stampede]

2) Lol, WoW. Move on. In seriousness, WoW is not a good game to draw any conclusions from whatsoever. Any kind of 'test' you do with WoW is bound to be flawed in execution or setup simply because of it being the abomination of code that WoW is and what it is trying to accomplish. That itself is disqualifying itself. Why anyone includes it in a benchmark is far beyond me, because any benchmark you do in WoW is completely unrepresentative of performance with the GPU; the GPU beyond like a r7 260x will never be a limiting factor.

Without a benchmark built into the game, any test done in WoW is Esoteric Recondite Arcana. But we know without any question of a doubt that most GPUs in the market is more than enough to handle WoW.

So if you're buying a GPU for WoW you might as well get one that is at least future proof and able to play other games as well. And yea the 8GB of memory on the 390 is marketing fluff but it's a lot better than the 3.5GB on the 970. Developers aren't going to limit their games to 3.5GB, but instead 4GB because a lot more graphic cards in the market have that memory configuration.

[Akainakali]

1) Indeed it does not. I realise you are trying to be snide, but it absolutely doesn't matter past a certain point, which both cards listed is well beyond.

Do or do not. There is no try.

Fortunately, you are the soul of polite discourse and would never think of telling someone something in a patronizing manner.

(was that better?)

2) Lol, WoW. Move on. In seriousness, WoW is not a good game to draw any conclusions from whatsoever. Any kind of 'test' you do with WoW is bound to be flawed in execution or setup simply because of it being the abomination of code that WoW is and what it is trying to accomplish. That itself is disqualifying itself. Why anyone includes it in a benchmark is far beyond me, because any benchmark you do in WoW is completely unrepresentative of performance with the GPU; the GPU beyond like a r7 260x will never be a limiting factor.

You really love the sound of telling yourself you're correct don't you.

You said there was no difference, but in my understanding of math 98 does not equal 175.

Is this fact unclear to you or do I need to break out the crayons?

Apparently this will also come as a shock, but some people want to know how a card performs on a game they play regardless of the relevance of the numbers to the real world.

Both maybe higher FPS than matters at that resolution, but saying WoW is purely CPU bound is ignoring the fact that not all cards perform the same on it.

Also while the FPS number for WoW may not be particularly relevant, they are generally consistent with how the cards perform on other games. So it is not without value as a rough guide to performance.

3) In tests you provided alone it shows that it severely changes the positioning between the 970 and the 290 (a weaker card than the 390) with 900p, to 1080p, draws out the newfound advantage in 1440p and beyond and thus the 970 should never be considered because its arguably stronger GPU is gimped by its membus.
I am hoping this was your point, because this is what you are proving.

Show me where the frame rates fall off a cliff due to the higher resolution, due to the crippled memory bandwidth.

If it was hitting a wall due to the memory bandwidth, you'd see a resolution where it's performance drops drastically vs the cards with a wider memory bus.

That simply doesn't happen in any of the results I've looked at, not even at 4k. The performance drops clearly, but it basically maintains the same relative performance to the other cards that you see at lower resolutions. I'm sure you can find SOME benchmark where it occurs if you dig enough though, but that gets back to the whole Graphics Engine vs system architecture.

The 290 vs a 970 in Crysis3, changes from a -2 fps deficit to a <+2 fps advantage going from 900p to 4k.

That's not what I consider a dramatic win or difference. The differences between the cards at the different resolutions, are largely the same.

Since clearly I have to be pedantically specific, "largely the same" does not mean they are identical.

Your clearly false claim that it's gimped by the memory bus simply doesn't hold up. It is not useless, it's performance does not fall off a cliff relative to the other cards when you get to higher resolutions.

That's what crippled means.

I was talking about the frame rate differences within the Crysis results, not between that and the WoW or any other game results.

Some games favor one architecture over another, a fact that seems lost on you as you congratulate yourself on how awesome you are. Does the 970 perform worse on Crysis3 relative to the AMD cards vs other games, yes clearly it does.

