Which is Faster for Solidworks: Vista or XP?

The debate rages on as to which operating system is better.  Well since we all use Solidworks, if we focus on performance for just that, which is faster?  Well over on the Solidworks forums, Mike Meuse posted a tale of his experiment to directly answer that question.  I think you may be suprized at his results.

Link to the original post: Battel of the operating systems

Quoted text of post:

Well, we are testing Vista 64 performance and Solidworks 64 performance for upcoming machine purchases. So as Engineering Manager I was the guinia pig (I don't do any "real" work anyway). I have been playing with Vista 64 Buisness and Solidworks 64 for a while with some minor issues. Mostly my issues are graphics related, a lag for the toolbar icons to respond when the mouse hovers over them but, only in large assemblies (1500 + parts). Otherwise it seems to run well. Before I get into the testing I will comment I like Vista 64, it performs well and has been very stable. Solidworks 64 also has been very stable, I have had fewer crashes running the 64 bit Solidworks. On to the testing...

So with all the comments on how slow Vista is I needed to verify for myself. THe first problem is my Dell Precision M90 was not entirely 64 bit ready....I needed a new processor so I purchased and installed a new Intel T7400 Core 2 Duo, 2.16 GHz 4 MB L2 cache. I chose this CPU because another M90 in the building has that cpu and I wanted to compare performance. I also purchased a new hard drive for the install so I could readilly switch back to XP any time I wanted. My first impressions after getting all the software loaded on was mostly favorable. I only have 1 program that has issues with Vista 64 (Microsoft Visual Studio 2003, figures Microsoft wouldn't build their own product compatible). My initial impression running Solidworks 64 is it appears pretty fast (once I corrected a minor configuration error). But still how does it really compare so on to the testing. I chose 2 types of tests the "Hole Punch" test and the SPEC.org Solidworks 2007 benchmark. I used the hole punch test because I saw it used as a performance measure here in the forums but, I felt it did not completely test the system performance. The 3 systems used:

1. Precision M90 - Vista 64 Buisness
Intel T7400 Core 2 Duo 2.16 Ghz 4 MB Cache
4 GB Ram
Nvidia FX 2500M 512 MB video card
250 GB 5200 rpm SATA HDD
Solidworks 64 SP 3.1

2. Precision M90 - XP 32 Pro
Intel T7400 Core 2 Duo 2.16 Ghz 4 MB Cache
4 GB Ram
Nvidia FX 2500M 512 MB video card
80 GB 5200 rpm SATA HDD
Solidworks 32 SP3.1

3. Precision M90 - XP 32 Pro
Intel T7400 Core 2 Duo 2.16 Ghz 4 MB Cache
2 GB Ram
Nvidia FX 2500M 512 MB video card
120 GB 5200 rpm SATA HDD
Solidworks 32 SP3.1

Performance testing:

System 1

Hole Punch Test 150.4 sec. (Without all the Vista Bling running)
Hole Punch Test 152.1 sec (With all Vista Bling running)
Solidworks Benchmark 2007 226.32 sec (Without Vista Bling running)

Benchmark breakdown:

Graphics 75.53 sec
CPU 63.77 sec
I/O 83.68 sec
Small case 8.14 sec
Tire 6.88
GTX assem 4.25 sec
GTX Transparent 9.36 sec
Engine 13.48 sec

System 2

Hole Punch Test 136.6 sec.
Solidworks Benchmark 2007 252.48 sec

Benchmark breakdown:

Graphics 94.75 sec
CPU 63.86 sec
I/O 93.87 sec
Small case 8.14 sec
Tire 5.63 sec
GTX assem 18.61 sec
GTX Transparent 18.57 sec
Engine 16.26 sec

System 3

Hole Punch Test 143.7 sec.
Solidworks Benchmark 2007 247.84 sec

Benchmark breakdown:

Graphics 84.52 sec
CPU 65.68 sec
I/O 97.64 sec
Small case 5.73 sec
Tire 3.77 sec
GTX assem 18.20 sec
GTX Transparent 18.18 sec
Engine 15.86 sec

Comments:

1. I am surprised by the differences between the hole punch test and the Benchmark test.
2. Vista 64 with Solidworks 64 did come out on top
3. The tests between system 2 and 3 surprised me a bit also. I ran the test a second time to make sure I did not make a mistake.
4. I did also run for comparison a Dell Precision 670 workstation with dual Zeons. But because of the poor performance I left it out.


