GPU Value Comparisons

[PhreeMe]

Go read up on galaxy's relationship/history with nvidia and as oem provider for other brands, then go look at kfa2's market where they did niche high end stuff. Both top notch companies that merged and due to a trademark dispute (samsung) now trade as galax.

Quick article I found http://www.bit-tech.net/news/hardware/2014/09/18/nvidia-partners-galaxy-and-kfa2-merge-to-fo/1

Thanks for the link.
I knew KFA2, Galaxy and Galax were in a relationship of sorts, and now I understand it

If I had a prior generation card (77 series) I'd most likely have gone for the R9390x. This time I wanted an actual jump and Fury was meant to be it. To correct that statement, it's not that Fury doesn't far outperform my 9280x, but I really would like to 'not have' the 30fps cap on prior versions of HDMI at >1080p resolutions. A very nitpicky reason to change allegiances, but sadly a valid enough reason for me

What keeps irking me about this choice, is that at 4k, the 980 Ti and the Fury X are basically equals. But what about down the road? Will they remain equals as graphical needs start advancing with new releases, or will HBM actually be the better long term choice?

I don't know how to compare those cards future strength. Bah humbug.

 

[Fulcrum29]

Thanks for the link.
I knew KFA2, Galaxy and Galax were in a relationship of sorts, and I understand it

If I had a prior generation card (77 series) I'd most likely have gone for the R9390x. This time I wanted an actual jump and Fury was meant to be it. To correct that statement, it's not that Fury doesn't far outperform my 9280x, but I really would like to 'not have' the 30fps cap on prior versions of HDMI at >1080p resolutions. A very nitpicky reason to change allegiances, but sadly a valid enough reason for me

What keeps irking me about this choice, is that at 4k, the 980 Ti and the Fury X are basically equals. But what about down the road? Will they remain equals as graphical needs start advancing with new releases, or will HBM actually be the better long term choice?

I don't know how to compare those cards future strength. Bah humbug.

The real question is how these cards will compare when we enter the DX12 and Vulkan revolution where Mantle has 'inspired' the development, HBM may then have an advantage...

 

[ponder]

The real question is how these cards will compare when we enter the DX12 and Vulkan revolution where Mantle has 'inspired' the development, HBM may then have an advantage...

I think when it comes to dx12 there are bigger performace gains for amd compared to nvidia if I'm not mistaken and this was with the gcn archtecture.

Would be nice if someone could do some dx12 benchmarks even if it's just synthetic ones as I'm not aware or of any dx12 games out at the moment.

 

[Fulcrum29]

I think when it comes to dx12 there are bigger performace gains for amd compared to nvidia if I'm not mistaken and this was with the gcn archtecture.

Would be nice if someone could do some dx12 benchmarks even if it's just synthetic ones as I'm not aware or of any dx12 games out at the moment.

Which is a problem for both AMD (from a sales point of view) and their prospects. Timing is an issue… Valve only recently announced Source 2 (compatible with Vulkan) and released an beta update for Dota 2 utilising the Source 2 engine, the only games Valve mentioned to be updated is CS:GO and Dota 2. This is the first (under correction) games which will be benefitted by a ‘new’ API which is utilising the essential mechanics behind Mantle. CryEngine (BUILD 2015) will be the first engine to utilise DX12. Benchmarks will take time…

 

[CataclysmZA]

Would be nice if someone could do some dx12 benchmarks even if it's just synthetic ones as I'm not aware or of any dx12 games out at the moment.

I plan to do a bunch of these before I end off my testing, with a Windows 7 vs 8.1 vs 10 match-up to see how the performance differs between the OSes when using the same hardware and drivers. I expect some interesting results, especially for AMD.

Also, I have these sitting in my kitchen.

View attachment 229776

Its going to be the most hectic six weeks of my life.

Edit: not sure why the forum switches the photo sideways, it was definitely taken in the correct orientation. Meh.

 

[Fulcrum29]

I plan to do a bunch of these before I end off my testing, with a Windows 7 vs 8.1 vs 10 match-up to see how the performance differs between the OSes when using the same hardware and drivers. I expect some interesting results, especially for AMD.

Also, I have these sitting in my kitchen.

View attachment 229776

Its going to be the most hectic six weeks of my life.

Edit: not sure why the forum switches the photo sideways, it was definitely taken in the correct orientation. Meh.

Will you be doing 3DMark API Overhead Feature tests, because as I understand it is only a proof of concept test, or will you be doing day-to-day testing thus apply it to your daily user experience?

 

[Rickster]

 

[CataclysmZA]

Will you be doing 3DMark API Overhead Feature tests, because as I understand it is only a proof of concept test, or will you be doing day-to-day testing thus apply it to your daily user experience?

That's one part of the testing that I could do, but it probably won't feature in testing suites for Windows 10 for more than two months after launch. Once 3DMark12 comes out with the next version of those benchmarks, I might be able to switch to that. I do wish more DX12 tests existed at this point, developers have had more than enough time with a solid base to work from to make their final versions of the software. All Microsoft is doing now is tweaking the UI and base services and software and nothing hardware-related is going to change from today until the July release.

Performance-wise now, though, there are definitely noticeable improvements from AMD's side when it comes to scheduling and how the driver works. There's less overhead overall on DX11 games and that has some improvements ranging from minute to decent depending on the game being tested.

