I am probably the wrong person to ask about this, but here goes. I have had two modern experiences and only one of them was Gigabyte. My Aorus Pro B450I m-ITX card paired with a 3400G performed adequately, I did have some problem OC:ing the RAM to 3200 but was otherwise fine. The stock 2666 didn’t lose me much FPS, but would probably have cared about it more if the system was more high end. For my purposes it was a pretty solid workhorse for AM4.
I am currently rocking an ASRock B650I Lightning paired with a 7900, no complaints here at all. It does what it is supposed to do, end of story.
For both systems, the only thing I bothered to change was the virtualization enabling and I left most settings at stock, BIOS was and is fine in both systems. The only thing I have done with my current setup is enable EXPO and Virtualization.
Gigabyte Aorus is pretty good, but not perfect - 3.5/5. ASRock punch above their weightclass, 4/5 if you will. Both are solid choices, but perhaps not 10/10. Asus high end is pretty good, but expensive and poor customer support pulls down their grades.
Also, there are two kinds of people - those who got screwed over by $BRAND and those who are yet to be screwed over by $BRAND. Given enough time all brands suck, but they also improve both relatively and absolutely (or they perish), so any experiences that are 3+ years old should be taken with a large pinch of salt. “Never” is a really long time to hold a grudge.
Not quite the right question since TDP is a marketing number only loosely correlated with actual power draw (especially for Intel). For a quiet build (< ~25 dB(A)) with common ambients (20-30 °C) around 160-170 W is practical with dual towers well matched to chip and socket. That’s not NH-D15 G1 or G2 in AM5.
So if you want air cooling 9900X would be the default. To chip in on the fear of AIO thing, nobody’s reported a leak in any thread I can think of. Leaks from custom loops, yes, but that’s pretty much always because either the builder didn’t use a leak tester (duh) or neglected loop maintenance.
That said, if you look hard enough probably you can find an unprovoked AIO leak or maybe somebody who ran one to failure because they never checked it. But you can also find failed air coolers. ¯\_(ツ)_/¯
Lian Li cases are supposed to be quite innovative and the supplied fans are of course an advantage. However, I’ve also come across several reports that the Lancool 207 isn’t entirely without issues, particularly when it comes to cable management, which is said to be a bit fiddly.
Yes, the North XL is definitely overpriced. Still, I specifically wanted a case with a wood aesthetic, without any RGB and deliberately went for a larger form factor to avoid any space constraints – whether for cable management or future upgrades. Thanks to its size, the North XL gives me more flexibility, whether I end up using an AIO or eventually switch back to air cooling.
I don’t really care about the interior aesthetics, even if I ended up choosing a glass side panel. Unless I end up with some RGB components – in that case, I’d actually prefer a closed case. That said, I’ve also read that the North /North XL tend to be noisier, since the wood can act as a sound amplifier. Admittedly, I’d like to avoid that as well.
So, if you know of any other cases with a wood look in general, ample space and better value for money, please let me know.
Again, read the OPs requirements, it’s not a dedicated gaming platform. 3DVcache accelerates many other workloads outside of the realm of gaming.
I am running an EPYC7473X specifically for the 768MB of 3dvcache as the performance improvement for some of my workloads is insane (especially with VMs), even though I am running at reduced clocks because I am using a server part.
And if you do consider gaming, upgrading your GPU later and being bottle necked because you didn’t opt for the x3d part is a good reason to get the x3d part even if right now you can’t take full advantage of it. Your GPU upgrade would become much more expensive because you’re now also buying a CPU to fix your prior mistake.
As for overpriced… I don’t think you understand the additional costs involved in producing the x3d part, it’s not just enabling some extra cache on the chip that was locked out. An extra die has to be produced and bonded to the part, which means an entire extra production line to make the cache dies, along with extra steps to assemble the final part, and the added cost of the extra silicon area.
You can kind of work around it for now, but if you need to stop the guest VM and start it again (reboots are usually ok), or take the GPU back so the host can use it, you will need to reboot the entire PC.
This is how I am currently operating with a RX9070XT as my guest GPU.
Check the Hardware Unboxed video linked by wertigon above. With the RX 9070 (non-XT), at 1440p only Warhammer 40,000 SP2 (1 out of 4 games tested) was faster (granted a whopping +31 % on Medium settings or +13 % on Ultra settings) using the 9800X3D compared to the 7600X.
The three other games had practically unchanged average and 1 % lows at 1440p. So, yeah, very dependent on the specific game and graphics settings.
OP seems to be aiming for an RX 9070 XT though which should increase the utility of the 9800X3D.
It’s interesting to see the TechPowerUp tests on AM4. It seems the extra cache makes a huge difference much earlier there. Perhaps that’s what we’ll see on AM5 too with the games that will be current at the end of AM5’s life.
