To Z390, or to X570?
To Z390, or to X570?
Scenario: audio production setup
Studio surroundings with dual 1080p displays and one 4K TV. Focus on X570 Pros (3700x) – budget-friendly option. Interest in PCIe 4.0 motherboards, especially the 3070 for gaming. Discuss trade-offs between newer components and older models like Z390 Pros (9900k). Consider energy efficiency, PCIe lanes, and integration needs. Mention the limitations of rear Thunderbolt 3 support and the impact of 14nm architecture. End with a practical takeaway and a humorous closing remark.
It has been this way for a while now, since the price difference of $180 has been in effect.
It's probably not necessary for the 30-series cards to use PCIe 4.0 for optimal speed, and even then, only specific models like the 3090 would benefit from it. A 3070 won't be able to fully utilize PCIe 3.0.
I need to confirm I’m looking for an ATX board. I often add extra cards for different purposes. That’s fine. Unfortunately, none of the ASRock X570 boards appeal to me due to their limited connectivity options. This is helpful to know. In this scenario, a Z390 wouldn’t be too bad. According to the details, a Z390 paired with a 9900k should handle workstation tasks better. On paper it seems so, but CPU UserBenchmark shows mixed results—higher clock speeds are listed but the score for workstations is lower. The page description at the bottom mentions that heavy users should consider the 3700X instead because it excels with 64-core processing. My DAW is Reaper, which doesn’t need 64 cores natively, and even more demanding plugins don’t really push CPU limits. Removing that caveat, I think a Z390/9900k setup could work well for me. For just a bit more cost compared to the X570 or 3700X path, I’d get better performance without worrying as much about pushing RAM speeds for speed. My current 5820k (the X99 was a big help back then) is still at the bottom of its line and provides 28 PCIe lanes. The 9900k gives only 24 lanes, so if I want two NVMe SSDs plus an extra USB card and aim for a 3070 at 16x speed, I might hit some limits. I’d love to hear others’ thoughts on needing many lanes in this situation.
Intel hasn’t adopted PCIe 4.0, so Gen4 NVMe M.2 SSDs rely on PCIe 3.0 or newer interfaces. For devices like the 970 Pro or Evo Plus, bandwidth limitations may still exist due to older controller support, potentially causing bottlenecks unless the drive uses a modern interface.
Good to know. Gen4 M.2 bandwidth would be nice for loading some of my heavier tools (Omnisphere and Analoglab are notoriously heavy, Kontakt can be heavy but even my X99 system handles it well over a SATA connection) but isn't really necessary as I'm handling them to a fairly satisfactory extent over a SATA SSD (860 EVO) so I know that even a Gen3 M.2 NVMe would already offer a significant performance increase. I think what I'm mostly concerned about, in order, is: 1. Raw CPU horsepower. As software emulation of analog hardware gets better, it becomes significantly more demanding on CPU resources. I can peg my 5820k pretty easily playing back larger sessions with multiple instances of higher-end emulations, even at a 4.4GHz OC. This is my draw to the 9900K, because it offers the biggest performance increase over what I currently have, whereas the 3700X only offers a mild to moderate performance increase. 2. Thunderbolt 3. Whether native motherboard integration or add-in card, I absolutely require it. 3. Memory optimization. I'm not exactly talking about trying to crank massive OC's on memory; I'm currently running 2400MHz DDR4 with no issues, and am upgrading to 3600MHz DDR4. This will already be a 50% speed increase. I just want stability and timing optimization, which has improved quite a bit already. I'm not concerned about achieving optimization on either platform, but it is important. 4. PCIe bandwidth & lanes. This is my draw to the 3700X. 24 PCIe Gen4 lanes is equivalent to 48 Gen3 lanes, which is literally double the PCIe bandwidth and would offer me access to Gen4 NVMe SSDs which would yield significantly faster loading times for heavier project assets. Right now trying to load a larger / heavier project can take several minutes (and may sometimes outright crash my system). Gen4 M.2 could significantly reduce that loading time. I have considered the possibility of simply upgrading the RAM and SSD setup on my current system and waiting until Intel jumps on the PCIe 4.0 train so I don't have to choose between raw CPU horsepower upgrade or PCIe bandwidth upgrade.