However your argument was that it's crippled, it clearly isn't.

The 970 is not the most awesome card for 1440p resolutions, but the games where it's frame rates are low at 1440p are typically games where even SLI/crossfire top of the line cards can struggle to produce decent frame rates.

It is more than sufficient for most people running at 1440p. This does not mean it will lead the pack in all games. Some games it's going to suck on, but that is much more due to the general architecture than some imaginary crippling limitation.

It's not a quad SLI Titan X setup, but it's not supposed to be.

- - - Updated - - -

Entirely false.

Look here:
http://www.sweclockers.com/test/2019...an-x/6#content

You can see that at 1080p the 970 is actually ahead of the 290x by 3FPS. At 1440p and 4K, they are about equal with the 290x pulling ahead by 1 FPS. The 290x simply performs better at higher resolutions then the 970. Since the 390x is basically a 290x with faster memory, the 390x will pull ahead even more.

So you consider a 1-3 FPS difference to be a crippling disadvantage then?

Other than that see above.

[Lathais]

So you consider a 1-3 FPS difference to be a crippling disadvantage then?

Other than that see above.

I never said it was a crippling disadvantage. I simply said that your statement that I bolded was entirely false, which it is. There is a measurable difference in which one performs better at 1080p but the other performs better at 1440p and 4K. The difference does not remain the same, one overtakes the other.

[Akainakali]

I never said it was a crippling disadvantage. I simply said that your statement that I bolded was entirely false, which it is. There is a measurable difference in which one performs better at 1080p but the other performs better at 1440p and 4K. The difference does not remain the same, one overtakes the other.

The memory bus on a 970 doesn't hurt it's performance relative to the 512 bit bus on the AMD cards at 1440p vs 1080p or even going up to 4k. The relative performance differences (differences in frame rates) are still essentially the same at each resolution, even on a game like Crysis 3. I'm sure you can find edge cases of games where it does, but it simply isn't the factor you seem to think it is.

I'm afraid arguing in english does not work like you think it does. You have to read ALL the words, at least within a paragraph before you go making statements like "completely false". Due to readability and grammar, you aren't supposed to have a single sentence contain all of your statements or points.

There were other important parts of the statements bolded, italicized and underlined, that you ignored. I am unimpressed with your reading comprehension.

Since as we are being pointlessly pedantic and seizing on parts of my post while ignoring the rest of it.

My statement was about the significance of the difference in the memory bus width.

It was NOT about the absolute performance of the two cards. I did not state there was absolutely no differences in the FPS performance at different resolutions, just that the memory bus width wasn't hurting that performance.

You ignored my statements about "Essentially the same" (it means that differences exist, but the differences aren't significant) and that you can probably find edge cases where there was one.

Apparently so you could say my statement was completely false, presumably to enable yourself to claim some sort of victory.

For it to be completely false you would have to have show that the FPS change was SOLELY due to the memory bus difference and not due to the GPU architecture, driver or ALL of the myriad of other factors that come into generating the frames.

So, in the spirit of being pointlessly pedantic and seizing on specific parts of a post.

You have proven this difference is due solely to the memory bus how?

I'm afraid that showing there are minor differences in the frame rate doesn't isolate out the effect of the memory bus.

However, I will freely concede that I was insufficiently specific and there by left open room for people who obsess over tiny difference to state that I was COMPLETELY and utterly wrong.

If it makes you happy, I will be more specific. A +/- 1 to 3 frame difference in performance RELATIVE to another card is insignificant. I'd say even up to about 5 FPS isn't worth worrying about. Either the frame rates are high enough that it doesn't make a difference or they are low enough that it isn't helping.

I'd say about a 10-15 FPS difference for moderate frame rates (up to 60 FPS) or something on the order of 20% difference in large FPS numbers shows significant performance differences.

[Vash The Stampede]

I noticed the OP hasn't responded since the first post. May want to see what the OP says.