That still gives you allmost a year of extra development time, because your utilizing most of your stuff, all your doing in that year of time is making sure you get as high framerate and as low amount of bugs as possible, the game engine, the story, the ai, everything is allready finished, so you basically got 1 whole year to port and tweak.
Cross platform development and choice of 'Lead system' *Spinoff*
- Thread starter iceberg187
- Start date
That still gives you allmost a year of extra development time, because your utilizing most of your stuff, all your doing in that year of time is making sure you get as high framerate and as low amount of bugs as possible, the game engine, the story, the ai, everything is allready finished, so you basically got 1 whole year to port and tweak.
The PC version of Oblivion (I assume the Xbox 360 initial release wasn't much different) has absurd inefficiencies in its renderer, as documented in this reverse engineering report (sorry, Russian):
http://www.gamedev.ru/community/gamedev_lecture/articles/?id=25
It shouldn't be used as an example, at least not in a tech-focused discussion.
http://www.gamedev.ru/community/gamedev_lecture/articles/?id=25
It shouldn't be used as an example, at least not in a tech-focused discussion.
Yeah, from Gametrailers and Gamersyde you can usually download 20 to 60 minutes of 720p gameplay videos, and it's almost like playing the game for yourself. And it's exactly the hardcore gamers who download these, and they're the ones to start spreading the word about games.
Neither in an art-focused discussion, IMHO
Honestly, I'm still surprised how much some people have liked its graphics...According to Games radar PS3 version of DMC4 is slightly better ("crispier textures") than X360:
http://www.gamesradar.com/gb/ps3/ga...7122116016612075&releaseId=200609151285614010
It's interesting for three reasons:
- Capcom made two x360 games based on framework engine (dead rising and lost planet), so they have experienced team...
- there is no framwork based games for PS3, Lost Planet PS3 port which is during production has worse textures than x360 original.
- PS3 has smaller amount of Ram so in typical situation xbox version of dmc4 should have better textures.
What's the reason for this?
Do you think taht PS3 is leading platform for this game?
Other thing is that DMC4 is not the most graphically impressive title..
http://www.gamesradar.com/gb/ps3/ga...7122116016612075&releaseId=200609151285614010
It's interesting for three reasons:
- Capcom made two x360 games based on framework engine (dead rising and lost planet), so they have experienced team...
- there is no framwork based games for PS3, Lost Planet PS3 port which is during production has worse textures than x360 original.
- PS3 has smaller amount of Ram so in typical situation xbox version of dmc4 should have better textures.
What's the reason for this?
Do you think taht PS3 is leading platform for this game?
Other thing is that DMC4 is not the most graphically impressive title..
Im shure it is becouse of the use of anisotropic filtering and not higher resolution for textures. That of course if it really is a difference...
RSX simply has more texture filtering units (24) than Xenos (16). I believe it was recommended in one of the GDC presentations for X360 devs to go bilinear + AF. Trilinear + AF seems to be asking too much of it.
Texture fetch limited *I think*.
From Xbox 360 GPU Performance Update :
Bits from Slides 23-25 (excuse the pun):
Texture Cache
IIRC, RSX's internal caches were increased from G70. I think that was mentioned on the forums before. I can't remember where though...
Bits from Slides 23-25 (excuse the pun
):Texture Cache
- Texture cache is 32KB, 16-way set associative
- Optimized for bilinear filtering, not for holding entire textures
- Trilinear threshold can noticeably improve cache performance for trilinear
- Trilinear is 2 bilinear fetches & a blend
- Trilinear threshold can noticeably improve cache performance for trilinear
- Trilinear fetch becomes a bilinear fetch near whole mip levels
- Lack of mipmaps, or mip biasing
- Do not expose mip bias in tools
- Consider aniso filtering instead of mip biasing
- Uncompressed textures
- Somewhat unavoidable for full screen effects
- Volume textures
- Be very careful with your sampling patterns
- Bad access patterns
- Too many simultaneous textures
- Sparse sampling patterns
IIRC, RSX's internal caches were increased from G70. I think that was mentioned on the forums before. I can't remember where though...
IIRC, RSX's internal caches were increased from G70. I think that was mentioned on the forums before. I can't remember where though...
Its here :
Originally Posted by Barbarian
By now, all that you need to know about RSX is already in the public domain. The clock, the bus, the number of pixel and vertex pipelines, etc. Accept that it is what it is.
The texture caches that you talk about, indeed they were increased, only to cope with the higher latency when pulling data from XDR memory. And just so you don't get any ideas, the texture buffers were increase from 48k to 96k. Nothing ground breaking.
Also the shaders got a couple of extra instructions - a fast vector normalize and some extra texture lookup logic. Nothing fancy again.
