Not true.For the case of Uncharted 2, the FMV cutscenes would have to be compressed further
Assembly programming is a religion now?Sounds like religion lol.
UC2 had tons of physics... what are you gonna do with them? XBOX 360 didn't have 150 Gflops of compute power. It has nothing to do with DVD vs Blu-Ray.Witcher 3 was ported to the Switch. It’s a matter of optimization.
PS4 is not a regular PC architecture, it uses a customized GCN GPU that tries to replicate Cell SPUs:First parties really pushed that hw back then, but since Uncharted Collection runs on PS4, Cell magic clearly can be ported to sorta regular PC architecture
1) Ridiculous statement. Source?1) I mean even Titanfall runs better on X360 than on XB1!
2) The developers who really compared both (at the end of the gen when all platform specific tricks were used) found out X360 was quite more powerful.
Witcher 3 is a physics-heavy game?Did you see Witcher 3 on the Switch?!
So yeah, Uncharted 2 on the 360 was fucking possible
Golden Abyss is even worse than UC1 on PS3. It used pre-rendered backgrounds in real-time gameplay.Would love to know if Uncharted on the Vita is using the same engine as the PS3. Is this known?
So it is the Uncharted engine? I always thought they just took the Syphon filter engine or smth similar and tweaked it.Golden Abyss is even worse than UC1 on PS3. It used pre-rendered backgrounds in real-time gameplay.
Vita is definitely not a portable PS3. Better than the PS2, worse than the PS3 in terms of compute.
It's a custom engine made by Bend, but they also had some help from ND:So it is the Uncharted engine? I always thought they just took the Syphon filter engine or smth similar and tweaked it.
Of course the Vita’s not a PS3.
1h and 8 min read. You just saved my evening. Thanks!It's a custom engine made by Bend, but they also had some help from ND:
Postmortem: Sony Bend Studio's Uncharted: Golden Abyss
In this incredibly detailed postmortem, Sony Bend dives deeply into the the development of Uncharted: Golden Abyss for the PlayStation Vita, explaining precisely what roadblocks held back the system's premier launch title.www.gamedeveloper.com
I membaIf they could get MSG4 on 360 I'm sure they could find a way here if they wanted. Multiple discs and an install.
2) Quite the opposite.
Late 7th gen multiplatforms run better on PS3 with better graphics settings.
You posted the cinematic at 10fps less. The in-game graphics are working much harder on the PS3 with more geometry and shader precision, as was true of pretty much every multi-plat without a "better or equal clause" that xbox relied on all gen to feed DF negative PS3 coverage, but naturally a cinematic favours more memory in a single location, so there's that. But MGS5 isn't a patch on the fx in MGS4, anyway.That's not necessarily true.
I think you're referring to GTA V here primarily. Other than a ground texture blur difference, the texture quality, resolution, performance etc is the same. The only real major example that was clearly better on PS3 was FF13 where it ran at a higher resolution and better performance, but that was mostly pulled back and they became identical with XIII-2 and Lightning Returns. There's other examples like Portal 2 where you can say the PS3 might be better because of MLAA vs 360's older AA solution, but other than that they had the same resolution and performance.
But there were other big late-gen games that ran notably better on 360 as well. Phantom Pain (already posted on the last page) ran long stretches of 10~ fps faster on 360 and Crysis 3 ran not only a higher resolution, but also higher average performance.
That is a very selective take on the features. The Xenos was 8bit(24bit) only when the RSX was 10bit(30bit) and did HDR as mentioned in the Drake's fortune making of. The Xenos was hard limited to 500-600Mpolys/s like all ATI GPUs of the time, but could render optimally regardless of data supplied or polygon ordering, where as the RSX could reach 1.1Billion polys/s but struggled with poorly supplied data, and needed optimal quad meshes ordered to optimise self culling...
on paper Cell Processor is crazy powerfull but at same time Xbox 360 also has better gpu.
..
