In an enlightened society it is NOT the government's job to determine for me what news I should or should no be allowed to see. Maybe China should try joining the 21st century instead of repressing people's free will.
> Americans think they're free
We actually have a choice in what to watch. If I want to watch China Uncensored then I can.
Maybe China's government should try fixing the root problem instead of knee-jerk reaction to the symptom.
At least in America:
* If I want comedy I can watch the news (Faux News, etc.) * If I want the news I can watch the comics (LastWeekTonight, etc.)
The GTX 1060 is a Tier 4 GPU. Granted, the OP probably should be using a GTX 1080 for 4K res but we don't know what game and video settings the OP is using. Maybe the 1060 is good enough for their needs. Like you, I'm very skeptical, but unlike you, I'm willing to give them the benefit of the doubt instead of prejudging before we have all the facts.
Regardless, this is just another sign that the OP is clueless.
Sadly, MS doesn't care about musicians anymore.:-/ When was the last time you actually heard them talk about kernel latency? MS has embraced mediocrity for so long that they wouldn't know the first thing about inspiring greatness. Hell they STILL don't understand UI's -- they just copy the lastest fad of the decade.
While all the cool kids are using DAWs (Digital Audio Workstation) and VSTs (Virtual Instrument) it seems like most of the creative types migrated over to the OSX. Apple _used_ to care with its Jam Packs. At least they still ship GarageBand (last time I checked) but it seems like no one cares about having a fully digital learn-to-play tutorial.
Harmonics _completely_ squandered their opportunity with RockBand and they were even designing / selling their own hardware. If they can't even recognize the need, and instead half-ass it, I don't expect anyone else will either.
MIDI has been a forgotten step child -- which sucks.
Time to buy those old MIDI devices off eBay while we still can.
The problem is the OP didn't show he actually researched the problem and thus sounds completely clueless. This site jumps on people who can't even take 2 seconds to actually "Think, McFly!" about what they are asking.
If he had prefaced his sub-text with something like -- "I just started learning about video streaming and it seems complicated to me" -- then more people would be willing to give them some slack.
The fact that he took the time to post his question on/. BUT couldn't be arsed to spend the time to learn about:
* Mbps (Mega bits-per-second) and * File Size
shows that he isn't actually using his brain.
There is a reason RTFM exists, or the modern vernacular: LMGTFY.
/sarcasm Gee, if only there was a way to modify the game Options of the Video Settings to chose between Quality or Performance.Nah, that scalability thing will never catch on.
Maybe he is streaming a ~10-year old game such as L4D or Minecraft with radius of 4?:-)
The 1060 isn't a potato (nor is it beefy) -- video settings exist for a reason. The OP didn't say what game(s) they are streaming.
1. Assuming Nothing! 2. Measure / Verify! Documentation can be, and is, wrong.
Back in the mid 90's I was working on an OpenGL game on some Indy SGI machines for my graphics course. Since float32 sin() / cos() performance was crap in those days, let alone float64 I used a 1.10 fixed point lookup table for a float32 fast_sin() and fast_cos() lookup. Now a 1.8 fixed point table of 256 entries would have been faster but didn't have the required precision I needed (360/256 = 1.45 degrees per entry), so I ended up with 1024 entries as a compromise between speed and precision (360/1024 = 0.3515625 degrees per entry)
Anyhoo, I ran my benchmark:
float sum = 0.f;
timer_start();
for( i = 0 ; i < N; i++ )
sum += fast_sin( (float) i );
timer_end();
and noticed that it was taking zero seconds!?!? I thought "That can't be right --I MUSThave a bug."
I double checked, and triple checked the table and fast_sin() call. All the values were correct. There was no bug.
So I increased my benchmark so I was doing 10,000,000 fast_sin() calls and noticed that I _finally_ had a blip on the total time! Something like 0.001 seconds, whereas as the standard C double sin() was taking 10+ second. The floating-point performance on the Indy was SO bad that even hitting slow memory was STILL faster!
