AMD Execution Thread [2023]

Status
Not open for further replies.
It's a smart move. Everyone who wants to buy this game is going to do so whether it has DLSS or not. Bethesda launches with FSR2... AMD is a happy sponsor... makes the game run better than it otherwise would... tech sites benchmark, test, and compare.. and then in the weeks following launch, DLSS2 is added and they get the name in the headlines again... and tech sites again return to test FSR2 vs DLSS2. Then after that comes FSR3 and DLSS3 support and then it happens again.

:)(y)
Or BGS is crunching itself to death given Starfield was already delayed, so upscaler-wise they did the bare minumum mandated by their partnership with AMD (or they ship FSR2 on consoles and it was a lower effort still, even).
 
Given the magnitude of the game would AMD or Nvidia really not be willing to send in their own army of engineers to implement their platform specific features and optimizations?
 
Given the magnitude of the game would AMD or Nvidia really not be willing to send in their own army of engineers to implement their platform specific features and optimizations?
if true then wouldn't we have FSR 3 at launch instead of FSR 2 ? I mean this is the biggest get for AMD and you'd think they want their best on it ?
 
Starfield is said to have almost no bugs and be one of the most polished BGS games ever. Doesn't look like they were crunching or anything.
 
Given the magnitude of the game would AMD or Nvidia really not be willing to send in their own army of engineers to implement their platform specific features and optimizations?
I believe AMD did send engineers to work on the FSR implementation based on Todd Howard video comments.
Todd also mention how AMD engineers were all in their code base helping with FSR 2 in starfield.
As for not including FSR 3 not much is known other that Bethesda is not among the listed FSR 3 partners or games purported to provide future support.
 
The 7800xt offers basically the same experience as the 6800xt which has been available at this price for over a year. There is no market for this card that wouldn’t have already bought a 6800xt. It’s hardly more power efficient at a meager 10% reduction.
I specifically waited to see what else AMD would release, and then I would get the choice between the 6800XT or whatever equivalent they would release. I'm still rocking an R9 Fury, so I had no trouble being patient for a little while longer.

I expect the 7800XT to be slightly faster than a 6800XT, although I am still a bit skeptical. RDNA3 doesn't seem to be that much faster per CU than RDNA2, and it's 60 CUs vs 72 CUs, which seems unbelievable for it to surpass the 6800XT.
But even if it performs the same or even slightly slower than a 6800XT, I'd take RDNA3 over RDNA2 in this case because of the features under the hood. It has more bandwidth, more AI stuff, and less power consumption. I don't need the USB-c on the card either.
 
Q2 2023 marketshare numbers are in, NVIDIA sets at 87%, AMD at 10% and Intel at 3%.

Total shipment increased in Q2 compared to Q1, where marketshare numbers for NVIDIA stood at 88% and AMD at 8%. So the market is rebounding.


Total-PC-dGPU-Market-Segment-_-JPR-_2.jpg
 
Last edited:
if true then wouldn't we have FSR 3 at launch instead of FSR 2 ? I mean this is the biggest get for AMD and you'd think they want their best on it ?

FSR 2 (and DLSS 2/reconstruction) at this point are rather established technologies. FSR 3 is technically still under development by AMD (and not launched until after Starfield), implementing that would be an entire different level in terms of complications. I doubt AMD's parternship is enough (paid) to delay something like Starfield to accommodate FSR 3's timeline.

Also given the history one might also wonder how frame gen decoupled from game logic may have additional complications for Starfield.

I believe AMD did send engineers to work on the FSR implementation based on Todd Howard video comments.

As expected given the partnership. But the context of that comment was really directed at the idea at the development resources needed being a justification of which technology to implement, when in all likelihood barring agreements preventing so, all IHVs would likely be rather generous in terms of committing their own resources.

This is an entire other level of discussion on this issue but I don't believe in the reality of truly neutral software/optimizations, and I feel people are really reaching when they try to justify reasons for what would be so called "unbiased" developer choices when that is fantasy given the realities of the marketplace. It's often done from a reverse justification perspective as well.

Take the Starfield situation (which I haven't waded into personally) in all likelihood your existing feelings towards the parties involved likely highly influence ones perspective on this, and really I feel most people likely then try to fit the narrative to suit that feeling.
 
Last edited:
FSR 2 (and DLSS 2/reconstruction) at this point are rather established technologies. FSR 3 is technically still under development by AMD (and not launched until after Starfield), implementing that would be an entire different level in terms of complications. I doubt AMD's parternship is enough (paid) to delay something like Starfield to accommodate FSR 3's timeline.

