Training a vision classification model on COCO data or running opencv on a gpu are excellent places to gain some experience. Why on earth would someone new to the field want to start on the bleeding edge?
https://pytorch.org/get-started/previous-versions/
I would recommend pytorch and you can see they have older binaries, but those lack support.
# CUDA 11.8
conda install pytorch==2.2.1 torchvision==0.17.1 torchaudio==2.2.1 pytorch-cuda=11.8 -c pytorch -c nvidia
# CUDA 12.1
conda install...
My 5950x is under water and fairly good overclocker on 128 gb of decent ram. She does anything I want within the 128 gb of ram she has. I do a lot of simulation and CVML work and I can easily use up 128 gb of ram. The other big point for my work is SIMD registers. The AVX2 intrinsics are okay...
You have a lot of compute for minecraft. If you wanted to do anything with scientific computing the AVX512 intrinsics would be a nice stepup over the 5950x AVX2. Note: I am running the same rig basically, and mine does everything I want. I honestly don't see the need for more compute, and if I...
https://docs.omniverse.nvidia.com/isaacsim/latest/features/physics/index.html
I was recently at the 2024 Nvidia GTC conference in San Jose and spent some time talking to Nvidia reps about Isaac Sim and Ominverse and it seems the software and Physics acceleration is not dead.
The idea is to...
It won't matter if gamers don't like the top teir Blackwell prices as there are plenty of people willing to pay a LOT for the latest GPUs. I am sure Nvidia will try to stay competitive in the value sector re 5060 gpus. Which, if they are still competitive to the competitors, is that really unfair?
I currently am running an RTX 3090 FE and a RTX A5000 for model development and some other work. I really do not need the features of the A5000 gpu and would rather run NVLink on a pair of 3090s so I either want to buy a used but fully working and good condition 3090 FE (has to be FE so I can...
VID is vcore and I try to set that as low as possible. The lower I can set my vcore, the higher my PBO is able to overclock. 1.47 seems crazy high and would probably make your cpu hot AF.
I get consistent 8 inch groups with my 18 inch DD using 55gr federal 556 at 600 yds. I can hit a 12 inch steel plate at 800 yds as well, but that's getting harder. Now if I step up to 77gr match grade bthp I can hit steel at 800 without too much effort. Past 800 the slightest gust of wind seems...
Doesn't bother me amigo...I am a product of Sandhill Ft. Benning...I was just fishing for other 11b types @ [H] (figured the comment would draw more out).
I have said it before...hardware matters, but giving developers support matters more. If all that mattered was speed everything would be done on specialized FPGA circuits. That's how NVIDIA beats out AMD in terms of CUDA. It drives me nuts when people jump on NVIDIA for that sort of support, but...
Dangit....I wish there were a better way to deal with errors in a post and yet let people delete their post. Like a version history for post that cannot be deleted. Then I could edit my post with having to make a whole new post, and yet not remove the original.
I am having a hard time understanding what you wrote and what Thomas Pornin wrote. Never the less, thanks for the post.
To me it seems to me (I could be wrong) that based on what T.P. wrote there is still a race to be had for some algorithms even with QC. If that is true, that means the...
I feel like you a missing something very fundamental about computing.
While possible to simulate a 1000 core processor in a single thread, and so possible to write threaded algorithms against the simulated multicore cpu and thus to test the performance improvements you could get with more...
Sufficient for what?
Why is RSA encryption hard to break? You need to ask this question to understand the problem.
The RSA algorithm is public knowledge so everyone is free to attempt to crack it. The reason is is that it is more difficult to factor numbers into primes than it is to find...
The question does not make sense as you state it. A Quantum Algorithm is very different from a traditional computer algorithm so comparing what one does vs what the other does....well, doesn't make much sense.
In terms of theory of computation all computers with sufficient memory and time...
No...I was (rather poorly) talking about the fact that having multiple cores does not necessarily mean faster computing. Parallelism applies to a lot of algorithms, but not all of them. To say more cores means your applications will run faster is misleading and false. The same thing (only to a...
No idea what you are talking about....
I hope they get to work, but I am rather certain the impacts will be purely for the advancement of the scientific community. I do not every see this kind of computing as replacing (perhaps augmenting) digital computers. It certainly will not make my...
All I said was that I don't see think it is right to state the two ideas of faster is accurate. We have a lot of cores, but that does not mean "faster" just because it has more cores.
I would say that it is an accurate description of the current state of affairs. If you want to let me know...
Solving a problem with a quantum computer involves using quantum gates to encode the problem. Then it pretty much just takes a single go and the problem is solved if you did it right.
Quantum gates are sort of like traditional logic gates, but a lot more complex...
If by "faster" computing you mean higher fps for video games, or super fast boot times or ...well just about anything that the average user will use his or her computer for, then no.
Other wise if you mean certain specific algorithms (certainly more and more as the subject matures) being able...
This complex idea about using quantum states to encode algorithms is being projected down to one metric, speed. And the general public that reads this gains in ignorance because they believe it is accurate to describe quantum computers as "fast."