AMD’s first crack at Nvidia hampered by half-baked training software, says TensorWave boss

After some teething pains, TensorWave CEO Darrick Horton is confident that AMD's Instinct accelerators are ready to take on large-scale AI training. "There were definitely some bumps in the road with the first generation product," Horton, whose rent-a-GPU outfit was among the first to throw its weight behind the Nvidia …

[Gene Cash]

AMD let down by software?

As anyone who's used their drivers: "say it isn't so!"

[that one in the corner]

There are better ways to do the training?

Software can be improved? There will be new ways to do the training?

For Heaven's sake man, don't say these things out loud, you never know who may be listening!

Why, if the Chinese ever find out ...

[anothercynic]

Re: There are better ways to do the training?

You read my mind. I read that Tencent piece and went "what if they don't bother with Nvidia but maybe someone else... like AMD?"

And BAM - someone else has the same brainwave. Have a beer (later, when it's more appropriate than before 9am) :-)

[rgjnk]

Bubble economics

So is the story at the end that their services aren't going to drive them financially, they've already milked the VCs and their plan/desperate hope is that someone will lend them money secured against a rapidly depreciating stack of GPUs?

It's a bold strategy. Let's see if it pays off for 'em.

BTW nothing wrong at all with the AMD gear, but their problem in this field as with others using GPU is that Nvidia is where the ecosystem support is. Often even if you'd love to move you just can't. The debate for buyers & investors is how long that lockin is sustained.

[Kevin McMurtrie]

More power, more cooling

It's like we're in the 1950s, struggling to build larger and larger computers out of bulky glass tubes and electromagnets. What's the next tech to cause its extinction?