Also given the history one might also wonder how frame gen decoupled from game logic may have additional complications for Starfield.



As expected given the partnership. But the context of that comment was really directed at the idea at the development resources needed being a justification of which technology to implement, when in all likelihood barring agreements preventing so, all IHVs would likely be rather generous in terms of committing their own resources.

This is an entire other level of discussion on this issue but I don't believe in the reality of truly neutral software/optimizations, and I feel people are really reaching when they try to justify reasons for what would be so called "unbiased" developer choices when that is fantasy given the realities of the marketplace. It's often done from a reverse justification perspective as well.

Take the Starfield situation (which I haven't waded into personally) in all likelihood your existing feelings towards the parties involved likely highly influence ones perspective on this, and really I feel most people likely then try to fit the narrative to suit that feeling.

Would it need to be delayed. Isn't the first game with FSR 3 intergration at the end of Sept?
 
Would it need to be delayed. Isn't the first game with FSR 3 intergration at the end of Sept?
It doesn't mean (especially the few first) integrations to new engines would go quickly or smoothly enough this close to release
 
It doesn't mean (especially the few first) integrations to new engines would go quickly or smoothly enough this close to release
of course but again read what I was responding too

arandomguy said:
Given the magnitude of the game would AMD or Nvidia really not be willing to send in their own army of engineers to implement their platform specific features and optimizations?

Starfield is the biggest title AMD has to show off FSR so if they were sending their team to optimize for their hardware wouldn't FSR 3 be at the top of their priority list ? Hell even saying FSR 2 at launch with FSR 3 support coming after launch would have been the smart move.
 
Starfield is said to have almost no bugs and be one of the most polished BGS games ever. Doesn't look like they were crunching or anything.

Uh, what do you think they were doing the entire past year when the game was delayed due to it not being in a shape suitable for releasing to the public? If anything, the fact it is relatively bug free (possibly, I'll wait til it's out in the wild before giving Bethesda a pass on relative bugginess) shows just how much crunch went on in the past year.

Regards,
SB
 
Would it need to be delayed. Isn't the first game with FSR 3 intergration at the end of Sept?
Starfield is Sept 1st.

The amount of engineering resources is also likely magnitudes different as FSR 3 is not ready technology. Committing to it is also another level of risk due to that.

Also I have to add again given historically how game logic works for BGS game there could be other issued with frame gen.

The other thing is BGS isn't really ever know to push graphics technology but really world scope in its games. They're also seem more on the "mercenary" side with these partnerships. In all likelihood they aren't going to be a suitable partner for something like launching FSR3.

And let's face it the primary goal for Starfield from a cross marketing stand point is as a Xbox showcase.
 
Last edited:
Starfield is said to have almost no bugs and be one of the most polished BGS games ever. Doesn't look like they were crunching or anything.
What they actually said was 'least buggy Bethesda release', not that it'll have little or no bugs. Cuz it absolutely will have plenty of bugs still. Not because they're just a terrible developer or 'lazy', just the nature of the games they make.

Also, none of this tells us anything about working conditions there. I suspect there's very, very few major AAA studios that dont crunch to some degree, at least within the last 6-12 months of development.
 
Starfield is Sept 1st.

The amount of engineering resources is also likely magnitudes different as FSR 3 is not ready technology. Committing to it is also another level of risk due to that.

Also I have to add again given historically how game logic works for BGS game there could be other issued with frame gen.

The other thing is BGS isn't really ever know to push graphics technology but really world scope in its games. They're also seem more on the "mercenary" side with these partnerships. In all likelihood they aren't going to be a suitable partner for something like launching FSR3.

And let's face it the primary goal for Starfield from a cross marketing stand point is as a Xbox showcase.

Right but they just announced a bunch of games that are getting FSR 3 support and none of them are Starfield. That is my point.
 
Right but they just announced a bunch of games that are getting FSR 3 support and none of them are Starfield. That is my point.

I'm kind of lost in the trail of this conversation here.

I think you're initial response was to my comment that AMD (or Nvidia) likely would provide ample resources to implement FSR2 (and etc.) as opposed to BGS having to rely on inhouse development resources given the scope of the game sponsorship or not. Especially given that if one were already implemented I just don't think it's a likelihood of internal resources being the driving factor of implementing the other. If we want to go into the specifics are we really going to pretend that Nvidia would not be willing to put their own engineers into implementing DLSS post launch if BGS can't spare their own? Well of course that's only assuming BGS allows them to.

I'm not seeing how FSR 3 plays into this? FSR 3 itself is not ready, it's not a simple matter of AMD committing to help BGS implement FSR 3 when they themselves do not even have FSR 3 ready. Notice all those FSR 3 game are either already released or far out from release.

