Hacker News Reader

sabalabaon 2080 Ti TensorFlow GPU Benchmarks

Not in theory, in practice, these are actual benchmarks. Your central argument is that prices within the first few weeks of a new product launch are higher than MSRP. That point will likely not hold true 30 days from now. It makes perfect sense to compare the products based on their MSRPs.

sabalabaon 2080 Ti TensorFlow GPU Benchmarks

The TL;DR on this is that the 2080 Ti is the most cost effective GPU on the market today for deep learning.

It is:

- 37% faster than the 1080 Ti with FP32, 62% faster with FP16, and 25% more expensive.

- 35% faster than the 2080 with FP32, 47% faster with FP16, and 25% more expensive.

- 96% as fast as the Titan V with FP32, 3% faster with FP16, and ~1/2 of the cost.

- 80% as fast as the Tesla V100 with FP32, 82% as fast with FP16, and ~1/5 of the cost.

sabalabaon 2080 RTX performance on Tensorflow with CUDA 10

https://lambdalabs.com/blog/2080-ti-deep-learning-benchmarks...

These numbers match up with the performance that we’ve measured in our own tests that were posted last week. The Titan V is simply too expensive for Deep Learning. The 2080 TI is, by far and away, the best GPU from a price/performance perspective.

As mentioned in the article, only possible reason that you might want a Titan V is if you care about FP64 performance: i.e., nobody training neural networks.

sabalabaon Rethink Robotics shuts down

A CEO's job is to set the overall vision, recruit amazing talent, and make sure there's money in the bank.

If a company wants to change the world, they can't forget #3. In fact it's probably the most important because #1 and #2 are not possible without #3.

So, how does one raise $149.5M and fail? The Fatal Pinch.

https://avc.com/2010/08/what-a-ceo-does/ http://www.paulgraham.com/pinch.html

sabalabaon Facebook Network Breach Impacts Up to 50M Users

Looks like they're clearing out compromised sessions. https://twitter.com/facebook/status/1045722820582432768

I was logged out of Facebook about 30 minutes ago.

sabalabaon RTX 2080 Ti Deep Learning Benchmarks

The training accuracy won't match. Here's an excerpt from a presentation on FP16 training to give you an idea of the loss in accuracy:

    Mode                     | Top1 accuracy, % | Top5 accuracy, %
    -------------------------+------------------+---------------------
    Fp32                     | 58.62            | 81.25
    Mixed precision training | 58.12            | 80.71
    FP16 training            | 54.89            | 78.12

[1] http://on-demand.gputechconf.com/gtc/2017/presentation/s7218...

sabalabaon RTX 2080 Ti Deep Learning Benchmarks

yes

sabalabaon RTX 2080 Ti Deep Learning Benchmarks

Stephen from Lambda here. These are, to our knowledge, the first public Deep Learning benchmarks for the 2080 Ti on actual hardware. We've provided all of the code and tooling to reproduce these results yourself so, if you have a 2080 Ti or another GPU you want to benchmark, please feel free to reproduce these results yourself:

https://github.com/lambdal/lambda-tensorflow-benchmark

sabalabaon Building your own deep learning computer is 10x cheaper than AWS

In terms of number 6. The Linux Deep Learning framework installation can happen in one line with Lambda Stack:

https://lambdalabs.com/lambda-stack-deep-learning-software

sabalabaon Amazon's Bezos Launches $2B Fund to Help the Homeless

1. Usually will end when you pass away.

2. (Especially if establishing a foundation.) Will outlive you and continue on for potentially hundreds of years into the future. For example:

https://en.wikipedia.org/wiki/Ford_Foundation

https://en.wikipedia.org/wiki/Carnegie_Endowment_for_Interna...

https://en.wikipedia.org/wiki/Rockefeller_Foundation

sabalabaon Ask HN: Who is hiring? (September 2018)

Deep Learning Hardware Company

Lambda Labs | Director of Sales | San Francisco, CA (Dogpatch) | Onsite | Full Time

* Build and manage a team of sales representatives

* Attract, hire, on-board, and retain top sales talent

* Partner with executive team to build sales organization

Required Experience:

* 5+ years of sales leadership experience

* Bachelor's Degree in technical field (e.g. CS, Physics, EE, Math, Stats, etc.) a plus

===

Lambda Labs | System Software / GPU Cloud Infrastructure Engineer | San Francisco, CA (Dogpatch) | Onsite | Full Time

* Architect and author GPU cloud management software

* Build reliable, scalable, highly available systems

Required Experience:

* Bachelor's degree in technical field (CS, Physics, EE, Math, Staistics, etc.)

