R9 380.... buy it over a 960, seriously.

[MattLantis]

Not sure if this was said, but I don't feel like wading thru all the responses, but WHY THE FUCK IS HE BENCHMARKING 1440p? Those cards obviously aren't meant to handle them.

[Mikjul]

Not sure if this was said, but I don't feel like wading thru all the responses, but WHY THE FUCK IS HE BENCHMARKING 1440p? Those cards obviously aren't meant to handle them.

380 handles it surprisingly well.

[Tim Drake]

I think you're referring to windows because in linux it's nvidia monopoly in the driver part. amd's drivers are garbage.

Why in the name of all that exists would I be talking about fucking linux?

[othertomperson]

The 380 isn't even on the benchmarks you are pointing to. However, the 285 is and that is enough considering the 380 is a rebranded 285. From the benchmarks you have cited they perform within 4% of eachother.

[don_svetlio]

The 380 isn't even on the benchmarks you are pointing to. However, the 285 is and that is enough considering the 380 is a rebranded 285. From the benchmarks you have cited they perform within 4% of eachother.

380 is a higher clocked 285 and has 2 extra gigs of usable memory - it is around 10% faster than the 960 and no, when both are OCed it's still around 5% faster.

[ccaian]

The music told me to fuck off. That was not very nice. 

[Lays]

Still no benches from the godly 1500 mhz 280x

 

[Mikjul]

Still no benches from the godly 1500 mhz 280x

 

Yeah i am intrigued in seeing GCN on air at 1500mhz.. surely 980ti's stock would have a few issues with that.

[Lays]

Yeah i am intrigued in seeing GCN on air at 1500mhz.. surely 980ti's stock would have a few issues with that.

 

Na, 1500 mhz 280x is only like ~14k GPU score on firestrike, stock 980 ti gets like 16k or something like that.

(For the most part, realworld performance can be gathered by just looking at GPU score on firestrike, since faster GPU's always get higher GPU score)

[othertomperson]

380 is a higher clocked 285 and has 2 extra gigs of usable memory - it is around 10% faster than the 960 and no, when both are OCed it's still around 5% faster.

 

It's higher clocked by 75 MHz, and there is a 4GB version but the base version is 2GB. These are identical to the 960 -- it also has a 4GB version.

 

Also are you aware that 5% of 60fps is 3fps? You'll get bigger variation than that between runs with the same GPU in the same game. You're not doing much to counter the "they're the fucking same" claim.

[don_svetlio]

It's higher clocked by 75 MHz, and there is a 4GB version but the base version is 2GB. These are identical to the 960 -- it also has a 4GB version.

 

Also are you aware that 5% of 60fps is 3fps? You'll get bigger variation than that between runs with the same GPU in the same game. You're not doing much to counter the "they're the fucking same" claim.

the fact that the 960 does not gain anything from the extra memory as it's 128-bit buss doesn't suffice for the memory to be used.

Fact of the matter is that benches show the 380 ahead every time.

[othertomperson]

the fact that the 960 does not gain anything from the extra memory as it's 128-bit buss doesn't suffice for the memory to be used.

Fact of the matter is that benches show the 380 ahead every time.

 

This bullshit again? The 960 uses a 128-bit bus because Nvidia have always favoured lossless compression algorithms over brute forcing it with a wide bus. People said the same thing about the 780's bus ffs. The 960 is perfectly capable of using 4GB vram if you really must have ultra textures with everything else set to medium for some reason because both of these cards are low end.

[don_svetlio]

This bullshit again? The 960 uses a 128-bit bus because Nvidia have always favoured lossless compression algorithms over brute forcing it with a wide bus. People said the same thing about the 780's bus ffs. The 960 is perfectly capable of using 4GB vram if you really must have ultra textures with everything else set to medium for some reason because both of these cards are low end.

IIRC people ran benches with the 2G and 4G versions of the 960 and there was no difference even when the card was pushed harder whereas cards like the 280 and 380 4G had quite the improvement.

[othertomperson]

IIRC people ran benches with the 2G and 4G versions of the 960 and there was no difference even when the card was pushed harder whereas cards like the 280 and 380 4G had quite the improvement.

 

It will depend on what you are pushing. The most obvious one is resolution, but neither the 380 nor 960 should really be used for 1440p. You can get bigger textures with 4GB but the benefit of doing so at such a low resolution is limited. 2GB is about right for both of these cards -- if you want something better I maintain that you are better off making the jump up to 390 or 970. You want some more GPU performance to go with the vram.

[don_svetlio]

It will depend on what you are pushing. The most obvious one is resolution, but neither the 380 nor 960 should really be used for 1440p. You can get bigger textures with 4GB but the benefit of doing so at such a low resolution is limited. 2GB is about right for both of these cards -- if you want something better I maintain that you are better off making the jump up to 390 or 970. You want some more GPU performance to go with the vram.

[don_svetlio]

By all means - argue with benchmarks and numbers - it's your word vs the numbers. I wonder who's the right one.

[-rascal-]

-snip-

 

The last image was interesting...to me at least.

R9-380 2 GB: 28.0 / 29.5

R9-380 4 GB: 36.0 / 37.0

 

28 versus 36...29.5 versus 37.

[MattLantis]

380 handles it surprisingly well.

it really wasn't. It was getting only a few more frames than the other card, and none of them were above 60 FPS

[Atomicnerd64]

Why in the name of all that exists would I be talking about fucking linux?

hey there center of the universe, I didn't know that what only mattered to you has to be the only thing that matters to everyone else. some of us use linux a lot so bad driver support in linux affects us as much as garbage drivers in windows affects you. and don'r be ignorant of alternative operating systems. and again, the fact that you aren't talking about linux doesn't mean that it isn't worth being mentioned and talked about when comparing drivers and graphics card performance.

