|OT| Horizon Zero Dawn - You Have My Bow
- Thread starter Ex-User (119)
- Start date
The patch is live, but I'm not at the right PC (or the right workplace 
) to test it right now. I'll report back when I can.
) to test it right now. I'll report back when I can.
Pre-Load is up. I repeat you can start preloading.
For real this time.
For real this time.
I was expecting a bit more at what is supposedly PS4 render settings, the original present (medium settings).Looking at some of pre-patch benches, it seems a bit hungrier than I expected.
The patch does not appear to change the behaviour regarding V-sync and fullscreen etc. (so the hints I wrote in my article are still valid), but it does improve frametime consistency for me, particularly at 4k with ultimate settings:
Pretty good reduction in frametime spikes.
Pretty good reduction in frametime spikes.
Sure, but the collectable plants are easy to identify after you know them, I turned it off permanently early on, it really bothered me. But yeah, the UI customization is great.
Unexpected to be sure. I feel like I haven't seen lengthy shader compilations outside of emulators since Splinter Cell 2, or was it 3.
CoD MW does it, although during menus. It can take a few minutes. I expect it's relatively common on PC where they can't just ship compiled shaders, or games just hitch it up.
Unexpected to be sure. I feel like I haven't seen lengthy shader compilations outside of emulators since Splinter Cell 2, or was it 3.
Yeah virtual threads still good. I still wonder at people not buying CPUs without em, although I only learned this once and for all prior to my most recent upgrade.
Me choosing 8700k over 9700k
.
To me those framerates don't seem bad at all 4k + ultimate, but is this on a 3080 Titan Ti Ultra or what?The patch does not appear to change the behaviour regarding V-sync and fullscreen etc. (so the hints I wrote in my article are still valid), but it does improve frametime consistency for me, particularly at 4k with ultimate settings:
Pretty good reduction in frametime spikes.
UPDATE - found the link!
Shader compilation is fine. Death Stranding was (properly) doing it as well, but during the initial first gameplay.
What is more of a problem are the following points:
What is more of a problem are the following points:
Sounds like the game needs more time in the oven.
2080Ti according to the article.
Didn't the Quantic Dream ports also do the same ? I remember reading that somewhere.
Unexpected to be sure. I feel like I haven't seen lengthy shader compilations outside of emulators since Splinter Cell 2, or was it 3.
Perhaps not?
I can't think of a game I've played with shader pre-caching and significant wait time prior to playing. I think maybe Doom 2016 when you switch from default OGL to Vulkan, but that was still relatively fast rather than 10-15mins. Not that this is much of a big deal as a one off thing. Would be nicer if this was just part of the download process, but I'm sure there's technical reasons for this.
I wonder how little Jim Ryan payed for this port to be created. They should have delayed it.
Everything seems to be that Kojima Productions cared more about PC than Guerrilla. I hope that's not the case, as in, I hope Guerrilla cares about the PC version of Horizon as KojiPro cared about the PC version of DS which is great version.
Everything seems to be that Kojima Productions cared more about PC than Guerrilla. I hope that's not the case, as in, I hope Guerrilla cares about the PC version of Horizon as KojiPro cared about the PC version of DS which is great version.
I'd rather test the game out myself and see how it behaves. Looks like quite varied reviews from people. I wouldn't be so sure they "cheaped out" on their port as such.
I'll refund it if the experience is not fun for me. Plenty of other games I'd taken on instead if it doesn't feel worth the price
DS is rendering a barren grassland without grass shadows... don't get me wrong it looks awesome but there is a lot more going on in HZD.
I never expected nor am I expecting the same performance in HZD as in DS to begin with.
But 50 to 65 fps on a 2080Ti at 1080p in scenarios with lots of foliage? And we aren't even talking about full ultra settings because Ambient Occlusion is sitting at medium.
The GPU benchmark scene they are using
Maybe Ultra settings are doing something that is over the top? I don't know. A good way to judge the quality of the port would be to run it on PS4Pro like hardware at PS4Pro like settings. That would be a RX570, GTX 970?
We'll see once the game is out.
I never expected nor am I expecting the same performance in HZD as in DS to begin with.
But 50 to 65 fps on a 2080Ti at 1080p in scenarios with lots of foliage? And we aren't even talking about full ultra settings because Ambient Occlusion is sitting at medium.
The GPU benchmark scene they are using
Maybe Ultra settings are doing something that is over the top? I don't know. A good way to judge the quality of the port would be to run it on PS4Pro like hardware at PS4Pro like settings. That would be a RX570, GTX 970?
We'll see once the game is out.
caughs in Forzas
Although in case of FH4 ... UWP/Xbox app managed to crash during it at least once
I think you have the right of it. Many more material permutations, many more object types and rendering techniques, more of everything.
It seems High is the best bang for your buck performance-wise and I'd wait to see DF's video on the topic before crying foul at the performance hit ^^.