Mike

Comments

[sikiş izle]

Graphics 75.53 sec
CPU 63.77 sec
I/O 83.68 sec
Small case 8.14 sec
Tire 6.88
GTX assem 4.25 sec
GTX Transparent 9.36 sec
Engine 13.48 sec

[sikiş izle]

1. I am surprised by the differences between the hole punch test and the Benchmark test.
2. Vista 64 with Solidworks 64 did come out on top
3. The tests between system 2 and 3 surprised me a bit also. I ran the test a second time to make sure I did not make a mistake

[Luciano]

I already knew it, but whatever thanks.
http://www.rapidsharemix.com

[cheap computers]

It performs well and has been very stable.

[Guest]

As to this test - the net results were close enough that the differences were practically irrelevant - and probably related more to the different hard drives used (80/120/250).

Hard drive performance is very important (since next to the user devices and removeable media it is typically the slowest element) - only *identical* hardware gives a valid comparison. Even same size/RPM drive may have dramatically different seek/read/write speeds and internal caching.

In Vista, flash drives can be used as hard-drive cache - this may make a significant difference (note: even though model is in memory - the OS is already swapping things to disk).

More memory actually often *slows* down a system - when unused (typical) the system still has to juggle the memory (as the memory is not associative). This can lead to huge slow downs (I've seen 4~5x performance improvements by *removing* excess identical RAM).

64-bits ability to *directly* access >4GB is not its main advantage (in fact can cause slow downs) - its the instruction set itself. Unfortunately, to realy see this advantage code must be properly structured (highly unsual, though compilers may optimize - an experienced Assembly programmer is required - also unusual).

[John Howard]

My new system (VistaBiz64/Intel Core i7/SW2009 SP3.0) has a compatibility problem with Excel--Excel stops working as the amount of SW files/history gets larger. This crashes the Tables functions embedded Excel apps, as well as preventing simultaneous use of Excel. Any suggestions?

[francis]

Hi i have a problem with solidwork 2008 or windows xpsp2?

In my window xp sp2 i have the configrations of intel(celeron)cpu with 430@1.81GHz processer,2.49 GB of ram,is this configration is ok to use it?

My problem is when is use solidworks 2008 it is very slow because of this configration or some oyher problems?

please replay me its very urgent........

regards

Francis

[Mike Meuse]

So, what have I learned from this experiment? First, be more concerned about spelling in my forum posts (lol). Next, based on data collected on various systems the O.S. is not where the performance is gained or lost. Improvement in performance gains start to decrease as you increase the RAM (e.g. the increase in performance between 2 GB and 4 GB is greater than the performance gain between 4 GB and 8 GB RAM). The largest gains in performance were made with CPUs and Graphics cards. The newer CPUs and the higher end Quadro Graphics cards made the greatest gains.

Mike Meuse

[that_guy]

isn't this a bit of an apples/oranges kind of comparison? The one little caveat of the whole 64 vs. 32 question is that there is no comparable difference between the same hardware running different OSs. If anything, we've all learned that 64 has more issues due to driver support and inherently they offer slightly lower performance due to the memory registers.

That being said, I would expect a 32/64 with same hardware and 4 GB to have comparable performance, the true advantage of 64 bit comes at greater than 4 GB or RAM where 32 bit systems can't play.

Perhaps the better comparison is 32 XP vs 32 Vista or 64 XP vs. 64 Vista?