Pretty sure this is in reference to the Async Compute Engines in the PS4, which the PS4 had 8 of, GCN 1.0 didn't have these but they were added in GCN 2.0.PS4 is not a regular PC architecture, it uses a customized GCN GPU that tries to replicate Cell SPUs:
You posted the cinematic at 10fps less. The in-game graphics are working much harder on the PS3 with more geometry and shader precision
A better example is the absence of the cutting edge PS3 light probe gathering system used in Split Second missing from the 360 - which became a staple technique from then on in the industry for Indirect lighting - and the game tearing with soft v-sync, and being sub-HD, and being less that 30fps frequently on 360, despite them being huge fans of the 360 from their article praise, and 360 render farm they had, and the Sony/PS3 criticisms their studio tech lead wrote in develop magazine throughout that gen, but still managed native 720p with solid frame-rate, hard vsync, and cutting edge lighting feature on the SPUs, because reality was the PS3 in the hands of skilled developers was significantly more powerful, like all the first party games showcased.
The particle buffer has also been cut down. Both versions have low-resolution smoke, but the PS3's alpha buffer seems to be running with a pared-down horizontal resolution, leading to the occasional odd-looking artefact you don't see on 360.
For example, the Xbox 360 version features a higher level of post-processing in environments at further distances compared to the PS3 game.
Engine stress can see tearing on 360 but very rarely do we see any kind of drop in overall frame-rate. PS3 has more tearing in similar scenarios and can be prone to dropping frames.
the 360 game is 1280x672 (as confirmed in the DF tech interview), up against native 720p on PS3.
as was true of pretty much every multi-plat without a "better or equal clause" that xbox relied on all gen to feed DF negative PS3 coverage, but naturally a cinematic favours more memory in a single location, so there's that. But MGS5 isn't a patch on the fx in MGS4, anyway.
There are also misleading figures flying around as to max theoretical compute ceilings, apparently RSX operated at different clock speeds for pixel and vertex pipelines. Pixel side is 211 GFlops/4.4 Gpixels at 550Mhz and Vertex one is 40 GFlops at 500Mhz, so 251 GFlops in total. X360 Xenos is at 240 Glops/4 Gpixels (unified).That is a very selective take on the features. The Xenos was 8bit(24bit) only when the RSX was 10bit(30bit) and did HDR as mentioned in the Drake's fortune making of. The Xenos was hard limited to 500-600Mpolys/s like all ATI GPUs of the time, but could render optimally regardless of data supplied or polygon ordering, where as the RSX could reach 1.1Billion polys/s but struggled with poorly supplied data, and needed optimal quad meshes ordered to optimise self culling.
Equally, for all the Xenos Esram performance for fill-rate, the size of the memory meant it lost lots of efficiency beyond its intended 1024x768 frame-buffer size, causing real performance efficiency problems for double buffering native 1280x720 at full colour and z precision - impacting fog equations - or with HDR, or native with stereoscopic 3D, when they eventually revised 360 hardware with hdmi 1.2, which were all easy for the RSX, which excelled even further when triple buffering.
And then there's the issue of the Xenos not being able to do true sRGB gamma correction, even when eating up shader cycles to mimic it, which was a free ASIC feature on the RSX.
The Xenos had some brute force strengths but was old hat because of the missing features and lack of ability to properly support 720p.
It's not just GTA V.That's not necessarily true.
I think you're referring to GTA V here primarily. Other than a ground texture blur difference, the texture quality, resolution, performance etc is the same. The only real major example that was clearly better on PS3 was FF13 where it ran at a higher resolution and better performance, but that was mostly pulled back and they became identical with XIII-2 and Lightning Returns. There's other examples like Portal 2 where you can say the PS3 might be better because of MLAA vs 360's older AA solution, but other than that they had the same resolution and performance.