* Prior experience building system software or in a SRE-like role

* Linux experience, Bash/Shell scripting

Send work experience and resume to work@lambdal.com.

sabalabaon A Conversation with Paul Graham [video]

Writing software on windows is like using chopsticks with mittens. You can do it, it’s not pretty, and it’s not the use case the mitten sewer had in mind during development.

Read the Unix Programming Environment to see the exact opposite mindset for OS design.

sabalabaon My Experience and Advice for Using GPUs in Deep Learning: Which GPU to get

The 2080Ti numbers are likely going to be a lot lower than that.

We’ve benched the 1080Ti vs the Titan V and the Titan V is nowhere near 2x faster at training than the 1080Ti as suggested in that graph. We observed a 30% to 40% speedup during our benchmarking:

https://deeptalk.lambdalabs.com/t/benchmarking-the-titan-v-v...

This is consistent with the 32% increase in FP32 flops from 11.3TFlops for the 1080Ti to 15TFlops for the Titan V. Additional speedups can be explained by the increase in memory bandwidth for HBM2 and the mixed precision fused multiply adds provided by the TensorCores.

Thus, given the quoted 13Tflop numbers for the 2080Ti, I would expect the 2080Ti to present something more like a 15-20% speedup over the 1080Ti. So 2080Ti is less bang for your buck. But benchmarking is the only way to tell what’s better on a FLOPS/$ basis.

sabalabaon Nvidia announces RTX 2000 GPU series with ‘6x more performance’ and ray-tracing

Most people I know who publish in NIPS/ICLR train their models using between 1-4 GPUs under their desk or 8 GPUs in a server. I would argue that these folks make competitive models and "are successful in their business" with only 1-8 GPUs. While having access to hundreds of GPUs helps you explore model space more quickly, it's not a hard pre-requisite to success in Deep Learning.

sabalabaon Nvidia announces RTX 2000 GPU series with ‘6x more performance’ and ray-tracing

"Reality", in our case, is based on conversations with a few dozen customers who use our products as well as the actively publishing researchers we know personally. The majority of sampled customers are training convnets of some fashion, the most common being Yolo/SSD and ResNet.

sabalabaon Nvidia announces RTX 2000 GPU series with ‘6x more performance’ and ray-tracing

It's far from a nonsense purchase for Deep Learning professionals. At https://lambdalabs.com nearly every one of our customers is calling us up interested in the 2080Ti for Deep Learning. The reality is that very few people training neural networks need more than 11GB.

sabalabaon Dijkstra: Reasoning About Programs (1990) [video]

Cool side note, introducing Prof. Dijkstra was a younger Eric Schmidt during his tenure as a software engineering manager at Sun. I wonder where all of the other lectures are? Somebody should repost them.

sabalabaon Google announces a new generation for its TPU machine learning hardware

Choice in the marketplace is still choice. And, Google might eventually decide to sell this to outside companies. Eventually the desire to build competitive advantages for Google Cloud might be surpassed by the desire to compete with NVIDIA. Commoditizing the complement is a powerful strategy.

sabalabaon Google announces a new generation for its TPU machine learning hardware

I'm not as pessimistic. 3dfx and Matrox didn't look like easy to knock out opponents in the 90s and yet here we are. Markets are powerful incentive mechanisms.

"Slowly grinds the mill of the gods, but it grinds fine."

sabalabaon Google announces a new generation for its TPU machine learning hardware

This is great news. It’s extremely important to the research community that large companies enter into the DL silicon space to contend with NVIDIA’s monopoly.

NVIDIA is now exerting pricing power to the point where they’ve decided to train their sales people to disregard the metric that is most important to customers: cost of training. Talk with one of their enterprise sales people and you’ll find they’ll say things like “FLOPS / $ doesn’t matter” to justify a 10x increase in price for their TESLA line. As history has shown, a monopolist sows the seeds of their own destruction and, by disregarding the metrics that matter, they alienate their customers.

