NVIDIA Tesla C1060 work on Mastercam?

[Newbieguy]

I just saw this promotion from NVIDIA about the Tesla C1060 bringing the power of a super-computer to a desktop size:

 

http://www.tigerdirect.com/applications/se...42-_-components

 

But since I'm not a computer geek, I'm wondering if this is something that would work hand in hand with Mastercam? It would be awesome to post "benchmark" times of 5 or 10 seconds - but I'm not willing to spend the $10,000 to be the beta tester. I do know that Mastercam will not take advantage of dual processors (at least it wouldn't in "X" - dual core yes, dual processor, no), so I'm guessing the C1060 would also be worthless...

 

Does anyone know?

[The_Evil_Machinist]

I've already tried it. Id doesnt work because mastercam needs cuda. Now a cluster computer does increase the benchmark time in mastercam. Go to the benchmark and read pages 24 and up. I have the benchmark record. :pats self on back:

[Newbieguy]

Thanks a bunch Evil - very informative reading. (I guess I should have went there first)

 

Now I have to follow that thread to see how your race against MurrayMold goes!!!

[The_Evil_Machinist]

quote:

---

Now I have to follow that thread to see how your race against MurrayMold goes!!!

---

.

He hasn't challenged me lately.

.

And I just got more computers to be recognized on my cluster network. I could prolly go down to 20 sec if I have it writing to a ram disk or a solid state hd instead of a regular hard drive.

[[email protected]]

I still have the best time with one computer

[The_Evil_Machinist]

quote:

---

I still have the best time with one computer

---

.

And? lol

[Guest CNC Apps Guy 1]

quote:

---

I could prolly go down to 20 sec if I have it writing to a ram disk or a solid state hd instead of a regular hard drive.

---

SSD's are amazing. Friend if mine has one in his laptop that is similar to mine. His boots up in a fraction of the time mine does (and mine is not slow by any stretch). He only runs CATIA so I don't have a benchmark. His battery life increased also. Good all the way around.

[The_Evil_Machinist]

quote:

---

His battery life increased also. Good all the way around

---

.

they have some problems and a good raid 0 array can almost beat them, but I want to raid 0 a pair of ssd's to see what or if they can do.

[Guest CNC Apps Guy 1]

quote:

---

...and a good raid 0 array...

---

Loose a drive in the array in that configuration and you're dead in the water. Data gone. I'd consider that "some problems" too wouldn't you?

[The_Evil_Machinist]

quote:

---

Loose a drive in the array in that configuration and you're dead in the water. Data gone.

---

.

And thats why I have a Apple Timecapsule hooked up to it and a mobile me cloud server! keep your data somewhere else. I'd do that even if I had only 1 hd.

[Steve Hattori]

A year of mcam crunching and daily transport on my laptop with Intel's X25M ssd, and not one problem. Plus, its dual-booted for XP64 and Vista64. All data stored elsewhere, just the apps and the pagefile on the ssd. Faster than my WD raid 0 on the desktop. Heard of good luck with 2 X25m's in raid 0, and nearly twice as fast...I may do that on my laptop. Sad that the C1060 doesn't work with mcam- I'm having trouble with a big file causing 2-3 HOUR calc times on 5x paths. A supercomp would help.

Evil's cluster is very appealing, but I haven't figured out how to get it to work with Windows instead of a Mac driving it.

Any advice?

[Guest CNC Apps Guy 1]

I use Carbonite. I like the idea of an offsite backup better than something "local".

 

Just a preference.

[The_Evil_Machinist]

quote:

---

Evil's cluster is very appealing, but I haven't figured out how to get it to work with Windows instead of a Mac driving it.

Any advice?

---

.

Yea, Get A Mac! lol

.

You can build a cluster system using windows. But It requires a bunch of 1u servers all using the same hardware. Gets pricey. Now if mastercam rewrote the multi threading manager in cuda, we could all just throw that tesla card in and crunch!

[Matthew Hajicek - Singularity]

If they rewrote in CUDA I bet a little 260GT card would be more horsepower than I'll need for a while.

[The_Evil_Machinist]

quote:

---

If they rewrote in CUDA I bet a little 260GT card would be more horsepower than I'll need for a while.

---

.

The 260 isnt a little card... If they rewrote it for cuda everyone could have my level times without a cluster on that "little" 260. I have a thread in the enhancement forum. If you want to see power unleashed, voice your opinion on that thread.

[[email protected]]

quote:

---

my level times without a cluster

---

You mean like mine

[The_Evil_Machinist]

quote:

---

You mean like mine

---

.

Yea, except without the cost of yours!

.

When are we gonna benchmark battle?

[[email protected]]

Thought we already did

Mines not gonna be any faster than it was that 1st day I ran it. If anything its gonna be slower.

[The_Evil_Machinist]

quote:

---

Mines not gonna be any faster than it was that 1st day I ran it. If anything its gonna be slower.

---

.

But because of the cache amount you have, with bigger files you could beat me. I'm actually intrigued by the advantages/disadvantages of cluster computers vs truly powerful computers.

[[email protected]]

quote:

---

with bigger files you could beat me

---

Okay, I will dig through my big files and find one that I can put on the ftp.

I'm leaving for the day now, I'll start looking tomorrow

[The_Evil_Machinist]

cool

[Steve Hattori]

Evil I am curious why Murray's dual-quad-Xeon is so much faster than your dual-quad OC QX system?

[Matthew Hajicek - Singularity]

quote:

---

The 260 isnt a little card...

---

It's little with respect to my wallet. I got it a few months ago for $150. Compare that to the cost of a Tesla.

[The_Evil_Machinist]

quote:

---

Evil I am curious why Murray's dual-quad-Xeon is so much faster than your dual-quad OC QX system?

---

.

Xeon processors cache is bigger and its just how the cpu threads. Thats why I cluster. To pseudo up the cache while using processing cycles to help out.

.

quote:

---

It's little with respect to my wallet. I got it a few months ago for $150. Compare that to the cost of a Tesla.

---

.

Money doesn't have anything to do with the performance of the card. It's just a well designed gpu.

[The_Evil_Machinist]

My cluster. It's super hacked together! The macbook pro is used as the main host controller, the netbook is used as a node and my siggy comp is a node but its running windows. The router is a 24 port server switch I got at a pc scrap yard for 20 bucks. From my macbook I can associate where each packet goes and what it will do. The windows comp runs mastercam and when the multithreading manager is active, it sends the data to the macbook, distributes to other comps. then sends it back to the win pc. When I started this, it sounded way inefficient. But using xgrid, and my friend that works for apple (the reason its running properly), I can utilize multiple computers as one entity. I just put in the netbook as a node, but it doesnt give much power. maybe a 3% boost. lol.

.

 

[ 10-01-2009, 12:08 AM: Message edited by: The_Evil_Machinist ]