I like how scalable the options seem to be but it seems like the granularity isn't really there. Not sure that there's a different slider for grass, characters, trees, objects and terrain for instance.
Last edited:
- Thread Starter
- #77
i mean who doesnt just enable AF at driver level now anyway cant remember the last time i needed a game to have the setting.
and yes forcing AF in driver works for horizon
and yes forcing AF in driver works for horizon
weeeeeeeeeeeeell ... if you do that for Forza Horizon 4, the game actually disables AF altogether for some reason
only the in-game setting actually works, lol
I feel like it's strange that some people seem to think that my results are fundamentally different from those reported by others. It might be that they are only skimming the article and not realizing that all my CPU measurements are designed to be fully CPU-limited? (as in, rendering at 50% of 1080p)
My 4k results seem to match quite well with other 2080ti results (as long as the benchmarks are done in the same circumstances -- no in-game V-sync or framelimiter, exclusive fullscreen, 16xPCIe).
My 4k results seem to match quite well with other 2080ti results (as long as the benchmarks are done in the same circumstances -- no in-game V-sync or framelimiter, exclusive fullscreen, 16xPCIe).
Some options are very coarse-grained while others allow finer adjustment. I have a screenshot of the full set in the article.
Last edited:
Can't speak for others, but: No, not really.
If there are differences it will be related to the benchmarked location. One of the reasons why I think it is important that those locations need to be shown with a video or at least with a pic like in your article.
Anyway, what I wanted to actually ask: PCGH.de is using medium Ambient Occlusion for their benchmarks: Do you remember maxed out AO being that much more demanding?
Oh no.
People will refund the game, it will be "review bombed" and people will not buy it. Not that I blame them.
Incompetence killing the Sony PC dream before it really started?
People will refund the game, it will be "review bombed" and people will not buy it. Not that I blame them.
Incompetence killing the Sony PC dream before it really started?
Maybe a review bomb will make Sony not just use PC as a "once in a while trash bin" for quick money and understand they actually need to put in some work with a PC version.
Damn, it's a shame the port seems to be a mess, the game certainly deserves better, I hope they fix it.
Durante's port report reads much more positive than some concerns I've seen... I'll just wait and see how it runs. Besides, i don't need everything maxed out. If it runs smoothly, I'm perfectly fine with mid-high settings.
For what it's worth, I thought it was well written and gave me good set of info both on CPU and GPU performance. Thank you.
- Thread Starter
- #90
my best comparison for this is odyssey, due to the amount of grass trees lod etc, which all seem better detailed than in ody too,
and heres some comparisons
horizon maxed pc
horizon original quality preset (ps4 settings)
pc max
ps4
ac ody
obviously we can see things vary between different places for horizon between ps4/pc. but even at ps4 settings the detail and qualklity is far higher than that of ody (pcmaxed)
i also want to add something none of the horizon screens can show and thats dynamic foliage for when you move through it etc, the ps4 base version only had dynamic for the tall grass parts because its gameplay required, where as the pc versions does this for all grass.
in no way am i trying to hide if the port is fully optimized etc but we all know how crap ody perfs and in comparison looks alot worse too.
and heres some comparisons
horizon maxed pc
horizon original quality preset (ps4 settings)
pc max
ps4
ac ody
obviously we can see things vary between different places for horizon between ps4/pc. but even at ps4 settings the detail and qualklity is far higher than that of ody (pcmaxed)
i also want to add something none of the horizon screens can show and thats dynamic foliage for when you move through it etc, the ps4 base version only had dynamic for the tall grass parts because its gameplay required, where as the pc versions does this for all grass.
in no way am i trying to hide if the port is fully optimized etc but we all know how crap ody perfs and in comparison looks alot worse too.
I really feel like Alex is overstating his case there.
The game is easily one of the best-looking foliage-rich open world games with full day/night cycle that exists at this point. Yes, it could probably perform better, but I also feel like it's far from the disaster he paints. Could it perform better with more optimization? Most likely, yes, but the same is true to varying extents for every game ever made. The extent is probably a bit larger in this one than some others, but hardly catastrophic.
I'd also like to make a remark on 4k performance. The PS4 Pro runs the game at a pretty decent 30 FPS (not 100%, but also not with very frequent drops) at checkerboarded 4k30. That's 1/4 of the shaded pixels needed for native 4k at 60 FPS on PC. With the "Original" settings, I measured 62 FPS average and 54 FPS in the 1% lows on a 2080ti. Now, obviously that's a very different architecture, but my 2080ti runs at clocks in this game which put it at ~15 TFLops, and a PS4 Pro has 4.2 TFlops. 15/4 is 3.75, quite a bit less than 4.2, so if it uses the 2080ti's potential hardware performance about as well as the PS4 Pro's, then it's pretty understandable that it won't hold 60 FPS at native 4k quite as well as the PS4 Pro holds 30 FPS at checkerboarded 4k.
Dictator any thoughts on that comparison (which I realize is very rough, but I still feel somewhat applicable)?
