Dell M6400 Processor Choice

Discussion created by Peter Darby on Feb 4, 2009
Latest reply on Feb 5, 2009 by Christopher Mueller

Like • Show 0 Likes0
Comment • 7

Hi all,
I've been contemplating getting a mobile workstation. For better or worse I'm looking at Dell; good performance from them in the past. I'm after solid performance over a three year payback. And looking at getting a better graphics card or additional ram with money 'saved'.

What processor to select? Is there a bottleneck that lays hidden in the choice between cache ram speed/size, graphics card size or a disc speed that needs to be addressed to solid performance for the money spent?

The lower priced/spec'ed processors have a lower cache. Is that all there is in the decision? The easiest additional spend is the $130; the $380 not so bad; the $690 seems a bit steep.

intel processor summmary specs:
http://en.wikipedia.org/wiki/L...ore_2_microprocessors

Dell reference:
http://configure.us.dell.com/d...c=us&l=en&s=bsd&cs=04

Dell's options in processors are (Coretm 2 Duo ) in $USD:
P8400 (2.26GHz, 3M L2 Cache, 1067MHZ) Dual Core [Included in Price]
P8600 (2.40GHz, 3M L2 Cache, 1067MHZ) Dual Core [add $60]
P8700 (2.53GHz 3M L2 Cache, 1066MHz) Dual Core [add $60]
T9400 (2.53GHz, 6M L2 Cache, 1067MHZ) Dual Core [add $130]
T9550(2.66GHz 6M L2 Cache, 1066MHz) Dual Core [add $130]
T9600 (2.80GHz, 6M L2 Cache, 1067MHZ) Dual Core [add $380]
T9800 (2.93GHz 6M L2 Cache, 1066MHz) Dual Core [add $380]
X9100 (3.06GHz, 6M L2 Cache, 1067MHZ) Dual Core [add $690]
Quad QX9300 (2.53GHz, 12M L2 Cache, 1066MHz FSB) [add $890]

Realise this is a bit detailed, but if i'm going to commit to a high end mobile workstation I dont want to throw the money away and live with the consequences for three years.

Thanks in advance

Outcomes

7 Replies

Jim Zink Feb 4, 2009 1:25 PM
Get the fastest processor you can afford. Quad core if you do lots of rendering or FEAl, otherwise dual core. The CPU will have the greatest impact on overall performance. Your paying big bucks for the M6400 and it would be "penny wise, pound foolish" to cripple it with a slow CPU.

If cost is an issue, check out the M6300s on the Dell Outlet. The M6300 is a real workhorse and there are some terrific deals on refurbished units. They still come with a 3 year warranty.
Like • Show 0 Likes0
Actions
Christopher Mueller Feb 4, 2009 1:28 PM
I'll tell you which one I would pick, the T9550 or the T9400. Here is the reasoning:

The Quad gets booted for two reasons, cost and practicality. Most of the functions in Solidworks cannot take advantage of quad cores, so you are better off getting the best dual core you can afford, which would probably be the X9100, but I could't tell you for sure, and that one is expensive too.

The P series laptop processors are probably darn good, but are not typically what users on this site are putting in their workstations. (I believe they are more for home systems, not office).

The benchmark spreadsheet (click here) shows us that the T9300, T9500, and T9600 are all very good choices and get the rebuild completed in 99, 94, and 86 seconds. You should definitely splurge the extra 130 for a T processor, but I agree with you that you can lay some more money on graphics instead of getting the most expensive processor.

Also, I have an opinion about the "three year payback". If you do choose a cheaper workstation, then I think you can adjust your timeframe downward, like say 2 to 2.5 years. After all, if you "save" $600 on this purchase, can't you more easily motivate using that money you "saved" to make your next purchase happen a little sooner?

Good luck whatever you choose.
Like • Show 0 Likes0
Actions
- 1-HPZFJU @ Christopher Mueller on Feb 4, 2009 2:06 PM
  I ended up getting the x9100 with the fx2700 512meg video card. I'm running 32 bit xp with 4 megs of ram for now. As others have said, for now a fast dual core is the way to go. I wouldn't have gotten the x9100 over the T9800, but the funding was there. There's no real gain in just a 100 Mhz or so. Battery life sucks compared to lesser units, but that's the trade off. This sucker is HEAVY. The brick that comes with it is huge. Those are the negatives.
  
  The number pad is a gift from god, or some other lesser deity if you want to get technical. Others I work with have the 6300 and like it, but once you use the numberpad it's hard to use another laptop that doesn't have it.
  
  That said, my use for this machine is designing and programming for a five axis mill. The mill doesn't have network connections. I use this machine mostly at a docking station, then bring it to the mill with me to upload files after running them through Mastercam. Every MHz counts when generating toolpaths, so that drove the processor decision even more than solidworks did. This laptop allows me to edit without having to go back to the office about a quarter mile walk away. If you do most of your work at a dock, and just like to have the portable function for support roles, the 6300 might be a better deal if money is an issue.
  
