VehiclesFashionRecipesBlogsHuntTravelsSportFunHandmadeITEducation
Mini-Games
x

x
zakruti.com » IT - Software » Gamers Nexus
Intel's Major Overhaul for CPU & GPU Benchmarking GPU Busy & Pipeline Technical Discussion

Intel's Major Overhaul for CPU & GPU Benchmarking GPU Busy & Pipeline Technical Discussion

FBTwitterReddit

video description

Rating: 4.5; Vote: 2
This video talks about new tools added to the Open Source PresentMon initiative, adding the capability to monitor a new metric called GPU Busy. In the video, we explain the rendering pipeline for frames, including discussion about game engines (e.g. Unreal Engine 5) CPU, GPU, and DirectX or API involvement in taking data and composing a frame presented to the player. This is a technical discussion with Tom Petersen, Senior Fellow (engineering) at Intel, who explains the new tools and how they can be used by end users and reviewers alike. As these are completely vendor agnostic and open source, we can apply these to reviews of all CPUs or GPUs (in Windows), and likewise the users can run PresentMon for all hardware at home. Intel hardware is not required. As background, PresentMon has already been around for many years now and is what many reviewers use for their benchmark and reviews process.
Date: 2023-08-18

Comments and reviews: 19


5:00 interesting seeing this visualized. As an Overwatch player I ve known for years that GPUs were almost never bottlenecking game performance. My friend used a 960 and still got 300+ avg fps. Game was always bottlenecked by CPU timing and seemingly more importantly ram timing. People I knew who could get 450+ avg fps did it by tuning up their ram timings to be as fast as possible. One person I knew who got 450+ fps avg swapped out his normal 2080 for one of their friends 1080 or 1060 that they were getting rid of, and there was zero difference in performance. He said he may have felt a minuscule input lag increase but that he thought it was coming from another thing he screwed up and not the card.
reply

16:28 So is this here basically proof that esport people who tune their game down to potato graphics in 1080p are actually making their game feel worse and run worse? I imagine for ideal play, you would want the ideal frametime with gpu load. If this is showing that running on potato mode at 1080p opposed to ultra is actually worse, I will laugh hysterically.
EDIT: I guess that how they say later on, having infinitly powerful CPUs and GPUs kinda nullifies this and this only really applies to mid range, low range hardware end users. Still super interesting. Awesome video

reply

This is really exciting to me. Mainly because the average PC gamer doesn't really understand the render pipeline and they don't really understand bottlenecks, so they turn to something like bottleneck calculator, and get very much misinformed. So I can see that this could be potentially used to objectively show a CPU bottleneck, or pin down what is causing an issue. Perhaps take some confusion out of it and give them a clear indicator and understanding of what's going on.
reply

Freakin sweet tech magic, i love it! Seems like an executable that you can run whilst playing a game that gathers data and presents a suggestion at the end for optimisations/things to check ( your 4K ultra hi-resolution horse anus texture for Skyrim is causing graphics card stutter ) would be brilliant for the average end-user. Love seeing that there's STILL areas people can dive into and start looking for ways to improve. Interesting stuff! :)
reply

People are sleeping on Arc's AI cores, XeSS comes close to DLSS at times, despite beinh only the first iteration of it. Not only that, but raytracing is not that far behind, even beating AMD.
Some might say that Intel should've focused on taking on AMD before going after Nvidia, but the fact that they have (Nvidia) as their target means they are aimining high.
Anyway, I still find Arc to be an interesting piece of hardware worth thinkering with.

reply

Yes more open source projects! I have dive down into OpenGPU stuff earlier (and made some fun projects using some of the AMD tools) and am happy to see Intel doing some more OS initiatives. Will definitely look in to PresentMon SDK and see what fun stuff I can make on my spare time.
If possible, perhaps in the future, I would love to see Intel be a part of OpenGPU. One could wish right? :) As companies compete together instead of against.