[Nena Trinity]

R9 380 > GTX 960

[Tim Drake]

hey there center of the universe, I didn't know that what only mattered to you has to be the only thing that matters to everyone else. some of us use linux a lot so bad driver support in linux affects us as much as garbage drivers in windows affects you. and don'r be ignorant of alternative operating systems. and again, the fact that you aren't talking about linux doesn't mean that it isn't worth being mentioned and talked about when comparing drivers and graphics card performance.

When I talk about a gaming GPUs gaming performance / in drivers, problems with games etc I'm not going to be talking about Linux.

[Tim Drake]

it really wasn't. It was getting only a few more frames than the other card, and none of them were above 60 FPS

The 380 is literally cheaper for more frames.

Fanboy alert.

[othertomperson]

By all means - argue with benchmarks and numbers - it's your word vs the numbers. I wonder who's the right one.

 

I don't trust your wccf source that says that a 280 offers up to 160% the performance of a 960. That has not been corroborated by any other source I have seen so I am calling bullshit there.

 

Your second source is a synthetic benchmark that is designed to test GPU performance and you are making the claim that because the exact same GPUs with different amounts of vram have the same result that the vram is unusable? Which is nonsense -- the point of that benchmark is to test the GPU performance, not be bottlenecked by RAM.

 

Your third source says that in one game the 380 does a few fps better. It's worth pointing out that both GPUs handle Very High with average FPS well above 60. Otherwise they are within 5-6 fps of eachother.

 

Your other instantaneous results (not averages so what they actually tell you is extremely limited) show a few FPS difference in all but one. Only the last one implies that a 4GB card actually has any benefit whatsoever, and even then you have only compared 2GB vs 4GB 380s and not done the same with 960s so you haven't proven your claim that in scenarios in which there is a benefit for more vram that the 960 is in any way impeded.

 

Even so if I were getting 26 or even 36 fps I would be wanting to turn down the settings anyway. Vram usage will naturally fall when you do that anyway since the things that require more GPU performance will also, in real world, non-synthetic scenarios, also want more vram.

 

While the benchmarks and numbers you provide don't actually support most of the claims you are making in the way you think they do, what they do imply to me is that the difference between the 380 and the 960 is generally quite small and in cases where you want more than 2GB vram you are much better off going for a more powerful GPU alongside it in either the 970 or the 290/390.

[don_svetlio]

I don't trust your wccf source that says that a 280 offers up to 160% the performance of a 960. That has not been corroborated by any other source I have seen so I am calling bullshit there.

 

Your second source is a synthetic benchmark that is designed to test GPU performance and you are making the claim that because the exact same GPUs with different amounts of vram have the same result that the vram is unusable? Which is nonsense -- the point of that benchmark is to test the GPU performance, not be bottlenecked by RAM.

 

Your third source says that in one game the 380 does a few fps better. It's worth pointing out that both GPUs handle Very High with average FPS well above 60. Otherwise they are within 5-6 fps of eachother.

 

Your other instantaneous results (not averages so what they actually tell you is extremely limited) show a few FPS difference in all but one. Only the last one implies that a 4GB card actually has any benefit whatsoever, and even then you have only compared 2GB vs 4GB 380s and not done the same with 960s so you haven't proven your claim that in scenarios in which there is a benefit for more vram that the 960 is in any way impeded.

 

Even so if I were getting 26 or even 36 fps I would be wanting to turn down the settings anyway. Vram usage will naturally fall when you do that anyway since the things that require more GPU performance will also, in real world, non-synthetic scenarios, also want more vram.

 

While the benchmarks and numbers you provide don't actually support most of the claims you are making in the way you think they do, what they do imply to me is that the difference between the 380 and the 960 is generally quite small and in cases where you want more than 2GB vram you are much better off going for a more powerful GPU alongside it in either the 970 or the 290/390.

First of all - VRAM usage has nothing to do with how powerful the card actually is. I suggest you read @D2ultima 's guide on the subject .

Secondly, the third bench shows a difference of over 10 fps - and that's something I would consider seeing as games will only get more demanding as time goes on. In 2 years, if the 960 is getting 32 fps and the 380 is getting 45 fps - I'd say people who bought a 380 would be much happier than others.

Look at the last image - see the frame time drops with the 2GB cards? Even the 3GB 280 has a drop - the only card that avoids it is the 380 4GB model - when you run out of memory it's not a pleasant thing. Thing is, the 960 doesn't really make good use of the extra memory as it is unable to because of the tiny bus it has. The 380 also has similar compression to that of the 960 but uses a bus that is double the size of the 960. The reason for that is because everyone knows that 2 GB is the bare minimum nowadays - face it. Nvidia have made and will always continue to make cards that are VRAM starved - Look at the 680(770), great card, very good today, but it has too little VRAM to really stay relevant much longer - it has the horsepower but it lacks the VRAM, same for the 670 and the 780 - fact of the matter is that the 960 is an overpriced toy for people who desperately want Nvidia.

[Atomicnerd64]

When I talk about a gaming GPUs gaming performance / in drivers, problems with games etc I'm not going to be talking about Linux.

you are not going to be talking about it. fine. you choose to ignore a part of the pc diy community. but don't throw shit at me or try to act like linux isn't important in the gaming market. valve will surely disagree with you. don't want to talk about it, fine, your right, everything cool. but do not play down it's importance.