• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Like a Dragon: Infinite Wealth appears to be well optimized on PC, doesn’t have stutters, runs with over 90fps at Native 4K/Max on NVIDIA RTX4090

Draugoth

Gold Member
Like-a-Dragon-Infinite-Wealth-new-feature-1038x576.jpg
SEGA has just released Like a Dragon: Infinite Wealth on PC. DSOGaming did preliminarly benchmarking

For these initial tests, I’ve used our main PC system to replay the Batman Arkham games. So basically, I used an AMD Ryzen 9 7950X3D, 32GB of DDR5 at 6000Mhz, and NVIDIA’s GeForce RTX 4090. I also used Windows 10 64-bit, and the GeForce 551.23 driver. Moreover, I’ve disabled the second CCD on our 7950X3

Now the good news here is that the NVIDIA GeForce RTX 4090 can push over 90fps at all times at Native 4K/Max Settings. Not only that but we did not experience any stutters while exploring the city. Yes, you won’t even experience any shader compilation stutters. After all, the game will compile its shaders the first time you launch it.
 

kruis

Exposing the sinister cartel of retailers who allow companies to pay for advertising space.
It looks like that framerate number of 90FPS on a 4090 was reached without using either DLSS or FRS, since they wrote:

What’s also interesting here is that Like a Dragon: Infinite Wealth supports both AMD FSR 3.0 and NVIDIA DLSS 3. As such, we’ll be sure to test them and share our opinion in a separate article. And yes, there is also support for Intel XeSS.

So in real world situations the framerate will be much higher than that because who'd want to leave DLSS turned off if you've got a 144+ HZ display?
 

Bojji

Member
I mean a 4090 is a literal console generation ahead with like 5x+ the raster compute of a PS5 and the PC version's presumably not adding many (if any) bells and whistles over the console versions. I should hope it can hit 4K/90 if the PS5 version runs at presumably 1440p+/60.

4090 is 3.2x more powerful than what is inside ps5.
 

Bulletbrain

Member
The Dragon Engine ranks very high in my "looks to performance" ratio. It looks good most of the time, sometimes can look great even (character rendering and night time city rendering), very scalable and runs well. Too bad RGG is thinking of giving it up.
 

Haint

Member
4090 is 3.2x more powerful than what is inside ps5.
Most 4090's full time clock to around 2800mhz out of the box, which is a genuine 45TF part in the old single FP32 instruction per clock sense. That's a 4.5x gap, without even addressing the grand canyon difference in bandwidth, the architectural gains over the aged RDNA2 in PS5, or the AI tensor cores that alone outsize the PS5s entire GPU die area.
 
Last edited:

Bojji

Member
Most 4090's full time clock to around 2800mhz out of the box, which is a genuine 45TF part in the old single FP32 instruction per clock sense. That's a 4.5x gap, without even addressing the grand canyon difference in bandwidth, the architectural gains over the aged RDNA2 in PS5, or the AI tensor cores that alone outsize the PS5s entire GPU die area.

Teraflop number is bullshit since Ampere (and RDNA3) so it doesn't really matter.

Overall performance is what I have said:

jYCogFD.jpg
 

Nvzman

Member
From the hour I've played it, its definitely a little heavier than Yakuza 7 (which makes sense, given its been 4 years), but it is still very optimized, the Steam Deck can manage it alright with some FSR in there for 40fps and even the new Intel Arc iGPU for meteor lake can do pretty decently at 1080p with XeSS set to balanced and medium settings on both machines.
 
Last edited:

Haint

Member
Teraflop number is bullshit since Ampere (and RDNA3) so it doesn't really matter.

Overall performance is what I have said:

jYCogFD.jpg

The "bullshit" teraflop number for the 4090 is actually around 100TF, which is the one that combines FP32 and INT32 calculations in a single instruction. 45TF is the old method and "real" TF number that also calculates the PS5 at 10TF.

The compute advantage is very real, the problem is it doesn't translate linearly to a 1:1 performance scale, and a large multi-game average obviously demonstrates the metric shit ton of variables as much as it does performance. A hypothetical 45TF PS6 for example will have the same problem in that you also won't see 5x higher frame rates or resolutions.
 
Last edited:

Alexios

Cores, shaders and BIOS oh my!
I mean, for ~3 months there's been a sizeable playable demo that is perfectly good like most of the other recent games in the series too, what did you think was gonna happen between then and now?
 
Last edited:

Killer8

Member
I will likely pick this up on PC this time around but will still wait for comparisons. Funny thing about Like a Dragon was that I preferred the visuals on PS5 over the PC and Xbox versions (the latter actually had an unresolved bug where clouds were missing, which I don't think was ever fixed).



The PS5's slightly more crushed contrast and different handling of shadows made it look a bit better to my eyes than the other versions, although you could probably just toggle the gamma to match it.
 
Last edited:

Bojji

Member
The "bullshit" teraflop number for the 4090 is actually around 100TF, which is the one that combines FP32 and INT32 calculations in a single instruction. 45TF is the old method and "real" TF number that also calculates the PS5 at 10TF.

