• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

10GB vram enough for the 3080? Discuss..

Status
Not open for further replies.
I don't see how RT performance is at all related to infinity cache, infinity cache helps normal games cause they can get bandwidth starved with rdna2 cards slower vram - but RT tends to be less affected, its performance is more or less directly related to how many RT cores you stick on the card
 
I would not expect much more of an increase in 6000 series ray tracing than we have now because of infinity cache. What will make an impact is upscaling which is yet to be used on the 6000 series, then we will get a good idea how the 3000 series and 6000 series compare

It has been said that AMD has provided their current code for upscaling to some game developers. https://youtu.be/LZC3wxwMY_U?t=679

I expect the Radeon 7000 series to be much better at ray tracing than the 6000 series with better hardware support in the gpu

I can’t see AMD doing well on AI upscaling (compared to Nvidia), although by the time it’s usable and more widely available we will have the new series, likely with some dedicated hardware.
 
I don't see how RT performance is at all related to infinity cache, infinity cache helps normal games cause they can get bandwidth starved with rdna2 cards slower vram - but RT tends to be less affected, its performance is more or less directly related to how many RT cores you stick on the card

Sorry but did you read the thread

"Ray tracing basically works by having dedicated hardware perform calculations of how the light rays behave, using a technique known as bounding volume hierarchy (BVH) traversal. Performing that task is very memory-intensive, which is why VRAM demands leap up when you enable ray tracing in a game. AMD says it’s able to keep “a very high percentage of the BVH working set” directly inside the Infinity Cache, reducing latency and improving overall performance."
 
Sorry but did you read the thread

"Ray tracing basically works by having dedicated hardware perform calculations of how the light rays behave, using a technique known as bounding volume hierarchy (BVH) traversal. Performing that task is very memory-intensive, which is why VRAM demands leap up when you enable ray tracing in a game. AMD says it’s able to keep “a very high percentage of the BVH working set” directly inside the Infinity Cache, reducing latency and improving overall performance."


Interesting. Why didn't AMD give infinity cache to the consoles though, will any developers bother optimising for a piece of hardware a small handful of pc gamers have and no one else
 
Interesting. Why didn't AMD give infinity cache to the consoles though, will any developers bother optimising for a piece of hardware a small handful of pc gamers have and no one else

I am sure anyone pc gaming for any length of time will have heard of poor ports to PC of console games, lets hope that the developer tools make it easy for the infinity cache to work well. I suspect that the infinity cache was not in the PS5 to keep costs down as it takes up a significant amount of die size
 
Interesting. Why didn't AMD give infinity cache to the consoles though, will any developers bother optimising for a piece of hardware a small handful of pc gamers have and no one else

Because the consoles chips are semi-custom, while the GPUs are inside job.

Semi-custom means that Microsoft and Sony decide what they need, not AMD, AMD only gives the desired IP that's needed and insures that the chips can be made.
 
Sorry but did you read the thread

"Ray tracing basically works by having dedicated hardware perform calculations of how the light rays behave, using a technique known as bounding volume hierarchy (BVH) traversal. Performing that task is very memory-intensive, which is why VRAM demands leap up when you enable ray tracing in a game. AMD says it’s able to keep “a very high percentage of the BVH working set” directly inside the Infinity Cache, reducing latency and improving overall performance."
What AMD say in their blurb and what is actually working well in practise are two different things.
 
jays 2 pence snipped

got a feeling he will be making a follow up to that video as im sure others have proven afterburner memory measurement's arnt that accurate for whats actually being used. something to do with how games load in to gpu memory rather than how its actually used.
 
got a feeling he will be making a follow up to that video as im sure others have proven afterburner memory measurement's arnt that accurate for whats actually being used. something to do with how games load in to gpu memory rather than how its actually used.


It was a very basic quickly thrown together video from Jay, not well thought out and not actually testing vram properly
 
Care to expand on this?

from my limited understanding (and jay sort of mentions it in the video) afterburner measures the loading of stuff in to memory and then sits there rather than showing a live memory usage, and theres something about how games/drivers pre allocates how much memory it needs and if its at certain limits the software blocks stuff rather than the game running then stuttering, i think doom eternal did this not sure if there's ways around it though now. again this is from my limited understanding, think gamers nexus or digital foundry said it originally a month or so ago.
 
Here comes a question about Afterburner;

5uF4vGS.png

Now this got me confused. Which one is allocated, which one is actually used? I gather "dedicated" is actually used, but is it or is it not?

Lld0p08.png

Task manager reports Crysis Remastered uses 7.7 GB of dedicated memory. Afterburner per process dedicated memory says 5.2 GB. Afterburner also tells me a total of 7.6 of dedicated vram usage. So I gather 7.8 gb total me usage, and 0.2 gb of allocation.

Yet, there's inconsistency between task manager's dedicated Crysis stat and ab's dedicated Crysis stat? Is the 2.5 GB difference actual allocation?

Can someone clarify what is actually used and how much is allocated?
 
I would be curious to see a poll in this thread on how many have a 3080. Of those who have one, how many experience that they've maxed out the GPU in terms of VRAM.

I have one and for 1440p and the games that I play I would never need more than 10gb. That's just me though.
 
My real world experience was the 3080 was that personally never had any issues with VRAM not being enough at 4k. Not once. I sold it and got a 3090 for the frames not the VRAM as it never once worried me it wouldnt be enough.

Would it be nice to have it? Sure, more is always better (unless you are talking years in prison :D) but I genuinely think its not an issue and wont be for a while. Looking at it from the other side though, it IS cutting it a little close to the bone and while it may be enough now its not giving you enough of a buffer to be comfortable. We ARE talking ifs and buts there though at the end of the day.
 
Here comes a question about Afterburner;

Now this got me confused. Which one is allocated, which one is actually used? I gather "dedicated" is actually used, but is it or is it not?

Task manager reports Crysis Remastered uses 7.7 GB of dedicated memory. Afterburner per process dedicated memory says 5.2 GB. Afterburner also tells me a total of 7.6 of dedicated vram usage. So I gather 7.8 gb total me usage, and 0.2 gb of allocation.

Yet, there's inconsistency between task manager's dedicated Crysis stat and ab's dedicated Crysis stat? Is the 2.5 GB difference actual allocation?

Can someone clarify what is actually used and how much is allocated?

7.7 is what you have after windows has done its borg thing. The game is using 5Gb. The difference is what is not being used. That's how I interpret it.
 
Status
Not open for further replies.
Back
Top Bottom