Windows XP is still king

[Mcinwwl]

7 minutes ago, NoelC said:

I always thought that the OS is spreading the work out on multiple cores, assigning threads to different logical processors when the kernel gets control during a system call - not some automatic processor feature. 

I just wanted to ask about the same.

46 minutes ago, FranceBB said:

@mciwnn... Intel CPUs have the ability to scale and redistribute the work on other cores automatically, which is why, if you run a single thread application on an Intel CPU, you will see every core throttling between 60 and 76%, while on AMD you will see Core 0 at 100%, and the other at 1-2℅.

I hadn't been expecting application, that's explicitly told to use single core to run on multiple - that's why I did it back then. If you say taht's possible, It's cool, but my experience looks more like Noel's.

[FranceBB]

@NoelC and @Mcinwwl yes and no: AMD Bulldozer's architecture uses Clustered Multithreading and has these things called "clusters" which it uses to process stuff. These clusters have two integer (aka ALU) units, one floating point unit (aka FPU), and share an execution engine (aka EX, the "do stuff" part). This basically makes the "cluster" (aka "core") equivalent to a dual-core processor in integer math, and a single-core processor in floating-point math. Now, having double the APUs is great for heavily multithreaded applications and there is a measurable advantage in certain types of applications. Anyway, the Bulldozer "clusters" share FPUs and L2 caches, and this causes single-threads to process slower since they have to "wait" for shared resources (aka serial, or in-order, processing). Hence, Bulldozer "clusters" have slower single-threaded performance as they get stuck in the queue. The Nehalem (Intel) architecture and its children use Symmetric Multithreading, which uses two identical logical processors per "core" - similar to an AMD "cluster," but with 100% of the same resources available to both processors. Each "core" has the equivalent of dual-core APU and dual-core FPU resources, and also shares an execution engine. But, since the processors in the "core" do not share resources in the same way as Bulldozer, it doesn't get stuck waiting for stuff to do. Let's suppose you have an 8 core/cluster AMD and a 4 core/8thread Intel. Your operating system really only sees 8 logical processors when it has to assign threads, so it assigns 1 per cluster, even though the cluster can really do two threads (if they're integer math). When things get stuck, this pretty much makes each "cluster" half the FPU performance of an equivalent Intel "core" and reduces the ALU performance advantage. Intel is faster per "core" in single-thread performance than AMD "clusters" due to their architectural differences. Additionally, Intel's improvements to their execution engines and resource scheduling have caused their 4 "core" (8 thread) processors to eliminate the advantage of 8 "cluster" processors in multi-threaded performance. Fortunately, AMD is abandoning clustered multi-threading in favour of simultaneous multi-threading (like HyperThreading on Intel). In other words, AMD processors have to "wait a lot" for shared resources in single-threaded applications, Intel processors don't, and even though threads are assigned by the OS, the threads a CPU shows to the OS and the way it shares resources across cores are a key point for single thread application. That's pretty much it and sorry for the wall of text. 

[Dibya]

MY msi Skylake board was blown now planning to move kabylake i7 7700k 4.5Ghz let see how Xp react with Kabylake . how it can react you guys think ?

One more great piece of  news: i got intel usb 3.0 driver work .

also implimented added a new function

int __stdcall IdnToAscii(int a1, int a2, int a3, int a4, int a5)
{
  int v5; // eax@2
  DWORD v7; // ST10_4@7

  if ( a1 & 0xFFFFFFFC )
  {
    SetLastError(0x3ECu);
  }
  else
  {
    v5 = RtlIdnToAscii(a1, a2, a3, a4, &a5);
    if ( v5 >= 0 )
      return a5;
    v7 = RtlNtStatusToDosError(v5);
    SetLastError(v7);
  }
  return 0;
}

[NoelC]

The only thing I was going to mention was the USB 3 compatibility issue, but it sounds like you're past that.

I use the PassMark PerformanceTest database to gauge what processors are giving the best performance in real systems.  For example:

https://www.cpubenchmark.net/high_end_cpus.html

The above page shows the i7-7700k is no slouch, scoring 12,277.  Xeons do better, probably because they're more optimized more for multi-processing with e.g. bigger caches.

-Noel

[Dibya]

8 hours ago, NoelC said:

The only thing I was going to mention was the USB 3 compatibility issue, but it sounds like you're past that.

I use the PassMark PerformanceTest database to gauge what processors are giving the best performance in real systems.  For example:

https://www.cpubenchmark.net/high_end_cpus.html

The above page shows the i7-7700k is no slouch, scoring 12,277.  Xeons do better, probably because they're more optimized more for multi-processing with e.g. bigger caches.

-Noel

They are optimized for server . I cannot afford xeon , I am not a grown up guy so i need helps from my parents .

[Damnation]

how did you manage to get usb 3.0 to work?

[Dibya]

22 minutes ago, Damnation said:

how did you manage to get usb 3.0 to work?

putting missing functions inside their respective dll/exe/sys

[Damnation]

1 hour ago, Dibya said:

putting missing functions inside their respective dll/exe/sys

can i try this mod out for myself?

Edited February 7, 2017 by Damnation

[jumper]

IdnToAscii isn't missing (it's in Normaliz.dll, a redist.). To what DLL did you add it?
On the other hand, RtlIdnToAscii _is_ missing from Kernel32.dll in XP, so that code won't work anyway.

[Dibya]

4 hours ago, jumper said:

IdnToAscii isn't missing (it's in Normaliz.dll, a redist.). To what DLL did you add it?
On the other hand, RtlIdnToAscii _is_ missing from Kernel32.dll in XP, so that code won't work anyway.
 

RtlIdnToAscii added to ntdll then added a import to kernel32.dll so my function can call it .

7 hours ago, Damnation said:

can i try this mod out for myself?

Sorry Bro I donot want to maKe your system unstable . I will release once i all up with my patches . Wanna be a beta tester of ExtendedXP?

[Damnation]

3 hours ago, Dibya said:

RtlIdnToAscii added to ntdll then added a import to kernel32.dll so my function can call it .

Sorry Bro I donot want to maKe your system unstable . I will release once i all up with my patches . Wanna be a beta tester of ExtendedXP?

sure, I'd like to be a beta tester, where can i sign up?

[Dibya]

12 hours ago, Damnation said:

sure, I'd like to be a beta tester, where can i sign up?

I have no website . I will pm you when ready that's all

[Dibya]

this is what speaks about xp .

This is how we will keep microsoft from destroying xp.

 

Edited November 1, 2017 by Dibya

[sdfox7]

Windows XP market share rose again in October:

NetMarketShare: Windows 10 sees its slowest growth in months

Meanwhile, Windows 7 still holds half of the market with just over two years to go until support ends. 

I think Windows 7's sheer market share is going to force Microsoft's hand to extend Windows 7's support. If not, we will have a lot of unsupported corporate machines on the Internet come January 2020.

Edited November 1, 2017 by sdfox7

[Dibya]

On softpedia news and other sites where they write about xp . Previously I seen people going against xp and praising 7/10 . But now over months xp supporter are increasingly significantly . One 10 guy and all other xp one linux . Its awesome.