New Computer for Win7 and Revit

[Anonymous]

OK, I'll ask this again another way, maybe making it fun this time...

if you were going to buy a new computer that would run 64bit Win7 Pro,
Acad 2010, and Revit and *cost was not an object* what would you buy?

I've spent almost two weeks now doing research and as each day goes by I
become more and more confused about what to do. I'm a drafter not a
computer tech. I need some help here please
thanks,
Dave
DDP

[Anonymous]

Hi Dave,

i have used the Cadalyst 2008 AutoCAD testing routine for some time to
compare different CPUs and graphics cards. I have only had limited
access to an i7 machine 940 2.66 GHz CPU, but it was substantially
slower than my P5 E8600 3.3GHz CPU - by about the same ratio as the CPU
speed ratio. I therefore conclude that the inherent performance of each
core in the i7 series is much the same as the dual core P5 series. The
fact that the i7 has eight cores does not help AutoCAD over and above 2
cores except for rendering. At the moment the pricing for i7 processors
is very high - especially the higher speed models.

[Anonymous]

George Drayton wrote:
> Hi Dave,
>
> i have used the Cadalyst 2008 AutoCAD testing routine for some time to
> compare different CPUs and graphics cards. I have only had limited
> access to an i7 machine 940 2.66 GHz CPU, but it was substantially
> slower than my P5 E8600 3.3GHz CPU - by about the same ratio as the CPU
> speed ratio. I therefore conclude that the inherent performance of each
> core in the i7 series is much the same as the dual core P5 series. The
> fact that the i7 has eight cores does not help AutoCAD over and above 2
> cores except for rendering. At the moment the pricing for i7 processors
> is very high - especially the higher speed models.

thanks for the info George, that's helpful
Dave

[Anonymous]

George,

That's rather weird. In every test I've seen, the i7 trounces all others with
extreme prejudice. Even with the clock difference, I would expect the results to
be at least very close.

When you ran your CADalyst benchmark, did you ensure that Vertical Sync was
disabled on both machines? That has a HUGE impact on performance in this
particular benchmark.

Also, was amount of RAM the same on both? Same OS? Same version of AutoCAD?


Matt
matt@stachoni.com


On Sat, 7 Nov 2009 03:41:14 +0000, George Drayton
wrote:

>Hi Dave,
>
>i have used the Cadalyst 2008 AutoCAD testing routine for some time to
>compare different CPUs and graphics cards. I have only had limited
>access to an i7 machine 940 2.66 GHz CPU, but it was substantially
>slower than my P5 E8600 3.3GHz CPU - by about the same ratio as the CPU
>speed ratio. I therefore conclude that the inherent performance of each
>core in the i7 series is much the same as the dual core P5 series. The
>fact that the i7 has eight cores does not help AutoCAD over and above 2
>cores except for rendering. At the moment the pricing for i7 processors
>is very high - especially the higher speed models.

[sssteve72]

I have not run the cadalyst tool that George talked about but I did just run the passmark benchmark on both my computer here at the office and the new machine I built at home. My machine here at the office is 2 years old (i think). I did run my machine on Windows 7 64bit for a while during the beta phase and it seemed to run fine to me. I don't use CAD at home so as far as CAD goes I don't have experience but the benchmark numbers I find interesting. While the Xeon is notably an older model this still falls inline with other benchmarks I have seen.

http://www.cpubenchmark.net/ <-- has a listing of tested benchmarks with its software. And that is the software I used to test.

Here are the two machines:

Dell 690 with XP pro SP3 32bit
Xeon 5160 @ 3.0Ghz
4gig of Ram DDR2 PC5200


Asus P6T Deluxe with XP Pro SP3 32bit (capable of 64bit though not using at the moment)
i7920 @ 2.67Ghz
3gig of Ram DDR3 PC3 10666

[Anonymous]

I'm pretty sure that i did but it was a while ago.
--
======================================================
George Drayton CD-CAD Ltd Christchurch New Zealand.

[Anonymous]

I downloaded this program and it looks quite good, but there is still
the question - how does the computer's actual AutoCAD performance stand
up. I have never found the synthetic testing programs measure
parameters that match with AutoCAD performance. AutoCAD is very
dependent on double-precision floating point calculation performance and
this parammeter influences real life AutoCAD performance. This process
is hardly used by common applications so the weighting of a performance
benchmark will general not match parameters of interest to AutoCAD - in
my view.
As a consequence I tend to value the Cadalyst benchmark although I tend
just to look at the total elapsed time for the tests rather than
individual parameters.
--
======================================================
George Drayton CD-CAD Ltd Christchurch New Zealand.

[sssteve72]

I don't know if it is the same or not but the Passmark Benchmark does give a specific result for floating point math.

I agree with you though. The Cadalyst benchmark is probably the way to go and a total elapsed time is probably what I would look at as well.