Spoiler-free Sonic Mania discussion thread

Cooljerk · Aug 29, 2017

MartiusR said:

Cooljerk said:

MartiusR said:

Of course it has. The "encyclopedical" meaning of optimization in case of software is considered as lowering the use of resources and increasing efficiency. In case of computer software resources are CPU, RAM, etc. (literally, resource is not processor itself, but the computing power of processor - the same thing with RAM etc.) If minimal requirements are significantly increased (in this case they are, aside from the fact that even those recommended are nothing special for even 4-years old PC), it's FAR from something, which could be considered as optimization. In other words - we were treated with big, juicy lie, with this reason of delay. The cooperation with various hardware you've mentioned - I would consider it rather in categories of compability than optimization - perfect example are games which are cooperating with almost every possible graphic card, processor etc, but are running terribly due to lack of optimization.
Click to expand...

I guess this is how laymen think optimization works, lol

Let me give you a real world example of optimization that doesn't at all adhere to your definition - architecture-specific morton encoding. Morton encoding is a useful way to project a multi-dimension space onto a single-dimension array to speed collision detection. Morton encoding is done by representing X,Y,Z coordinates in binary, then interleaving their bitfields. The resultant bitstring acts as a z-ordered set that can be used as a map to determine quadrants of the screen where the point resides (quite like balanced octrees).

Morton encoding is a normally expensive process depending on how you achieve it. Nested for loops, for example, will yield an n^2 time. One can optimize this operation by, for example, using a pre-computed look-up table, which approaches N^2logN. Thus, yes, in that specific case, optimization sped up the process on the same hardware. However, the implimentation of a lookup table drastically (by several orders) increases the memory footprint of your program.

However, using processor-specific inline assembly is ALSO optimization. For example, pre-haswell X86 processors and ARM processors have a "multiply no carry" instruction (PCLMULQDQ on X86, vmul_p8 on arm), and a property of multiply no carry is that a value multiplied by itself will result in the original string interleaved by 0's. Bitshifting after multiply no carry and bitmasking can accomplish the same encoding in NlogN time - even faster than a lookup table. But in this case, "optimization" increased the hardware requirements.

And to go a step further, prcoessors with BMI2 instruction sets (read haswell and better processors) have two instructions - Parallel Bit Deposit (PDEP) and Parallel Bit Extract (PEXT) which are built specifically for hardware bit interleaving. Using these instructions, you approach LogN as your run time.

Each one of these optimizations did the exact opposite of what you claimed - they improved performance but required better hardware each time. Which is usually how optimization works - it's about using the hardware available smarter.

TL;DR: your perspective on optimization is all wrong.
Click to expand...

The fact that you're showing example of optimization which is not connected with lowering system requirements, doesn't mean that what I've written is not correct anymore. Maybe it's working good as eristic tricks for someone, who is not reading carefully whole conversation, but you won't fool me with such tricks - lowering system requirements and better use of resources is repeated in many books representing various aspects of software and hardware.
Click to expand...

"To optimize" means to make optimal. The way you make something optimal is to use the hardware available as best as possible. Things which are general are usually not optimal, and vice versa. It's really that simple.

McAleeCh · Aug 29, 2017

Really can't wait to finally play this - have to go out this evening so I won't be able to get started until tomorrow, but I've got tomorrow booked off so I can have a proper Mania marathon! Don't usually take time off for video games, but had a spare day of holiday so decided to go for it. Anyone else taken time off work for this, either now or for the console release?

Deef · Aug 29, 2017

That's what I should have done with the console release.

Anyway, crossing my fingers it has 144Hz support!

MartiusR · Aug 29, 2017

Cooljerk said:

MartiusR said:

Cooljerk said:

MartiusR said:

Of course it has. The "encyclopedical" meaning of optimization in case of software is considered as lowering the use of resources and increasing efficiency. In case of computer software resources are CPU, RAM, etc. (literally, resource is not processor itself, but the computing power of processor - the same thing with RAM etc.) If minimal requirements are significantly increased (in this case they are, aside from the fact that even those recommended are nothing special for even 4-years old PC), it's FAR from something, which could be considered as optimization. In other words - we were treated with big, juicy lie, with this reason of delay. The cooperation with various hardware you've mentioned - I would consider it rather in categories of compability than optimization - perfect example are games which are cooperating with almost every possible graphic card, processor etc, but are running terribly due to lack of optimization.
Click to expand...

I guess this is how laymen think optimization works, lol

Let me give you a real world example of optimization that doesn't at all adhere to your definition - architecture-specific morton encoding. Morton encoding is a useful way to project a multi-dimension space onto a single-dimension array to speed collision detection. Morton encoding is done by representing X,Y,Z coordinates in binary, then interleaving their bitfields. The resultant bitstring acts as a z-ordered set that can be used as a map to determine quadrants of the screen where the point resides (quite like balanced octrees).