With the console versions of Assassin's Creed Rogue, there's a definite sense that the conversion work across both platforms isn't as closely matched as 2013's Black Flag. The Xbox 360 version is softer and noticeably fuzzier than the PS3 release: while both versions utilise a form of FXAA that attempts to mimic traditional multi-sampling style coverage across edges (but considerably blurring the image in the process), the PS3 version renders natively at 720p whereas a sub-HD resolution is in place on the Microsoft console. Pixel counting puts the ballpark native resolution on the 360 at around 1200x688.
Beyond the framebuffer set-up, we find the core art and most of the effects work is interchangeable between PS3 and 360, although there are some unexpected differences between the two platforms that were not present in Black Flag. For one, SSAO is present on PS3, helping to add depth to characters and the environment, while on Xbox 360 the effect is completely absent, lending more brightly lit scenes a generally flatter appearance. Secondly, in most cases we find that streaming is generally slower on 360, with low resolution textures left on-screen (sometimes without normal maps) for a few seconds during changes in camera angles in some cut-scenes, and when transitioning to gameplay. The situation is much improved on PS3, where the majority of the best quality assets are usually loaded in before the scene begins.
The primary focus of this patch, then, was clearly performance. The patched version now runs very close to its target 60fps in even the most demanding scenes, with v-sync engaged at all times. This level of performance exceeds both the original HD release and the original PlayStation 2 version, though cut-scene performance sees less of an increase, with the original version occasionally besting the updated release (at the expense of visual quality). One of the key changes implemented by HexaDrive to solve the performance problem is the utilisation of the PlayStation 3's SPUs to handle the workload previously designed for PlayStation 2's vector units. ZOE 2 pushed the PlayStation 2 to its limits and the RSX simply doesn't have the muscle on its own to power through it without help. HexaDrive balanced performance across the entire system, including the SPUs, in order to avoid bottlenecks, resulting in a 10x increase in performance across the board.
It's not just GTA V.
Far Cry 3:
Assassin's Creed Rogue:
I remember saying three years ago that we wanted to create something revolutionary, but in reality we couldn't really do that because of the CPU. We're using the Cell engine to its limit., actually. Please don't get me wrong, I'm not criticizing the PS3 machine, it's just that we weren't really aware of what the full-spec PS3 offered - we were creating something we couldn't entirely see.
adamsapple
Without us having these discussions per game at the time of these games releasing with non-DF sources, I'm not sure what I'm looking at, now, or would be arguing over. I did own a 360 at the time of the Split Second releasing and it felt like a significant downgrade in the demo compared to the PS3 AFAIR, even ignoring the lighting technique.
Although it is interesting you didn't comment on the light probe gathering technique on the SPUs which was very significant and way beyond what the 360 could do, and is a bit of alarm bells when it couldn't even manage 720p for the game either without.
It makes the lighting significantly more coherent - ie the game looks better with it, hence why the technique was reworked for general indirect lighting in games since.....
Because there's nothing tom comment about ?
The picture was just so you'd know where to look in the video. At Full HD or above it is noticeable and indicative of either lower resolution being upscaled more, or lower precision shader work.The 4 way picture you've shown shows no perceptible difference between the PS3 and 360 versions. Only anecdotal ones at best.
If you were someone that knew a lot more than DF, you wouldn't think that.And later you've posted a comparison from completely different places and times of day, which doesn't really show anything.
The failure to compare like for like in that moment is l on Konami - probably as it doesn't show the game favourable in this marketing video - but the day time shot is more forgiving of lower frequency in a quick wipe, and only exposes the short range of shader fx running on the 360 version easily when viewing as a still because of the forward casting shadows. But half way up the image in the background the dynamic lighting/shaders are all disabled and we're just looking at baked textures, and the graphical composition workout is harder on the PS3 shot, anyway so both images seem representative of what Konami think each system can manage well with lighting and shaders to use for marketing.
So rather than acknowledge that the PS3 version looked exceptional in the shots Konami chose and the 360 version doesn't look so great compared to even other games, like say the fidelity in MGS4 that looks exceptional, it is just the 360 being unfairly treated, yes?Ok, so there's no differences visible in DF's captures or contents.
There's no difference and it's Konami's marketing fault for not showing it.
