I'm using a 7700x and the damn thing is a nightmare in terms of cooling, had to set an aggressive fan curve with minimal fan smoothing, else it would go from 40'c idle to 70-80 in a matter of a few seconds.
Yes and no. I delidded my 7950X for a lark and it now sits at 82 under load rather than 95, but actual performance gains in Cinebench is just a few percent.
this is the woo im talking about man the laptops are primo but the mac mini costs $500. its like saying the mac studio is absurdly cheap at $2,999 because the old mac pros started at $4,999. you can always construct the perfect normie but they're in this weird price point where most people with a budget or specific goal dont seem likely to buy one
This is why I've given up trying to get the temps low for me as long as it isn't thermal throttling or crashing idrc at this point.
I've seen the 9950x suffer from the same thing with similar aio's, so I'm sure its normal behavior.
Intel & AMD's mobile chips are desktop chips scaled down to mobile. As such, they have many of the same design problems when it comes to power efficiency, particularly relying on high boost clocks rather than high bandwidth to achieve high performance.
Yeah, but try downloading a game from steam on one. I used to have an old dramless nvme and games would take forever. The disk activity would be this sawtooth pattern that would slow down to write speeds of like 10kbs before the cache cleared and it would jump back up to a 100mbs then drift back down. It got worse the more the drive filled up. It was worse than a HDD.
I wish U.3 was standard on the consumer level and not just for servers. They are the same footprint as a 2.5 inch drive, but have incredible capabilities at PCIe speeds.
Only in the sense that the AM5 lid is immensely bad for thermals (circa 20C on its own)
I assumed I put too little thermal paste on even though I clearly didn't. Outside of my thermal OCD, it's fine. AMD just loves fucking with people like that.
It's a "new" Latin American poor man's exclusive, derivative of the previous limited edition Micro Center exclusive 5600X3D. There may only be tens of thousands of either of them. I'd like to know more about AMD's AM4 strategy. How many new CPUs are being made? Are they spillover from chiplets being produced to satisfy Epyc Milan demand? Are new Cezanne dies being used to make the 5705G, or is it old inventory with a new label slapped on it?
I wouldn't be surprised if there were a ton more 5500X3D chips than 5600X3D. Sell the highest binned chips to a smaller, higher paying market, and stockpile the cheaper ones to sell in South America along with the 8GB 9060. My guess is that they have been stockpiling 5500X3Ds for years, and that they haven't been actively manufacturing consumer grade AM4 stuff for a while. But who knows.
Only 20% raster for next gen consoles? Is that normal or is that pretty bad for new gens now? I’m assuming they will also lean heavily on improved upscaling and frame gen, but ideally consoles can finally move past 30 fps “quality” modes and 60 fps with upscaling becomes the standard and then they can advertise higher fps with frame gen.
+20% raster per compute unit. 64 CUs of RDNA4 (9070 XT) roughly matched 96 CUs of RDNA3 (7900 XTX). So you could get that same level of performance with ~54 CUs of UDNA1. This is shit math, of course.
No way its that big at best it'll be 60 cus. Otherwise the ps6 will be 900+ minimum.
(The die sizes and nodes are estimates and guesses, Soyny doesn't tell AFAIK.)
The UE Cycle - new version comes out, runs like crap, everyone forgets old version used to run like crap and retroactively praises it, new version cooks for a while and stops running like crap, and then new new version comes out.
I kinda miss the big fat yaoi hands UE3 games had.
I think it's that TSMC's N7 node is still live is getting very high yields, so there's still profit to be made by fabricating chips on it, especially since there really isnt much functionality in the last several years that matters in the consumer CPU space.