Morton encoding is a normally expensive process depending on how you achieve it. Nested for loops, for example, will yield an n^2 time. One can optimize this operation by, for example, using a pre-computed look-up table, which approaches N^2logN. Thus, yes, in that specific case, optimization sped up the process on the same hardware. However, the implimentation of a lookup table drastically (by several orders) increases the memory footprint of your program.

However, using processor-specific inline assembly is ALSO optimization. For example, pre-haswell X86 processors and ARM processors have a "multiply no carry" instruction (PCLMULQDQ on X86, vmul_p8 on arm), and a property of multiply no carry is that a value multiplied by itself will result in the original string interleaved by 0's. Bitshifting after multiply no carry and bitmasking can accomplish the same encoding in NlogN time - even faster than a lookup table. But in this case, "optimization" increased the hardware requirements.

And to go a step further, prcoessors with BMI2 instruction sets (read haswell and better processors) have two instructions - Parallel Bit Deposit (PDEP) and Parallel Bit Extract (PEXT) which are built specifically for hardware bit interleaving. Using these instructions, you approach LogN as your run time.

Each one of these optimizations did the exact opposite of what you claimed - they improved performance but required better hardware each time. Which is usually how optimization works - it's about using the hardware available smarter.

TL;DR: your perspective on optimization is all wrong.
Click to expand...

The fact that you're showing example of optimization which is not connected with lowering system requirements, doesn't mean that what I've written is not correct anymore. Maybe it's working good as eristic tricks for someone, who is not reading carefully whole conversation, but you won't fool me with such tricks - lowering system requirements and better use of resources is repeated in many books representing various aspects of software and hardware.
Click to expand...

"To optimize" means to make optimal. The way you make something optimal is to use the hardware available as best as possible. Things which are general are usually not optimal, and vice versa. It's really that simple.
Click to expand...

Exactly - and what is happening when your software is better using available resources on some configuration(s) - the same, which previously were insufficient to be considered as "minimal" or "optimal" for program to run? System requirements ("real") are lower (vide casus with Witcher 1, which after many patches and improvements after release had lower "real" system requirements, where game was running fine.

LordOfSquad · Aug 29, 2017

McAleeCh said:

Really can't wait to finally play this - have to go out this evening so I won't be able to get started until tomorrow, but I've got tomorrow booked off so I can have a proper Mania marathon! Don't usually take time off for video games, but had a spare day of holiday so decided to go for it. Anyone else taken time off work for this, either now or for the console release?
Click to expand...

I took the day off on the 15th for it. First time I took a day off specifically for any game, and it was way worth it. Got a migraine though, figures.

Cooljerk · Aug 29, 2017

MartiusR said:

Cooljerk said:

MartiusR said:

Cooljerk said:

MartiusR said:

Of course it has. The "encyclopedical" meaning of optimization in case of software is considered as lowering the use of resources and increasing efficiency. In case of computer software resources are CPU, RAM, etc. (literally, resource is not processor itself, but the computing power of processor - the same thing with RAM etc.) If minimal requirements are significantly increased (in this case they are, aside from the fact that even those recommended are nothing special for even 4-years old PC), it's FAR from something, which could be considered as optimization. In other words - we were treated with big, juicy lie, with this reason of delay. The cooperation with various hardware you've mentioned - I would consider it rather in categories of compability than optimization - perfect example are games which are cooperating with almost every possible graphic card, processor etc, but are running terribly due to lack of optimization.
Click to expand...

I guess this is how laymen think optimization works, lol

Let me give you a real world example of optimization that doesn't at all adhere to your definition - architecture-specific morton encoding. Morton encoding is a useful way to project a multi-dimension space onto a single-dimension array to speed collision detection. Morton encoding is done by representing X,Y,Z coordinates in binary, then interleaving their bitfields. The resultant bitstring acts as a z-ordered set that can be used as a map to determine quadrants of the screen where the point resides (quite like balanced octrees).

Morton encoding is a normally expensive process depending on how you achieve it. Nested for loops, for example, will yield an n^2 time. One can optimize this operation by, for example, using a pre-computed look-up table, which approaches N^2logN. Thus, yes, in that specific case, optimization sped up the process on the same hardware. However, the implimentation of a lookup table drastically (by several orders) increases the memory footprint of your program.

However, using processor-specific inline assembly is ALSO optimization. For example, pre-haswell X86 processors and ARM processors have a "multiply no carry" instruction (PCLMULQDQ on X86, vmul_p8 on arm), and a property of multiply no carry is that a value multiplied by itself will result in the original string interleaved by 0's. Bitshifting after multiply no carry and bitmasking can accomplish the same encoding in NlogN time - even faster than a lookup table. But in this case, "optimization" increased the hardware requirements.

And to go a step further, prcoessors with BMI2 instruction sets (read haswell and better processors) have two instructions - Parallel Bit Deposit (PDEP) and Parallel Bit Extract (PEXT) which are built specifically for hardware bit interleaving. Using these instructions, you approach LogN as your run time.

Each one of these optimizations did the exact opposite of what you claimed - they improved performance but required better hardware each time. Which is usually how optimization works - it's about using the hardware available smarter.

