Thread replies: 10
Thread images: 3
Thread images: 3
Anonymous
AMD Introduces Radeon Instinct Machine Intelligence Accelerators 2016-12-13 03:03:19 Post No. 57973664
[Report] Image search: [Google]
AMD Introduces Radeon Instinct Machine Intelligence Accelerators 2016-12-13 03:03:19 Post No. 57973664
[Report] Image search: [Google]
File: da1512d2bea4c844c18f4e9e05485f4e79729e99.png (846KB, 816x3341px) Image search:
[Google]
846KB, 816x3341px
AMD Introduces Radeon Instinct Machine Intelligence Accelerators
Anonymous
2016-12-13 03:03:19
Post No. 57973664
[Report]
>AMD Introduces Radeon Instinct Machine Intelligence And Deep Learning Accelerators
>AMD is announcing a new series of Radeon-branded products today, targeted at machine intelligence ( AI) and deep learning enterprise applications, called Radeon Instinct.
As its name suggests, the new Radeon Instinct line of products are comprised of GPU-based solutions for deep learning, inference, and training. The new GPUs are also complemented by a free, open-source library and framework for GPU accelerators, dubbed MIOpen. MIOpen is architected for high-performance machine intelligence applications, and is optimized for the deep learning frameworks in AMD’s ROCm software suite, which we recently talked about here.
http://hothardware.com/news/amd-introduces-new-radeon-instinct-machine-intelligence-and-deep-learning-accelerators
>>
>>57973664
Nvidia getting wreckt, no wonder Google is cutting a massive deal with AMD over GPU's
>>
File: 1476835566278.jpg (116KB, 850x1020px) Image search:
[Google]
116KB, 850x1020px
>>57973664
hawt
>>
>>57974339
Fuck off, this isn't a GPU thread, stop with the nvidia vs amd shit
>>
>>57976597
This is literally AMD answer to the new Pascal Teslas and Nvidia focus on Deep Learning you fucking retard
Down to the point to launching their own machine learning libraries
>>
File: 1469679477452.png (297KB, 1200x1200px) Image search:
[Google]
297KB, 1200x1200px
>>57976525
>>
basically machine learning hardware is a fuckton of cores; each core running shitty-performance floats.
So not doubles.
that is, the normal operations used for floating points are approximated and shortened to make the process faster, if not as accurate.
This is acceptable in machine learning because unsurprisingly it's all about statistics; and the occasional outlier doesn't matter much.
That's what this is, in a nutshell.
Trash hardware for trash usage at trash price.
>>
>>57978546
>When nvidia is better
AHAHA AMDPOORFAGS CANT INTO TESSLATION AND MUH DEEPLURNINS
>When nvidia is worse
LOL DEEP LEARNING IS A MEME AND SO IS TESSELLATION AND ANYTHING ELSE NVIDIAS BAD AT
>>
>jewgle is run by poojeet
>radeon is poojeet gpu
Nothing to see here. Move along.
>>
>>57979912
But it's an East Asian...
Thread posts: 10
Thread images: 3
Thread images: 3