Support NeoGAF

LelouchZero · Jun 25, 2017

Regular viewers of this series will know that it's sometimes a struggle to get a great 4K experience from GTX 970, involving lots of testing and settings tweaks. But what if the game did the settings adjustments for you on the fly? Wouldn't that be awesome? Well the Forza games on PC show you just how cool that actually is.

Richard runs Forza Horizon 3 at 4K 30 fps with the dynamic high quality settings enabled. Gameplay offers a locked 30 fps however has stutters during "super detailed close-ups" which only occur in engine driven cut-scenes.

Forza Motorsport 6: Apex runs at a locked 4K 60 fps with the dynamic high quality settings enabled as well.

This is an excellent showing for the Forza Motorsport engine, the dynamic optimization capabilities of the engine is really impressive and is able to scale the visuals up and down in order to hit the frame-rate target. It also scales up really well from Consoles to PCs.

The GTX 970 released on September 18th, 2014 and is almost a 3 years old GPU at this point. It is over 2 years, 9 months old at the time of creating this thread.

Richard also expresses his desire to test Forza Motorsport 7 when it releases on the Xbox One and on Windows 10 PCs, and wonders if it can run at 4K 60 fps on a GTX 970 without 4K textures, or with 4K textures on a RX 480, 580 and even an R9 390 which is a refreshed R9 290 with 8GB of memory.

I'm really interested in seeing how it performs on an R9 290 and 290X, the 290X is a 5.6 teraflop GPU that released on October 24th, 2013 and is over 3 years and 8 months old, a 66MHz boost in clock speed will put it over 6TFs, it will be fascinating to see how it performs!

hardcastlemccormick · Jun 25, 2017

I played Horizon 3 at 4k/30 when I had a GTX 970 and it was a lot of fun. The 4K resolution bump really makes a difference when seeing down the road at distant objects and other cars.

I tested the game prior to the optimization patch, looks like they got better results in their testing. Good sign.

napata · Jun 25, 2017

LelouchZero said:
Source

I'm really interested in seeing how it performs on an R9 290 and 290X, the 290X is a 5.6 teraflop GPU that released on October 24th, 2013 and is over 3 years and 8 months old, a 66MHz boost in clock speed will put it over 6TFs, it will be fascinating to see how it performs!

In theory it should be about 5-10% slower at 6 TFs than the XBOX X, given that that's around the difference in "IPC" between GCN 2.0 & 4.0.

Does anyone still have that benchmark that compared all the different GCN architectures with the same TFs? I think it was German.

grmlin · Jun 25, 2017

I tried the Horizon 3 Demo in 4K on my RX 470 and was pretty impressed on how good it looked and ran. Pretty cool.

Sad Affleck · Jun 25, 2017

VRAMgate aside the GTX 970 belongs to the graphics card hall of fame. Amazing performance even at a resolution it was never designed for.

Xyber · Jun 25, 2017

napata said:
In theory it should be about 5-10% slower at 6 TFs than the XBOX X, given that that's around the difference in "IPC" between GCN 2.0 & 4.0.

Does anyone still have that benchmark that compared all the different GCN architectures with the same TFs? I think it was German.

Nvidia Teraflops and AMD Teraflops can't be compared like that, so doesn't even work in theory. AMD usually needs more TF to get the same in-game performance as Nvidia.

480 and 1060 are roughly similar in performance, the 480 comes in at 5,8TF and the 1060 at 3,8TF.

Edit: Oh, you were probably talking about the AMD cards mentioned in that post. Just glanced over it and only saw the video link so thought you were talking about the 970.

LelouchZero · Jun 25, 2017

hardcastlemccormick said:
I played Horizon 3 at 4k/30 when I had a GTX 970 and it was a lot of fun. The 4K resolution bump really makes a difference when seeing down the road at distant objects and other cars.

I tested the game prior to the optimization patch, looks like they got better results in their testing. Good sign.

I played some of Horizon 3 at 4K 30 as well, the image quality is really great! I prefer 1080p 60 though for the fluidity.

napata said:
In theory it should be about 5-10% slower at 6 TFs than the XBOX X, given that that's around the difference in "IPC" between GCN 2.0 & 4.0.

Does anyone still have that benchmark that compared all the different GCN architectures with the same TFs? I think it was German.

This one?
GCN 1 vs GCN 3 vs GCN 4 (Thread)
AMD Polaris: The fourth generation GCN delivers so much power

It appears that the performance difference varies depending on the game and the workload subjected to the GPU. It's very interesting to see the performance across different versions of GCN.

Xyber said:
Nvidia Teraflops and AMD Teraflops can't be compared like that, so doesn't even work in theory. AMD usually needs more TF to get the same in-game performance as Nvidia.

480 and 1060 are roughly similar in performance, the 480 comes in at 5,8TF and the 1060 at 3,8TF.

I believe they're talking about the R9 290X that I mentioned which is based on GCN 2, while the RX 480 is based on GCN 4 which others higher performance per compute unit compared to the previous versions of GCN.

napata · Jun 25, 2017

Xyber said:
Nvidia Teraflops and AMD Teraflops can't be compared like that, so doesn't even work in theory. AMD usually needs more TF to get the same in-game performance as Nvidia.

480 and 1060 are roughly similar in performance, the 480 comes in at 5,8TF and the 1060 at 3,8TF.

Huh? I never mentioned Nvidia. Not that you are wrong or anything.

Also no 1060 is really 3.8 Tflops. Pretty sure all of them boost past 1.8 ghz. Mine boosts close to 1.9 ghz at stock clocks if it isn't too hot outside, which makes it a 4.8 Tflops card. Really, all Pascal cards have much higher Tflop values than advertised. I don't know why Nvidia undersells their boost clocks so much.

LelouchZero said:
This one?
GCN 1 vs GCN 3 vs GCN 4 (Thread)
AMD Polaris: The fourth generation GCN delivers so much power

It appears that the performance difference varies depending on the game and the workload subjected to the GPU. It's very interesting to see the performance across different versions of GCN.

Yes, that's what I was looking for. Thanks. It was a pretty interesting article but I forgot to save it. Didn't think about using Neogaf search.

Edit: It does seem to make a lot of difference depending on the game. Almost 40% in TW3 and less than 10% in FC: Primal. I guess that makes sense given that games are all coded differently. The same happens with Nvidia where sometimes a 1060 beats a 980 by 5% and sometimes the opposite happens.

Gestault · Jun 25, 2017

Performance checks like this for lower-end options on PC are a great idea for DF content, especially with a game series over a few different iterations.

Sad Affleck · Jun 25, 2017

Gestault said:
Performance checks like this for lower-end options on PC are a great idea for DF content, especially with a game series over a few different iterations.

Yeah, it's a great series. Arkham Knight was the most impressive so far given the initial state of the port.

Gren · Jun 25, 2017

Didn't know the Forza games had this dynamic feature. It's pretty cool.

Support NeoGAF

Digital Foundry: 4K On A Budget: GTX 970 vs Forza Horizon 3 + Forza Motorsport 6 Apex

LelouchZero

Member

hardcastlemccormick

Banned

napata

Member

grmlin

Member

Sad Affleck

Member

Xyber

Member

LelouchZero

Member

napata

Member

Gestault

Member

Sad Affleck

Member

Gren

Member

Similar threads