THEN I tried to take advantage of quadrant symmetry since you only need to store 0 degrees... 90 degrees for a cos lookup table (since sine is just is sin(x) = cos( x - 90 degrees) but it WASN'T any faster due to have to perform extra calculations!
A smart programmer would NOT waste their time optimizing with bit-twiddling when a smarter algorithm would do the job.
However, using a better algorithm is INSUFFICIENT when you are trying to optimizing for speed. You MUST use Data-Orientated Design if you want maximum performance. Simply knowing which algorithm is better will NOT do it. Why?
Because HOW you access the cache lines matter.
Two algorithms could have the exact O() performance --in theory, but in practice have DRASTICALLY different real-world performance.
Sum of LEN numbers, stride: 0
for( size_t i = 0; i < LEN; i++ )
nSum += pData[ i ];
Sum LEN numbers; stride: constant
for( size_t col = 0; col < STRIDE; col++ )
for( size_t i = 0; i < LEN; i += STRIDE )
nSum += pData[ i + col ];
Summing 1,073,741,824 elements (int32) for a total size of 4GB, have these BOTH showing O(n) performance (in theory) but with a stride of 32 the latter is seven times slower (in practice)
BOTH theory and implementation are important.
People who say that only theory matters are retards.
> Edsger W. Dijkstra... in fact shouldn't, involve actual computers at all.
Computer Science is an applied science..
Dijkstra was an idiot who thought that only theory should be taught.
* In theory performance shouldn't matter * In practiceit does.
Implementation details do matter regardless of many fucking cluesless profs try to handwave them. For example, how do you sort your data when it fit into available RAM? There is a reason why Map Reduce was invented.
Focus solely on theory is the wrong approach. There are 3 types of optimizations that a programmer needs to understand.
2. Algorithmic Spending time to optimize a bubble sort is a complete waste of time when you could use mergesort, quicksort, etc.
3. Macro-optimization (or cache-orientated) aka (Data-Orientated Design) Techniques such as Memoization exist for a reason.
A good programmer learns HOW to optimize. i.e.
Code Clinic 2015: How to Write Code the Compiler Can Actually Optimize
Ignoring optimization doesn't make it go away. That's how we end up with bloated crap where a user is forced to download a 50 MB file for a bloody printer driver.
Thanks for tracking down that Open Source Minecraft "promise" from Notch. I remembered reading it but couldn't remember where.
Yeah, I guess everyone found what it took for Notch to sell out his values: $2.5 Billion. (It's almost as if when Microsoft asked Notch how much he wanted for Mojang he probably jokingly said: 2^31. MS thought about it and replied "How about we round it up for an cool $2.5B?" But again MS is a dumb-ass company that paid $8.5 Billion for Skype.)
Ironically, Notch is not sleeping fine -- probably because he recognizes he's a sell out -- he lived the American (software) dream: Started a company and got bought out by a bigger one for an ungodly amount of money. And now he realizes that materialism is shallow. Go figure.
"Hanging out in ibiza with a bunch of friends and partying with famous people, able to do whatever I want, and I've never felt more isolated,"
The sad part is Minecraft has only scratched 1% of its potential yet Mojang doesn't really seem to understand what the community wants:
* A modding API (people are sick and tired of de-compiling the Minecraft source code that breaks mods *every* version) * Depth of items and crafting. It is ironic that Terraria (pardon the pun) has way more depth, both literally, and figuratively with crafting then Minecraft.
/sarcasm "But, hey, we get tridents in 1.13!"
Big Fucking Deal.
Where is the ability to have different colored wooden chests? Or barrels?
I'm surprised someone hasn't made a Terraria + Minecraft clone yet.
You are spot on that people use x86 / Windows to run existing desktop apps. Unless ALL their software is ported over (unlikely) there is just too much momentum to switch.
