Thank you very much for some of the interesting links. They are quite helpful for me as I too was planning on getting a computer system for myself in the near future.
By the way, what budget are you aiming at? I think according to this http://www.trustedreviews.com/reviews/nvidia-geforce-gtx-1080-ti-performance-and-overclocking-page-2 the GTX 1080ti has quite a few more advantages than 1080 in performance and the extra 150$ seems worth spending (again only when it is restocked). On that note, anyone knows how long does it take to re-stock 1080ti ? I think since the past two weeks it has been out of stock. Any other place where one can get 1080ti?
When working with live audio and video for real-time processing, how best to handle the input to the black box?
Would digital signal processing (DSP) music software take advantage of GPU? I really miss MAX, and wish I could have been able to use it on the NeXT with real-time frequency modulation (FM) synthesis on a card back in the day. I just wrote them about how fast.ai might use a front-end to help automate the process of finding the best slope for the loss function.
As my plan unfolds for the brain in a box, so far there are a number of custom setups we should strive for:
Thinking of potential builds, for the professional is the black box ssh I/O to crunch numbers 24/7, the type of system that might reverse engineer planarian regeneration after the boss takes the first stab at it, security bots needing real-time I/O, and game bots where “While in Rome” I have a chip with which to toy.
“Garbage in, garbage out” they say in computer programming. After teaching 39 generations of fruit flies how to count, the 40th were born already knowing how. This works in people, too, be it a predilection for music or domestic violence. Intelligence, or lack thereof, is inherited. Now how do you program it?
Here’s your storage medium.
QUERY: I know there’s a web page listing all the DNA shorthand codes. I’m wracking my brain trying to locate it again.
The 1950’s found Robert G. Heath putting brain implants into human subjects (not a nice guy, by the way), while the 1960’s had Dr. José Delgado at Yale make them remote-controlled, his famous front page New York Times article letting folks know the future (whether they’d realize it or not).
Fast forward to NASA’s 2004 subvocal speech BCI that I heard Delta Force uses which is being tallied at the moment for popularity points among a commercial product.
In related news are modulated microwave bursts. (My condolences go out to our diplomats overseas currently experiencing “health attacks” as the assaults read in the papers.)
We’re at war, sir, despite outward appearances, and terrorist activities are thwarted thanks to AI pattern-spotting software that needs to be improved upon.
All it’d take is taking out the NVIDIA factory(s) to shut down the US AI lead.
Taking out a few factories would not really change any fundamental flaws in AI strategy. For example, preventing computer imports into Russia during the cold war, actually had such a profound effect on their maths research that it put them in the lead for many years.
It doesn’t really make sense to answer a complex question with simple cliches.
In short, this is not the forum to sell your stuff. If someone wanted an NVIDIA (and they are for the moment the only game in town), there are a hundred websites where they can be found.
This is questionable. The top AI scientists still reside and work in the US. The vast majority of money thrown at AI comes by US companies. (Please note I am european).
I agree, it is questionable, which is why I said ‘arguably’.
You make a good point about the amount of money being thrown at AI. However the money being thrown at AI in the US is tactical with what seem to be short-term return expectations. They’re after a quick buck, and the overall strategies don’t seem cogent. They are just as likely to pull the funding if there’s no short-term return and this is not the basis for a solid AI strategy that will lead to ‘leader’ status.
Secondly, if you compare what China is doing (different populations, different time scales, different target areas), then I see that the east and west are working to different cost functions. I think China is possibly leading now.
Thirdly, AI is now almost democratised. Anyone and everyone can pick up the basics, and implement something quite impressive on a $20 piece of hardware using open source frameworks. The US as a body historically led when the world was more ignorant of technology. The delta between the bleeding-edge and good-enough is not really substantial enough these days.
The US does a good job of supporting that democratisation of AI specifically because of its business model and if the east were to lead, then I don’t see the world following along as it’s not an eastern model to spread the technological know-how as it has been done in the west. The West isn’t altruistic. That model just supports a capitalist and entrepreneurial practise. Whether that helps bring the US back to a predominantly leading model is interesting, and I hope it does. However, to compare current US investment to AI leadership is too simplistic a metric in my view.
Thanks for the detailed post and also the enlightening discussions. I have prepared two builds which I am sharing below. Can you please provide your suggestions. I have some specific questions regarding the builds which I am providing with the build configs. But, before that let me provide my requirements/type of ML works I intend to do.
I have already bought 2 1080Ti GPUs and a Samsung 860 EVO 500 gb around 3 months back. So, they are sitting idle now. Now that I have some more money, I want to buy the other parts of a DL machine. My budget is around Rs. 100,000 [Rs is the Indian currency].
The machine will be in the server room of my institute. So, the cheapest cooler [whatever noise level] and cabinet is what I would prefer.
I shall mostly run RL codes [both training and inference] on images. Later, I might do some classification work on videos [but this is a distant possibility at this moment].
I don’t plan to expand this machine beyond 2 GPUs.
My builds provide some prices which does not have web links. This is from https://mdcomputers.in – a local but reputable vendor. I could not find how to link their product pages to pcpartpicker. Otherwise, I would have done that. So, you have to believe my words about the price.
My first build uses core i5 9600K and a compatible motherboard. As my budget allows me to spend some more money, should I go for i7 8700K? This change along with a change of motherboard costs me Rs. 6000 more [still remaining inside the budget]. Core i7 processor supports hyperthreading. From time to time, I plan to use multiple threads for preprocessing in Tensorflow. Or, instead of upgrading the processor, should I just go for RAM upgrade to say a total 48 GB [and stick to core i5 9600K]? This will cost me around Rs. 12,000 more [and so I am remaining within the budget]. And later, I can pull out one memory to be used in some other machine.
Build with core i5 9600K – https://in.pcpartpicker.com/user/dasabir/saved/fgQ299
Build with core i7 8700K – https://in.pcpartpicker.com/user/dasabir/saved/bD7J8d
While looking for the option of core i7 8700K, I came across core i7+ 8700 [https://mdcomputers.in/intel-core-i7-8700-bo80684i78700.html]. I see that this will cost me Rs. 11,000 more over my core i5 9600K build. I am not sure what is the difference between an i7 8700K and i7+ 8700 (other than the frequency/speed). Here is the comaprison link – https://ark.intel.com/compare/126684,140642 . Will i7+ 8700 require different motherboard? It says the box includes NVME 3.0 x 2, does it help me? Also the i7+ processor includes a 16 GB optane memory. Will it be of any help (e.g., keeping the OS there)? Also does optane memory occupy PCIe lanes? Any suggestion on this would be great to have.
My second build is with AMD processors. I tried with AMD Ryzen 7 2700X. The price is coming around the same as the core i5 9600K build. It does have 8 cores compared to 6 cores for the intel processors, but does AMD have hyperthreading? I am not sure. Also it does not have MKL, is intel MKL going to be crucial for deep learning?
Build with AMD Ryzen 7 2700X – https://in.pcpartpicker.com/user/dasabir/saved/3ddTBm
Though people say number of PCIe lanes are not that important especially with 2 GPUs, I just tried my luck with an AMD threadripper processor. As expected, it is overbudget. But, if you say, it is worth spending this much money, I might also go for it.
Build with AMD Threadripper 1900X – https://in.pcpartpicker.com/user/dasabir/saved/73mhyc
