Xbox One X's Project Cars 2 running 4K (update)

I know you didn't. I was explaining the point of mentioning my lowest framerates under extreme load in comparison to GS.

Again, my videos disproves your assertion of the GTX 1070 sitting around 52fps with MSAA disabled.

I linked benchmarks from 2 different review sites which do not support your claim of GTX 1070 being steady 4k/60fps performer. 46fps and 52fps avg, respectively.

Well take a look at my video. It performs way better than that.

@quadknight said:

? http://gamingbolt.com/project-cars-2-wont-run-at-4k-or-60fps-on-xbox-one-x

So much for equalling a GTX 1070 lol.

?????

Its possible to have the 1070 performance GPU and also be bottlenecked by CPU in CPU intensive games..........like Racing games.

The CPU would bring the frame rate down significantly since it cant feed 60 frames to the GPU.

Your post confuses me.

Not only that but those bench's were taken when the game shortly after release, the game has had time to mature and patched and driver updates to iron out some of the issues. Also Ive noticed in ron's latest video example with 1070 staying below 60fps look at the gpu usage..... when it in the 70's to mid 80's fps was 60 or below when gpu usage went about 90% framerate went above 60 fps. This clearly shows the game in general does not full utilize cpu nor gpu at all times..... Every video Ive seen that shows 1070 staying above 60 fps, usage was high.

I took a look at my GPU settings in the Nvidia control panel. I have some settings enabled that are disabled by default that I turned on when playing ME:A. Fast Vsync, MFAA enabled (improves MSAA performance), and threaded optimization is on, not auto. It's possible that these video share similar settings that allow them to run the game much better and others probably don't. I'm going to play around with this a little later.

Racing games shouldn't be CPU intensive..... when you only throw vast majority of the work loads onto one or two threads, is just poor work. PS4 and X1 got special treatment with the original Project cards vs pc version when it first released.

Also forcing the game to run at high resolutions shows down the rate that the cpu has to send frame data to gpu. Which should lower cpu usage.

I dont follow your points so ill respond to the parts that i understand.

Racing games ARE CPU intensive. The only genre that is probably more CPU intensive is probably RTS.

Racing games can afford to look pretty because everything from the physics simulation, ai of the cars, streaming the track, netcode if online. The CPU has to do all of that every frame and also handle draw calls. Very rarely is a racing game GPU bound unless its something like split second where you have tons of explosions and effects and background objects loaded into memory.

Also, increasing resolution is a GPU intensive action, not CPU.

Running at a higher resolution doesnt automatically lower CPU usage (if thats what youre saying, im not sure)

@Zero_epyon:

If you get a chance run fraps benchmark for min, max, avg with 4K/Ultra/MSAA-enabled/SMAA-ultra on dry track full grid, then wet track full grid.

Then disable cores on your i7 so you simulate an i3. Re-run the same benchmarks.

---

Thanks for a being a good sport. I have only a RX 480 and GTX 1060, so I'd need you to do the tests for me whenever you get a chance.

Most racing games are not cpu intensive, in the regard that all the cpu resources or threads are being used. Most racing games only use one or two threads fully and leave the rest underutilized. Only reason why racing games in general are cpu demanding is because of lack of proper multithreading and load balancing for cpu resources. the same problem with most RTS games they put most of the load onto one or two threads. Most online FPS shooters with 40+ players are more cpu demanding than racing games.

Actually cpu and gpu usage are directly tied to each other even with racing games. Lower the resolution/settings the less frame data gpu has to process which means gpu gets the each frame job done faster. This causes the gpu to ask for more data from the cpu at a faster rate which increases cpu load and if the has to wait for the cpu to send data, this one of main reasons why we see gpu usage drops and frame rates drop because cpu isn't feeding gpu fast enough. Now when high resolution/settings are used it takes the gpu longer to get done with each frame lowering the rate the cpu has to send gpu data. That lowers cpu usage and forces most of the bottleneck to the gpu and not the cpu at lower resolutions.

They are not repeated. Each video has different AA settings and some has different tracks and different time of day.

Zero_epyon's track, AA settings and car selection yields superiority for MSI Gaming 980 Ti with MSAA enabled over GTX 1070.

GTX 1070 has less robust 4K 60 fps result than my GTX 1080 Ti.

My saying regularly hitting didn't mean to say occasionally hitting. I'm saying more often than not, the frame rate is 60 or higher. There are multiple videos in this thread that appear to have a 60 or higher average on dry track, and there isn't a need to drop "a whole host of other things" like the developer said would be needed for the X1X.

In terms of pure horsepower, I'd put the GPU of the X1X with the RX580, but with the customization and a bit of optimization, I expect better than RX580 results, just not to the level of the GTX 1070. (assuming it's not a hack job of a port)

what an uninteresting story

RX-580's horsepower is gimped by memory bandwidth.

Dry track's frame rate result can be influenced by the car selection with large transparency rear windows, track's object density, time of day, enabling "real weather" feature instead of static weather and AA solution. LiveTrack relates to "real weather" and wet track features.

So, in other words, a PC with a GTX 980ti also outperforms the X1X.

And NVIDIA was correct to claim a GTX 1080 because that is what they used and they made no other claim.

Underclocking your GTX 1080ti will diminish its performance. If you underclock it enough, it could be outperformed by a GTX 1070. Changing types and levels of anti-aliasing can affect the frame rate for both wet and dry track races. If you hold a normal ice cube directly against the GPU chip, it will cool the GPU, but most board manufacturers recommend against doing this.

1. Borderline. My arguments are NOT about X1X beating XYZ, but for GPU group placement i.e. RX-480/GTX 1060 group or the next GTX 1070/GTX 980 Ti/Fury X group. My 980 Ti is not reference i.e. it's 7.7 TFLOPS factory settings against 6 TFLOPS reference.

2. I don't see reference GTX 1070 being a robust 4K 60 fps card.

1. Then you should probably mention the point you're trying to make, particularly in a thread that's not specifically about that point. But I can save you time. For legacy titles, it will change depending on the work load and game/driver updates as to whether the GTX 980ti or the GTX 1070 have the upper hand. For future titles, the GTX 1070 will have a growing advantage, for the most part, with the occasional exception. Both are above the RX 480/GTX 1060 tier.

2. That has no bearing on what NVIDIA claimed.

Hahahahaaaaaaaaaaaaaaaa.

You are getting owned on this thread and you simply can't grasp it..

The xbox one X would have to give up many things to be able to run the game at 4k 60FPS and even then is not a warranty 60FPS.

You make a false claim and you didn't back it up,Nvidia doesn't say there ''YOU NEED A 1080 FOR 4K 60FPS'' YOU DID.

So either back it up or admit that you invented that lemming.

@tormentosasylumni

Project Cars 2 with GTX 1080's 4K 60 fps example. This is not GTX 1080 Ti.

http://wccftech.com/project-cars-2-first-impressions-strong/

Meanwhile, the graphics of Project CARS 2 already look amazing. Running on an i7 6700K and Palit GTX 1080 Super JetStream, the game ran smoothly (even more than seen in the captured footage below) with max settings and native 4K resolution

Your post lacks info. No frame counter, GPU usage and whether the frame-rate is capped at 60 or unlocked. You may also want to include the point you're trying to make.