But you know there's a difference, which you've yet to highlight in any direct comparison.
Got it.
Although this is from the original uncharted, and this was on the disc or a spare disc that came with the original PS3 game IIRC, and most of it still applies. I've timestamped the video at the part that is most relevant to the technical discussion - in regards of the 360's GPU and what its version would have lost - but the whole video is still a good watch IMO.
So rather than acknowledge that the PS3 version looked exceptional in the shots Konami chose and the 360 version doesn't look so great compared to even other games, like say the fidelity in MGS4 that looks exceptional, it is just the 360 being unfairly treated, yes?
This thread is about Uncharted 2, and as shown in the making video of Uncharted 1, it has fx that are miles better than that 360 MGS5 marketing footage, but feel free to drink the DF cool aid that the 360 was on par or better than the PS3 hardware, and had results to rival PS3 first party games or exclusives.
You posted images that aren't even native resolution for the games on PS3 and 360, so could be mislabelled Xbox1 for all I know. I choose footage from a reliable source wanting to sell all copies on all platforms and their choosen footage shows what it shows. Or are you saying Konami's chosen 360 clip isn't fair to say it is representative of what the hardware can do? Or are you saying they didn't choose something that looks as good as it could - so as not to sell 360 copies?Why would I acknowledge something that hasn't been presented? You chose screens from two completely different locations and different times. I posted picture(s) from the exact same places which show no difference.
I want you to SHOW me the differences, don't TELL me about it.
Show, don't tell.
You're more focused on "teh DF bias" than an actual discussion.
You posted images that aren't even native resolution for the games on PS3 and 360, so could be mislabelled Xbox1 for all I know. I choose footage from a reliable source wanting to sell all copies on all platforms and their choosen footage shows what it shows. Or are you saying Konami's chosen 360 clip isn't fair to say it is representative of what the hardware can do? Or are you saying they didn't choose something that looks as good as it could - so a not to sell 360 copies?
I didn't say you did, I just implied that after all these years - unless it was an archive.org link from the day they posted it - the reliability of the info isn't assured, and that's even assuming I thought Rich wasn't in it for clicks and actual cares about faceoffs beyond his business interest/popularity.sigh ...
1. The images are from the Digital Foundry article.
2. Why would I go through the trouble trying to pass XBO screens as X360? Are you saying that because before today you didn't actually know what the X360 version of the game looked like?
3. The 'chosen' Konami clip shows two completely different locations and times of day, you cannot draw a direct 1 : 1 comparison with them because they're not from the same places
You are the one talking about some kind of perceived light probe and geometrical improvements on PS3. Find me some relevant comparisons that show them, don't post one screen from the day time in one place and night time in another and say "see".
Also, the images you're posting aren't even from Phantom Pain, they're from Ground Zeroes.
Please reply to me when you have actual comparisons that show the differences you're talking about, or don't quote me.
Pal Engstad works at Apple as a senior software architect:This is one of the videos I was thinking of, it was a different era... Do any of those devs even work at naughty dog anymore? or did they leave after druckman took over?
We also don't see deep dives like this into new tech, it was such a common thing in the 360/ps3 gen. So much was changing. Look at early games from that gen compared to 2013 games. Way different in so many ways.
Great believer in the slogan: Keep it Simple and Fast.
Specialties: Assembly programming, programming, optimization, graphics engines, computer graphics, animation systems, compression, software engineering, management and hiring.
If Gears of War 3 fully utilizes 3 PPU cores (and I don't mean floating point operations by that, but more general purpose code), then there's no way the PS3 will be able to run it at acceptable framerates. And that's fine by me.That's just like me saying that PS3 wouldn't be able to handle Gears 3 because that one dev-build video that showed Gears 3 running on the PS3 runs like shit. Of course additional optimizations and dedicated work would iron out things like that, just like how if Naughty Dogs were a multi-platform developer making Uncharted 2, they would use the 360's strengths to get the same, or a very close version, of the game on that hardware.