Also some studios have closer partnerships as opposed to just one off deals. I don't think BGS has any long standing relationship wiht either IHV, as mentioned they feel more to me like essentially a "mercenary" in these types of deals. Whomever pays plays. They aren't reliant on graphics technology marketing to sell their games either. FSR3 isn't going to make people (well a significant amount of) more interested in Starfield.
 
Yeah I assumed FSR 3 was more of a timing issue with Starfield. AMD likely would have wanted Starfield to be the big seller for FSR 3 but it wasn't ready 6-12 months ago or whenever BGS started implementing FSR2.
 
Yes, beside double issue, RDNA3 has doubled register file and LDS size, iirc. That's raising a standard which was set with PS4 and has not changed since that (idk about NV).
That's a big optimization potential - bad occupancy due to register pressure and being tight on LDS should be gone, so we can crank things up.
I can increase the quality of my GI for example, something that just more cores does not really allow. I need more LDS.

There is an 2x increase in L0/L1 cache size (32/256 KB), and 1.5x increase in L2 cache size (6 MB); the bandwidth is also higher due to 1.5x wider links.

L3 cache, i.e. "Infinity Cache", is smaller for higher-end parts (64/80/96 MB) but larger for the low-end Navi 33 (48 MB), and L3 cache bandwidth is 1.5-2x higher (up to 3.5 TB/s).

Navi 31/32 also have 1.5x more VGPRs, 1536 bytes or 384 FP32 registers per 'wave' (thread), but Navi 33 stays at 1024 Bytes or 256 FP32 registers, as shown by the code commit to the AMDGPU LLVM frontend.

LDS size did not change though, it's the same 128 KB per WGP (or 64 KB per CU) as in RDNA1, but latency has been dramatically improved. Curiously GDS got shrunk to 4 KB.


Then of course there are supercalar dual-issue ALUs capable of processing FP32 and FP32/Int8 register / immediate operands in one clock, using either V_DUAL instructions in VOPD encoding for Wave32 threads, or Wave64 threads which would transparently dual-issue all VOPD instructions as proved by Chips&Cheese benchmarks below (but bizarrely Wave64 mode is not supported in HIP/ROCm due to hardware limitations, and there are further limits such as source VGPRs from different banks and cache ports, even/odd register numbers etc.).

In practice Wave32 is the prevalent mode, so dual-issue relies on compilers/optimizers which do a very poor job of automatically scheduling parallel multiple-operand instructions (Intel and HP will testify to the spectacular failure of the much touted Explicitly Parallel Instruction Computing (EPIC) architecture for the Itanium (IA64) processors, which relied on exactly the same kind of compiler optimizations in order to extract 6-way parallelism from its VLIW instruction bundles).


AMD cites "17.4% architectural improvement clock-for-clock", so with a 8% frequency increase that should have resulted in a 25% higher performance, but so far such improvements mostly manifest in compute-limited tasks, like heavy levels of raytracing.

2nzd2VPZ7VyshqC32tXE6c.jpg



RDNA3 doesn't seem to be that much faster per CU than RDNA2, and it's 60 CUs vs 72 CUs, which seems unbelievable for it to surpass the 6800XT.
RDNA3 has a superscalar "dual-issue" scheduler (or argualby twice the number of ALUs in a CU/WGP), higher VGPR counts and higher clocks - so in theory this should compensate for less CUs, but in practice, only manifests in raytracing-heavy titles and synthetic benchmarks.


But even if it performs the same or even slightly slower than a 6800XT, I'd take RDNA3 over RDNA2 in this case because of the features under the hood.
It's equally important that the Navi 32 package is only 346 mm2, with the main GCD taking 200 mm2 and the rest taken by four MCDs, which are shared with Navi 31 - that's comparing to monolithic 520 mm2 die in Navi21, which only sold for $650 in the defective bin of 72 CUs, and originally was $999 in the defect-free bin of 80 CUs for the RX 6900 XT.

So RX 7800 XT should be far more proftitable to AMD and partners, and they would be able to offer additional incentives in the tune of at least $25-30 by this Christmas season, and probably even more by next Summer.
 
Last edited:
Right but they just announced a bunch of games that are getting FSR 3 support and none of them are Starfield. That is my point.
Not only is Starfield missing among the games getting future FSR 3 support but also Bethesda is not listed as one supporting future FSR 3 partners. Maybe it's contract related and implementation might require a revised (new) contract as opposed to just implementing FSR 3 at a later date?
 
Last edited by a moderator:
Status
Not open for further replies.
Back
Top