Like most evenings after work, Markus was on the computer when he stumbled upon an indie game he hadn't tried before. It was called Infiniminer. Markus downloaded the game, installed and clicked it into motion, and then almost fell off his chair. "Oh my God," he thought. "This is genius."
In early May 2009, Markus uploaded a video recording (above) of a very early version of Minecraft on YouTube. It didn't look like much more than a half-finished system for generating worlds and Markus gleefully jumping around inside it, but still, the essence of it hinted at how the game might look when it was done.
"This is a very early test of an Infiniminer clone I'm working on. It will have more resource management and materials, if I ever get around to finishing it," -- Notch, May 2009
Zach, the creater of Infiminer, said this about Minecraft:
"The act of borrowing ideas is integral to the creative process. There are games that came before Infiniminer, and there are games that will come after MineCraft. That's how it works."
I agree 100% that Apple has completely lost their way W.R.T. computer hardware.
When they:
* put out a new computer that is slower the previous model (the Mac Mini fiasco) * make it impossible to upgrade RAM and SSD because they are soldered in (WTF!?), etc.
you quickly realize Apple is all about streamlining their products at the expense of versatility.
Seriously, who the fuck would buy an Mac Pro -- when the hardware is so over-priced and out-dated it isn't even funny -- so why even continue selling it?
They are out of touch with what geeks want.
* Make gamepad support a 1st class citizen on the iPad / iPhone because some touch SUCKs for certain games * Allow me to hook up a wired (*gasp*!) keyboard and monitor to the iPad * Allow me to listen AND charge my phone at the SAME time -- like they fucking used to. * Provide a fucking MicroSD ram slot * Allow me to use my iPhone / iPad as a generic storage device
Instead we get "consume only" devices with a hard-one for everything wireless. Wireless is great -- but give me the OPTION of wired.
Apple only needs to do one thing to regain their focus:
How about making it _easy_ to _create_ content -- you know, like what the Mac became originally famous for.
Now this I definitely could see as being a/the problem!
Maybe a quick test would be to rename the Windows ARK executable and compare the visual output with the vanilla filename? If they differ then you know that the Windows Driver is in "automatic run-time patch mode" for the game that isn't getting applied on the Linux side.
Another test would be to do single "frame capture" -- dump all the D3D calls and see how they are getting translated to the OpenGL backend. That would help narrow down where the problem is.
It sucks that of the 4 companies involved: Microsoft (D3D), Epic, nVidia, Studio Wildcard -- that no one wants to take the time to figure this shit out.
Slashdot Mirror
I wasn't talking about _external_ hardware (which you are correct) but about Microsoft's support for it.
> Protectionism is a GOOD thing.
Only cowards censor.
In an enlightened society it is NOT the government's job to determine for me what news I should or should no be allowed to see. Maybe China should try joining the 21st century instead of repressing people's free will.
> Americans think they're free
We actually have a choice in what to watch. If I want to watch China Uncensored then I can.
Maybe China's government should try fixing the root problem instead of knee-jerk reaction to the symptom.
At least in America:
* If I want comedy I can watch the news (Faux News, etc.)
* If I want the news I can watch the comics (LastWeekTonight, etc.)
The GTX 1060 is a Tier 4 GPU. Granted, the OP probably should be using a GTX 1080 for 4K res but we don't know what game and video settings the OP is using. Maybe the 1060 is good enough for their needs. Like you, I'm very skeptical, but unlike you, I'm willing to give them the benefit of the doubt instead of prejudging before we have all the facts.
Regardless, this is just another sign that the OP is clueless.
Sadly, MS doesn't care about musicians anymore. :-/ When was the last time you actually heard them talk about kernel latency? MS has embraced mediocrity for so long that they wouldn't know the first thing about inspiring greatness. Hell they STILL don't understand UI's -- they just copy the lastest fad of the decade.