Not true.
ND used MPEG-2 for cutscenes on PS3, not MPEG-4 (H.264).
They had ample of space on BD-ROM, but it's not like they had unlimited amounts of compute for more elaborate decompression (H.264) or dedicated video co-processors.
Assembly programming is a religion now?
Fine, I'll take it!
UC2 had tons of physics... what are you gonna do with them? XBOX 360 didn't have 150 Gflops of compute power. It has nothing to do with DVD vs Blu-Ray.
UC3 even had a real-time ocean sequence with real-time physics that was actually playable, not a cutscene:
On a PC you'd need a GTX 280 for something remotely equivalent:
Cell with a GeForce 8 GPU would have been nuts!
PS4 is not a regular PC architecture, it uses a customized GCN GPU that tries to replicate Cell SPUs:
Not to mention PS4 is like 5-10 times more powerful than the PS3... maybe XBOX ONE could run UC2-3 with no cutbacks on physics, but definitely not the 360.
1) Ridiculous statement. Source?
XB1 is far more powerful than the X360. Just because it failed commercially, it doesn't mean it's weaker. That's why it supports 360 BC easily.
2) Quite the opposite.
Late 7th gen multiplatforms run better on PS3 with better graphics settings.
Witcher 3 is a physics-heavy game?
Golden Abyss is even worse than UC1 on PS3. It used pre-rendered backgrounds in real-time gameplay.
Vita is definitely not a portable PS3. Better than the PS2, worse than the PS3 in terms of compute.
Forums just fight for them now.Let’s go back in time, shall we back when developers and publishers were console warring.
I remembered this article stating that Uncharted 2 would be impossible to port on Xbox 360 due to many technical factors.
Sourcel
How true are these statements?
Do you miss the old days where developers were taking shots at one another?
Like this example from Sucker Punch:
Was that infamous 2 or the first one? I can't believe they put that in there, its hilarious, Ring ring electronics , talk about a dig on the competition but subtle. Masterfully done.Let’s go back in time, shall we back when developers and publishers were console warring.
I remembered this article stating that Uncharted 2 would be impossible to port on Xbox 360 due to many technical factors.
Sourcel
How true are these statements?
Do you miss the old days where developers were taking shots at one another?
Like this example from Sucker Punch:
I agree with the premise of that, but putting aside the difficulty of finding a single player gaming workload that would map perfectly to those 2 extra 2way cores, I'm still not entirely sure that it couldn't be mapped to 4 or 5 SPUs, where 1 of the 4 or 5 was a branch predictor and prefetcher to feed the others and get the same throughput......
If Gears of War 3 fully utilizes 3 PPU cores (and I don't mean floating point operations by that, but more general purpose code), then there's no way the PS3 will be able to run it at acceptable framerates. And that's fine by me.
Agreed.I say bring back the glorious PS360 era! Less crunch, less wokeness, more games, more profitability, more bravado, less drama, more exclusives on both platforms... what's not to love from that era? Gaming is no longer the same these days.
1) $300 loss, but yeah, I agree.1) Sony was losing over $200 on every PS3 console. So something like an 8800GTX in the PS3 was definitely out of the question.
2) However, I think that Sony has gimped the RSX too much (2x less memory bandwidth compared to the 7800GTX) making it more comparable to the 7600GT. PC games that run like a dream on the 7800GTX run with serious problems on the PS3.
3) If I remember correctly, Crysis 1 had some nice water physics, and that game ran quite well on my 8800Ultra at maxed out settings, so I don't think 280GTX would be needed to achieve similar effects as on PS3. I bought the PS3 and PC (Q6600 + 8800Ultra) in the same month and the gap between my PC and PS3 was just insance. Games that run at 20-30fps on the PS3 run at 50-60fps at 1440p on my 1440p.
