.NET 4 Rocks!
-
.NET4 rocks! I had this impression but now I'm just upgrading an old performance intensive demo from .NET 1 => .NET 2 => .NET 3 => .NET 4 => .NET4 + Parrallel it takes 40% time less when I change the runtime from 3 to 4!!! Yoohoo! Now trying parallel extension.. it's more tricky, sometimes it's slow things down!!! Investigating....
A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station.... _________________________________________________________ My programs never have bugs, they just develop random features.
-
.NET4 rocks! I had this impression but now I'm just upgrading an old performance intensive demo from .NET 1 => .NET 2 => .NET 3 => .NET 4 => .NET4 + Parrallel it takes 40% time less when I change the runtime from 3 to 4!!! Yoohoo! Now trying parallel extension.. it's more tricky, sometimes it's slow things down!!! Investigating....
A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station.... _________________________________________________________ My programs never have bugs, they just develop random features.
We found some of the issues when one loop is waiting on the processing from another loop. It seems to slow things right down. Apart from that, .NET 4 is a vastly improved product - partly, I suspect, down to the pared down runtime.
I have CDO, it's OCD with the letters in the right order; just as they ruddy well should be
Forgive your enemies - it messes with their heads
-
.NET4 rocks! I had this impression but now I'm just upgrading an old performance intensive demo from .NET 1 => .NET 2 => .NET 3 => .NET 4 => .NET4 + Parrallel it takes 40% time less when I change the runtime from 3 to 4!!! Yoohoo! Now trying parallel extension.. it's more tricky, sometimes it's slow things down!!! Investigating....
A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station.... _________________________________________________________ My programs never have bugs, they just develop random features.
never underestimate the powers of pure C++ ... in general 40 to 50 % faster than any runtime.
-
never underestimate the powers of pure C++ ... in general 40 to 50 % faster than any runtime.
You make a statement like that, you've got to be able to prove it. Go on, back it up.
I have CDO, it's OCD with the letters in the right order; just as they ruddy well should be
Forgive your enemies - it messes with their heads
-
never underestimate the powers of pure C++ ... in general 40 to 50 % faster than any runtime.
yes, I was often told! but it doesn't appeal to me much... I also thought of having just a small module with just the algorithm in C++, but the deployment problem haunt me! (you know, I need a 32bit, 64bit version, no version for Silverlight, multiple installer, etc...) + plus if I plug external C# function in the middle of it, I suspect it will slow thing down, interop tend to do that! :(
A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station.... _________________________________________________________ My programs never have bugs, they just develop random features.
-
never underestimate the powers of pure C++ ... in general 40 to 50 % faster than any runtime.
federico.strati wrote:
never underestimate the powers of pure C++ ... in general 40 to 50 % faster than any runtime.
At two syllables rather than three, C# is 33% faster than C++ just in terms of saying the name. That's got to count for something? Right? -Rd
-
never underestimate the powers of pure C++ ... in general 40 to 50 % faster than any runtime.
I agree with Mr O'Hanlon; Lets see a code samples. ;)
Simon
-
I agree with Mr O'Hanlon; Lets see a code samples. ;)
Simon
Exactly what I thought... it is the beginning of a flame war... Anyway, managed code is inherently slower than unmanaged. You got other benefits, like platform indipendence and so on... Too tired to present an example, but I suspect serialisation would be an excellent candidate.
-
never underestimate the powers of pure C++ ... in general 40 to 50 % faster than any runtime.
You're right as long as MFC stays away from us.
-
You make a statement like that, you've got to be able to prove it. Go on, back it up.
I have CDO, it's OCD with the letters in the right order; just as they ruddy well should be
Forgive your enemies - it messes with their heads
-
never underestimate the powers of pure C++ ... in general 40 to 50 % faster than any runtime.
In a lot of usages, pure speed is not the only deciding factor. Assuming your ascertation is more than simple hypothesis, let us consider for one moment the advantages of ‘slower’ managed code. 0. Memory is, on the whole, MANAGED for you. Thus the reliance on the coder getting it right is removed. BIG bonus time. 1. Development time is, I am sorry to say, quicker using C# [for example] than C++ 2. Developers are cheaper for modern managed languages over older unmanaged platforms. So, from these simple three points you can see why managed not-det environment is preferred. The customer – that’s the guy paying for the work – get’s more bangs for their buck and doesn’t notice that the response is 200ms against 120ms. To clarify my position, we use TCL scripting, C#, C++ and Java all in one huge big mofu application. Depending on what is being done, different tools are chosen for different jobs. For desktop UI, C# rocks. For processing 10 kt/m C++ is great. For site modifiable interactions and configuration TCL is the Daddy. To keep the Hippies happy Java is the way man.