While all the cool kids are using DAWs (Digital Audio Workstation) and VSTs (Virtual Instrument) it seems like most of the creative types migrated over to the OSX. Apple _used_ to care with its Jam Packs. At least they still ship GarageBand (last time I checked) but it seems like no one cares about having a fully digital learn-to-play tutorial.
Harmonics _completely_ squandered their opportunity with RockBand and they were even designing / selling their own hardware. If they can't even recognize the need, and instead half-ass it, I don't expect anyone else will either.
MIDI has been a forgotten step child -- which sucks.
Time to buy those old MIDI devices off eBay while we still can.
There were MANY social media sites BEFORE and AFTER facebook.
Social Media Timeline
Mark has/had an arrogant attitude towards his users:
Only an idiot goes after free advertising.
Greater Internet Fuckwad Theory (GIFT)
On a more serious note Portrait of a Troll: Q&A with Dr. Erin Buckels
The problem is the OP didn't show he actually researched the problem and thus sounds completely clueless. This site jumps on people who can't even take 2 seconds to actually "Think, McFly!" about what they are asking.
If he had prefaced his sub-text with something like -- "I just started learning about video streaming and it seems complicated to me" -- then more people would be willing to give them some slack.
The fact that he took the time to post his question on /. BUT couldn't be arsed to spend the time to learn about:
* Mbps (Mega bits-per-second) and
* File Size
shows that he isn't actually using his brain.
There is a reason RTFM exists, or the modern vernacular: LMGTFY.
/sarcasm Gee, if only there was a way to modify the game Options of the Video Settings to chose between Quality or Performance. Nah, that scalability thing will never catch on.
Maybe he is streaming a ~10-year old game such as L4D or Minecraft with radius of 4? :-)
The 1060 isn't a potato (nor is it beefy) -- video settings exist for a reason. The OP didn't say what game(s) they are streaming.
/sarcasm /cynical Technology is great! Just put an ankle ball-and-chain on them -- they can't fall in if they can't get close to the pool !
I'm a big fan of Michael Abrash. Great reminders!
1. Assuming Nothing!
2. Measure / Verify! Documentation can be, and is, wrong.
Back in the mid 90's I was working on an OpenGL game on some Indy SGI machines for my graphics course. Since float32 sin() / cos() performance was crap in those days, let alone float64 I used a 1.10 fixed point lookup table for a float32 fast_sin() and fast_cos() lookup. Now a 1.8 fixed point table of 256 entries would have been faster but didn't have the required precision I needed (360/256 = 1.45 degrees per entry), so I ended up with 1024 entries as a compromise between speed and precision (360/1024 = 0.3515625 degrees per entry)
Anyhoo, I ran my benchmark:
float sum = 0.f;
timer_start();
for( i = 0 ; i < N; i++ )
sum += fast_sin( (float) i );
timer_end();
and noticed that it was taking zero seconds!?!? I thought "That can't be right --I MUSThave a bug."
I double checked, and triple checked the table and fast_sin() call. All the values were correct. There was no bug.
So I increased my benchmark so I was doing 10,000,000 fast_sin() calls and noticed that I _finally_ had a blip on the total time! Something like 0.001 seconds, whereas as the standard C double sin() was taking 10+ second. The floating-point performance on the Indy was SO bad that even hitting slow memory was STILL faster!
THEN I tried to take advantage of quadrant symmetry since you only need to store 0 degrees ... 90 degrees for a cos lookup table (since sine is just is sin(x) = cos( x - 90 degrees) but it WASN'T any faster due to have to perform extra calculations!
Your reminder about Michael Abrash's points:
1. Assume Nothing!
2. Measure / Verify!
are great timeless reminders.
Can we just ban web/print/radio advertising already?
If your product isn't good enough to sell by word-of-mouth alone then it is probably crap.
Advertisers have ZERO respect for your time or money so why do we put up with this shit??
> No idea if your code examples are formatted wrong
They are formatted correctly.