So don't look for any miracles here. If you want to feel better, understand this - noone has ever used the 'lowly' 7800GT's to it's full potential yet. That's the beauty of fixed hardware, you know all that the hardware is capable of and you don't have to worry about compatibility with 5 year old crapy cards. So that's where the true power of RSX lies - in developer's talent to extract incredible effects from a very capable and robust piece of silicon.
And here:
http://forums.e-mpire.com/showpost.php?p=1250089&postcount=200
THE MOSTLY COMPLETE AND PROBABLE FINAL RSX SPECS
RSX
Core Frequency - 500MHz
Memory Frequency - 650MHZ
Bus Size: 128BIT
Pixel Shaders - 24
Vertex Shaders - 8
ROPS - 8
Total Texture Cache Per Quad of Pixel Pipes (L1 & L2) - 96KB
Post Transform & Lighting Cache - 63 Max Vertices
*A few extra shader instructions - Extra Texture Lookup Logic & Fast Vector Normalize
*FLEX IO interface to CPU (Much Faster)
7800GTX
Core Frequency - 430
Bus Size: 256BIT
Memory Frequency - 600MHZ
Pixel Shaders - 24
Vertex Shaders - 8
ROPS - 16
Total Texture Cache Per Quad of Pixel Pipes (L1 & L2) - 48KB
Post Transform & Lighting Cache - 45 Max Vertices
PCI BUS interface to CPU (Much Slower)
NOTES: About the RSX.
Total Texture Cache Per Quad of Pixel Pipes - (L1 and L2) 96KB total Texture Cache - Previously 48K
(L1 only available to Pixel Shaders)
Post Transform and Lighting Vertex Cache - 63 Max Vertices - Previously 45 Vertices
(Cache located after Vertex Shader and before the triangle setup and before the Rasterizer.)
Vertex shader --> Post Transform and Lighting Vertex Cache 63MAX ---> Triangle Setup
Texture Lookup Logic to help RSX transport data from XDR.
About DMC4...in videos(gamersyde.com,gametrailers.com etc) i dont see any diferencies.
Last edited by a moderator:
I would attribute a lot to personal fanboyism on part of the reviewer.
I've watched the comparison video posted at gametrailers several times, and I can't find what they're talking about; if anything, on several scenes the textures seem to be in favor of the 360 version.
On the other hand, I've read tons of fanboy yammer about how this is a "playstation game" at heart, and even if it were exactly the same, it would be better on a "playstation family platform". Blech.
The comparison video is garbage when it comes to judging the quality of textures. I'm a 360 owner and only care about playing the game on the 360. When looking at the quality of the video I can tell that they not only did a poor job of capturing video from the 360 but an even worse job capturing video from the PS3. The PS3 part seems to be more compressed than the 360 part. The video does a great job showing people that the game runs smoothly on both consoles but it wouldn't be fair to judge the quality of textures from it.
The only difference I could spot through out the whole thing is at 1:18. It should look the same on both sides of the screen but it doesn't.
Hm, strange. If I was them I would capture using the least compressing codec possible, splice the two videos in this form, and compress the result as one video, using the same codec with the same settings.
A possible source of different compression would be the different brightness/gamma of the two halves.
Edit: Speaking of differences, there's considerably detail in the bump map for the red scarf thing around 1:23 in the 360 version.
Last edited by a moderator:
MazingerDUDE
Regular
IMO, GT comparison videos are untrustworhty.
I captured few shots from the video, and compared it to shots taken from their ealier PS3 video
http://www3.telus.net/public/dhwag/DMC1.jpg
http://www3.telus.net/public/dhwag/DMC2.jpg
As you can see in the comparison video, PS3 side is in considerably less quality, leaving some of the details out from their previous video. (ie. bumps on red scarf)
Only real difference I see is the lack of AA in PS3 version (check out the gun part)
Got the wrong pictures up. I tried to edit, but it seems like edit function is disabled on my posts
Mod Note: Please try to refrain from posting multiple >200kB images. Indeed there do exist folks who are not so privileged with high bandwidth connections. The first two screenshots do not demonstrate anything worthy, so I've converted them to links for brevity. I've also fixed your double img posts for DMC3.JPG and DMC4.JPG and merged your second post.
You'll have the privilege to edit posts once you post more btw.
As I said before if someone is stupid enough to run the risk of being caught after having leaked informations under NDA on a public forum he/she should at least reveal proper information, not some kind of dumbed down and wrongly reinterpreted version of the real deal.
Ugg. The video in their comparison is so compressed that it's difficult to make any judgment. It does look the the blocking is worse on the PS3 side as well. It's hard to make a conclusive statement about the bumps in the scarf because I can easily image the detail has been filtered out during the capture/encode. Still, the fact remains that in motion (and the game moves really fast!) it's all but impossible to see any difference. Both versions look great and seem to run well. Well done, Capcom.
