EFLOPS is a unit of the speed of computer systems, equal to one quintillion floating-point operations per second.

EFLOPS表示每秒执行一千万亿次的运算。

So what did we do? We're using the fact that, you know, we have a gigaflop sitting here.

所以我们怎么做的呢？我们用到了千兆次运算。

Is ithat close to like a megaflop or something?

它接近百万次运算吗？

Plus, rounding errors are inherent in floating point computations.

另外，舍入误差是计算中固有的。

Several methods have been developed to represent floating point numbers.

已经开了几种方法表示数。

It'll be a native array of floats.

它是一个原生的数组。

A good GPU can do teraflops today.

如今，优秀的 GPU 可以实现万亿次运算。

There's an infinite number of floating point real numbers in the world.

世界上有无数的实数。

This time I'm using floats, but let's go ahead and just run this.

这次我用的是数，但让我们继续运行这个。

Let's just store the float A in the first element of the native array.

我们数 A 存储在本机数组的第一个元素中。

If it only has a finite number of bits, 32, via which to represent a float.

如它只有有限个位， 32表示数。

These are called " floating point" numbers, because the decimal point can float around in the middle of number.

这些被称为“”数字，因为小数可以在数字中间动。

Remember the fields have to be blitable, bytes, floats, integers, stuff like that.

请记住，字段必须是可直接传送的、字节、数、整数等。

There was float, which is a real number with a decimal point.

有一个数，是一个带小数的实数。

And I'm doing some floating point arithmetic, ultimately.

最后，我在做一些运算。

I'll invoke instantiateEntity and then pass in a float 3, 4, 0, 4.

我调用 instantiateEntity，然后传入数 3、4、0、4。

Percent f backslash n, because I'm printing a float this time.

%f反斜杠n，因为我这次打印的是数。

So in this case, I'll have a native array of floats, and let's just call this result.

因此，在这种情况下，我拥有一个本机数组，我们只需调用此结即可。

They are giant, inscrutable matrices of floating point numbers that we nudge in the direction of better performance until they inexplicably start working.

它们是无法参透的大型数矩阵，我们在朝着更好的性能试探着前进，直到它们莫名其妙地运行起。

Or maybe floating point values, like 3.14159, or other such values.

或者是值，比如3.14159，或者其他类似的值。