- Joined
- Oct 18, 2024
- Messages
- 21 (0.20/day)
System Name | Office case + Dremel = gaming |
---|---|
Processor | Ryzen 5 5600x |
Motherboard | Asus Prime b450m-A II |
Cooling | Thermalright Assassin X 120 SE |
Memory | Corsair Vengeance 2x8+2x16=48GB 3600 Mtps |
Video Card(s) | Msi Aero OC Gtx 1080ti |
Storage | crucial nvme ssd 1 tb pcie 3.0 |
Display(s) | Minifire 180hz Full HD IPS |
Case | Asus Aspire m3201 |
Audio Device(s) | xbox wireless headset (over usb) |
Power Supply | MSI MAG A650BN 650w non-modular |
Mouse | TMKB M1SE (Pink) |
Keyboard | TMKB T98SE (Brown tactile switches) |
VR HMD | Oculus quest 2 |
Software | windows11 |
Benchmark Scores | P106-90 6GB in Steel Nomad: 550 (Vega 8 for display out) |
Thanks everyone for your replies. I do have a simple pcie x1 to x16 adapter which I’m currently using, a mining riser to be precise as my case is mATX. The reason I‘m interested in this is the that the thing im plugging in, a weird gpu originally for mining, is already limited to 1/16 of the normal pcie speed it should have. Plugging it into an x1 connection limits it to 1/64 which as you may guess does have a very mild impact on performance.
I did some more research on my own as well and it seems that a specialised chip for this doesn‘t exist and won‘t exist in the future either. It’s a niche and complicated part that barely anyone needs so even if it‘s possible it’s probably not commercially viable.
It would have to independently negotiate the pcie speed with the motherboard and gpu during post, combine or split all sent packages, speed or slow down the connections and somehow simultaneous manage the timing of two different frequencies. Additionally it would need a buffer for incoming and outgoing packets, introducing delay and therefore reducing the actual performance gain. This only gets more difficult the more lanes and pcie generation differences the pcie slot and device have.
Thanks again for your time and suggestions.
I did some more research on my own as well and it seems that a specialised chip for this doesn‘t exist and won‘t exist in the future either. It’s a niche and complicated part that barely anyone needs so even if it‘s possible it’s probably not commercially viable.
It would have to independently negotiate the pcie speed with the motherboard and gpu during post, combine or split all sent packages, speed or slow down the connections and somehow simultaneous manage the timing of two different frequencies. Additionally it would need a buffer for incoming and outgoing packets, introducing delay and therefore reducing the actual performance gain. This only gets more difficult the more lanes and pcie generation differences the pcie slot and device have.
Thanks again for your time and suggestions.