GPUs & CPUs & Enthusiast hardware: Questions, Discussion and fanboy slap-fights - Nvidia & AMD & Intel - Separe but Equal. Intel rides in the back of the bus.

My 7800x3d idles at 20 watts according to hwinfo. You sure you haven't set a bad power plan in windows? Or you running the integrated gfx?
I'm running the igpu with a dgpu. I'm not sure if that's what it is as the igpu is unused and it never has any load placed on it. My amd laptop with a more powerful apu can run the browser and light applications at 5-15w. so I'm starting to think its an optimization issue as @The Ugly One said. On desktop 30 - 40w idle doesn't matter much but on laptop its very bad, so they have to modify the power setting for the chip to be more conservative on laptop. I will disable the igpu and test it tho
 
Writer is probably a currycel

No, the writer is most definitely an engineer from the 1990s. Pajeets don't even know Fortran.

I'm running the igpu with a dgpu. I'm not sure if that's what it is as the igpu is unused and it never has any load placed on it. My amd laptop with a more powerful apu can run the browser and light applications at 5-15w. so I'm starting to think its an optimization issue as @The Ugly One said. On desktop 30 - 40w idle doesn't matter much but on laptop its very bad, so they have to modify the power setting for the chip to be more conservative on laptop. I will disable the igpu and test it tho

Idle power on my i9-12900 just 5W.
 
Update on the PC fans:
20250207_142224.jpg
There is no such thing as too much airflow.
 
That is what we call FROSTY. What core are you cooling with that wind farm?
7800X3D. My issue is the 6750XT due to some weird coil whine at high temps. From where I am from, the afternoon becomes heatwave central. Thank goodness the P12 Max I got was Rev. 2 (FDB).
Always envious of people with good cable management. Mine is a disaster and haunts me every time I am unfortunate enough to have to open my case up for whatever reason.
Eh, could have been worse. I blame Corsair for adding whatever the fuck that SATA cable is.
20250207_231946.jpg
 
Last edited:
Intel Nova Lake preliminary desktop specs list 52 cores: 16P+32E+4LP configuration

This may be the next big Intel desktop CPU in 2026 (Panther Lake is a mobile Lunar Lake sequel). I'm interested to see if 4x next-gen LPE-cores pay off and let it idle at 1 Watt or something.

It may be using two 8P+16E tiles. No hyperthreading, just corespam.

NVIDIA Cuts GeForce RTX 4060 Supply By Up To 60% As The GPU Gets Close To EOL

4060 set to disappear, but nobody will care if the 5060 delivers +5-10% price/perf.

HP Debuts HP EliteBook X Laptops Featuring 128 GB LPDDR5X-8533 Memory, The First In The Market With Strix Point

Wccfkek forgot how to do math (what a surprise), but I guess this could be one of the first products actually using 32-bit 32 GB (256 Gb) LPDDR5X modules, which have been previously announced. What I want to know is why there's been no evidence of 192-256 GB Strix Halo products.
 
Other leakers have backed MLID's claim of 12-core Zen 6 CCDs shared between mainstream mobile (Medusa Point), premium mobile (Range and Halo) and (up to 24-core) desktop:

AMD Medusa Point/Ridge/Halo and Range may share the same 12-core Zen6 CCD (archive)

Probably non-C Epyc too. And since they're all the same, presumably any product could come with 3D V-Cache, although it's not confirmed for anything like the Point APUs. Article says TSMC N2 for the CCD, but MLID thinks N3.

In the same livestream:
* He promises renders for some of these Zen 6 products within a week or two.
* Medusa Point on FP10 probably has 16 CUs again, but has a total die area of around 275mm^2 (~75mm^2 is the CCD), which is large for the mainstream APU and could mean some Infinity Cache or a big NPU on the I/O die.
* Source says 5070 Ti is around +5-10% raster, +8-15% raytracing than the 4070 Ti Super, and supply will be limited at launch (Feb 20).
* Source says 5070 is around +10-20% raster, barely faster in raytracing than the 4070. He guesses that Nvidia may try to "flood" the market with 5070 in early March to counter RDNA4, but doesn't know.
* RDNA4 launching early March with good supply and performance.
* He updated his GPU rank chart: https://www.mooreslawisdead.com/rank-charts (archive) (ghost)
 
Last edited:
It still amazes me that you can build a half decent AM4 rig cheaper than an AM5 rig. I put together an 8 core 5700x build with a 4060 in it for around $1100. Honestly, I think that's what I'm going to do for my next build. Hopefully by around November / December I'll be able to do it.

Right now I'm running a Ryzen 5 5600 + RX 580. I mostly play a lot of older games and retro stuff, and I have a PS5 slim for anything new that catches my attention. It's working out pretty well so far. Life's pretty great when you stop worrying about bullshit console slapfights and just game for the sake of it.
 
It still amazes me that you can build a half decent AM4 rig cheaper than an AM5 rig. I put together an 8 core 5700x build with a 4060 in it for around $1100. Honestly, I think that's what I'm going to do for my next build. Hopefully by around November / December I'll be able to do it.

Right now I'm running a Ryzen 5 5600 + RX 580. I mostly play a lot of older games and retro stuff, and I have a PS5 slim for anything new that catches my attention. It's working out pretty well so far. Life's pretty great when you stop worrying about bullshit console slapfights and just game for the sake of it.
Make sure whatever you buy supports PCIE 4.0 as the 4060 is significantly bottlenecked if you put it into a PCIE 3.0 slot.
 
  • Informative
Reactions: Betonhaus
Make sure whatever you buy supports PCIE 4.0 as the 4060 is significantly bottlenecked if you put it into a PCIE 3.0 slot.

Not really. I've seen a lot of videos with the 4060 in PCIE 3, and the difference is virtually negligible.
 
Back