I find it crazy Banjo Kazooie Nuts & Bolts was impossible on PS3 due to the physics.
i miss uncharted
after working in games for a while alot of programmers go into software programing. it less stressful/easier better quality of life (depending on the software) . its like nba players becoming sports commentaters.Pal Engstad works at Apple as a senior software architect:
Less crunch, higher pay... why work at ND anymore? His resume is filled up.Pål-Kristian Engstad - Senior Software Architect - Apple | LinkedIn
A veteran in Computer Graphics and Game Programming · Great believer in the slogan: Keep it Simple and Fast.<br><br>Specialties: Assembly programming, programming, optimization, graphics engines, computer graphics, animation systems, compression, software engineering, management and hiring. ·...www.linkedin.com
I feel like this is a dying breed of old school Demoscene programmers... at best we can hope one day the AI will fill their niche.
If Gears of War 3 fully utilizes 3 PPU cores (and I don't mean floating point operations by that, but more general purpose code), then there's no way the PS3 will be able to run it at acceptable framerates. And that's fine by me.
I say bring back the glorious PS360 era! Less crunch, less wokeness, more games, more profitability, more bravado, less drama, more exclusives on both platforms... what's not to love from that era? Gaming is no longer the same these days.
I guess the other point about the Xenon compared to Cell BE is that the general purpose PPE cores in both systems can only reach 40-50% throughput even in the most optimal workloads, and the SPUs can easily exceed that efficiency upto 80-90% in optimal real tasks IIRC which in the second half of the generation was common in first party games. but still probably not.There are also misleading figures flying around as to max theoretical compute ceilings, apparently RSX operated at different clock speeds for pixel and vertex pipelines. Pixel side is 211 GFlops/4.4 Gpixels at 550Mhz and Vertex one is 40 GFlops at 500Mhz, so 251 GFlops in total. X360 Xenos is at 240 Glops/4 Gpixels (unified).
Whole system for PS3: CELL BE 192+RSX 251=443 GFlops. Whole system for X360: Xenon 115+Xenos 240=355 GFlops. Not that it matters more than architectural differences.
They had a shit ton of effects that were added to the engine in comparison with Drake's Fortune, once they started using the SPUs.
DoF, Per-Object Motion Blur, SSS, the perfect AO, simulations, the constant streaming.. all thanks to Cell.
Thanks for the insight. I also had the higher SIMD saturation in mind leading to much higher efficiency making 1:1 compute comparisons somewhat misleading (as it tends to happen). And just to confirm the theoretical max; is the official figure for CELL PPU 33.8 GFlops or 25.6 GFlops (just like the SPEs)? Is it the exact same core as the Xenon one? I've had some difficulties to find the official Sony/IBM figures concerning those points.I guess the other point about the Xenon compared to Cell BE is that the general purpose PPE cores in both systems can only reach 40-50% throughput even in the most optimal workloads, and the SPUs can easily exceed that efficiency upto 80-90% in optimal real tasks IIRC which in the second half of the generation was common in first party games. but still probably not.
Cell BE wiki "Tests by IBM show that the SPEs can reach 98% of their theoretical peak performance running optimized parallel matrix multiplication"
So not sure if the Xenon with VMX128 figure is correct. But going by 25.6Gflops/s per PPE core , in optimal real situations (for the 2nd & 3rd PPEs anyway) the Xenon was no more than 38.4Gflops total, and the Cell BE was no more than 12.8Gflops + 6x 80% SPU = 136.6 Gflops.
Given that the main PPE in both the Xenon and Cell wouldn't be doing vector maths exclsuively and would mostly be doing main program flow control, the 12.8Gflop is still overly inflated for the comparison, and the real comparison would really be Xenon 25.6Gflops vs Cell BE SPUs 122.8Gflops for helping their respective GPUs in optimal situations
Which give a comparison of 25.6+240 = 264.6Gflops + main PPE for Xbox 360 vs 122.8 + 251 = 373.8Gflops + main PPE for PS3. And obviously we aren't counting that a 7th working SPU inside the PS3 could have been recording TVTVTV, as it was on my unit frequently while playing games like AC1/2, MGS4, UC1-3,etc