960 4GB Thoughts

[vectorx5000]

How good is the SSC for $180? It seems like a great deal.

[Atmos]

Gross.

 

Get a 380 instead, same price, better all around performance, and it can actually use more than 2gbs of vram, thanks to it not being gimped with a 128bit bus, and instead having an actual modern bus of 256bit

[The_Stig]

A 4gb 380 would be better, but thats a pretty good deal.

[vectorx5000]

Alright. Just saw it and wondered.

[Jacktastic-Mofo]

That's a really good deal, also the 960 is a great card! I know from personal experience. 

[don_svetlio]

Crappy card - 380 is more powerful and has the memory bandwidth to properly support 4GB. 380 also supports DX12 properly. Power Color 380 is currently 175$

[Notree]

That's a really good deal, also the 960 is a great card! I know from personal experience. 

Not when comparing it to a 380.

[Jacktastic-Mofo]

Not when comparing it to a 380.

It performs very similarly to a 380, and the 380 only pulls ahead by a few frames in certain titles. The 380 is better mainly for it's ability to handle higher resolution gameplay.

[Decon]

4gb 960 is gimped BS needs to stop...

[Notree]

It performs very similarly to a 380, and the 380 only pulls ahead by a few frames in certain titles. The 380 is better mainly for it's ability to handle higher resolution gameplay.

You said it yourself. So why buy a 960 again? 

[Jacktastic-Mofo]

You said it yourself. So why buy a 960 again? 

There are certain NVIDIA specific features that I really enjoy such as DSR, Shadowplay, Gamestream, Game Optimization, and Geforce Now.

[Hiebly]

Gross.

 

Get a 380 instead, same price, better all around performance, and it can actually use more than 2gbs of vram, thanks to it not being gimped with a 128bit bus, and instead having an actual modern bus of 256bit

 

What on earth are you talking about... the memory bus doesn't affect whether the VRAM benefits you or not

[Notree]

There are certain NVIDIA specific features that I really enjoy such as DSR, Shadowplay, Gamestream, Game Optimization, and Geforce Now.

AMD has VSR, GVR,Broadcast,AMD gaming evolved.. All features AMD have also. If you said cuda and physx you might have a point.

[Hiebly]

AMD has VSR, GVR,Broadcast,AMD gaming evolved.. All features AMD have also. If you said cuda and physx you might have a point.

 

Shadowplay is superior to AMD's solution. It has higher image quality and works in EVERY DirectX tower as well as allowing desktop recording, whereas AMD's only works with titles that Raptr (or plays.tv or whatever) has made a compatibility layer for. Unless you want to use a VCE-compatible release of OBS but I've always had lots of issues getting OBS to work properly.

[Notree]

Shadowplay is superior to AMD's solution. It has higher image quality and works in EVERY DirectX tower as well as allowing desktop recording, whereas AMD's only works with titles that Raptr (or plays.tv or whatever) has made a compatibility layer for. Unless you want to use a VCE-compatible release of OBS but I've always had lots of issues getting OBS to work properly.

Wheres your evidence? DirectX tower ? Not sure what titles it doesn't work with. Can you name a few? And obs is by far better then both. If you cant get it to work properly your doing something wrong.

[Atmos]

What on earth are you talking about... the memory bus doesn't affect whether the VRAM benefits you or not

 

However, it does. Memory bus has a direct effect on how well the memory modules perform. Hence why 4gbs @ 4096bit of hbm can keep pace with 12gbs of gddr5 @ 384bit on the titan x at 4k.

 

Bus makes a big difference, and while nvidia has made leaps and bounds with their transfer of textures to and from memory, just increasing the memory bus on the card would yield it even more performance. There plainly was no reason to gimp its bus so hard, when they could have just kept with a 256bit as is standard and had extra performance. But when posed the question they back off with "it was to keep costs down" bullshit excuses. 

 

when all is said and done, you're looking at 4 year old technology, being driven by current software, which is just disgusting in terms of performance. Step it up nvidia, make hardware to suite your software, stop cutting corners for the sake of power efficiency, or style, or cost. If AMD can do it, you can too.

[Kinda Bottlenecked]

Good card. Not as good as the 380.

[Hiebly]

However, it does. Memory bus has a direct effect on how well the memory modules perform. Hence why 4gbs @ 4096bit of hbm can keep pace with 12gbs of gddr5 @ 384bit on the titan x at 4k.

 

They're completely unrelated. Bandwidth and quantity are independent, one doesn't make up for the other or depend on the other. 4GB of HBM doesn't "keep pace" with 12GB of GDDR5. The HBM is faster than GDDR5 in throughput, but in any situation where 4GB is not enough the fact that it is HBM makes no difference, the performance will crash and burn.

[Atmos]

They're completely unrelated. Bandwidth and quantity are independent, one doesn't make up for the other or depend on the other. 4GB of HBM doesn't "keep pace" with 12GB of GDDR5. The HBM is faster than GDDR5 in throughput, but in any situation where 4GB is not enough the fact that it is HBM makes no difference, the performance will crash and burn.

for normal textures, and files yes. but a lot of the reason that we have for games so much memory is not becuase every file on vram is being used, but because a lot of it is waiting to be offloaded or imported onto the vram. the actual file size isnt that large, which is why right now, 4gbs of hbm bottlenecks as much as anywhere from 8-12gbs of gddr5.and which is why the fury series is the only gpu that performs better comparitively the higher the resolution, and the more it can make use of its very, very fast memory. but according to you hbm means nothing in games, it might as well be gddr5 and all the benchmarks and every result of testing it should be scrapped.

[xAcid9]

Great deal if you can't find a 4GB 380 for around the same price +-$15.

[Hiebly]

for normal textures, and files yes. but a lot of the reason that we have for games so much memory is not becuase every file on vram is being used, but because a lot of it is waiting to be offloaded or imported onto the vram. the actual file size isnt that large, which is why right now, 4gbs of hbm bottlenecks as much as anywhere from 8-12gbs of gddr5.and which is why the fury series is the only gpu that performs better comparitively the higher the resolution, and the more it can make use of its very, very fast memory. but according to you hbm means nothing in games, it might as well be gddr5 and all the benchmarks and every result of testing it should be scrapped.

 

Lol so you ignore what I'm saying and then put words in my mouth... goodbye