World War 3 Download Size

[Breanna Mangels]

It is hard to represent our spherical world on flat piece of paper. Cartographers use something called a "projection" to morph the globe into 2D map. The most popular of these is the Mercator projection.

Every map projection introduces distortion, and each has its own set of problems. One of the most common criticisms of the Mercator map is that it exaggerates the size of countries nearer the poles (US, Russia, Europe), while downplaying the size of those near the equator (the African Continent). On the Mercator projection Greenland appears to be roughly the same size as Africa. In reality, Greenland is 0.8 million sq. miles and Africa is 11.6 million sq. miles, nearly 14 and a half times larger.

This app was created by James Talmage and Damon Maneice. It was inspired by an episode of The West Wing and an infographic by Kai Krause entitled "The True Size of Africa". We hope teachers will use it to show their students just how big the world actually is.

Let's assume that we have a single node with 4 GPUs, and we want to run our script on those 4 GPUs (i.e. one process per GPU). In such a scenario, what are the rank world size and rank? I often find the explanation for world size: Total number of processes involved in the job, so I assume that that is four in our example, but what about rank?

You can think of world as a group containing all the processes for your distributed training. Usually, each GPU corresponds to one process. Processes in the world can communicate with each other, which is why you can train your model distributedly and still get the correct gradient update. So world size is the number of processes for your training, which is usually the number of GPUs you are using for distributed training.

Rank is the unique ID given to a process, so that other processes know how to identify a particular process. Local rank is the a unique local ID for processes running in a single node, this is where my view differs with @zihaozhihao.

I think group should be understood correctly first. It can be thought as "group of processes" or "world", and one job is corresponding to one group usually. world_size is the number of processes in this group, which is also the number of processes participating in the job. rank is a unique id for each process in the group.

I am new to the rust language. When i compare my rust hello world program binary with c/c++ hello world program binary i found huge size difference. My rust binary size is 3.1MB and c/c++ binary size is 128KB. I tried google to know why the the size is so huge. And i come to know that rust by default do the static linking due to this the size is increasing. Can any one please correct me if i am wrong. And if this the case please tell me how can i reduce my binary size for hello world program. Using cargo i tried both develop and release binary and in both the case the binary size is 3.1MB.

@MaximilianKoestler thank you very much for your help. Now i am able to create my binary with 1.2MB. Now all most it reduced to around 70% after adding below option in Cargo.toml which one i found from you git hub.

I wonder if I am understanding the concept behind the world size correctly. Am I right to assume that if I set world width and length to 10k and polar distance to 5k, that the poles would be right on the edge of my map to the south and to the north? Like this:

As for how the poles are positioned on a 10k map with 5k pole-equator distance? I actually don't know. I wondered this before, but didn't arrive at a conclusive answer. One one hand, it makes intuitive sense to align the climate to a world edge that way; on the other hand, the 0,0 coordinate is always at a latitude of 45 north, and you'd intuitively expect there to be half a world in each direction. You cannot have both at the same time. And then there's the fact that there's two coordinate systems in VS - one absolute (used in the data files), and one relative to the world spawn (displayed to the player). It gets confusing.

thanks a lot for your reply. I actually did it (but in survival mode). When I reached the southern edge it did not look like a pole to me. I am going to do some further testing in creative mode. I just abandoned my 60h+ save file because I realized that I had a 1 million world with 100k equator. I hate this to be the standard setting. I would not see any southern biome in forever. When you first start the game you have no clue what it all means (and I guess I still don't really understand it).

of course you could manipulate the world in having a single southpole and a single northpole, just start in the tropics, that'll set the middle to somewhere around the equator. (in that 10k world length 5k distance to pole scenario)

But yeah, if the world is centered on the player spawn, then that is precisely what I would expect to see. You can further research this using the command /wgen pos latitude at your northern and southern world borders to print your current latitude into chat.

