One of the biggest questions is why the Quadro RTX 6000? Few things:
1. Cost it has the same performance as the 8000. The difference is 8 more GB of RAM that comes at a steep premium. Cost is important to us as it allows us to be at a more affordable price point.
2. We have all heard or used the Tesla V100, and it's a great card. The biggest issue is that it's expensive. So one of the things that caught our eye is the RTX 6000 has a fast Single-Precision Performance, Tensor Performance, and INT8 performance. Plus the Quadro RTX supports INT4. https://www.nvidia.com/content/dam/en-zz/Solutions/design-vi... https://images.nvidia.com/content/technologies/volta/pdf/tes... Yes, these are manufactures numbers, but it caused us pause. As always, your mileage may vary.
3. RT cores. This is the first time (TMK) that a cloud provider is bringing RT cores into the market. There are many use cases for RT that have yet to be explored. What will we come up with as a community?!
Now with all that being said, there is a downside, FP64 aka double precision. The Tesla V100 does this very well, whereas the Quadro RTX 6000 does poorly in comparison. We think although those workloads are important, the goal was to find a solution that fits a vast majority of the use cases.
So is the marketing true to get the most out of MI/AI/Etc? Do you need a Tesla to get the best performance? Or is the Tesla starting to show its age? Give the cards a try I think you'll find these new RTX Quadros with Turning architecture are not the same as the Quadros of the past.
Two of my colleagues use high-end AMD GPUs to train RNNs and transformers with tensorflow-rocm. There are still some nasty bugs (e.g. [1]), so it is currently not for everyone. However, given how far they have come compared to 1-2 years ago, it is very likely that in a year or so, they are a real competitor to NVIDIA for compute. That competition was long needed.
[1] https://github.com/ROCmSoftwarePlatform/tensorflow-upstream/...
This is incorrect. The RTX 6000 has 24GB of VRAM and is $4000, and the RTX 8000 has 48GB of VRAM (double the amount) and is $5500. Is it worth the price increase? For a lot of people I know it is.
Also, the RTX Titan is $2500 and is identical to the RTX 6000 (at the chip level) and also with 24GB of VRAM, with the only difference being software enabling of additional H.264/5 encoding features on the Quadro. Definitely not worth the cost increase, especially for anyone doing ML.
[1]: http://fortune.com/2018/01/07/nvidia-consumer-video-cards/
Your knowledge is incomplete. T4 has been available in google cloud for many months.
These are great improvements but are virtually worthless if linode didn't change their behavior.
As far as standards go, we use Linode and all of our customers (some of them quite demanding about internal security details) have been satisfied with the various acronyms they are accredited with... Although I understand this does not necessarily guarantee anything about response behavior, so interested to hear about past incidents.
We made some improvements to our disclosure / Bug Bounty program last year and launched this on HackerOne. The community and quality of submissions has been great. More information: https://blog.linode.com/2018/05/16/linodes-new-bug-bounty-pr...
We've also been making ongoing improvements to our application security and security infrastructure through the implementation of a DevSecOps culture. This is something we take very seriously.
GTX1080 for 100$ a month. Grantend, it is older, but it still works for DL. Let's say you do 10 experiments a month for ~20 hours. Thats 0.5$/hour and I don't think it is 3 times faster.
If you then want to do even more learning the price goes even down.
//DISCLAIMER: I do not work for them, but used it for DL in the past and it was for sure cheaper than GCP or AWS. If you have to do lots of experiments (>year) go with your own hardware, but do not underestimate the convenience of >100MByte/s if you download many big training sets.
It is not a server card, however, it is much faster than any old AWS instances for 1k$/m (if you happen to be an AWS user and did not want to upgrade because of the price going up 3x) TBH, 100 bucks per month is free, while most of the researches do not have 1k$/m for a server, it is cheaper to buy hardware and put Linux on it.
There are of course other options and Linode is kinda late to the party, but I am happy they made this move.
Considering their main competitor, DO, Vultr, UpCloud, none of the them offers any GPU instances, I don't think they are late at all. If not the first for their market segment.
Also check availability shows a 5 day wait current: “EX51-SSD-GPU for Falkenstein (FSN1): Due to very high demand for these server models, its current setup time is approximately up to 5 workdays.*” Or maybe there are other regions/dcs.
But you can just ask them.
I have to say that not everything was 100% smooth - sometimes the proprietary NVidia driver crashed (you have to use the right CUDA and driver combination) my Linux instance and hanged the system, so I had to hard-reboot it (which is supported via their admin console) which takes some minutes. However that's not their fault as I heard the driver is a big pile of crap shit anyway because NVidia is too embarrassed to post it to LKML.
Or is that prohibited in US only?
Not deep dived into it but maybe using nouveau instead of GeForce works around that restriction.
You are allowed to use the driver in data centres for cryptocurrency usage. The EULA limited datacenter usage hasn’t really been challenged in court yet. Both sides would have an argument. NVidia are using the Eula to limit an activity that a user would be allowed to do if the location that activity was different (and not even talking type of industry here, though that’s prob in the Eula too) On the other hand, it’s nVidia’s software, they are free to license it how they like.
One thing I noticed when recently trying to get a GPU cloud instance, the high core counts are usually locked until you put in a quota increase. Then sometimes they want to call you.
So I wonder if Linode will have to do that or if they can figure out another way to handle it that would be more convenient.
I also wonder if Linode could somehow get Windows on these? I know they generally don't do anything other than Linux though. My graphics project where I am trying to run several hundred ZX Spectrum libretro cores on one screen only runs on Windows.
Linode skews more towards smaller scale customers with many of their offerings so I think the GPUs here make sense. The real test will be how often they upgrade them and what they upgrade them too.
Also seems to be a lot cheaper than AWS counterpart.
A P100 on GCP is $1.46/hr alone, so maybe Linode is a good deal if the performance is indeed comparable.
edit: could be wrong thought I read of AWS being .65 dollars an hour for deep learning GPU use. edit2: Did a quick look, the .65 dollars doesn't include the actual instance, so its around 1.8 an hour on the low end, I think this cheaper.
p3.2xlarge has NVIDIA Tesla V100 GPU which is NVIDIA's most recent deep learning GPU, but it's $3.06/hr.
That said, AWS is among the most expensive providers if you just need a deep learning GPU (but obviously AWS offers a lot of other useful things). For example, OVH Public Cloud has Tesla V100 for $2.66/hr. And comparable NVIDIA GPUs that are not "datacenter-grade" should be even cheaper; AWS, GCP, Azure, etc. are unable to offer them because of contracts when they buy e.g. the Tesla V100.
This is a good deal.
The best time to mine is during the drops, not the highs, unless you follow buy high, sell low and don't believe the market will correct for the better again (which it has).
Of course, it depends on electricity prices, but it is profitable to mine ethereum, especially if you know how to tune the cards to maximize hash/consumption.
That said, mining is competitive and difficult and unless you are going to go really large, don't bother. If you are interested in learning about it, definitely experiment though don't expect to make a lot of money.
