| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Help Conquer Cancer
Thread: Faster crunching |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 26
|
|
| Author |
|
|
Ingleside
Veteran Cruncher Norway Joined: Nov 19, 2005 Post Count: 974 Status: Offline Project Badges: 
|
I can't help wondering, if equivalent work went into speeding up the Linux work units that is going into the Windows units, just how fast would they be? If the changes is just a switch from a mediocre Windows-compiler to a Windows-compiler that is much better at generating highly optimized code, it's not certain anything at all can be gained for Linux...  "I make so many mistakes. But then just think of all the mistakes I don't make, although I might." |
||
|
|
Dark Angel
Veteran Cruncher Australia Joined: Nov 11, 2005 Post Count: 728 Status: Offline Project Badges:
|
If it's just a switch to a better compiler then you'd be fairly right. If work has gone into the code to make better use of Windows particular memory management systems and processor scheduling, for example, then the same optimisation techniques could be applied in a Linux context. It really depends on exactly what is being done.
---------------------------------------- Currently being moderated under false pretences |
||
|
|
armstrdj
Former World Community Grid Tech Joined: Oct 21, 2004 Post Count: 695 Status: Offline Project Badges: 
|
The perfrormance improvements are due to a compiler upgrade on Windows. This should make windows closer to equal with Linux.
Thanks, armstrdj |
||
|
|
Dark Angel
Veteran Cruncher Australia Joined: Nov 11, 2005 Post Count: 728 Status: Offline Project Badges:
|
So this will sort out the credit issue or will it make it substantially worse?
----------------------------------------Currently being moderated under false pretences |
||
|
|
TimAndHedy
Senior Cruncher Joined: Jan 27, 2009 Post Count: 267 Status: Offline Project Badges:
|
The perfrormance improvements are due to a compiler upgrade on Windows. This should make windows closer to equal with Linux. Makes you wonder how after 20+ years of x86 the compiler could suddenly get 40% faster. Is there really that big of a difference. You are using the microsoft compiler, correct? Nothing else changed? |
||
|
|
Dark Angel
Veteran Cruncher Australia Joined: Nov 11, 2005 Post Count: 728 Status: Offline Project Badges:
|
In all fairness the x86 compiler has had to update many times and include instruction sets for newer processors that never existed in the early models.
----------------------------------------Currently being moderated under false pretences |
||
|
|
TimAndHedy
Senior Cruncher Joined: Jan 27, 2009 Post Count: 267 Status: Offline Project Badges:
|
Yes but the lowest common denominator factor still applies.
I doubt they are using any new instructions for this. |
||
|
|
Dark Angel
Veteran Cruncher Australia Joined: Nov 11, 2005 Post Count: 728 Status: Offline Project Badges:
|
Well the lowest common denominator for a x86 compiler would be an 8086 processor, which I suspect won't run anything currently on the project list. For that matter I suspect we can rule out 2, 3 and 486 processors as well so that cuts out a few more. On that basis alone we can assume that the compiler should be running nothing less than full 32bit addressing and instruction sets without needing to allow for 8 or 16bit instructions and addressing ... which Windows still includes. The lowest common denominator here is a lot lower than you might think. Updating the compilers has the potential to remove compatibility with quite a number of older processors and OS versions. As an example, I have a Pentium Pro 200MHz machine running Windows 2000 here that returned a HCC unit in the last six months (just for a laugh. It took close to a week to finish it.) Removing compatibility (by including things like SSE2,3,4, larger memory addressing etc) for processors and OSs of that age isn't going to stop a lot of people from contributing but will allow newer machines to use more powerful instruction sets and thus speed up calculations substantially.
----------------------------------------Currently being moderated under false pretences |
||
|
|
TimAndHedy
Senior Cruncher Joined: Jan 27, 2009 Post Count: 267 Status: Offline Project Badges:
|
I do wish they would have two versions of apps.
1. Current support everything in the last 20 years. 2. Support only systems built in the last 3-5 years. A lot of power is wasted and I hate wasting potential, especially for something like this that runs 24/7 on my systems. |
||
|
|
KWSN - A Shrubbery
Master Cruncher Joined: Jan 8, 2006 Post Count: 1585 Status: Offline |
I understand your wanting support for older systems, but realistically speaking, it's just not worth the time of the coders or the scientists. One i7 replaces somewhere between the equivalent of 8 and 25 systems that are older than 5 years.
----------------------------------------In a perfect world with unlimited resources, we could have apps for everything. But then in a perfect world, we wouldn't need to run this project.  Distributed computing volunteer since September 27, 2000 |
||
|
|
|