Here are open source projects that you can contribute to to break the monopoly:

ROCm: https://rocm.github.io/

MIOpen: https://github.com/ROCmSoftwarePlatform/MIOpen

TensorFlow: https://github.com/tensorflow/tensorflow

sabalabaon Hard to Believe: Film Explores Chinese Regime’s Killing of Prisoners for Organs

Epoch Times is founded by and run by Falun Gong members. Just keep in mind that anything they publish is extremely biased towards an anti-PRC stance. In the same way that you should keep in mind that consuming media from Voice of America or RT is respectively biased towards a pro-American and pro-Russian stance.

sabalabaon Ask HN: Who is hiring? (April 2018)

Lambda Labs | Director of Sales | San Francisco, CA (Dogpatch) | Onsite | Full Time

We're looking for a Director of Sales who can help build Lambda's sales org. You will:

* Build and manage a team of sales representatives

* Attract, hire, on-board, and retain top sales talent

* Partner with executive team to build sales organization

Required Experience:

* 5+ years of sales leadership experience

* Bachelor's Degree in technical field (e.g. CS, Physics, EE, Math, Stats, etc.) a plus

===

Lambda Labs | System Software / GPU Cloud Infrastructure Engineer | San Francisco, CA (Dogpatch) | Onsite | Full Time

We're looking for a system software engineer who wants to build a deep learning GPU cloud management system. You will:

* Architect and author GPU cloud management software

* Build reliable, scalable, highly available systems

Required Experience:

* Bachelor's degree in technical field (CS, Physics, EE, Math, Staistics, etc.)

* Prior experience building system software or in a SRE-like role

* Linux experience, Bash/Shell scripting

===

We're located in San Francisco 5 minutes away from the 22nd street Caltrain.

Send work experience and resume to work@lambdal.com.

https://lambdal.com for more info about our company.

sabalabaon NVIDIA Develops NVLink Switch: NVSwitch, 18 Ports For DGX-2

The bad news is that a lot of system builders like us (we build pre-built systems pre-installed with Deep Learning software http://lambdalabs.com) are now used to paying more than MSRP for our GPUs. MSRP for a 1080Ti is supposed to be $699. They haven't been available even in large bulk purchases at that price for a while now. I don't see it going back down any time soon.

sabalabaon Amazon's policies promote counterfeiting

Yes, we looked into Brand Registry -- this is my first time seeing the Brand Gating, which looks like it would solve this problem. Thanks for bringing that to my attention.

In general, I think there's frustration for smaller companies and brands that Amazon (understandably) doesn't retain resources to support individually.

sabalabaon Amazon's policies promote counterfeiting

Since you work on the web services side it's forgivable that you would think that link to reporting infringement is an effective solution / something that Amazon pays close attention to. We've sold hundreds of thousands of dollars worth of product on Amazon, found a company selling a counterfeit version of our product, and submitted a complaint to that URL. That was on 4/5/2016. We have yet to receive a response. The only solution we had was to call up the infringer and tell them to stop, which they luckily did.

But, as I said, from the outside, that report link does seem like a viable "option to fight back"; in reality, it's a customer support black hole.

sabalabaon Cloud TPUs in Beta

Any plans to support AMD GPUs and the Radeon Open Compute project? The AI/ML community really needs viable alternatives to NVIDIA, otherwise they will continue to flex pricing power. Google, via TensorFlow, is in a phenomenal position to promote open source alternatives to the proprietary Deep Learning software ecosystem that we see today with CUDA/CuDNN.

sabalabaon The Anti-Bamboozler: Early writings of H. L. Mencken

“Every normal man must be tempted, at times, to spit on his hands, hoist the black flag, and begin slitting throats.” ― Prejudices: First Series

sabalabaon Nvidia’s New Policy Limits GeForce Data Center Usage

Developers should be mad about this. Here's where to start contributing to open alternatives:

ROCm: https://rocm.github.io/

OpenCL: https://www.khronos.org/opencl/

TensorFlow: https://github.com/tensorflow/tensorflow

Nouveau: https://nouveau.freedesktop.org/wiki/

sabalabaon Nvidia’s New Policy Limits GeForce Data Center Usage

This should be a rallying cry to the developer community to stop using software with restrictive licenses.

https://m.youtube.com/watch?v=9sJUDx7iEJw

“Join us now and share the software!”

In all seriousness anybody using CUDA should immediately start checking out and contributing to ROCm, AMD’s Open Source Compute Platform:

https://github.com/RadeonOpenCompute/ROCm

sabalabaon Nvidia’s New Policy Limits GeForce Data Center Usage

It's true, here is the updated license: http://www.nvidia.com/content/DriverDownload-March2009/licen....

Text:

  No Datacenter Deployment. The SOFTWARE is not licensed for datacenter deployment,
  except that blockchain processing in a datacenter is permitted.