TL;DR: your perspective on optimization is all wrong.
Click to expand...

The fact that you're showing example of optimization which is not connected with lowering system requirements, doesn't mean that what I've written is not correct anymore. Maybe it's working good as eristic tricks for someone, who is not reading carefully whole conversation, but you won't fool me with such tricks - lowering system requirements and better use of resources is repeated in many books representing various aspects of software and hardware.
Click to expand...

"To optimize" means to make optimal. The way you make something optimal is to use the hardware available as best as possible. Things which are general are usually not optimal, and vice versa. It's really that simple.
Click to expand...

Exactly - and what is happening when your software is better using available resources on some configuration(s) - the same, which previously were insufficient to be considered as "minimal" or "optimal" for program to run? System requirements ("real") are lower (vide casus with Witcher 1, which after many patches and improvements after release had lower "real" system requirements, where game was running fine.
Click to expand...

You've completely missed the point of my posts. That being that your seething claim that they lied because hardware requirements didn't lower is bunk, that you don't appear equipped to have a technical understanding of what optimization is, and that your bar for optimization being "lowered minimum specs" is a misreading of what optimization can be and often is. If you're classifying optimization solely as a method to lower minimum system requirements, then you flat-out don't know what optimization is. Indeed, optimization is often the opposite, it increases specificity and lowers generality.

And you're not really saying anything new in any of your posts.

Regarding Witcher 1: Those "patches" were merely ini file edits, i.e. fine tuning configuration files to turn down effects and buffers that were always, from the get-go, meant to be configurable. Which, ironically, is much closer to your original claims that one would consider them "in categories of compability than optimization." If this is what you think optimization is, no wonder you're confused.

Flygon · Aug 29, 2017

Deef said:

That's what I should have done with the console release.

Anyway, crossing my fingers it has 144Hz support!
Click to expand...

I wonder if my CRT will do 200Hz running 640*480? :v:

synchronizer · Aug 29, 2017

Frostav said:

It's not like the game is super buggy, to be honest.
Click to expand...

Sure, but the way Taxman phrases it ("In the future") makes it seem as though we will wait weeks or more.

Lilly · Aug 29, 2017

As said earlier in the thread, iirc, updates for console versions have to be certified before released. So even if they wanted to post an update now, they couldn't do it right away.

I think there's three minutes left before the game unlocks in my timezone. So hyped.

EDIT: Turns out it's still less than an hour away, what a tease. Be still, my heart. :specialed:

Crimson Neo · Aug 29, 2017

"This game will unlock in approximately less than an hour".

Ugh, Steam.

MainMemory · Aug 29, 2017

I guess the counter rounds down to the nearest hour, which sort of gives the false impression it releases an hour before it actually does. Why they can't just use a countdown in hh:mm:ss format, like the preorder timer immediately below it, I don't know. Also apparently you can still preorder the game for an hour after it's released.

Dark Sonic · Aug 29, 2017

MainMemory said:

Also apparently you can still preorder the game for an hour after it's released.
Click to expand...

I'm sorry. What?

How does that even work

ApocalypticSalad · Aug 29, 2017

MainMemory said:

I guess the counter rounds down to the nearest hour, which sort of gives the false impression it releases an hour before it actually does. Why they can't just use a countdown in hh:mm:ss format, like the preorder timer immediately below it, I don't know. Also apparently you can still preorder the game for an hour after it's released.
Click to expand...

It's far more egregious when they round down weeks and months the same way (which they do), but I suppose when the date's right above it's less likely to confuse. Also, it looks like they, as of a couple of minutes ago, just updated the pre-order timer to be correct by shaving an hour off of it.

Covarr · Aug 29, 2017

The Growler said:

I guess the Steam download will become available once the "humble bundle" pre-order sale ends? If so, let the countdown begin!
Click to expand...

Yes, the Steam store generally updates at 10am VALV^E Standard Time (Pacific). Sometimes they do midnight launches for really high profile games, but if it's first-party VALV^E stuff they are pretty much bound to screw it up and launch a few hours late. But for a 3rd-party game launching at the normal 10am PDT, that shouldn't be an issue at all.

tokumaru · Aug 29, 2017

Also noticed the pre-order/release time discrepancy. Anyway, guess I'll be home for the launch after all!

jubbalub · Aug 29, 2017

The preorder timer has been updated to reflect the release date - 15 minutes bois

MainMemory · Aug 29, 2017

Maybe they read my post. :v:

Deef · Aug 29, 2017

Terrible time for Australia

Willie · Aug 29, 2017

It says I can play the game now! Now I can enjoy the game on my own terms instead of having to re-accomplish the same things repeatedly on different devices from friends. Also will finally play Knuckles which I saved for the PC version outside of the game's first act.

ApocalypticSalad · Aug 29, 2017

Game's out. If you have it pre-ordered, the 'Install' button should be there if you check your library.

Spoiler-free Sonic Mania discussion thread

Useful Searches