> and no idea what a stride is, either
Array Stride
The fact that you don't even know what a Stride of an Arrayis shows you don't know what the fuck you are talking about.
> Click a random page on Amazone .... there wont be anything in the processor cache of the machine handling your request, relevant to your request.
Total nonsense.
We are not talking about Amazone (2000)
A good programmer realizes that CPU optimization is only ONE part of optimization.
* Network
* Disk (Seek time, throughput, latency)
* Memory (Cache)
* CPU (Registers)
Listed from slowest to fastest -- the slowest access will have the biggest impact if we optimize for that.
> optimizing for the cash made
Maybe you try optimizing for the instruction cache and the data cache instead.
> was the early 1990s.
That would explain why you don't know what the fuck you are talking about. Your optimization information is 20+ years out of date.
Here's a modern clue stick:
Pitfalls of Object Oriented Programming
You are completely missing the point.
A smart programmer would NOT waste their time optimizing with bit-twiddling when a smarter algorithm would do the job.
However, using a better algorithm is INSUFFICIENT when you are trying to optimizing for speed. You MUST use Data-Orientated Design if you want maximum performance. Simply knowing which algorithm is better will NOT do it. Why?
Because HOW you access the cache lines matter.
Two algorithms could have the exact O() performance --in theory, but in practice have DRASTICALLY different real-world performance.
Sum of LEN numbers, stride: 0
for( size_t i = 0; i < LEN; i++ )
nSum += pData[ i ];
Sum LEN numbers; stride: constant
for( size_t col = 0; col < STRIDE; col++ )
for( size_t i = 0; i < LEN; i += STRIDE )
nSum += pData[ i + col ];
Summing 1,073,741,824 elements (int32) for a total size of 4GB, have these BOTH showing O(n) performance (in theory) but with a stride of 32 the latter is seven times slower (in practice)
BOTH theory and implementation are important.
People who say that only theory matters are retards.
^^ THIS.
Only a complete noob is using a GPU to mine.
* Mining Hardware Comparison
* Non-specialized Hardware comparison
I never said learning theory didn't matter. Of course theory matters.
But implementation details ALSO matter.
Try actual reading instead of an ignorant knee-jerk reaction.
> Edsger W. Dijkstra ... in fact shouldn't, involve actual computers at all.
Computer Science is an applied science. .
Dijkstra was an idiot who thought that only theory should be taught.
* In theory performance shouldn't matter
* In practice it does.
Implementation details do matter regardless of many fucking cluesless profs try to handwave them. For example, how do you sort your data when it fit into available RAM? There is a reason why Map Reduce was invented.
Focus solely on theory is the wrong approach. There are 3 types of optimizations that a programmer needs to understand.
1. Micro-optimization: Bit-Twiddling
I.e.
https://graphics.stanford.edu/...
2. Algorithmic
Spending time to optimize a bubble sort is a complete waste of time when you could use mergesort, quicksort, etc.
3. Macro-optimization (or cache-orientated) aka (Data-Orientated Design)
Techniques such as Memoization exist for a reason.
A good programmer learns HOW to optimize. i.e.
Code Clinic 2015: How to Write Code the Compiler Can Actually Optimize
Ignoring optimization doesn't make it go away. That's how we end up with bloated crap where a user is forced to download a 50 MB file for a bloody printer driver.
Get off your fucking high horse already and stop being a cultural snob. No one died and made you king.
Logicomix: An epic search for truth is 352 pages.
Watchman is 448 pages
No one gives a fuck how long a graphic novel is -- only if they were entertained.
Let me guess, you were probably one of those snobs who thought "talkies" (talking movies) were ruining movies via a focus on dialogue would subvert the unique aesthetic virtues of soundless cinema.
You condemn yourself with your ignorance.
Could have been worse -- Canadian Beaver
The Nightmare Fuel (in Spirited Away) is pretty weird too.
