Justify new workstations 1

[GhostTX]

So I started at a new company and they have some pretty pathetic computers. So I went to SPECapc and downloaded their benchmarking software for what we use (Solidworks in this case) and had all our engineers run it.

In short, the conclusion, on average, our PCs are 52% slower than Dell's Precision 390 (cost about $3K).

What's a good way to show the cost of us running slower? Or what's a good way to justify the new computers. I have a burden rate, but I can't just arbitrarily say we'll save 1/2 that rate.

Any tips would be great!

 

[MadMango]

Why not? If it takes you twice as long to complete a task, that's wasted time.

Break down tasks; rebuilds, opening assemblies, saves, drawing changes... estimate how often these tasks are done on a monthly basis and use an estimated hourly rate. Then use the same data with the 50% time reduction against the rate. Factor in the cost of new computers and you should have a pretty sizable ROI.

"Art without engineering is dreaming; Engineering without art is calculating."

Have you read faq731-376 to make the best use of Eng-Tips Forums?

 

[swertel]

Doing calcs like MadMango states is great for accountants, and sometimes management, but it only gives half the story.

I always like calling the boss over to tell him you need to ask him a question. They always come a bit later than you want. When he arrives at your desk, tell him to hold on a second, I need to rebuild the model to show you this. He'll soon get so bored watching the hourglass with you he'll understand why you want a new computer and can't believe that he is wasting that much of his designers' time each day.

As you're waiting, tell him that you do this 5, 6, 10, 12 times a day. He'll get the hint really quick.

--Scott

http://wertel.eng.pro

 

[bvanhiel]

Only 52% slower? They must be last years computers as that's hardly what I'd call "pathetic".

It really depends on how much of your time is spent waiting. If all you're doing is regenerating your model all day, then you'll see that 50% will be a huge timesaver. I suspect, however, that you regen less than you think, and that all you'll be doing is quickly adapting to a new expectation of speed that will lead to the exact same frustration level.

I'd estimate that I'm using a similar performing computer. I have access to a box that is supposedly 2x better. I hardly notice the difference when I'm using it. Hence I'm waiting for at least a 3x or 4x performance difference before I try justifying an upgrade.

-b

 

[MadMango]

Percentages are misleading. If it takes me 6mins to open a complex model, and 3mins for each regen (say 5x a day), that's 21mins wasted per user, 3 users and you've just wasted a man-hour. But if it only took 2mins to open and 10sec to regen, then the 50% increase isn't that much of a gain. The OP needs to do more in-depth, real world (your own models and work flow) time studies to realize potential gains.

"Art without engineering is dreaming; Engineering without art is calculating."

Have you read faq731-376 to make the best use of Eng-Tips Forums?

 

[mechengdude]

I beleive swertel is on to something. In my experience if you are at a company that doesn't keep the engineer's computers reasonably up to date then you have a bigger problem than just out dated computers.

Justifying computers or a new software package on paper has always been an exercise in futility and something managment has you do to keep you out of their hair.

hope I'm wrong in your case and you get some nice new Dell Quad processor machine with 8 gig RAM.

 

[GhostTX]

My boss is fully aware of our issues. Unfortunately, he needs a cost justification in order to sell his boss. Otherwise, we're just a bunch of whiney engineers.

I have done times on my own PC with open and save times anywhere from 3-5 minutes. We're talking assemblies with 50-100 parts. We have P4 3Ghz with 1GB RAM and Quadro FX 1400. We managed to bump up our ram to 3GB, 'cause that was using "petty cash" within our budget. That helped with managing drawings (dragging views, balloons, etc) and helped with regens, and mating. However, we still have performance hits with assemblies (75+ parts). Fortunately, we do mostly sheetmetal, so parts themselves are simplistic.

 

[mechengdude]

Engineers are overhead (ie) a necesary evil for the company to make and sell a product for a profit. Typically engineers time in terms of productivity is not viewed the same as a production workers that makes 10 wiggets a day.

Based on that logic don't be suprised if upper management looks at those numbers and says "while they are waiting for assemblies to load they can be reading a report or doing some other work."

Sorry to be a pessimist but if your boss is fully aware of the issues he should stand up for his folks and get the right tools in to do the job.