  The following Dell link might be helpful if you have to justify your choice:
  http://www.dell.com/content/to...?c=us&cs=04&l=en&s=bsd
  Like • Show 0 Likes0
  Actions
  - 1-HPZFJU @ 1-HPZFJU on Feb 4, 2009 2:42 PM
    Following up earlier, I ran the Punch holder benchmark, sp 2009 sp1 Dell m6400 x9100
    
    Feature Statistics:
    
    Punch Holder, sw2009 2/4/2009 10:39:21 AM
    
    Features 36, Solids 1, Surfaces 0
    Total rebuild time in seconds: 80.24
    
    Time % Time(s) Feature Order
    
    59.77 47.95 LPattern5
    17.08 13.70 #8-32 Tapped Hole1
    6.91 5.55 LPattern4
    5.24 4.20 5/16-18 Tapped Hole1
    3.29 2.64 #10-32 Tapped Hole1
    1.73 1.39 Hole4
    1.69 1.36 3/32 (0.09375) Diameter Hole2
    0.60 0.48 1/4 (0.25) Diameter Hole1
    0.60 0.48 3/8 (0.375) Diameter Hole1
    0.56 0.45 11/32 (0.34375) Diameter Hole1
    0.56 0.45 7/32 (0.21875) Diameter Hole1
    0.25 0.20 Chamfer2
    0.23 0.19 Cut-Extrude5
    0.21 0.17 LPattern1
    0.19 0.16 Chamfer1
    0.16 0.13 Extrude1
    0.16 0.13 Hole2
    0.14 0.11 Sketch25
    0.10 0.08 5/32 (0.15625) Diameter Hole1
    0.06 0.05 3/16 (0.1875) Diameter Hole1
    0.06 0.05 Cut-Extrude2
    0.06 0.05 LPattern2
    0.04 0.03 Hole1
    0.04 0.03 Cut-Extrude6
    0.04 0.03 Sketch5
    0.04 0.03 Cut-Extrude1
    0.04 0.03 Sketch14
    0.04 0.03 Sketch1
    0.02 0.02 Hole3
    0.02 0.02 3/32 (0.09375) Diameter Hole1
    0.02 0.02 1/8 (0.125) Diameter Hole1
    0.02 0.01 Sketch17
    0.02 0.01 Cut-Extrude4
    0.00 0.00 Sketch20
    0.00 0.00 Sketch26
    0.00 0.00 Cut-Extrude3
    Like • Show 0 Likes0
    Actions
- Dale Dunn @ Christopher Mueller on Feb 4, 2009 2:11 PM
  
  Originally posted by: Christopher Mueller...
  Also, I have an opinion about the "three year payback". If you do choose a cheaper workstation, then I think you can adjust your timeframe downward, like say 2 to 2.5 years. After all, if you "save" $600 on this purchase, can't you more easily motivate using that money you "saved" to make your next purchase happen a little sooner?
  ...
  
  This has been on my mind for a while. When I hear opinions on how often a workstation should be replaced, I frequently hear after 3 years or after processor speed has doubled. Has anyone ever attempted a method of evaluation for how long a new workstation takes to pay for itself? A 3 year life aligns very roughly with platform refreshes, but may make no sense at all for computationally sensitive jobs like ours. I have a suspicion that $600 savings now will not cover the loss of performance over even 2 years. If you bill at say, $50/hour, $600 will be quickly gone in time lost to a processor that isn't the fastest available. For example, I can lose an hour exporting a new revision to pdf. I can't get any other work done during that time. It's lost money.
  
  I suspect that a more realistic workstation turnover period could be more like 1.5 years for a heavy SW user using the fastest systems, but there is so far no way to reasonably measure the effect of workstation performance on productivity in terms that would be useful in making ROI evaluations.
  Like • Show 0 Likes0
  Actions
  - Bruce Fisher @ Dale Dunn on Feb 4, 2009 2:34 PM
    There's a saying in the machine shop that you don't have enough time to make it right, but you always have time to make it over. The corollary here is that you can't get the money for the faster PC, but you always have the money to burn on the extra time it takes for the slower PC to get the work out.
    
    It seems to me that getting two years service out of a heavily used SW workstation is reasonable, especially if you work in an office environment where you can replace your workstation and recycle the old one to a secretary or sales person.
    
    I run benchmarks when I'm starting to look at replacement, then do some very conservative calculations to determine how much "real" time savings the faster PC could provide. As Dale pointed out, you could quickly pay for the capital outlay at $50/hr. There is also an intangible time savings that you get because you are better able to keep your train of thought when you don't have to wait on the PC.
    
    Just my 2 cents
    Like • Show 0 Likes0
    Actions
  - Christopher Mueller @ Dale Dunn on Feb 5, 2009 5:16 AM
    
    Dale Dunn wrote:
    
    I have a suspicion that $600 savings now will not cover the loss of performance over even 2 years. If you bill at say, $50/hour, $600 will be quickly gone in time lost to a processor that isn't the fastest available. For example, I can lose an hour exporting a new revision to pdf. I can't get any other work done during that time. It's lost money.
    
    That's exactly the kind of calculation that needs to be done in advance, but is extremely hard to do. You are looking at 12 hours of CPU time that has to be saved in order for the X9100 to pay for itself compared to the T9400. That processor saves 14 seconds in the punch holder rebuild. But that means you need to do the equivalent of rebuilding the punch holder 3000 times in two years for the X9100 to be worth it. For some heavy users it might be a no brainer to get the X9100, but for the users at my company I don't think it would pay off.
    
    We spent 11.000 swedish crowns (about $1100) on our present system, which all the users are happy with. I probably could have spent 20.000 if wasn't knowledgable enough about processors and just bought the most expensive Xeon that I could find. We got a great system for the price. I think if I go to my boss after the system is 2 years old, he would let us replace the systems, because he no longer sees workstations as a huge investment they once were. Now they are more like a beefed up home pc. It feels really good that we won't be sitting on a system for 3 years. It looks like when we turn over there should be a workstation out there with a basic i7 chip in it for about $1000.
    Like • Show 0 Likes0
    Actions

Dell M6400 Processor Choice

Outcomes

Related Content

Recommended Content