I Build Supercomputers in My Spare Time
parallella.org
I Build Supercomputers in My Spare Time
1–10 of 34 posts
Re: I Build Supercomputers in My Spare Time
#2For anyone running into the DB connection error and the page not loading:
http://webcache.googleusercontent.com/search?q=cache:http://...
Re: I Build Supercomputers in My Spare Time
#3Having read through the documents, I see that it is capable of ~200 GFLOPS. Obviously, this outshines my macbook performance of ~50GFLOPS, but it is substantially less than my workstations GPU, an Nvidia GTX 770, capable of ~3000 GFLOPS.
I suppose my question is why would I choose to build something like this over using a GPU?
Re: I Build Supercomputers in My Spare Time
#4Re: I Build Supercomputers in My Spare Time
#5Having read through the documents, I see that it is capable of ~200 GFLOPS. Obviously, this outshines my macbook performance of ~50GFLOPS, but it is substantially less than my workstations GPU, an Nvidia GTX 770, capable of ~3000 GFLOPS. I suppose my question is why would I choose to build something like this over using a GPU?
Here's some previous discussion:
https://news.ycombinator.com/item?id=4702456
In terms of raw GFLOPS, you shouldn't.
Parallela offers something halfway between a multi-CPU cluster, differing in terms of memory access, and a GPU, differing in that each core is a real core and can be independently making calls, branching, etc.
Another point to note is that every GPU architecture is different and that some support a different degree of control flow parallelism vs. data parallelism.
Ultimately it depends on the kernel you're working with - if you're to a point where you've got strictly linear SIMD and you depend entirely on floating-point math throughput or memory bandwidth, it wouldn't make sense not to use a GPU instead.
Re: I Build Supercomputers in My Spare Time
#6Having read through the documents, I see that it is capable of ~200 GFLOPS. Obviously, this outshines my macbook performance of ~50GFLOPS, but it is substantially less than my workstations GPU, an Nvidia GTX 770, capable of ~3000 GFLOPS. I suppose my question is why would I choose to build something like this over using a GPU?
I guess it is easier to program, I am only a simple web developer and i would be more productive programming on this than learning to use cuda.
Re: I Build Supercomputers in My Spare Time
#7I can see one fan at the top of the cylinder- is only one fan really sufficient? It looks like this would generate a large amount of heat.
I would be quite interested to see the temperatures this gets up to.
Re: I Build Supercomputers in My Spare Time
#8Wouldn't it be quite hard to benchmark something like Parallela as it contains an FPGA/ARM/Their own multi-core chip.
Re: I Build Supercomputers in My Spare Time
#9I can see one fan at the top of the cylinder- is only one fan really sufficient? It looks like this would generate a large amount of heat. I would be quite interested to see the temperatures this gets up to.
It's a pretty big fan. Also, the cylindrical design certainly helps airflow
Re: I Build Supercomputers in My Spare Time
#10"Error establishing a database connection" *scnr