Good luck

 

[epoisses]

Agree
The seconds and minutes argument just won't work. If I were an engineering manager with a tight budget I would grin and refer you to the "How much time do you actually spend working" thread.
I'd think it should be easier to find an "indispensable" new software package and then have the hardware upgraded to be able to run it.

 

[josephv]

Hello GhostTX:

Here is an idea, can you borrow a Dell Quad from a dealer? Then install SWX and open your assemblies in that machine, and show the time differences to management.

BTW many moons ago, I had a very slow PC when I first learned SWX. At the time it was frustrating, but I learned how to create simplified configs, lock external references, save on rebuild times by mirroring bodies and using geometry patterns. So when I finally got a fast PC, those assemblies rocked.

Anyways, hope all works out great for you.

 

[csd72]

52% slower, does that mean that they are running at 48% of the speed or do you mean 52% as fast?

The language doesnt make much of a difference in this case, but it could if the number was, say 10%.

If I said that it was running 10% slower, nobody would think it meant 1/10th of the speed.

csd

 

[GhostTX]

On average, our machines perform the benchmark in 303.23 seconds, the Precision 390 does the test in 143.19 seconds.

We already had SWX come to our plant to check out the peformance and they gave us the large assembly mode, configs, etc. Our biggest hit is we have a HUGE UGS file vault. We just went to SWX 1 year ago. The SWX admits the imported files hit the machines harder than native SWX files. That's where we're getting our biggest issues.

 

[MadMango]

The imported files should be that big of a hit (unless helix based). After importing to SW, save teh files as SW files. What version SW are you using?

"Art without engineering is dreaming; Engineering without art is calculating."

Have you read faq731-376 to make the best use of Eng-Tips Forums?

 

[ajack1]

Tell your boss that if he buys you a $3K PC and gives you a $10,000 raise, you will be happy for him to double your work and he can take any shortfall from your salary.

That way everyone is happy, right?

 

[GhostTX]

MadMango,

SWx 2007 SP2. We do save the UG parts as SW. I guess the problem is its one large surfacing, rather than a solid model. As said, even the ppl at SWX says that's probably not helping with performance issues, since SWX is handling each imported part as surfaces. :S

LOL @ ajack1!

 

[PSE]

It always seems tough to cost justify upgrading PCs as they seem to become outdated so rapidly. For all essential purposes they are a pure expense (likely from the bosses point of view). Software capability soon expands and slows you back down again as it demands more system resources.

A PC is a tool. These days it is a primary tool for doing the job. You have done some of the research that shows you can be more efficient with new hardware. This could in theory allow the department to handle additional load due to greater throughput. Is there additional work available in the "pipeline"? Somehow, upper management needs to recognize a benefit (assuming there is one from their point of view) to improved efficiency/flexibility (reduced time to market perhaps). Are there overtime premiums involved with the work.

See if you can enhance your cost justification numbers with the ability to address "larger" opportunities due to increased flexibility and speed.

Regards,

 

[cvg]

since computer hardware is depreciated for tax purposes in 5 years, your management would like to keep it for the full five years. Your IT staff would probably look at a slightly shorter period, maybe 3 - 4 years between replacements. This would mirror the likely software upgrades (Op System and modeling software) so that your system does not become obsolete. Unless your computer is approaching that age, not likely they will replace as a) capital cost would be quite high; b) they would have to admit that they bought the wrong equipment the first time. Perhaps part of your problem is the file server, does it have the capacity and bandwidth necessary? Maybe it is approaching it's useful life and could be replaced.

 

[GhostTX]

Oh, everybody knows the network is even more pathetic. In fact, there's an IT project to replace the backbone and lines in every plant. Of course, its stuck in limbo 'cause of the high cost.

An IT guy bought these computers 1.5 years ago for the engineers, and he bought desktops with the low end graphics card (Quadro 1400). Apparently, there was no engineering input on the purchase. SWX, for all its perks, does crap file size management. Models and drawings easily exceed several meg, where as the previous CAD system, Unigraphics, had smaller file size and was more efficient. The machines did fine on Unigraphics, but not anymore with SWX.

Somebody just has to realize they may have saved a penny on the front end, but now they're not on the backend. I've done a time study and we're easily losing nearly 24-40 minutes a day on this issue.

 

[ewh]

But think of the savings not paying those UG prices!

 

[ewh]

I was being facetious