Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Status
Not open for further replies.
No - HPE ProLiant ML350 Gen9 https://www.hpe.com/us/en/product-c...p.hpe-proliant-ml350-gen9-server.7271259.html

Rack-mount model. Dual socket E5-26xx v4. 2400 watt power (quad 800 watt hot plug supplies).

For dev/test I also have a bunch of DL380 Gen9 (similar dual socket E5-26xx v4, but in 2U chassis with support for two GPUs).

View attachment 697609

Dude, you got the real Iron.

I suppose I could ask what you are running with all that GPU power, but I doubt you would be at liberty to tell me.:oops:
 
Ah. I was looking for that picture when I was spec'ing out my Octane Render box recently.

The one I got is a SuperMicro 4028GR-TRT 4U box with Dual Xeon E5-26xx v4's and 8x GTX 1080's installed in it. Don't have any photos handy, but it sure puts our relatively new CPU render farm in the shade. A serious piece of kit and pretty inexpensive, all things considered.
 
New fluff: https://www.pcgamesn.com/nvidia/nvidia-volta-gpu-specifications
[doublepost=1493330875][/doublepost]
The one I got is a SuperMicro 4028GR-TRT 4U box with Dual Xeon E5-26xx v4's and 8x GTX 1080's installed in it.
Our jobs are a mix of CPU and GPU work. (Setup in CPU, analysis in GPU, consolidation in CPU.)

200 to 300 CUDA cores per Xeon thread is our current sweet spot (although several systems are at 80 CUDA cores per Xeon thread because their jobs are more CPU heavy). Eight GPUs per dual-socket system wouldn't be as cost effective for us as two dual-socket systems with four GPUs each.

And while SuperMicro makes some great systems, we're heavily invested in the HPE "lights-out" management eco-system - so bringing in random other servers would add to management complexity and cost. It's great to give the BIOS the URL of an ISO image, and boot from it.
 
Shouldn't it also be Nvidia's thing to make the web driver work on the first day of the Volta debut? And don't they (both) watch what we say here?
The first Volta chips could easily be $10K each - and might not be available as standard PCIe cards. (Same as the Pascal launch.)

Having drivers available when consumer Volta PCIe cards ship would be a nice goal, however. Perhaps Apple is eyeing Volta for the mMP....
 
Shouldn't it also be Nvidia's thing to make the web driver work on the first day of the Volta debut?

By all means feel free to contact Nvidia too. However, I recall Nvidia CEO's statement that Apple drivers are really hard to do without Apple's help.
 
We're a small, rogue research group in a big company. We don't let IT know what we're doing.
Same here… small film production film, in a +5000 employee company.
No one messes with our stuff!
[doublepost=1493398677][/doublepost]
If y'all send clear (official) feedback to Apple that full CUDA support for standard PCIe graphics cards on the mMP is essential - then this will happen.
How can this be done? Officially, I mean.
 
I won't be able to justify buying either Volta or Vega, but I am looking forward to a slight price drop for the Nvidia 10 series when the those big boys roll out.
 
I just ordered twenty GTX 1080 Ti cards.

I really hope that turns out to be a $14K mistake :D .
The first 14 came in...guess that I'll get a few days of Pascal goodness before the Volta news comes out. The other six should arrive Monday according to the tracking number.

1080ti.jpg

They're locked in a cabinet under some empty boxes....
titans.jpg
 
The first 14 came in...guess that I'll get a few days of Pascal goodness before the Volta news comes out. The other six should arrive Monday according to the tracking number.


They're locked in a cabinet under some empty boxes....

I need to follow around whatever UPS delivery truck goes to your place, and wait for it to hit a big bump in the road.
 
  • Like
Reactions: LightBulbFun
Status
Not open for further replies.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.