My main reason for asking is in the two worlds I've played the most in, they have had very different temperature patterns. On one, even in the peak of summer it never gets above 20-25 C, and in the winter gets down to -20, with snow falling as early as October and persisting until well into April. On another, it was getting well into the 30's by early summer, and didn't get snowfall until January, and didn't stay freezing for long. Both have bases built at spawn, and have default temperature settings in worldgen.

Yep. For some reason, the world temperature seems to get randomized and tied to the seed. But only if you don't click the advanced customize menu, because when you do it seems to get randomized regardless of the seed.

...Yeah, it's extremely weird. I did a whole systematic test series a few weeks ago and passed the results on to Tyron; it's probably not intended to be this way. I mean, for the 1st of May at 8:00 in the morning at 45 north, I got anywhere between -15C and +37C in my testing.

I read somewhere once (Can't remember where, or how old it was) that Spawn was at a random latitude between an Equator and a Pole to the North, I just assumed when I started with colder Spawns that I was closer to the pole than the equator, or visa versa for a warmer spawn area.... But is that not true? That some worlds are just colder on average all over?!

@Jessica O. Yeah, some worlds are just colder than others, and the spawn latitude is always fairly close to 45N. In fact, if you create a superflat world, it is always perfectly 45N. The temperature can still vary wildly.

It just appears to me that all the tutorials on UE5 World Partition are still using the old UE4 World Composition dev pipeline of using the UE4 Landscape size restrictions from the UE4 Landscape Technical guide instead of taking advantage, as I understand it, of the new capabilities of World Partition Landscape Size possibilities?

An imported heightmap larger than 4 km across needs to be broken into tiles, with a specific naming convention. Each tile must have a suffix that indicates its position in the map by showing what number it is in the x direction and what number it is in the y direction, as in the image

UE5 World Partition is still limited to the same Recommended Landscape Sizes from the documentation, and the equivalent large worlds based on those dimensions, such as 16x16 tiles of 1009x1009 tile size, or 8x8 tiles of 2017x2017 tile size.

This is because the Landscape actor uses power-of-two textures under the hood to manage all of the Sections and LODs. So these same design restrictions are imposed in UE5 as well.
Read the Landscape Technical Guide page to see how Sections and Components relate to GPU textures, and follow those guidelines when creating any terrains.

The UE5 Landscape interface for creating New Landscapes is limited to a maximum of 8161x8161, the same as UE4, when manually creating a new terrain.
To create terrains larger than 8km x 8km requires that you import an XY Tile Set.
So computer system main memory and video memory are determining factors for how large of a terrain world you can create without running out of memory.
Importing the tiles and creating the cells uses main memory and GPU memory, and for World Partition the amount of memory used is higher than UE4 World Composition.

I have spent the past few days digging into the World Partition system, and I have figured out their new World Partition world Landscape sizes. They are different than World Composition, my guess is to make the Section size a larger texture for fewer drawcalls on large terrains.

I will be adding the information that I have determined for valid world sizes to the Unreal Engine Landscape dialog in the TerreSculptor Terrain Software, and also put this information into the upcoming video series on my YouTube channel regarding World Partition.

However when I tested importing an even larger 24km terrain (3 x 3 - 8129 tiles) in UE4, save it and then open that level in UE5 to trigger the conversion steps, then finally convert to World partition the memory consumption seems less aggressive with hardly any shared memory being used:

UE4 World Composition used tiles of Recommended Landscape Sizes as defined in the UE4 Landscape documentation, such as 8x8 tiles of 1009x1009 tile sizes with shared edges. which would result in terrain Section sizes of either 63 or 127, and Components that are either 1x1 or 2x2 Sections.
World Partition does NOT do this! It uses an entirely new sizing method, based on Section sizes of 511 (not 63 or 127). My guess is to get the number of Drawcalls lower for rendering large terrains in World Partition.
This means that totally different Landscape sizes will have to be used for World Partition.

This also means that to create a tile set for import, you will have to determine what tile multiplier dimensions are valid. Most World Partition resolutions only have one tile size set value available for them, because they are such odd numbers.

c80f0f1006