Panic, Chaos, Destruction. My work here is done. or "Drink. Get drunk. Fall over." - P O'H
-
never underestimate the powers of pure C++ ... in general 40 to 50 % faster than any runtime.
Granted, 'managed' means some additional overhead which good old C or C++ avoid. But 40 - 50 %? First you must take a look at your program and identify the parts where the CPU is really put to work. And already 90% or more of your code have become totally uninteresting. After identifying the critical parts of your program, you need to take a good look at your algorithms at those locations. A poorly chosen algorithm can kill the performance, totally independent of the programming language, the optimizations the compiler may perform or the capabilities of the actual hardware. In C++ it has become unusual to optimize the code for a certain hardware. Nowadays there are just too many different CPUs, so you can't really know what awaits your program at runtime. Managed applications may actually have an advantage here since they are compiled just before execution. So where is this guaranteed performance boost in C++ supposed to come from? P.S: I still like working in C++ and also have done a bit of optimizing there, but I simply don't see that vast advantage in performance. You have to work hard to get it and with similar treatment a managed language may produce similar results.
A while ago he asked me what he should have printed on my business cards. I said 'Wizard'. I read books which nobody else understand. Then I do something which nobody understands. After that the computer does something which nobody understands. When asked, I say things about the results which nobody understand. But everybody expects miracles from me on a regular basis. Looks to me like the classical definition of a wizard.
-
never underestimate the powers of pure C++ ... in general 40 to 50 % faster than any runtime.
-
Not really no - it goes part of the way, but I'd like to see it with .NET 4 which is much better performing than 3.5 due to the leaner client framework. Certainly, on Windows the difference between C# and Visual C++ is nowhere near the 40-50% difference claimed by the OP.
I have CDO, it's OCD with the letters in the right order; just as they ruddy well should be
Forgive your enemies - it messes with their heads
-
Thanks, very interesting data. On Windows "C++ is %15.8 faster than .Net" On Linux "C++ is 22.% faster than Java" [Remember these are benchmarks over very specific high CPU usage problem sets. These are not general statements] Not quite the 40-50% originally said, but certainly interesting none the less. I'd be really interested to see the same tests repeated with .net 4.
Simon
-
Exactly what I thought... it is the beginning of a flame war... Anyway, managed code is inherently slower than unmanaged. You got other benefits, like platform indipendence and so on... Too tired to present an example, but I suspect serialisation would be an excellent candidate.
federico.strati wrote:
Exactly what I thought... it is the beginning of a flame war...
No, it's not. You made an unsubstantiated comment - I'd be more inclined to listen if you actually backed it up rather than making a random statement and then run off when challenged to provide proof. Don't present anecdotal evidence - be empirical. I just get sick and tired of language zealots claiming that language A is better in than language B without offering any qualifying evidence.
I have CDO, it's OCD with the letters in the right order; just as they ruddy well should be
Forgive your enemies - it messes with their heads
-
-
The test should be done against .NET4, which is way faster than .NET3, whereas C++ is only a little bit faster than .NET3! ;)
A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station.... _________________________________________________________ My programs never have bugs, they just develop random features.
-
federico.strati wrote:
Exactly what I thought... it is the beginning of a flame war...
No, it's not. You made an unsubstantiated comment - I'd be more inclined to listen if you actually backed it up rather than making a random statement and then run off when challenged to provide proof. Don't present anecdotal evidence - be empirical. I just get sick and tired of language zealots claiming that language A is better in than language B without offering any qualifying evidence.
I have CDO, it's OCD with the letters in the right order; just as they ruddy well should be
Forgive your enemies - it messes with their heads
Pete O'Hanlon wrote:
I just get sick and tired of language zealots claiming that
0is better in than1without offering any qualifying evidenceftfy
Panic, Chaos, Destruction. My work here is done. or "Drink. Get drunk. Fall over." - P O'H
-
Not really no - it goes part of the way, but I'd like to see it with .NET 4 which is much better performing than 3.5 due to the leaner client framework. Certainly, on Windows the difference between C# and Visual C++ is nowhere near the 40-50% difference claimed by the OP.
I have CDO, it's OCD with the letters in the right order; just as they ruddy well should be
Forgive your enemies - it messes with their heads
It's interesting that Mono appears to be quicker than .NET - 3.5 at least.