The game is easily one of the best-looking foliage-rich open world games with full day/night cycle that exists at this point. Yes, it could probably perform better, but I also feel like it's far from the disaster he paints. Could it perform better with more optimization? Most likely, yes, but the same is true to varying extents for every game ever made. The extent is probably a bit larger in this one than some others, but hardly catastrophic.
I'd also like to make a remark on 4k performance. The PS4 Pro runs the game at a pretty decent 30 FPS (not 100%, but also not with very frequent drops) at checkerboarded 4k30. That's 1/4 of the shaded pixels needed for native 4k at 60 FPS on PC. With the "Original" settings, I measured 62 FPS average and 54 FPS in the 1% lows on a 2080ti. Now, obviously that's a very different architecture, but my 2080ti runs at clocks in this game which put it at ~15 TFLops, and a PS4 Pro has 4.2 TFlops. 15/4 is 3.75, quite a bit less than 4.2, so if it uses the 2080ti's potential hardware performance about as well as the PS4 Pro's, then it's pretty understandable that it won't hold 60 FPS at native 4k quite as well as the PS4 Pro holds 30 FPS at checkerboarded 4k.
Dictator any thoughts on that comparison (which I realize is very rough, but I still feel somewhat applicable)?
I took a look and indeed he posted these strong words about the port. I do feel they are kind of justified when it comes to performance on mid-range hardware though.
That being said, if we're talking 1080p/60fps, I feel a 1060/4c8t CPU should suffice considering the base PS4 hardware. That is, unless there's a very specific kind of resource usage that was designed for the PS4 and they couldn't adapt to the PC in time. The PCIe large bandwidth usage suggests this to me.
Last edited:
I'm glad you ask me that, Mor, so, let's see, in the first place we have a giant banner prior the release of the game, quite fantastic I must say but, if you focus you can see a countdown there, oh my, how interesting, huh? let me tell you something, only 3 banners had countdowns for releases and those are (from less to more important)
-Jedi Fallen Order (the day EA announced their partnership with Valve for the comeback)
-Index pre-order (the first piece of hardware since Steam Controller completely manufactured by Valve)
-Half-Life: Alyx (the first entry of the series in 13 years and their flagship game for VR)
You can bet THIS is a big thing if there's a countdown banner.
Just Woah.
The performance definitely could be improved (and I am sure it will with patches), but it looks like I will get what I was wanting, which is 60fps at my resolution. Looking forward to the replay, got to around 90% finished on original PS4, can now finish it and try expansion for the first time.
Damn. The 'mid range' issues really bite on this title. No reason for someone with an rx580 or gtx 1060 to have a worse experience than I did with my PS4 slim that I bought for $150.
Damn that 4K > 1080p > 4K thing is ridiculous. Having seen the DF video, I have to revise my opinion. This is a really wobbly port in its current state. The engine was brilliantly designed but I don't get the feeling that the team working on the port really did a good job to provide meaningful options for PC and the performance is just a joke.
First, regarding PC-specific scalability, beside upped buffer resolutions and draw distances which respectively don't take much effort to implement, what is there beyond the physicalized foliage - which doesn't exactly stun visually? Basically my perception is that they took advantage of the foundational work GG did but didn't really do much in the way of additions nor optimization. Screwing up AO also kind of surprises me. It's not the easiest effect to plan around but countless ports have taken nvidia's implementation of GTAO and they worked just fine.
It's a far cry from what such a symbolic release should have been. There's little scalability, performance should be much higher (That 1060 perf at 1080p is not normal as shown by its AMD counterpart) and the frequent stuttering really is the worst part. It speaks to a port job that didn't really go further than an inch in the expected PC extra mile. This seems to have had very little done in the way of optimization beyond GG's original engine work. The graphical artifacts on the 1060 are unprofessional and show that they probably haven't touched the engine and rendering code much at all.
Charging 50 bucks for this kind of job on a game you can purchase at a fraction of that price on PS4 is unacceptable to me.
First, regarding PC-specific scalability, beside upped buffer resolutions and draw distances which respectively don't take much effort to implement, what is there beyond the physicalized foliage - which doesn't exactly stun visually? Basically my perception is that they took advantage of the foundational work GG did but didn't really do much in the way of additions nor optimization. Screwing up AO also kind of surprises me. It's not the easiest effect to plan around but countless ports have taken nvidia's implementation of GTAO and they worked just fine.
It's a far cry from what such a symbolic release should have been. There's little scalability, performance should be much higher (That 1060 perf at 1080p is not normal as shown by its AMD counterpart) and the frequent stuttering really is the worst part. It speaks to a port job that didn't really go further than an inch in the expected PC extra mile. This seems to have had very little done in the way of optimization beyond GG's original engine work. The graphical artifacts on the 1060 are unprofessional and show that they probably haven't touched the engine and rendering code much at all.
Charging 50 bucks for this kind of job on a game you can purchase at a fraction of that price on PS4 is unacceptable to me.