 

[ponder]

I plan to do a bunch of these before I end off my testing, with a Windows 7 vs 8.1 vs 10 match-up to see how the performance differs between the OSes when using the same hardware and drivers. I expect some interesting results, especially for AMD.

Would be nice if one could obtain demos for the new Deus Ex, Star Citizen, King of Wushu etc where one could test this but I doubt that's gonna happen.

My bet is still on AMD getting the biggest benefit out of DX12.

Do you get to keep those GPUs or do you have to return them after evaluation?

 

[CataclysmZA]

Would be nice if one could obtain demos for the new Deus Ex, Star Citizen, King of Wushu etc where one could test this but I doubt that's gonna happen.

My bet is still on AMD getting the biggest benefit out of DX12.

Do you get to keep those GPUs or do you have to return them after evaluation?

When those games come out, we'll definitely look at doing testing for that. I'm certainly moving over to Windows 10 for my main work rig.

I don't get to keep them all, sadly. If I'm able to pick one or two for future reference, it'd be the R9 390X and R9 380. I think Hawaii and Tonga are going to see some very substantial changes over the next couple of months with game performance in Windows 10.

 

[PhreeMe]

Would be nice if one could obtain demos for the new Deus Ex, Star Citizen, King of Wushu etc where one could test this but I doubt that's gonna happen.

My bet is still on AMD getting the biggest benefit out of DX12.

Agree with demo testing bit. The only thing that keeps holding me back from actually purchasing a new gfx card, is that I'd like to know what actually happens between nVidia and AMD when DX12 actually hits.

Ponder, I know AMD has got 12_0 support on their R9300 and Fury cards, but are you saying you expect them to perform better from Fury's perspective due to HBM? I know the features that appear in 12_1 are largely being downplayed, but surely that can't be discounted?

 

[CataclysmZA]

Ponder, I know AMD has got 12_0 support on their R9300 and Fury cards, but are you saying you expect them to perform better from Fury's perspective due to HBM? I know the features that appear in 12_1 are largely being downplayed, but surely that can't be discounted?

With Fiji, it isn't HBM that's going to benefit it, but the architecture itself. My guess is, looking at reviews online, that the Fury X isn't being used to its limit. I have a hunch that drivers based on DX11 are holding parts of it back and that it could end up being way better than Maxwell. The Fiji arch is a very wide design and one of the benefits of DX12 is better scheduling and better branch prediction algorithms, which would mean that more parts of the GPU can be put to work to run things that will only be used later, saving time and making better use of the hardware available.

 

[Gboss]

Is the R9 390 a good upgrade? Or should I hold out for pricing on the Fury Nano?

 

[CataclysmZA]

Is the R9 390 a good upgrade? Or should I hold out for pricing on the Fury Nano?

Depends on what you had before, really, as well as your chosen resolution.

 

[Gboss]

Depends on what you had before, really, as well as your chosen resolution.

I currently have a Ati 6870 that needs retiring. I recently upgraded my PC but held back on the GPU because I was waiting to see Ati's new R9 300 series cards before buying.

 

[Fulcrum29]

One thing about the AMD Fury series, is it only me or are AMD steering the design towards Steam PC (custom and OEM)? This would position them in an early exploitable market…

 

[Gboss]

One thing about the AMD Fury series, is it only me or are AMD steering the design towards Steam PC (custom and OEM)? This would position them in an early exploitable market…

The Fury range is remarkably smaller than previous cards, literally half the length thanks to stackable HBM, which would you could easily fit one into a SFF PC to be used as a Steam PC.
You might be onto something there..

 

[ponder]

One thing about the AMD Fury series, is it only me or are AMD steering the design towards Steam PC (custom and OEM)? This would position them in an early exploitable market…

AMD's linux drivers suck donkey balls so good luck on that.

 

[Fulcrum29]

AMD's linux drivers suck donkey balls so good luck on that.

AMD's latest drivers only benefits the Fury and 3xx. This brought up questions to why the 2xx isn't benefitting from the drivers as it is generically the same as the 3xx. Due to the Fury being earmarked for VR gaming, it can be assumed that they will be targeting the Steam platform.

 

[CataclysmZA]

I currently have a Ati 6870 that needs retiring. I recently upgraded my PC but held back on the GPU because I was waiting to see Ati's new R9 300 series cards before buying.

That's a good upgrade, you're looking at somewhere close to a 3x jump in performance. I myself went from a HD6870 1GB to a Radeon R7 265 2GB and it was worth it even though the performance jump was about 50%.

AMD's latest drivers only benefits the Fury and 3xx. This brought up questions to why the 2xx isn't benefitting from the drivers as it is generically the same as the 3xx. Due to the Fury being earmarked for VR gaming, it can be assumed that they will be targeting the Steam platform.

My guess is that because there are more voltage control points inside the refreshed chips, the power management features like frame-rate control don't work like they should. The memory controller also supposedly carries some enhancements to handle 6GHz chips on Hawaii.

Also, increased sales because there's higher tesselation performance that R9 290-series doesn't benefit from yet. Sneaky, but makes business sense. Apple does that all the time.

On the Linux side, the most recent improvements to the open-source and fglrx drivers have targeted Kaveri and Tonga. I'm guessing that the Linux driver team is much smaller, but there are definitely improvements going on that side.