Array
(
[content] =>
[params] => Array
(
[0] => /forum/whats-new/latest-activity?before_id=112203
)
[addOns] => Array
(
[DL6/MLTP] => 13
[Hampel/TimeZoneDebug] => 1000070
[SV/ChangePostDate] => 2010200
[SemiWiki/Newsletter] => 1000010
[SemiWiki/WPMenu] => 1000010
[SemiWiki/XPressExtend] => 1000010
[ThemeHouse/XLink] => 1000970
[ThemeHouse/XPress] => 1010570
[XF] => 2031070
[XFI] => 1060170
)
[wordpress] => /var/www/html
)
You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please,
join our community today !
JavaScript is disabled. For a better experience, please enable JavaScript in your browser before proceeding.
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an
alternative browser .
K
I’m not sure that’s lower than normal for a modern attention-based model. I think a typical model FLOPs utilization (MFU) number runs...
Wednesday at 7:40 AM
K
The best summary that I've seen published about AI data center failure rates is the following. (data published by Meta -- Nvidia GPUs...
Wednesday at 7:26 AM
C
The best summary that I've seen published about AI data center failure rates is the following. (data published by Meta -- Nvidia GPUs...
Wednesday at 6:21 AM
L
Meta were only on utilisation rate of 38% which is far lower than a normal training run
Wednesday at 5:50 AM
B
Anyone who is having their information used should look at a subscription model like these firms are also doing.
What do these tech...
Wednesday at 5:21 AM
S
Intel was the first to place all their free cash flow on the line in a transformation, but now in Silicon Valley the frontier AI...
Wednesday at 5:04 AM
B
Anyone who is having their information used should look at a subscription model like these firms are also doing.
What do these tech...
Wednesday at 4:50 AM
They probably should pay for it, but how do you price using a book or other copyrighted source for something that can be used by...
Wednesday at 3:49 AM
With this settlement, the legality of fair use in AI training continues to be uncertain.
Wednesday at 3:45 AM
Intel was the first to place all their free cash flow on the line in a transformation, but now in Silicon Valley the frontier AI...
Wednesday at 3:12 AM
B
Intel was the first to place all their free cash flow on the line in a transformation, but now in Silicon Valley the frontier AI...
Wednesday at 2:52 AM
B
Intel is back. Their fabs finally have customers, and big ones. Apple, SpaceX, Tesla, Nvidia and the Trump have all bet on the company...
Wednesday at 2:46 AM
B
The description of Kevork Kechichian demanding a "recovery plan" and the result is a schedule push-out, is priceless.
He views that as...
Wednesday at 2:37 AM
B
That must be really weird for the employees.
It seems like LBT's strategy is.. execute Pat's strategy, but do ir better. ? (i.e...
Wednesday at 2:36 AM
Key Points
- Elon Musk officially announced plans for his Terafab project in March to make chips for Tesla, SpaceX and xAI.
- New...
Wednesday at 2:19 AM
Like.
The best summary that I've seen published about AI data center failure rates is the following. (data published by Meta -- Nvidia GPUs...
Xebec replied to the thread AI Anthropic settles for $1.5B for copyrighted works of book authors.They probably should pay for it, but how do you price using a book or other copyrighted source for something that can be used by...
Daniel Nenni reacted to benb's post in the thread Tech Altar: Intel is back. Thank the old CEO. with