reply

Flip is at least in a sense of ancient system to switch between buffers. Basically the OS held multiple buffers (at least two - as double buffering), one for what is on the screen, and others for one being drawn by applications or the OS. Once the OS is ready to discard the current frame and goes to the next frame, it flips to the next buffer in the chain and put the old one to the back of the list.
I think it s also called Blit.

reply

This is really interesting, I game only in VR. I am forever changing my setting as I'm always limited by something and the fact I can deep dive a bit more to find out why would be really helpful.
I explanation about increasing the load for a better balance, finally explains something in my system I didn't understand. I reduced some settings and performance was worse. I now know why that happened

reply

This content brings up the discussion of how GPUs are presented and reviewed. I would love to see GPUs matched to appropriate hardware that better matches their use case. Do you expect to create content differently, given that you could now better match parts to work more efficiently with each other? i.e. mid-range CPUs for mid-range GPUs vs. always testing with top-spec CPUs?
Cheers
Rick

reply

Would be really great to be able to tell which part of the rending is killing the framerate. Is it the 3 million polygons character models or the 60k textures with pores on a atomic scale. Is it the shaders being too heavy or because there's a bazillion shaders to run. Would be fun to have the opposite of Nvidia meme RTX Remix to sell meme GPUs and have performance optimized remasters instead.
reply

I want y all s opinion on my build. Just curious if anyone has some criticism. List below
Ryzen 5 5800x3d
RTX 4080
MSI B550-F MPG
32Gigs 3200Mhz Corsair Vengeance 4sticks 8gb per slot
Corsair 360mm AIO.
Lian Li Lancool 2 Mesh 5 Fans 2 200mm covering the Radiator intake and 3 120mm exhaust.
850w Corsair PSU.
Monitor 175Hz 1440p Asus monitor. IPS

reply

Great news for more data - especially considering the recent minor controversy between Jeff from Craft Computing and way of the most test are run.
Main focus being - should all the consumer focused GPU tests be run on the top of the line hardware while most people do not run it.
And today we saw that even intel do not think they should run things on i7/i9 but on i5

reply

These videos with Tom have been great both because we learn cool information about how graphics work, and we get to see someone genuinely enthusiastic about the topic who is actively working to make it better and share that progress with us. If everyone who worked in these spaces was as enthusiastic as Tom and Steve, I think we'd all benefit. Great video!
reply

Sounds like this might help one to tune their graphics settings? Eg if you're getting bottlenecked by textures, you probably want to alter settings related to them (quality or perhaps anisotropic filtering?). As it is, it is sometimes difficult to tell what setting one should adjust to minimize quality loss while maximizing performance gain.
reply

Hmmm... how convenient: first tank your competition and then release a mayor product to replace them. But, hey, it was in the best interest of the community.
I'm positive that if it were been the opposite situation, he would at least give you a heads up and NOT stab your back. Cheers Steve.

reply

Amazing vid , im a 40 year old man and i feel like a schoolboy attending his only favourite class ( out of 15 ) again ... Love the feeling , and i will certainly love HUB 's reaction to this vid when they realise that their bechmarking time spending just when up about 40% .. perfect :P
reply

What would be interesting test is a comparing between the standard CPU/GPU handshake rendering process (that you showed) to the SOC approach like what Apple is doing where the GPU and CPU share the same memory (will that produce what you call the perfect world graph plot?).
reply

21:40 I love how intel is making this open source and cross platform, we need more of this. I was gonna say they're following in amd's footsteps but with the starfield fsr story that's up in the air. Keep doing it right, intel, and we'll notice and appreciate it.
reply

I don't know what scarier the fact that I enjoyed watching this, The fact that I understood it, or the fact that I'm excited about the possibilities this is the consumer. It like a irl RPG we have to find the right combo of drivers, gpu, cpu to min/max the PC
reply
Add a review, comment






Other channel videos