The compute advantage is very real, the problem is it doesn't translate linearly to a 1:1 performance scale, and a large multi-game average obviously demonstrates the metric shit ton of variables as much as it does performance. A hypothetical 45TF PS6 for example will have the same problem in that you also won't see 5x higher frame rates or resolutions.

You are right, I knew this 45TF number looked off to me, haha.

Yep, in theory 4090 is 4.5 more powerful than PS5 but real numbers will always be less than that. Same way Xbox is not automatically 20% better than PS5 in every game
 

Silver Wattle

Gold Member
The "bullshit" teraflop number for the 4090 is actually around 100TF, which is the one that combines FP32 and INT32 calculations in a single instruction. 45TF is the old method and "real" TF number that also calculates the PS5 at 10TF.

The compute advantage is very real, the problem is it doesn't translate linearly to a 1:1 performance scale, and a large multi-game average obviously demonstrates the metric shit ton of variables as much as it does performance. A hypothetical 45TF PS6 for example will have the same problem in that you also won't see 5x higher frame rates or resolutions.
Stop using compute to determine gaming performance, it's why we bench games.
 

//DEVIL//

Member
I don't consider this a super optimized port if it's just running 90 average on a 4090. Not sure how op considers this an accomplishment lol.
 

LiquidMetal14

hide your water-based mammals
I don't consider this a super optimized port if it's just running 90 average on a 4090. Not sure how op considers this an accomplishment lol.
Considering its running at what appears to be a damn near locked 4k120 without DLSS, I'd call it good. It's not clapping and standing ovation but giving it the credit it deserves.
 

Celcius

°Temp. member
Interesting that they disabled the second CCX of the cpu instead of using the Xbox game bar thing like AMD intended lol
 
Last edited:

OverHeat

« generous god »
Interesting that they disabled the second CCX of the cpu instead of using the Xbox game bar thing like AMD intended lol
I use process lasso when game bar don’t choose the right CCX way more convenient then going in the bios and disable it.
 
Good to hear that they have put some effort into getting it to run smoothly. So many games seem to need a good 3-6 months post launch to run smoothly on PC.
 

//DEVIL//

Member
Considering its running at what appears to be a damn near locked 4k120 without DLSS, I'd call it good. It's not clapping and standing ovation but giving it the credit it deserves.
i didn't watch the video. 120 native is good. 90 ? not so much lol.

this game, is still using the same engine from the PS4 days. maybe better graphics abit which is why at least should be 120 4k native.
 

LiquidMetal14

hide your water-based mammals
i didn't watch the video. 120 native is good. 90 ? not so much lol.

this game, is still using the same engine from the PS4 days. maybe better graphics abit which is why at least should be 120 4k native.
You're right but for what it's worth, has good scalability (Steam Deck verified again). It looks good and for what it does, I'm happy they put this kind of effort vs the budget limited Ys series (which I like as well).
 

TheStam

Member
I'm happy with the graphics but they are hardly jaw-dropping. Yakuza games always have a bit of PS3 /360 era flavor to me and I find it pretty charming in a weird way. Especially the environments and objects where geometry tends to be very basic and there is no ray-tracing or other advanced features except for upscaling techniques. I was surprised when I couldn't get 4k60 fps on my old 2080ti when Like a Dragon released in 2020 and Infinite Wealth looks better and runs about the same, so it's an improvement. IW runs great at around 120 fps* on my 4080 at 3400x1440p with DLAA but I feel that is to be expected for these sort of graphics. It also runs pretty well on the Steam Deck so I'm happy, but I don't think it deserves a medal for optimization as if Dragon engine was RE engine it's more of a serviceable and (so far) bug free PC port. No problems but nothing truly amazing. Plus points for including HDR this time though.

*Got to edit this, on the second map I'm now getting around 70 fps, so the article saying it's greatly optimized is a bit off I'd say.
 
Last edited:

TheStam

Member
Anyone try it on steam deck?

Yeah it runs pretty well, definitely playable and I will probably play most of the game on the Deck as JRPGs are so perfect on it. Performance is about the same as Yakuza: Like a Dragon I would say, but it has FSR 2 + 3. I found FSR 2 looked a bit better then FSR3 but not really sure. Frame gen didn't work well for me. I play it with a mix of high and medium and locked at 45 but I get drops to the 30s. Turn based game so frame rate doesn't feel superimportant to me. Second map seems to be a bit more taxing and I just got there so might have to change it up a bit. HDR is nice for the OLED.
 

RNG

Member
Can confirm, I play it in 4K with DLAA on with a 4090 with Frame-Gen off and it runs great. For some reason frame gen causes stuttering problems whenever I try to pickup items around the world and it won't even allow me to pick up some briefcases too.
 
Last edited:

Chastten

Banned
I love how the title basically explains in one sentence everything that's wrong with PC-gaming

I would hope a game can run smoothly on the fastest and most expensive card ever produced, that shouldn't even be a question.
 
Last edited:
Top Bottom