Maybe Hayao Miyazaki was on drugs at the time. Who knows? Either way, pretty crazy shit and I'm saying that as a fan. Beautiful movie.
Thanks for tracking down that Open Source Minecraft "promise" from Notch. I remembered reading it but couldn't remember where.
Yeah, I guess everyone found what it took for Notch to sell out his values: $2.5 Billion. (It's almost as if when Microsoft asked Notch how much he wanted for Mojang he probably jokingly said: 2^31. MS thought about it and replied "How about we round it up for an cool $2.5B?" But again MS is a dumb-ass company that paid $8.5 Billion for Skype.)
Ironically, Notch is not sleeping fine -- probably because he recognizes he's a sell out -- he lived the American (software) dream: Started a company and got bought out by a bigger one for an ungodly amount of money. And now he realizes that materialism is shallow. Go figure.
http://www.wired.co.uk/article...
The sad part is Minecraft has only scratched 1% of its potential yet Mojang doesn't really seem to understand what the community wants:
* A modding API (people are sick and tired of de-compiling the Minecraft source code that breaks mods *every* version)
* Depth of items and crafting. It is ironic that Terraria (pardon the pun) has way more depth, both literally, and figuratively with crafting then Minecraft.
Big Fucking Deal.
Where is the ability to have different colored wooden chests? Or barrels?
I'm surprised someone hasn't made a Terraria + Minecraft clone yet.
/sarcasm But this time it will be different!
*crickets*
No it won't.
You are spot on that people use x86 / Windows to run existing desktop apps. Unless ALL their software is ported over (unlikely) there is just too much momentum to switch.
> If you're Notch, and invented a new game genre
Notch did NOT invent a new genre; he even admitted he blatantly ripped off Infiminer
Zach, the creater of Infiminer, said this about Minecraft:
I agree 100% that Apple has completely lost their way W.R.T. computer hardware.
When they:
* put out a new computer that is slower the previous model (the Mac Mini fiasco)
* make it impossible to upgrade RAM and SSD because they are soldered in (WTF!?), etc.
you quickly realize Apple is all about streamlining their products at the expense of versatility.
Seriously, who the fuck would buy an Mac Pro -- when the hardware is so over-priced and out-dated it isn't even funny -- so why even continue selling it?
They are out of touch with what geeks want.
* Make gamepad support a 1st class citizen on the iPad / iPhone because some touch SUCKs for certain games
* Allow me to hook up a wired (*gasp*!) keyboard and monitor to the iPad
* Allow me to listen AND charge my phone at the SAME time -- like they fucking used to.
* Provide a fucking MicroSD ram slot
* Allow me to use my iPhone / iPad as a generic storage device
Instead we get "consume only" devices with a hard-one for everything wireless. Wireless is great -- but give me the OPTION of wired.
Apple only needs to do one thing to regain their focus:
How about making it _easy_ to _create_ content -- you know, like what the Mac became originally famous for.
Now this I definitely could see as being a/the problem!
Maybe a quick test would be to rename the Windows ARK executable and compare the visual output with the vanilla filename? If they differ then you know that the Windows Driver is in "automatic run-time patch mode" for the game that isn't getting applied on the Linux side.
Another test would be to do single "frame capture" -- dump all the D3D calls and see how they are getting translated to the OpenGL backend. That would help narrow down where the problem is.
It sucks that of the 4 companies involved: Microsoft (D3D), Epic, nVidia, Studio Wildcard -- that no one wants to take the time to figure this shit out.
Agreed 100%!
Originally, corporations could NOT own other corporations but I'm not sure what year that got hijacked.
This whole "I want to reap the benefits of a company but not have _any_ responsibility for when they are liable" has gotten WAY out of hand.
The fact that corporations are treated like people in the eyes of the Law just makes things worse.
Some interesting reading:
https://www.npr.org/2014/07/28...
http://reclaimdemocracy.org/co...
https://consumerist.com/2014/0...